fix: incorrect offset in fixUrl function

[szmarczak]

Fixes #1183

[B4nan]

cc @vladfrangu, looks like the new infinite scrolling implementation makes this test flaky. would be good to do something about that, better to have it skipped if it would be too flaky, it can easily just add noise instead of value that way

FAIL test/puppeteer_utils.test.js (7.79 s)
  ● Apify.utils.puppeteer › with launchPuppeteer › infiniteScroll() › stopScrollCallback works

    expect(received).toBe(expected) // Object.is equality

    Expected: true
    Received: false

      398 |                 // The test passes because the Infinite waitForSecs is broken by the callback.
      399 |                 // If it didn't, the test would time out.
    > 400 |                 expect(after).toBe(true);
          |                               ^
      401 |             });
      402 |         });
      403 |

      at Object.<anonymous> (test/puppeteer_utils.test.js:400:31)
          at runMicrotasks (<anonymous>)

[vladfrangu]

cc @vladfrangu, looks like the new infinite scrolling implementation makes this test flaky. would be good to do something about that, better to have it skipped if it would be too flaky, it can easily just add noise instead of value that way

FAIL test/puppeteer_utils.test.js (7.79 s)
  ● Apify.utils.puppeteer › with launchPuppeteer › infiniteScroll() › stopScrollCallback works

    expect(received).toBe(expected) // Object.is equality

    Expected: true
    Received: false

      398 |                 // The test passes because the Infinite waitForSecs is broken by the callback.
      399 |                 // If it didn't, the test would time out.
    > 400 |                 expect(after).toBe(true);
          |                               ^
      401 |             });
      402 |         });
      403 |

      at Object.<anonymous> (test/puppeteer_utils.test.js:400:31)
          at runMicrotasks (<anonymous>)

The issue isn't really the implementation of the scrolling, but the fact that it's not instant like window.scrollBy (to emulate a more human-like scroll), and past tweaking the delay between the scrolls.. I don't think (but do correct me if I'm wrong) there's anything more we can do.. 😢
From my testing, with stopScrollCallback to true it almost scrolls to the bottom at times (increasing the delay between scrolls to 150 made it almost perfect on my PC and in some CIs..). I guess we could increase the delays between scrolls to 250ms instead? Might be more ideal for users too I suppose

[B4nan]

Yeah either increase the timeout or disable the test. I already disabled most of the flaky tests in past weeks.

[vladfrangu]

Let's increase the timeout to 250ms and see what happens 👍