Convert Node Streams into ECMAScript-Observables
npm install @teppeis/stream-to-observableThis is a fork of SamVerschueren/stream-to-observable.
> Convert Node Streams into ECMAScript-Observables
Observables are rapidly gaining popularity. They have much in common with Streams, in that they both represent data that arrives over time. Most Observable implementations provide expressive methods for filtering and mutating incoming data. Methods like .map(), .filter(), and .forEach behave very similarly to their Array counterparts, so using Observables can be very intuitive.
Note: This module was forked from stream-to-observable and released under a different name due to inactivity.
``
$ npm install --save @teppeis/stream-to-observable
stream-to-observable relies on any-observable, which will search for an available Observable implementation. You need to install one yourself:
`
$ npm install --save zen-observable
or
`
$ npm install --save rxjs
If your code relies on a specific Observable implementation, you should likely specify one using any-observables registration shortcuts.
`js
const fs = require('fs');
const split = require('split');
const streamToObservable = require('@teppeis/stream-to-observable');
const readStream = fs
.createReadStream('./hello-world.txt', {encoding: 'utf8'})
.pipe(split());
streamToObservable(readStream)
.filter(chunk => /hello/i.test(chunk))
.map(chunk => chunk.toUpperCase())
.forEach(chunk => {
console.log(chunk); // only the lines containing "hello" - and they will be capitalized
});
`
The split module above will chunk the stream into individual lines. This is often very handy for text streams, as each observable event is guaranteed to be a line.
#### stream
Type: ReadableStream
Note:
stream can technically be any EventEmitter instance. By default, this module listens to the standard Stream events (data, error, and end), but those are configurable via the options parameter. If you are using this with a standard Stream, you likely won't need the options parameter.
#### options
##### await
Type: Promise
If provided, the Observable will not "complete" until await is resolved. If await is rejected, the Observable will immediately emit an error event and disconnect from the stream. This is mostly useful when attaching to the stdin or stdout streams of a child_process. Those streams usually do not emit error events, even if the underlying process exits with an error. This provides a means to reject the Observable if the child process exits with an unexpected error code.
##### endEvent
Type: String or false "end"
Default:
If you are using an EventEmitter or non-standard Stream, you can change which event signals that the Observable should be completed.
Setting this to false will avoid listening for any end events.
Setting this to false and providing an await Promise will cause the Observable to resolve immediately with the await Promise (the Observable will remove all it's data event listeners from the stream once the Promise is resolved).
##### errorEvent
Type: String or false "error"
Default:
If you are using an EventEmitter or non-standard Stream, you can change which event signals that the Observable should be closed with an error.
Setting this to false will avoid listening for any error events.
##### dataEvent
Type: String
Default:
If you are using an EventEmitter or non-standard Stream, you can change which event causes data to be emitted to the Observable.
- Overview
- Formal Spec
- egghead.io lesson - Video
- rxjs observables - Observables implementation
- zen-observables - Observables implementation
data events on the stream will be emitted as events in the Observable. Since most native streams emit chunks of binary data, you will likely want to use a TransformStream to convert those chunks of binary data into an object stream. split` is just one popular TransformStream that splits streams into individual lines of text.
It's important to note that using this module disables back-pressure controls on the stream. As such, it should not be used where back-pressure throttling is required (i.e. high volume web servers). It still has value for larger projects, as it can make unit testing streams much cleaner.
MIT