perf(e2e) : Optimize saved-replies test suite

[stefanbinoj]

ref #1064

Problem

• Previously, utils/replyHelper used promises with a timeout of 5000 ms to determine whether the modal was for creating or editing saved replies.

  const buttonPromises = [
    updateBtn
      .waitFor({ state: "visible", timeout: 5000 })
      .then(() => updateBtn)
      .catch(() => null),
    addBtn
      .waitFor({ state: "visible", timeout: 5000 })
      .then(() => addBtn)
      .catch(() => null),
    saveBtn
      .waitFor({ state: "visible", timeout: 5000 })
      .then(() => saveBtn)
      .catch(() => null),
  ];

  // Wait for any button to become visible
  const buttons = await Promise.all(buttonPromises);
  const visibleButton = buttons.find((btn) => btn !== null);

• As a result, the clickSaveButton function took 5 seconds timeout to execute, and it’s used in 13 places across the codebase.

Solution

• Instead of identifying the modal type via visible buttons, a more efficient approach is to distinguish it using the modal title and perform the action directly.

  const createDialog = page.locator('[role="dialog"]:has-text("New saved reply")');
  const editDialog = page.locator('[role="dialog"]:has-text("Edit saved reply")');

  const [isCreateContext, isEditContext] = await Promise.all([
    createDialog.isVisible({ timeout: 500 }).catch(() => false),
    editDialog.isVisible({ timeout: 500 }).catch(() => false)
  ]);

  if (isCreateContext) {
    await createDialog.locator('button:has-text("Add")').click();
  } else if (isEditContext) {
    await editDialog.locator('button:has-text("Update")').click();
  }

• This optimization reduces the total time from 5–10 seconds to under 500 ms. Saving couple of 5sec delays.

Local

Before	After

• Saves approximately 13 × 5 = 65 seconds overall.

AI Usage

• GPT (free) via chat.openai.com was used for formatting and fixing typos in the PR description.

Self Review

• All code was written by myself and I have reviewed all the code.

Live Stream Disclosure

• Viewed both live streams of antiwork and guidelines to follow.

[stefanbinoj]

@slavingia, could you please review? I’ve confirmed with the artifact from this PR’s Playwright E2E tests that this change reduces the runtime by 5 seconds across 13 tests.