Skip to content

Kruithne/spooder

BranchesTags

Repository files navigation

spooder · typescript license badge npm version bun

spooder is a purpose-built server solution that shifts away from the dependency hell of the Node.js ecosystem, with a focus on stability and performance, which is why:

  • It is built using the Bun runtime and not designed to be compatible with Node.js or other runtimes.
  • It uses zero dependencies and only relies on code written explicitly for spooder or APIs provided by the Bun runtime, often implemented in native code.
  • It provides streamlined APIs for common server tasks in a minimalistic way, without the overhead of a full-featured web framework.
  • It is opinionated in its design to reduce complexity and overhead.

The design goal behind spooder is not to provide a full-featured web server, but to expand the Bun runtime with a set of APIs and utilities that make it easy to develop servers with minimal overhead.

Note

If you think a is missing a feature, consider opening an issue with your use-case. The goal behind spooder is to provide APIs that are useful for a wide range of use-cases, not to provide bespoke features better suited for userland.

It consists of two components, the CLI and the API.

  • The CLI is responsible for keeping the server process running, applying updates in response to source control changes, and automatically raising issues on GitHub via the canary feature.
  • The API provides a minimal building-block style API for developing servers, with a focus on simplicity and performance.

CLI

The CLI component of spooder is a global command-line tool for running server processes.

API

spooder exposes a simple yet powerful API for developing servers. The API is designed to be minimal to leave control in the hands of the developer and not add overhead for features you may not need.

Installation

# Installing globally for CLI runner usage.
bun add spooder --global

# Install into local package for API usage.
bun add spooder

Configuration

Both the CLI and the API are configured in the same way by providing a spooder object in your package.json file.

{
	"spooder": {
		"auto_restart": 5000,
		"update": [
			"git pull",
			"bun install"
		],
		"canary": {
			"account": "",
			"repository": "",
			"labels": [],
			"crash_console_history": 64,
			"throttle": 86400,
			"sanitize": true
		}
	}
}

If there are any issues with the provided configuration, a warning will be printed to the console but will not halt execution. spooder will always fall back to default values where invalid configuration is provided.

Note

Configuration warnings do not raise caution events with the spooder canary functionality.

CLI > Usage

For convenience, it is recommended that you run this in a screen session.

screen -S my-website-about-fish.net
cd /var/www/my-website-about-fish.net/
spooder

spooder will launch your server either by executing the run command provided in the configuration, or by executing bun run index.ts by default.

{
	"spooder": {
		"run": "bun run my_server.ts"
	}
}

While spooder uses a bun run command by default, it is possible to use any command string. For example if you wanted to launch a server using node instead of bun, you could do the following.

{
	"spooder": {
		"run": "node my_server.js"
	}
}

CLI > Dev Mode

spooder can be started in development mode by providing the --dev flag when starting the server.

spooder --dev

The following differences will be observed when running in development mode:

  • Update commands defined in spooder.update will not be executed when starting a server.
  • If the server crashes and auto_restart is enabled, the server will not be restarted, and spooder will exit with the same exit code as the server.
  • If canary is configured, reports will not be dispatched to GitHub and instead be printed to the console; this includes crash reports.

It is possible to detect in userland if a server is running in development mode by checking the SPOODER_ENV environment variable.

if (process.env.SPOODER_ENV === 'dev') {
	// Server is running in development mode.
}

Note

SPOODER_ENV should be either dev or prod. If the variable is not defined, the server was not started by the spooder CLI.

CLI > Auto Restart

Note

This feature is not enabled by default.

In the event that the server process exits, regardless of exit code, spooder can automatically restart it after a short delay. To enable this feature specify the restart delay in milliseconds as auto_restart in the configuration.

{
	"spooder": {
		"auto_restart": 5000
	}
}

If set to 0, the server will be restarted immediately without delay. If set to -1, the server will not be restarted at all.

CLI > Auto Update

Note

This feature is not enabled by default.

When starting or restarting a server process, spooder can automatically update the source code in the working directory. To enable this feature, the necessary update commands can be provided in the configuration as an array of strings.

{
	"spooder": {
		"update": [
			"git reset --hard",
			"git clean -fd",
			"git pull origin main",
			"bun install"
		]
	}
}

Each command should be a separate entry in the array and will be executed in sequence. The server process will be started once all commands have resolved.

Important

Chaining commands using && or || operators does not work.

If a command in the sequence fails, the remaining commands will not be executed, however the server will still be started. This is preferred over entering a restart loop or failing to start the server at all.

You can utilize this to automatically update your server in response to a webhook by exiting the process.

server.webhook(process.env.WEBHOOK_SECRET, '/webhook', payload => {
	setImmediate(async () => {
		await server.stop(false);
		process.exit();
	});
	return 200;
});

Skip Updates

In addition to being skipped in dev mode, updates can also be skipped in production mode by passing the --no-update flag.

CLI > Canary

Note

This feature is not enabled by default.

canary is a feature in spooder which allows server problems to be raised as issues in your repository on GitHub.

To enable this feature, you will need to create a GitHub App and configure it:

1. Create a GitHub App

Create a new GitHub App either on your personal account or on an organization. The app will need the following permissions:

  • Issues - Read & Write
  • Metadata - Read-only

Once created, install the GitHub App to your account. The app will need to be given access to the repositories you want to use the canary feature with.

In addition to the App ID that is assigned automatically, you will also need to generate a Private Key for the app. This can be done by clicking the Generate a private key button on the app page.

Note

The private keys provided by GitHub are in PKCS#1 format, but only PKCS#8 is supported. You can convert the key file with the following command.

openssl pkcs8 -topk8 -inform PEM -outform PEM -nocrypt -in private-key.pem -out private-key-pkcs8.key

Each server that intends to use the canary feature will need to have the private key installed somewhere the server process can access it.

2. Add package.json configuration

"spooder": {
	"canary": {
		"account": "<GITHUB_ACCOUNT_NAME>",
		"repository": "<GITHUB_REPOSITORY>",
		"labels": ["some-label"]
	}
}

Replace <GITHUB_ACCOUNT_NAME> with the account name you have installed the GitHub App to, and <GITHUB_REPOSITORY> with the repository name you want to use for issues.

The repository name must in the full-name format owner/repo (e.g. facebook/react).

The labels property can be used to provide a list of labels to automatically add to the issue. This property is optional and can be omitted.

3. Setup environment variables

The following two environment variables must be defined on the server.

SPOODER_CANARY_APP_ID=1234
SPOODER_CANARY_KEY=/home/bond/.ssh/id_007_pcks8.key

SPOODER_CANARY_APP_ID is the App ID as shown on the GitHub App page.

SPOODER_CANARY_KEY is the path to the private key file in PKCS#8 format.

Note

Since spooder uses the Bun runtime, you can use the .env.local file in the project root directory to set these environment variables per-project.

4. Use canary

Once configured, spooder will automatically raise an issue when the server exits with a non-zero exit code.

In addition, you can manually raise issues using the spooder API by calling caution() or panic(). More information about these functions can be found in the API section.

If canary has not been configured correctly, spooder will only print warnings to the console when it attempts to raise an issue.

Warning

Consider testing the canary feature with the caution() function before relying on it for critical issues.

CLI > Canary > Crash

It is recommended that you harden your server code against unexpected exceptions and use panic() and caution() to raise issues with selected diagnostic information.

In the event that the server does encounter an unexpected exception which causes it to exit with a non-zero exit code, spooder will provide some diagnostic information in the canary report.

Since this issue has been caught externally, spooder has no context of the exception which was raised. Instead, the canary report will contain the output from both stdout and stderr.

{
	"proc_exit_code": 1,
	"console_output": [
		"[2.48ms] \".env.local\"",
		"Test output",
		"Test output",
		"4 | console.warn('Test output');",
		"5 | ",
		"6 | // Create custom error class.",
		"7 | class TestError extends Error {",
		"8 | 	constructor(message: string) {",
		"9 | 		super(message);",
		"     ^",
		"TestError: Home is [IPv4 address]",
		"      at new TestError (/mnt/i/spooder/test.ts:9:2)",
		"      at /mnt/i/spooder/test.ts:13:6",
		""
	]
}

The proc_exit_code property contains the exit code that the server exited with.

The console_output will contain the last 64 lines of output from stdout and stderr combined. This can be configured by setting the spooder.canary.crash_console_history property to a length of your choice.

{
	"spooder": {
		"canary": {
			"crash_console_history": 128
		}
	}
}

This information is subject to sanitization, as described in the CLI > Canary > Sanitization section, however you should be aware that stack traces may contain sensitive information.

Setting spooder.canary.crash_console_history to 0 will omit the console_output property from the report entirely, which may make it harder to diagnose the problem but will ensure that no sensitive information is leaked.

CLI > Canary > Sanitization

All reports sent via the canary feature are sanitized to prevent sensitive information from being leaked. This includes:

  • Environment variables from .env.local
  • IPv4 / IPv6 addresses.
  • E-mail addresses.
# .env.local
DB_PASSWORD=secret
await panic({
	a: 'foo',
	b: process.env.DB_PASSWORD,
	c: 'Hello person@place.net',
	d: 'Client: 192.168.1.1'
});
[
	{
		"a": "foo",
		"b": "[redacted]",
		"c": "Hello [e-mail address]",
		"d": "Client: [IPv4 address]"
	}
]

The sanitization behavior can be disabled by setting spooder.canary.sanitize to false in the configuration. This is not recommended as it may leak sensitive information.

{
	"spooder": {
		"canary": {
			"sanitize": false
		}
	}
}

Warning

While this sanitization adds a layer of protection against information leaking, it does not catch everything. You should pay special attention to messages and objects provided to the canary to not unintentionally leak sensitive information.

CLI > Canary > System Information

In addition to the information provided by the developer, spooder also includes some system information in the canary reports.

{
	"loadavg": [
		0,
		0,
		0
	],
	"memory": {
		"free": 7620907008,
		"total": 8261840896
	},
	"platform": "linux",
	"uptime": 7123,
	"versions": {
		"node": "18.15.0",
		"bun": "0.6.5",
		"webkit": "60d11703a533fd694cd1d6ddda04813eecb5d69f",
		"boringssl": "b275c5ce1c88bc06f5a967026d3c0ce1df2be815",
		"libarchive": "dc321febde83dd0f31158e1be61a7aedda65e7a2",
		"mimalloc": "3c7079967a269027e438a2aac83197076d9fe09d",
		"picohttpparser": "066d2b1e9ab820703db0837a7255d92d30f0c9f5",
		"uwebsockets": "70b1b9fc1341e8b791b42c5447f90505c2abe156",
		"zig": "0.11.0-dev.2571+31738de28",
		"zlib": "885674026394870b7e7a05b7bf1ec5eb7bd8a9c0",
		"tinycc": "2d3ad9e0d32194ad7fd867b66ebe218dcc8cb5cd",
		"lolhtml": "2eed349dcdfa4ff5c19fe7c6e501cfd687601033",
		"ares": "0e7a5dee0fbb04080750cf6eabbe89d8bae87faa",
		"usockets": "fafc241e8664243fc0c51d69684d5d02b9805134",
		"v8": "10.8.168.20-node.8",
		"uv": "1.44.2",
		"napi": "8",
		"modules": "108"
	},
	"bun": {
		"version": "0.6.5",
		"rev": "f02561530fda1ee9396f51c8bc99b38716e38296",
		"memory_usage": {
			"rss": 99672064,
			"heapTotal": 3039232,
			"heapUsed": 2332783,
			"external": 0,
			"arrayBuffers": 0
		},
		"cpu_usage": {
			"user": 50469,
			"system": 0
		}
	}
}

API > Serving

serve(port: number, hostname?: string): Server

Bootstrap a server on the specified port (and optional hostname).

import { serve } from 'spooder';

const server = serve(8080); // port only
const server = serve(3000, '0.0.0.0'); // optional hostname

By default, the server responds with:

HTTP/1.1 404 Not Found
Content-Length: 9
Content-Type: text/plain;charset=utf-8

Not Found

API > Routing

🔧 server.route(path: string, handler: RequestHandler)

Register a handler for a specific path.

server.route('/test/route', (req, url) => {
	return new Response('Hello, world!', { status: 200 });
});

🔧 server.unroute(path: string)

Unregister a specific route.

server.route('/test/route', () => {});
server.unroute('/test/route');

HTTP Methods

By default, spooder will register routes defined with server.route() and server.dir() as GET routes. Requests to these routes with other methods will return 405 Method Not Allowed.

Note

spooder does not automatically handle HEAD requests natively.

This can be controlled by providing the method parameter with a string or array defining one or more of the following methods.

GET | HEAD | POST | PUT | DELETE | CONNECT | OPTIONS | TRACE | PATCH

server.route('/test/route', (req, url) => {
	if (req.method === 'GET')
		// Handle GET request.
	else if (req.method === 'POST')
		// Handle POST request.
}, ['GET', 'POST']);

Note

Routes defined with .sse() or .webhook() are always registered as 'GET' and 'POST' respectively and cannot be configured.

Redirection Routes

spooder does not provide a built-in redirection handler since it's trivial to implement one using Response.redirect, part of the standard Web API.

server.route('/redirect', () => Response.redirect('/redirected', 301));

Status Code Text

spooder exposes HTTP_STATUS_CODE to convieniently access status code text.

import { HTTP_STATUS_CODE } from 'spooder';

server.default((req, status_code) => {
	// status_code: 404
	// Body: Not Found
	return new Response(HTTP_STATUS_CODE[status_code], { status: status_code });
});

API > Routing > RequestHandler

RequestHandler is a function that accepts a Request object and a URL object and returns a HandlerReturnType.

HandlerReturnType must be one of the following.

Type Description
Response https://developer.mozilla.org/en-US/docs/Web/API/Response
Blob https://developer.mozilla.org/en-US/docs/Web/API/Blob
BunFile https://bun.sh/docs/api/file-io
object Will be serialized to JSON.
string Will be sent as `text/html``.
number Sets status code and sends status message as plain text.

Note

For custom JSON serialization on an object/class, implement the toJSON() method.

HandleReturnType can also be a promise resolving to any of the above types, which will be awaited before sending the response.

Note

Returning Bun.file() directly is the most efficient way to serve static files as it uses system calls to stream the file directly to the client without loading into user-space.

API > Routing > Query Parameters

Query parameters can be accessed from the searchParams property on the URL object.

server.route('/test', (req, url) => {
	return new Response(url.searchParams.get('foo'), { status: 200 });
});
GET /test?foo=bar HTTP/1.1

HTTP/1.1 200 OK
Content-Length: 3

bar

Named parameters can be used in paths by prefixing a path segment with a colon.

Important

Named parameters will overwrite existing query parameters with the same name.

server.route('/test/:param', (req, url) => {
	return new Response(url.searchParams.get('param'), { status: 200 });
});

API > Routing > Wildcards

Wildcards can be used to match any path that starts with a given path.

Note

If you intend to use this for directory serving, you may be better suited looking at the server.dir() function.

server.route('/test/*', (req, url) => {
	return new Response('Hello, world!', { status: 200 });
});

Important

Routes are FIFO and wildcards are greedy. Wildcards should be registered last to ensure they do not consume more specific routes.

server.route('/*', () => 301);
server.route('/test', () => 200);

// Accessing /test returns 301 here, because /* matches /test first.

API > Routing > Fallback Handlers

🔧 server.handle(status_code: number, handler: RequestHandler)

Register a custom handler for a specific status code.

server.handle(500, (req) => {
	return new Response('Custom Internal Server Error Message', { status: 500 });
});

🔧 server.default(handler: DefaultHandler)

Register a handler for all unhandled response codes.

Note

If you return a Response object from here, you must explicitly set the status code.

server.default((req, status_code) => {
	return new Response(`Custom handler for: ${status_code}`, { status: status_code });
});

🔧 server.error(handler: ErrorHandler)

Register a handler for uncaught errors.

Note

Unlike other handlers, this should only return Response or Promise<Response>.

server.error((err, req, url) => {
	return new Response('Custom Internal Server Error Message', { status: 500 });
});

Important

It is highly recommended to use caution() or some form of reporting to notify you when this handler is called, as it means an error went entirely uncaught.

server.error((err, req, url) => {
	// Notify yourself of the error.
	caution({ err, url });

	// Return a response to the client.
	return new Response('Custom Internal Server Error Message', { status: 500 });
});

API > Routing > Slow Requests

🔧 server.on_slow_request(callback: SlowRequestCallback, threshold: number)

server.on_slow_request can be used to register a callback for requests that take an undesirable amount of time to process.

By default requests that take longer than 1000ms to process will trigger the callback, but this can be adjusted by providing a custom threshold.

Important

If your canary reports to a public repository, be cautious about directly including the req object in the callback. This can lead to sensitive information being leaked.

server.on_slow_request(async (req, time, url) => {
	// avoid `time` in the title to avoid canary spam
	// see caution() API for information
	await caution('Slow request warning', { req, time });
}, 500);

Note

The callback is not awaited internally, so you can use async/await freely without blocking the server/request.

🔧 server.allow_slow_request(req: Request)

In some scenarios, mitigation throttling or heavy workloads may cause slow requests intentionally. To prevent these triggering a caution, requests can be marked as slow.

server.on_slow_request(async (req, time, url) => {
	await caution('Slow request warning', { req, time });
}, 500);

server.route('/test', async (req) => {
	// this request is marked as slow, therefore won't
	// trigger on_slow_request despite taking 5000ms+
	server.allow_slow_request(req);
	await new Promise(res => setTimeout(res, 5000));
});

Note

This will have no effect if a handler hasn't been registered with on_slow_request.

API > Routing > Validation

🔧 validate_req_json(handler: JSONRequestHandler)

In the scenario that you're expecting an endpoint to receive JSON data, you might set up a handler like this:

server.route('/api/endpoint', async (req, url) => {
	const json = await req.json();
	// do something with json.
	return 200;
})

The problem with this is that if the request body is not valid JSON, the server will throw an error (potentially triggering canary reports) and return a 500 response.

What should instead happen is something like this:

server.route('/api/endpoint', async (req, url) => {
	// check content-type header
	if (req.headers.get('Content-Type') !== 'application/json')
		return 400;

	try {
		const json = await req.json();
		if (json === null || typeof json !== 'object' || Array.isArray(json))
			return 400;

		// do something with json.
		return 200;
	} catch (err) {
		return 400;
	}
})

As you can see this is quite verbose and adds a lot of boilerplate to your handlers. validate_req_json can be used to simplify this.

server.route('/api/endpoint', validate_req_json(async (req, url, json) => {
	// do something with json.
	return 200;
}));

This behaves the same as the code above, where a 400 status code is returned if the Content-Type header is not application/json or if the request body is not valid JSON, and no error is thrown.

Note

While arrays and other primitives are valid JSON, validate_req_json will only pass objects to the handler, since they are the most common use case for JSON request bodies and it removes the need to validate that in the handler. If you need to use arrays or other primitives, either box them in an object or provide your own validation.

API > Routing > Directory Serving

🔧 server.dir(path: string, dir: string, handler?: DirHandler)

Serve files from a directory.

server.dir('/content', './public/content');

Important

server.dir registers a wildcard route. Routes are FIFO and wildcards are greedy. Directories should be registered last to ensure they do not consume more specific routes.

server.dir('/', '/files');
server.route('/test', () => 200);

// Route / is equal to /* with server.dir()
// Accessing /test returns 404 here because /files/test does not exist.

By default, spooder will use the following default handler for serving directories.

function default_directory_handler(file_path: string, file: BunFile, stat: DirStat, request: Request): HandlerReturnType {
	// ignore hidden files by default, return 404 to prevent file sniffing
	if (path.basename(file_path).startsWith('.'))
		return 404; // Not Found

	if (stat.isDirectory())
		return 401; // Unauthorized

	return apply_range(file, request);
}

Note

Uncaught ENOENT errors thrown from the directory handler will return a 404 response, other errors will return a 500 response.

Note

The call to apply_range in the default directory handler will automatically slice the file based on the Range header. This function is also exposed as part of the spooder API for use in your own handlers.

Provide your own directory handler for fine-grained control.

Important

Providing your own handler will override the default handler defined above. Be sure to implement the same logic if you want to retain the default behavior.

Parameter Type Reference
file_path string The path to the file on disk.
file BunFile https://bun.sh/docs/api/file-io
stat fs.Stats https://nodejs.org/api/fs.html#class-fsstats
request Request https://developer.mozilla.org/en-US/docs/Web/API/Request
url URL https://developer.mozilla.org/en-US/docs/Web/API/URL 
server.dir('/static', '/static', (file_path, file, stat, request, url) => {
	// Implement custom logic.
	return file; // HandlerReturnType
});

Note

The directory handler function is only called for files that exist on disk - including directories.

Asynchronous directory handlers are supported and will be awaited.

server.dir('/static', '/static', async (file_path, file) => {
	let file_contents = await file.text();
	// do something with file_contents
	return file_contents;
});

API > Routing > Server-Sent Events

🔧 server.sse(path: string, handler: ServerSentEventHandler)

Setup a server-sent event stream.

server.sse('/sse', (req, url, client) => {
	client.message('Hello, client!'); // Unnamed event.
	client.event('named_event', 'Hello, client!'); // Named event.

	client.message(JSON.stringify({ foo: 'bar' })); // JSON message.
});

client.closed is a promise that resolves when the client closes the connection.

const clients = new Set();

server.sse('/sse', (req, url, client) => {
	clients.add(client);
	client.closed.then(() => clients.delete(client));
});

Connections can be manually closed with client.close(). This will also trigger the client.closed promise to resolve.

server.sse('/sse', (req, url, client) => {
	client.message('Hello, client!');

	setTimeout(() => {
		client.message('Goodbye, client!');
		client.close();
	}, 5000);
});

API > Routing > Webhooks

🔧 server.webhook(secret: string, path: string, handler: WebhookHandler)

Setup a webhook handler.

server.webhook(process.env.WEBHOOK_SECRET, '/webhook', payload => {
	// React to the webhook.
	return 200;
});

A webhook callback will only be called if the following critera is met by a request:

  • Request method is POST (returns 405 otherwise)
  • Header X-Hub-Signature-256 is present (returns 400 otherwise)
  • Header Content-Type is application/json (returns 401 otherwise)
  • Request body is a valid JSON object (returns 500 otherwise)
  • HMAC signature of the request body matches the X-Hub-Signature-256 header (returns 401 otherwise)

Note

Constant-time comparison is used to prevent timing attacks when comparing the HMAC signature.

API > Routing > WebSockets

🔧 server.websocket(path: string, handlers: WebSocketHandlers)

Register a route which handles websocket connections.

server.websocket('/path/to/websocket', {
	// all of these handlers are OPTIONAL

	accept: (req) => {
		// validates a request before it is upgraded
		// returns HTTP 401 if FALSE is returned
		// allows you to check headers/authentication

		// if an OBJECT is returned, the object will
		// be accessible on the websocket as ws.data.*

		return true;
	},

	open: (ws) => {
		// called when a websocket client connects
	},

	close: (ws, code, reason) => {
		// called when a websocket client disconnects
	},

	message: (ws, message) => {
		// called when a websocket message is received
		// message is a string
	},

	message_json: (ws, data) => {
		// called when a websocket message is received
		// payload is parsed as JSON

		// if payload cannot be parsed, socket will be
		// closed with error 1003: Unsupported Data

		// messages are only internally parsed if this
		// handler is present
	},

	drain: (ws) => {
		// called when a websocket with backpressure drains
	}
});

Important

While it is possible to register multiple routes for websockets, the only handler which is unique per route is accept(). The last handlers provided to any route (with the exception of accept()) will apply to ALL websocket routes. This is a limitation in Bun.

API > Server Control

🔧 server.stop(immediate: boolean)

Stop the server process immediately, terminating all in-flight requests.

server.stop(true);

Stop the server process gracefully, waiting for all in-flight requests to complete.

server.stop(false);

server.stop() returns a promise, which if awaited, resolves when all pending connections have been completed.

await server.stop(false);
// do something now all connections are done

API > Error Handling

🔧 ErrorWithMetadata(message: string, metadata: object)

The ErrorWithMetadata class allows you to attach metadata to errors, which can be used for debugging purposes when errors are dispatched to the canary.

throw new ErrorWithMetadata('Something went wrong', { foo: 'bar' });

Functions and promises contained in the metadata will be resolved and the return value will be used instead.

throw new ErrorWithMetadata('Something went wrong', { foo: () => 'bar' });

🔧 caution(err_message_or_obj: string | object, ...err: object[]): Promise<void>

Raise a warning issue on GitHub. This is useful for non-fatal issues which you want to be notified about.

Note

This function is only available if the canary feature is enabled.

try {
	// Perform a non-critical action, such as analytics.
	// ...
} catch (e) {
	// `caution` is async, you can use it without awaiting.
	caution(e);
}

Additional data can be provided as objects which will be serialized to JSON and included in the report.

caution(e, { foo: 42 });

A custom error message can be provided as the first parameter

Note

Avoid including dynamic information in the title that would prevent the issue from being unique.

caution('Custom error', e, { foo: 42 });

Issues raised with caution() are rate-limited. By default, the rate limit is 86400 seconds (24 hours), however this can be configured in the spooder.canary.throttle property.

{
	"spooder": {
		"canary": {
			"throttle": 86400
		}
	}
}

Issues are considered unique by the err_message parameter, so avoid using dynamic information that would prevent this from being unique.

If you need to provide unique information, you can use the err parameter to provide an object which will be serialized to JSON and included in the issue body.

const some_important_value = Math.random();

// Bad: Do not use dynamic information in err_message.
await caution('Error with number ' + some_important_value);

// Good: Use err parameter to provide dynamic information.
await caution('Error with number', { some_important_value });

🔧 panic(err_message_or_obj: string | object, ...err: object[]): Promise<void>

This behaves the same as caution() with the difference that once panic() has raised the issue, it will exit the process with a non-zero exit code.

Note

This function is only available if the canary feature is enabled.

This should only be used as an absolute last resort when the server cannot continue to run and will be unable to respond to requests.

try {
	// Perform a critical action.
	// ...
} catch (e) {
	// You should await `panic` since the process will exit.
	await panic(e);
}

🔧 safe(fn: Callable): Promise<void>

safe() is a utility function that wraps a "callable" and calls caution() if it throws an error.

Note

This utility is primarily intended to be used to reduce boilerplate for fire-and-forget functions that you want to be notified about if they fail.

safe(async (() => {
	// This code will run async and any errors will invoke caution().
});

safe() supports both async and sync callables, as well as Promise objects. safe() can also used with await.

await safe(() => {
	return new Promise((resolve, reject) => {
		// Do stuff.
	});
});

API > Content

🔧 parse_template(template: string, replacements: Replacements, drop_missing: boolean): Promise<string>

Replace placeholders in a template string with values from a replacement object.

const template = `
	<html>
		<head>
			<title>{$title}</title>
		</head>
		<body>
			<h1>{$title}</h1>
			<p>{$content}</p>
			<p>{$ignored}</p>
		</body>
	</html>
`;

const replacements = {
	title: 'Hello, world!',
	content: 'This is a test.'
};

const html = await parse_template(template, replacements);
<html>
	<head>
		<title>Hello, world!</title>
	</head>
	<body>
		<h1>Hello, world!</h1>
		<p>This is a test.</p>
		<p>{$ignored}</p>
	</body>
</html>

By default, placeholders that do not appear in the replacement object will be left as-is. Set drop_missing to true to remove them.

await parse_template(template, replacements, true);
<html>
	<head>
		<title>Hello, world!</title>
	</head>
	<body>
		<h1>Hello, world!</h1>
		<p>This is a test.</p>
		<p></p>
	</body>
</html>

parse_template supports passing a function instead of a replacement object. This function will be called for each placeholder and the return value will be used as the replacement. This function can be a Promise/async function.

const replacer = (placeholder: string) => {
	return placeholder.toUpperCase();
};

await parse_template('Hello {$world}', replacer);
<html>
	<head>
		<title>TITLE</title>
	</head>
	<body>
		<h1>TITLE</h1>
		<p>CONTENT</p>
		<p>IGNORED</p>
	</body>
</html>

parse_template supports optional scopes with the following syntax.

{$if:foo}I love {$foo}{/if}

Contents contained inside an if block will be rendered providing the given value, in this case foo is truthy in the substitution table.

An if block is only removed if drop_missing is true, allowing them to persist through multiple passes of a template.

parse_template supports looping arrays with the following syntax.

{$for:foo}My colour is %s{/for}
const template = `
	<ul>
		{$for:foo}<li>%s</li>{/for}
	</ul>
`;

const replacements = {
	foo: ['red', 'green', 'blue']
};

const html = await parse_template(template, replacements);
<ul>
	<li>red</li>
	<li>green</li>
	<li>blue</li>
</ul>

All placeholders inside a {$for:} loop are substituted, but only if the loop variable exists.

In the following example, missing does not exist, so test is not substituted inside the loop, but test is still substituted outside the loop.

<div>Hello {$test}!</div>
{$for:missing}<div>Loop {$test}</div>{/for}
await parse_template(..., {
	test: 'world'
});
<div>Hello world!</div>
{$for}Loop <div>{$test}</div>{/for}

🔧 generate_hash_subs(length: number, prefix: string, hashes?: Record<string, string>): Promise<Record<string, string>>

Generate a replacement table for mapping file paths to hashes in templates. This is useful for cache-busting static assets.

Important

Internally generate_hash_subs() uses git ls-tree -r HEAD, so the working directory must be a git repository.

let hash_sub_table = {};

generate_hash_subs().then(subs => hash_sub_table = subs).catch(caution);

server.route('/test', (req, url) => {
	return parse_template('Hello world {$hash=docs/project-logo.png}', hash_sub_table);
});
Hello world 754d9ea

Important

Specify paths as they appear in git, relative to the repository root and with forward slashes (no leading slash).

By default hashes are truncated to 7 characters (a short hash), a custom length can be provided instead.

generate_hash_subs(40).then(...);
// d65c52a41a75db43e184d2268c6ea9f9741de63e

Note

SHA-1 hashes are 40 characters. Git is transitioning to SHA-256, which are 64 characters. Short hashes of 7 are generally sufficient for cache-busting.

Use a different prefix other than hash= by passing it as the first parameter.

generate_hash_subs(7, '$#').then(subs => hash_sub_table = subs).catch(caution);

server.route('/test', (req, url) => {
	return parse_template('Hello world {$#docs/project-logo.png}', hash_sub_table);
});

🔧 get_git_hashes(length: number): Promise<Record<string, string>>

Internally, generate_hash_subs() uses get_git_hashes() to retrieve the hash table from git. This function is exposed for convenience.

Important

Internally get_git_hashes() uses git ls-tree -r HEAD, so the working directory must be a git repository.

const hashes = await get_git_hashes(7);
// { 'docs/project-logo.png': '754d9ea' }

If you're using generate_hash_subs() and get_git_hashes() at the same time, it is more efficient to pass the result of get_git_hashes() directly to generate_hash_subs() to prevent redundant calls to git.

const hashes = await get_git_hashes(7);
const subs = await generate_hash_subs(7, undefined, hashes);

// hashes[0] -> { 'docs/project-logo.png': '754d9ea' }
// subs[0] -> { 'hash=docs/project-logo.png': '754d9ea' }

🔧 apply_range(file: BunFile, request: Request): HandlerReturnType

apply_range parses the Range header for a request and slices the file accordingly. This is used internally by server.dir() and exposed for convenience.

server.route('/test', (req, url) => {
	const file = Bun.file('./test.txt');
	return apply_range(file, req);
});
GET /test HTTP/1.1
Range: bytes=0-5

HTTP/1.1 206 Partial Content
Content-Length: 6
Content-Range: bytes 0-5/6
Content-Type: text/plain;charset=utf-8

Hello,

API > State Management

🔧 set_cookie(res: Response, name: string, value: string, options?: CookieOptions)

Set a cookie onto a Response object.

const res = new Response('Cookies!', { status: 200 });
set_cookie(res, 'my_test_cookie', 'my_cookie_value');
HTTP/1.1 200 OK
Set-Cookie: my_test_cookie=my_cookie_value
Content-Length: 8

Cookies!

Important

Spooder does not URL encode cookies by default. This can result in invalid cookies if they contain special characters. See encode option on CookieOptions below.

type CookieOptions = {
	same_site?: 'Strict' | 'Lax' | 'None',
	secure?: boolean,
	http_only?: boolean,
	path?: string,
	expires?: number,
	encode?: boolean,
	max_age?: number
};

Most of the options that can be provided as CookieOptions are part of the standard Set-Cookie header. See HTTP Cookies - MDN.

Passing encode as true will URL encode the cookie value.

set_cookie(res, 'my_test_cookie', 'my cookie value', { encode: true });
Set-Cookie: my_test_cookie=my%20cookie%20value

🔧 get_cookies(source: Request | Response, decode: boolean = false): Record<string, string>

Get cookies from a Request or Response object.

GET /test HTTP/1.1
Cookie: my_test_cookie=my_cookie_value
const cookies = get_cookies(req);
{ my_test_cookie: 'my_cookie_value' }

Cookies are not URL decoded by default. This can be enabled by passing true as the second parameter.

GET /test HTTP/1.1
Cookie: my_test_cookie=my%20cookie%20value
const cookies = get_cookies(req, true);
{ my_test_cookie: 'my cookie value' }

API > Database Schema

spooder provides a straightforward API to manage database schema in revisions through source control.

Database schema is updated with db_update_schema_DRIVER where DRIVER corresponds to the database driver being used.

Note

Currently, only SQLite and MySQL are supported. This may be expanded once Bun supports more database drivers.

// sqlite example
import { db_update_schema_sqlite } from 'spooder';
import { Database } from 'bun:sqlite';

const db = new Database('./database.sqlite');
await db_update_schema_sqlite(db, './schema');
// mysql example
import { db_update_schema_mysql } from 'spooder';
import mysql from 'mysql2';

const db = await mysql.createConnection({
	// connection options
	// see https://github.com/mysqljs/mysql#connection-options
})

Important

MySQL requires the optional dependency mysql2 to be installed - this is not automatically installed with spooder. This will be replaced when bun:sql supports MySQL natively.

Database initiation and schema updating can be streamlined with the db_init_schema_DRIVER functions. The following examples are equivalent to the above ones.

// sqlite example
import { db_init_schema_sqlite } from 'spooder';
const db = await db_init_schema_sqlite('./database.sqlite', './schema');
// mysql example
import { db_init_schema_mysql } from 'spooder';
const db = await db_init_schema_mysql({
	// connection options
	// see https://github.com/mysqljs/mysql#connection-options
}, './schema');

Pooling

MySQL supports connection pooling. spooder allows you to create a connection pool instead of a single connection.

import { db_init_schema_mysql_pool } from 'spooder';
const pool = await db_init_schema_mysql_pool({
	// connection options
	connectionLimit: 10
});

const connection = await pool.getConnection();

Schema Files

The schema directory is expected to contain an SQL file for each table in the database with the file name matching the name of the table.

Note

The schema directory is searched recursively and files without the .sql extension (case-insensitive) will be ignored.

- database.sqlite
- schema/
	- users.sql
	- posts.sql
	- comments.sql

import { db_update_schema_sqlite } from 'spooder';
import { Database } from 'bun:sqlite';

const db = new Database('./database.sqlite');
await db_update_schema_sqlite(db, './schema');

Each of the SQL files should contain all of the revisions for the table, with the first revision being table creation and subsequent revisions being table modifications.

-- [1] Table creation.
CREATE TABLE users (
	id INTEGER PRIMARY KEY,
	username TEXT NOT NULL,
	password TEXT NOT NULL
);

-- [2] Add email column.
ALTER TABLE users ADD COLUMN email TEXT;

-- [3] Cleanup invalid usernames.
DELETE FROM users WHERE username = 'admin';
DELETE FROM users WHERE username = 'root';

Each revision should be clearly marked with a comment containing the revision number in square brackets. Anything proceeding the revision number is treated as a comment and ignored.

Note

The exact revision header syntax is ^--\s*\[(\d+)\].

Everything following a revision header is considered part of that revision until the next revision header or the end of the file, allowing for multiple SQL statements to be included in a single revision.

When calling db_update_schema_sqlite, unapplied revisions will be applied in ascending order (regardless of order within the file) until the schema is up-to-date.

It is acceptable to omit keys. This can be useful to prevent repitition when managing stored procedures, views or functions.

-- example of repetitive declaration

-- [1] create view
CREATE VIEW `view_test` AS SELECT * FROM `table_a` WHERE col = 'foo';

-- [2] change view
DROP VIEW IF EXISTS `view_test`;
CREATE VIEW `view_test` AS SELECT * FROM `table_b` WHERE col = 'foo';

Instead of unnecessarily including each full revision of a procedure, view or function in the schema file, simply store the most up-to-date one and increment the version.

-- [2] create view
CREATE OR REPLACE VIEW `view_test` AS SELECT * FROM `table_b` WHERE col = 'foo';

Schema revisions are tracked in a table called db_schema which is created automatically if it does not exist with the following schema.

CREATE TABLE db_schema (
	db_schema_table_name TEXT PRIMARY KEY,
	db_schema_version INTEGER
);

The table used for schema tracking can be changed if necessary by providing an alternative table name as the third paramater.

await db_update_schema_sqlite(db, './schema', 'my_schema_table');

Important

The entire process is transactional. If an error occurs during the application of any revision for any table, the entire process will be rolled back and the database will be left in the state it was before the update was attempted.

Important

db_update_schema_sqlite will throw an error if the revisions cannot be parsed or applied for any reason. It is important you catch and handle appropriately.

try {
	const db = new Database('./database.sqlite');
	await db_update_schema_sqlite(db, './schema');
} catch (e) {
	// panic (crash) or gracefully continue, etc.
	await panic(e);
}

Schema Dependencies

By default, schema files are executed in the order they are provided by the operating system (generally alphabetically).

If you have a schema file that depends on one or more other schema files to be executed before it (for example, using foreign keys), you can specify dependencies.

-- [deps] table_b_schema.sql, table_c_schema.sql
-- [1] create table_a
CREATE ...

Important

Cyclic or missing dependencies will throw an error.

Legal

This software is provided as-is with no warranty or guarantee. The authors of this project are not responsible or liable for any problems caused by using this software or any part thereof. Use of this software does not entitle you to any support or assistance from the authors of this project.

The code in this repository is licensed under the ISC license. See the LICENSE file for more information.

About

Purpose-built web server written for Bun

Topics

http http-server bun

Resources

Readme

License

ISC license
Activity

Stars

2 stars

Watchers

2 watching

Forks

0 forks
Report repository

Releases 77

4.5.9 Latest
Jan 23, 2025
+ 76 releases

Languages