Yet another fetch based HTTP library :)
Networking.ts comes with batteries included, so you can go straight ahead and try some of the existing public API clients. Here's an example on how to consume GitHub RAW API:
const client = new RawGitHubNetworkingClient({
repository: {
owner: 'web-pacotes',
repo: 'networking.ts',
ref: 'master'
}
});
// Get README.md content
const getEndpointResult = await client.get({ endpoint: 'README.md' });
// Tadaaaam!
console.info(getEndpointResult);
Creating new clients can either be done by extending the NetworkingClient
class or manually creating NetworkingClient
instance:
const client = new NetworkingClient({
baseUrl: new URL('https://api.my-awesome-service.com/v1/'),
fetchClient: window.fetch
});
// Make some requests...
const healthResult = await client.get({ endpoint: 'health' });
const authenticateResult = await client.post({
endpoint: 'auth',
body: { username: 'my-username', password: 'oops' }
});
// extract the result value
const hasAuthenticated = fold(
authenticateResult,
(l) => false,
(r) => r.status === 204
);
Want to go ahead and try the library on the web? We got you: https://joaomagfreitas.link/demo-networking-ts/
Networking.ts aims to be a custom HTTP client library, so it provides designed API functions for the major HTTP methods: get
, post
, put
, patch
and delete
. The library also follows a functional style and aims to be side effect free, by replacing all exception/error throws with an Either
monad. All client functions
return a Either<HttpRequestError, HttpResponse>
which describes that the result is either an request error or response result. To query the result value you can use the isLeft
, isRight
and fold
functions.
Another niche detail about the library, is that it does not rely only on the existing fetch
implementation. Instead, it allows library clients to pass a fetch
function that knows how to resolve requests based on the fetch
spec. This is really neat when making requests in the browser or in Svelte.js, which bundles a custom fetch
implementation.
There is support for interceptors which can act at three levels:
- request level, useful to include mandatory headers before a request is sent to the server.
- response level, useful to parse the response in a different type before a response is returned to the caller.
- request error level, useful to log error messages before the error is returned to the caller.
To create an interceptor you will have to extend either the Interceptor
, RequestInterceptor
, ResponseInterceptor
or ErrorInterceptor
classes. The major difference in all of them, is that the last ones only describe how to intercept a request/response/request error.
Additionally, you can make use of the existing AuthorizationInterceptor
class, which intercepts a request and appends a basic authorization header. Here's how it's useful in the ImgurNetworkingClient
:
class ImgurApiAuthorizationInterceptor extends AuthorizationInterceptor {
constructor(clientId: string) {
super({
scheme: 'Client-ID',
parameters: clientId
});
}
}
export class ImgurNetworkingClient extends NetworkingClient {
constructor({
clientId,
apiVersion,
interceptors,
fetchClient,
timeoutMS
}: ImgurNetworkingClientNetworkingClientPositionalProperties) {
super({
baseUrl: resolveUrl('https://api.imgur.com/', apiVersion),
fetchClient: fetchClient,
timeoutMS: timeoutMS,
interceptors: [
...(interceptors ?? []),
new ImgurApiAuthorizationInterceptor(clientId)
]
});
}
}
- client net (a set of clients which each one is chosen to be used with a criteria algorithm, like load-balancing)
- post form data
- websocket support
- response streaming
- requests redirection follow-up
Found any bug (including typos) in the package? Do you have any suggestion or feature to include for future releases? Please create an issue via GitHub in order to track each contribution. Also, pull requests are very welcome!
To contribute, start by setting up your local development environment. The setup.md document will onboard you on how to do so!