rxjs 如何期望 observable 拋出錯誤

Question

在我的 TypeScript 應用程序中，我有一個返回 rxjs Observable 的方法，在某些情況下，它可以返回throwError ：

import { throwError } from 'rxjs';

// ...

getSomeData(inputValue): Observable<string> {
  if (!inputValue) {
    return throwError('Missing inputValue!');
  }

  // ...
}

我如何編寫測試來涵蓋這個特定案例？

Answer 1

您可以使用 RxJS Marble 圖表測試來測試它。 就是這樣：

const getSomeData = (inputValue: string): Observable<string> => {
  if (!inputValue) {
    return throwError('Missing inputValue!');
  }

  // e.g.
  return of(inputValue);
};

describe('Error test', () => {

  let scheduler: TestScheduler;

  beforeEach(() => {
    scheduler = new TestScheduler((actual, expected) => {
      expect(actual).toEqual(expected);
    });
  });

  it('should throw an error if an invalid value has been sent', () => {
    scheduler.run(({ expectObservable }) => {

      const expectedMarbles = '#'; // # indicates an error terminal event

      const result$ = getSomeData(''); // an empty string is falsy

      expectObservable(result$).toBe(expectedMarbles, null, 'Missing inputValue!');
    });
  });

  it('should emit an inputValue and immediately complete', () => {
    scheduler.run(({ expectObservable }) => {

      const expectedMarbles = '(a|)';

      const result$ = getSomeData('Some valid string');

      expectObservable(result$).toBe(expectedMarbles, { a: 'Some valid string' });
    });
  });
});

有關如何編寫這些測試的更多信息，請查看此鏈接。

Answer 2

我想你的完整案例類似於這樣的東西

// first there is something that emits an Observable
export function doSomethingThatReturnsAnObservable() {
  return createSomehowFirstObservable()
  .pipe(
     // then you take the data emitted by the first Observable 
     // and try to do something else which will emit another Observable
     // therefore you have to use an operator like concatMap or switchMap
     // this something else is where your error condition can occur
     // and it is where we use your getSomeData() function
     switchMap(inputValue => getSomeData(inputValue))
  );
}
}

// eventually, somewhere else, you subscribe
doSomethingThatReturnsAnObservable()
.subscribe(
   data => doStuff(data),
   error => handleError(error),
   () => doSomethingWhenCompleted()
)

錯誤條件的測試可能看起來像這樣

it('test error condition'), done => {
   // create the context so that the call to the code generates an error condition
   .....
   doSomethingThatReturnsAnObservable()
   .subscribe(
      null, // you are not interested in the case something is emitted
      error => {
        expect(error).to.equal(....);
        done();
      },
      () => {
        // this code should not be executed since an error condition is expected
        done('Error, the Observable is expected to error and not complete');
      }
   )
})

Answer 3

除了拉戈曼的回答。 您可以簡化獲取 testScheduler 的方法。

describe('Error test', () => {  
  it('should throw an error if an invalid value has been sent', () => {
    getTestScheduler().run(({ expectObservable }) => { // getTestScheduler from jasmine-marbles package

      const expectedMarbles = '#'; // # indicates an error terminal event

      const result$ = getSomeData(''); // an empty string is falsy

      expectObservable(result$).toBe(expectedMarbles, null, 'Missing inputValue!');
    });
  });

  it('should emit an inputValue and immediately complete', () => {
    getTestScheduler().run(({ expectObservable }) => {

      const expectedMarbles = '(a|)';

      const result$ = getSomeData('Some valid string');

      expectObservable(result$).toBe(expectedMarbles, { a: 'Some valid string' });
    });
  });
});

Answer 4

另一種方法是檢查操作員是否顯示錯誤，如下所示對其進行管道傳輸。 我假設您正在使用 Chai。

import { catchError } from 'rxjs/operators';

it('must throw an error',  done => {
     getSomeData().pipe(catchError((e) => [e])).subscribe(e => {
        expect(e).to.be.an.instanceof(Error);
        done();
      })
})

Source 如何測試拋出錯誤的 RxJS 操作