Token holder

A token holder in Alette Signal is an access control helper, storing a value acting as a JWT or OAuth token and a value acting as a refresh token, while managing their lifecycle.

Token provider

A token provider is a synchronous or asynchronous function, returning a value acting as a JWT or OAuth token and a value acting as a refresh token to the token holder:

async ({
    id,
	prevToken,
	refreshToken,
	context,
	getCredentials,
	getCredentialsOrThrow,
}) => {
	const { accessToken, refreshToken } = await getToken.execute();
	return {
        token: accessToken,
        refreshToken
	};
}

TIP

Token providers can omit refresh tokens:

async () => {
    const { accessToken, refreshToken } = await getToken.execute();
	return accessToken;
}

Token credentials

Token credentials is an arbitrary data stored inside a token holder, and is used by token providers to obtain or refresh tokens.

async ({
	/* ... */
	getCredentials,
	getCredentialsOrThrow,
}) => {
    const { email, password } = await getCredentialsOrThrow();
	const { accessToken, refreshToken } = await getToken.execute({ 
		args: { email, password } 
    });

	return {
        token: accessToken,
        refreshToken
	};
}

Configuring token holders

To configure a token holder, call the token() function obtained from the Alette Signal core plugin, and pass a token provider to the .from() token holder builder method:

// ./src/api/base.ts
const core = coreApiPlugin();
export const { token } = core.use();

// ./src/api/auth.ts
// ...
const getToken = mutation(/* ... */)

export const jwtToken = token()
	.from(async () => {
        const token = await getToken.execute();
        return token;
	})
	.build();

Return a record from the token provider to save both the token and the refresh token inside the token holder:

export const jwtToken = token()
	.from(async () => {
        const { accessToken, refreshToken } = await getToken.execute();
        return {
            token: accessToken,
            refreshToken
		};
	})
	.build();

DANGER

1. If a non-string value is returned to act as a token, the TokenTypeValidationError fatal error will be thrown.
2. If a non-string value is returned to act as a refresh token, the RefreshTokenTypeValidationError fatal error will be thrown.

TIP

The initial token provider can be overridden:

export const thirdPartyToken = token()
	.from(() => {
        console.error('Third party token provider was not implemented.');
        return '';
	})
	.build();

// Overrides the initial token provider
thirdPartyToken.from(() => getToken.execute())

TIP

Alette Signal allows for multiple token holders in the same application:

export const jwtToken = token()
	.from(() => getToken.execute())
	.build();

export const aiKey = token()
    .from(() => process['env']['AI_KEY'])
    .build();

Configuring credential storage

To configure the credential storage of a token holder, pass a runtime schema implementing the Standard Schema interface to the .credentials() token holder builder method:

const jwtToken = token()
	.credentials(z.object({
		email: z.string(),
		name: z.string()
	}))
	.from(() => getToken.execute())
	.build();

To access token credentials, destructure the first token provider argument:

const jwtToken = token()
	.credentials(z.object({
		email: z.string(),
		password: z.string()
	}))
	.from(async ({ 
		getCredentials,
		getCredentialsOrThrow, 
    }) => {
        const { email, password } = await getCredentialsOrThrow();
        return getToken.execute({ args: { email, password } })
	})
	.build();

Setting token credentials

To set token credentials, call the .using() token holder method:

jwtToken.using({ 
	email: 'alette-signal@mail.com',
	password: '12345',
})

To set token credentials while accessing previous credentials, pass a function to the .using() token holder method:

jwtToken.using(({ previous, context }) => {
    const previousPassword = previous?.password;
    
    return {
        email: 'alette-signal@mail.com',
        password: previousPassword || '12345',
    };
})

TIP

The .using() token holder method can be used with UI forms:

// React component
const Form = () => {
    // Pseudo code
    const { subscribe, /* ... */ } = useForm();

    useEffect(
        () => subscribe(({ values: { email, password } }) => {
            jwtToken.using({ email, password });
		}),
		[]
	);
    
    return <form>{ /* ... */ }</form>
}

Obtaining tokens

To obtain a token from the server, invoke the token provider by calling the .get() method on the token holder:

const myJwtToken = await jwtToken.get();

TIP

1. Only a single .get() call is made, while other calls to the same .get() method are queued.
2. If the token provider resolves successfully, the token value is propagated to the queued .get() callers without invoking the token provider again:

const [
    token1,
    token2,
    token3,
    token4,
    token5,
] = await Promise.all([
    // Only one server request is made,
	// and every pending `get()` invocation 
	// receive the same token without 
	// calling the server again.
    jwtToken.get(),
	jwtToken.get(),
	jwtToken.get(),
	jwtToken.get(),
	jwtToken.get(),
])

Invalidating tokens

To refresh a token, call the .refresh() method on the token holder:

jwtToken.invalidate();

Next time the .get() method is called, the token is re-obtained:

const newToken = await jwtToken.get();

WARNING

The .invalidate() token holder method does not call the token provider automatically.

Refreshing tokens

To refresh a token in the background, call the .refresh() method on the token holder:

jwtToken.refresh();

To refresh and get the token simultaneously, call the .refreshAndGet() method on the token holder:

const newToken = await jwtToken.refreshAndGet();

Subscribing to token changes

To subscribe to token changes, call the .onStatus() token holder method:

const unsubscribe = jwtToken.onStatus({
    loading: async ({ context }) => {
        // ...
    },
    valid: async ({ context }) => {
        // ...
    },
    invalid: async ({ context }) => {
        // ...
    },
});

TIP

Token holders allow subscribing to a subset of the token status events:

jwtToken.onStatus({
    valid: () => {
        // ...
    },
});

TIP

Token status subscriptions can be used to synchronize UI and token updates:

// React component
const AuthScreen = () => {
    const [isTokenValid, setIsValid] = useState(false);

    useEffect(() => {
        return jwtToken.onStatus({
            loading: async ({ context }) => {},
            valid: async ({ context }) => {},
            invalid: async ({ context }) => {},
        })
    }, []);

    if (isTokenValid) {
        return <div>{ /*...*/ }</div>;
    }

    return <div>Unauthorized</div>;
}

Periodic token refresh

To set up periodic token refresh in the background, pass an interval value to the .refreshEvery() token holder builder method:

const jwtToken = token()
	/* ... */
    .refreshEvery("20 seconds")
	// or
    .refreshEvery(5000)
	// or
    .refreshEvery("1 hour")
	.build();

INFO

Periodic token refresh uses the token obtaining algorithm under the hood.

Converting tokens to headers

To convert a token to HTTP headers, use the .toHeaders() token holder method:

const tokenValue = 'hey';

const jwtToken = token()
	.from(() => tokenValue)
	.build();

/**
* authHeaders - { Authorization: `Bearer ${tokenValue}` },
* */
const authHeaders = await jwtToken.toHeaders()

To change how a token is converted to HTTP headers, pass a function to the .whenConvertedToHeaders() token holder builder method:

const tokenValue = 'hey';

const jwtToken = token()
	.from(() => tokenValue)
	.whenConvertedToHeaders(({ token, context }) => ({
        'X-XSRF-TOKEN': token
	}))
	.build();

/**
* authHeaders -
* { 
* 	'X-XSRF-TOKEN': 'hey' 
* },
* */
const authHeaders = await jwtToken.toHeaders()

INFO

Alette Signal automatically obtains tokens being converted into headers.

Token obtaining algorithm

Alette Signal token obtaining algorithm has 6 steps:

1. The .get() method of the token holder is called.
2. If a valid token is stored inside the token holder, the token provider is not called, and the valid token is returned.
3. If an invalid token is stored inside the token holder, the token provider is invoked.
4. If a token is absent from the token holder, the token provider is invoked.
5. If the token provider throws an error, the token stored inside the token holder is marked as invalid.
6. If the token provider succeeds, the token is marked as valid and replaces the old token stored inside the token holder.