A branch task decides at run time which of its downstream tasks run. Write it with @task.branch and return a task id, a list of task ids, or None:
@task.branch
def pick(logical_date=None):
return "weekend_report" if logical_date.weekday() >= 5 else "weekday_report"
Every direct downstream task that is not returned is skipped, and the skip spreads to the tasks after it. That includes a join task that comes after both paths: with the default rule it would be skipped, so give it trigger_rule="none_failed_min_one_success" (run if nothing failed and at least one path succeeded).
The returned ids must be direct downstream tasks of the branch, or the task fails with "Invalid tasks found". BranchPythonOperator does the same in the classic style.