Weenie Express

This package is meant to be used with Weenie. However, if you are not using Weenie but are interested in this functionality, there's nothing wrong with using it directly.

The only required config for this module is config.http.listener.port. You may also set config.http.listener.host to override the default host (localhost).

This module augments a standard express app with some convenient additional functionality, each piece corresponding to an optional config value. Specifically:

• config.http.logIncomingReqs (default true) - Emit a debug-level log message for each incoming request with the request method and URL.
• config.http.parseJson (default true) - Automatically parse bodies as JSON for the configured mime types (see next).
• config.http.jsonMimeTypes (default ['application/json', 'application/json-rpc']) - Which mime types to parse as JSON (only applicable if config.http.parseJson is true).
• config.http.handleErrors (default true) - Handle errors in our standard way. If a request provides an accept header for JSON-RPC, returns a JSON-RPC-formatted error. Otherwise, returns { t: "error"; error: HttpError; }.
• config.http.handleFallthrough (default true) - Automatically throw 404 error if a request is not handled by one your registered handlers.
• config.http.listenOnReady (default true) - This is very Weenie-specific. The Weenie framework provides a "service manager" that provides a whenReady promise as a dependency. If listenOnReady is set to true, then this library automatically listens on the configured port and interface when you call deps.svc.declareReady().
• config.mask500Errors (default true) - If true or set to a string, this will mask the text of any 500 errors that are caught. If set to true, a default string is used. If set to a string, that string is used as the message.
• config.errOnBlankPost (default true) - Sometimes users will use the wrong content-type header for a POST, PUT or PATCH request, and this will result in the request body not parsing and appearing blank. If this option is set to true, a detailed error is thrown when a POST, PUT or PATCH request has a blank body to try to help the user figure out what's wrong. If you have any POST, PUT or PATCH endpoints that do not require a body, then set this to false.

Returns the following dependencies:

• deps.http - The express app. You can use this to register endpoints and middleware.
• deps.getHttpServers - A function that can be used to return all active http servers (returned by express.listen). You can use this to shut down the HTTP servers on SIGINT or SIGTERM or in tests, etc.

Example

Following is a typical example of how you'd use this.

import { Weenie, express, logger, serviceManager } from '@wymp/weenie-framework';

const config = {
  logger: {
    logLevel: 'notice',
  },
  svc: {
    handleShutdown: true,
  },
  http: {
    listener: {
      port: 80,
      host: '0.0.0.0',
    },
    logIncomingReqs: true,
    parseJson: true,
    jsonMimeTypes: ['application/json', 'application/json-rpc', 'application/vnd.api+json'],
    handleErrors: true,
    handleFallthrough: true,
    listenOnReady: true,
    mask500Errors: 'Oops! There was an error',
    errOnBlankPost: true,
  },
}

const deps = Weenie({ config })
  .and(logger)
  .and(serviceManager)
  .and(express)
  .done((d) => {
    // Use the service manager's shutdown tasks to shut down any HTTP servers that we've opened
    d.onShutdown(async () => {
      const servers = d.getHttpServers() ?? [];
      await Promise.all(
        servers.map((s) => new Promise<void>((res) => s.close(res)))
      );
    });

    return {
      log: d.log,
      svc: d.svc,
      http: d.http,
    }
  });

// Do other things ....
// For example:
deps.http.get('/', (req, res, next) => {
  deps.log.notice('Yaaaaaay');
  res.json({ status: 'ok' });
});

// Now declare that we're ready for business. Because of our configuration, this will start our HTTP server.
deps.svc.declareReady();