如何用Jest覆盖类方法try/catch中的catch分支测试？

First, let's recap your code for clarity:

Your class method:

get () { 
  try { 
    return 'value' 
  } catch (e) { 
    return 'value2' 
  } 
}

Your existing test:

test('the get method retrieves value', () => { 
  const value = MyClass.get() 
  expect(value).toBe('value') 
})

The core issue here is that the try block in your get() method currently has no code that can throw an exception—returning a string will never trigger an error. That's why your coverage report shows the catch branch is uncovered. Here's how to fix this, depending on your actual implementation:

Case 1: Your real get() method has dependencies (the example is simplified) #

If the return 'value' is a stand-in for actual logic that could fail (like calling an API, accessing a database, or invoking another service), you need to mock that dependency to throw an error in your test.

For example, if your get() method actually calls a dataService.fetchValue() function, here's how you'd test the catch branch with Jest:

// Mock the dependency to throw an error
jest.mock('./dataService', () => ({
  fetchValue: jest.fn().mockImplementation(() => {
    throw new Error('Test error trigger');
  })
}));

test('the get method returns value2 when an error occurs', () => {
  const value = MyClass.get();
  expect(value).toBe('value2');
});

This forces the try block to fail, sending execution into the catch branch and covering that code.

Case 2: Your get() method is exactly as written (no failing logic) #

If the try block truly only contains return 'value', then the catch branch is unreachable code—there's no scenario where that return statement would throw an error. In this case, you have two practical options:

• Remove the unnecessary try/catch block from your class method (since it serves no functional purpose here).
• If you must cover it for compliance reasons, you can use a testing tool to manipulate execution flow (though this is a workaround for code that shouldn't exist). For example, with Jest, you could spy on and mock internal parts of the method, but this tests artificial behavior rather than real-world scenarios.

内容的提问来源于stack exchange，提问作者tommyd456