How to sum inner property of nested observable arrays, using Angular/RXJS?

Question

I am having trouble getting the sum (or any reduction) of an inner number property of an Observable within another Observable.

I have an Observable array of "Account" objects (Observable<AppAccount[]>).

export interface AppAccount {
    _id?: string;
    name: string;
}

And an Observable array of "Balance" objects, each with an accountId. many balances can be associated with an account (sorted/filtered by date, but that piece is removed for brevity)

export interface AccountBalance {
    _id?: string;
    accountId: string;
    amount: number;
}

I have a helper method which returns just the last Balance object's Amount for a given Account.

getLastAmount(account: AppAccount): Observable<number> {
    return this.balanceService.balances$.pipe(
      map(balances => {
        let last = balances.filter(balance => {
          return balance.accountId === account._id;
        }).sort().pop();
        //console.log(last)
        return last ? last.amount : 0;
      }),
      tap(amount => console.log(`getLastAmount() => ${amount}`)),
    );
  }

Now I am trying to write a method which will loop through the Accounts, call getLastAmount() for each, and then sum them all and return an Observable. This is what I have managed so far:

getTotalBalance(accounts$: Observable<AppAccount[]>): Observable<number> {
    return accounts$.pipe(
      map(accounts => from(accounts)),
      mergeAll(),
      mergeMap(account => this.getLastAmount(account)),
      reduce((sum, current) => {
        console.log(`${sum} + ${current}`);
        return sum + current;
      }, 0)
    );
  }

But this seems to never return, and gets stuck in an infinite loop??

With just one account and one balance associated, with the balance having an 'amount' of '10', I get this from my console log: '0 + 10' over and over, and the network log also confirm it is calling getBalances() continuously.

Am I on the right track? Is there a better way? Why does this RXJS pipe get stuck in a loop?

EDIT: I've made some changes based on picci's suggestions:

getTotalBalance(accounts$: Observable<AppAccount[]>): Observable<number> {
    return accounts$.pipe(
      map(accounts => accounts.map(account => this.getLastAmount(account))),
      concatMap(balances$ => { console.log('balances$', balances$); return forkJoin(balances$); }),
      tap(balances => console.log('balances', balances)),
      map(balances => balances.reduce(
        (amountSum, amount) => {
          console.log(`${amountSum} + ${amount}`)
          amountSum = amountSum + amount;
          return amountSum
        }, 0))
    );
  }

But this is still not returning, or the pipe is not completing? I've made a stackblitz here: https://stackblitz.com/edit/angular-rxjs-nested-obsv If you check the console output, it seems to not get any further than the forkJoin call...

Answer 1

Well - first I don't think you should be using obsevables like that.

If you only need to the totalBalance you could use something like this (:

  private appAcount$ = from<AppAccount[]>([
    { _id: '1', name: 'user-1' },
    { _id: '2', name: 'user-2' },
    { _id: '3', name: 'user-3' },
  ]);

  // this would be your http call
  public getBalances(accountId: string): Observable<AccountBalance[]> {
    const ab = [
      { _id: '1', accountId: '1', amount: 100 },
      { _id: '2', accountId: '2', amount: 200 },
      { _id: '3', accountId: '2', amount: 300 },
      { _id: '4', accountId: '3', amount: 400 },
      { _id: '5', accountId: '3', amount: 500 },
      { _id: '6', accountId: '3', amount: 600 },
    ];

    return of(ab.filter(x => x.accountId === accountId));
  }

  lastAmounts$: Observable<AccountBalance[]> = this.appAcount$
    .pipe(
      switchMap(appAccount => 
        this.getBalances(appAccount._id)
          .pipe(
            // replace this with your date filter
            map(balances => [balances[balances.length - 1]]) 
          )
      ),
      scan((acc, c) => [ ...acc, ...c ])
    );

  totalBalance$ = this.lastAmounts$
    .pipe(
      map(x => x.reduce((p, c) => p += c.amount, 0))
    )

If you only need the total balance, the you could just subscribe to the totalBalance$ observable.

Let me say though, that I wouldn't recommend doing an HTTP call for each appAccount if you can do a batch fetch of all AccountBalances of all the appAccounts you have - this way you can just use combineLatest for both appAccounts$ and balances$.

Answer 2

If I understand right, you could proceed like this

// somehow you start with the observable which returns the array of accounts
const accounts$: Observable<AppAccount[]> = getAccounts$()
// you also set the date you are interested in
const myDate: Moment = getDate()

// now you build the Observable<number> which will emit the sum of the last balance amounts
const amountSum$: Observable<number> = accounts$.pipe(
  // you transform an array of accounts in an array of Observable<number> representing the last Balance amount
  map((accounts: Account[]) => {
    // use the getLastAmount function you have coded
    return accounts.map(account => getLastAmount(account, myDate))
  }),
  // now we trigger the execution of the Observable in parallel using concatMap, which basically mean wait for the source Observable to complete
  // and forkJoin which actually executes the Observables in parallel
  concatMap(accounts$ => forkJoin(accounts$)),
  // now that we have an array of balances, we reduce them to the sum using the Array reduce method
  map(balances => balances.reduce(
    (amountSum, amount) => {
      amountSum = amountSum + amount;
      return amountSum
    }, 0)
  )
)

// eventually you subscribe to the amountSum$ Observable to get the result
amountSum$.subscribe({
  next: amountSum => console.log(`The sum of the last balances is: ${amountSum}`),
  error: console.err,
  complete: () => console.log("I am done")
})

There may be other combinations that bring to the same result, but this seems to work and can be checked in this stackblitz.

If you are interested in some frequent patterns of RxJS with http calls, you may want to read this blog.