Filter not null filters out all emissions that are null
or undefined
// No emissions if there's a network error
this.http.get(...).pipe(
filterNotNull(),
catchError(err => of(null)
)
Performs the given mapping operation and if the result wasn't null
or undefined
triggers an emission.
// No emissions on even numbers
of(1,2,3,4,5,6).pipe(
mapNotNull(it => it%2===0 ? null : it)
)
// Emits: 1, 3, 5 and completes
From a given stream returns three streams:
data$
the data stream itself. If an error occurs doesn't emit errors nor dataerror$
only emits an error on the regular pipeline if the source stream emits an errorloading$
emitstrue
when the source observable is still processing and false when it sends data, an error or completes
It's really useful to use this one to match your components full state:
<!-- component.html -->
<div>
<loading-spinner *ngIf="loading$|async"></loading-spinner>
<app-error *ngIf="error$|async as err" [error]="err></app-error>
<div *ngIf="data$|async as data"> {{data | json }}</div>
</div>
@Component({...})
export class YourComponent implements OnInit {
loading$: Observable<boolean>
error$: Observable<any>
data$: Observable<Type>
constructor(
private actRoute: ActivatedRoute,
private apiService: ApiService
) {
const dataStream = this.actRoute.paramMap.pipe(
map(params => params.get('id')),
switchMap(id => this.apiService.get(id))
)
const { data$, error$, loading$ } = streamsOf(dataStream)
this.data$ = data; this.error$ = error; this.loading$ = loading
}
}
From this example, how does it look?
This code...
@Component({
selector: 'app-root',
templateUrl: './app.component.html',
styleUrls: ['./app.component.css']
})
export class AppComponent {
loading$: Observable<boolean>;
obs$: Observable<any>
error$: Observable<any>
constructor(
httpClient: HttpClient
) {
const {data$, loading$, error$} = streamsOf(
httpClient.get('https://pokeapi.co/api/v2/pokemon/mew').pipe(
delay(1000)
)
)
this.obs$ = data$;
this.error$ = error$;
this.loading$ = loading$
}
}
results in...
This code...
@Component({...})
export class AppComponent {
loading$: Observable<boolean>;
obs$: Observable<any>
error$: Observable<any>
constructor(
httpClient: HttpClient
) {
const {data$, loading$, error$} = streamsOf(
httpClient.get('https://pokeapi.co/api/v2/pokemon/mew').pipe(
delay(1000), // Simulate latency ⌛
// force an error 💥
switchMapTo(throwError({message: 'Something really unexpected happenned', status: 500}))
)
)
this.obs$ = data$;
this.error$ = error$;
this.loading$ = loading$
}
}
results in...
This code...
@Component({...})
export class AppComponent {
loading$: Observable<boolean>;
obs$: Observable<any>
error$: Observable<any>
constructor(
httpClient: HttpClient
) {
const {data$, loading$, error$} = streamsOf(
httpClient.get('http://httpstat.us/500').pipe(
delay(1000), // Simulate latency
ngDefaultRetryLogic()
)
)
this.obs$ = data$;
this.error$ = error$;
this.loading$ = loading$
}
}
results in...
Adds retry logic with a certain criteria which is passed by the RetryOptions
interface.
retryWith
(options). It's a wrapper that hides a little bit the RxJS boilerplate that one needs to write in order to get retries on a streamdefaultRetryLogic()
is a wrapper on top of the previous operation settingretries: 2, delay: 500ms
you may pass a callback to determine whether a retry should happen based on the error that just got emitted.ngDefaultRetryLogic()
yet another wrapper on the previous one. This one assumes that the shape of the error is of anHttpErrorResponse
and it lets you to pass a list of http error codes where to perform a retry; by default it only retries on a 500 error code.
throwError(new Error('Something bad')).pipe(
// Only retry if the error message is 'Something bad'
defaultRetryLogic(err => err.message === 'Something bad')
).subscribe({
// After three retries will print "crash" and the error message
error: err => console.log('crash!', err)
})
this.http.get(...).pipe(
ngDefaultRetryLogic() // Only retries if the backend returns a 500
)