Unable to write multiple request intercepts to pact file

[kand]

Hello! I have a test that's something like this:

describe('test some UI', () => {
  before(() => cy.setupPact('consumer-ui', 'provider-api');

  it('User is able to do a process that causes two requests I want to record', () => {
    cy.intercept('POST', '/api/thing-one').as('thing-one');
    cy.intercept('POST', '/api/thing-two').as('thing-two');

    // ... interact with UI

    cy.usePactWait('thing-one').then((xhr) => {
      expect(xhr.response.statusCode).to.eq(200);

      // ... UI is in a new state, interact with it again to get the second request
      cy.usePactWait('thing-two').then((xhr) => {
        expect(xhr.response.statusCode).to.eq(200);
      });
    });
  });
});

At this point, I have a pact file that only includes the last interaction for thing-two and not the interaction for thing-one. I believe this is because constructPactFile is excluding "duplicate" interactions based on the test title, which in this case, the test has two interactions under the same test title, so the first one gets excluded/overwritten.

Am I using the Cypress adapter correctly? If so, is there a reason I can only write one interaction per test? Or is this a bug in the adapter?

Thanks!

[kand]

I think the interface would make intuitive sense if the formattedAlias was included in testCaseTitle here:

pact-cypress-adapter/src/index.ts

Line 60 in ac177d7

testCaseTitle: `${testCaseTitle}`,

Which would match the behavior of using cy.usePactWait with an array of aliases.

[nivedhasamy]

I am facing the same issue. Did you manage to resolve this?

[kand]

Because of the issue in the adapter code I referenced in my last comment, I think the only way to resolve this is to separate the tests. So instead of

Test
- setup scenario A
- test result A
- setup scenario B now that we're in A
- test result B

I split it into two tests:

Test A
- setup scenario A
- test result A

Test B
- setup scenario A
- setup scenario B now that we're in A
- test result B

Kind of a bummer since this is not how Cypress works. Probably better test design though.

[mefellows]

This sounds like a bug to me and is unexpected behaviour, although I take the point around better test design.

I think that's an unreasonable expectation of users to re-write such tests, so I think we should fix the problem here.