abck | Library which makes generating akamai cookies | Plugin library

Windows Solution

If you are running on Windows then CTRL-C interrupt handling does not seem to work too well with multiprocessing pools. The following is a bit clumsy but seems to work after a fashion.

The idea is to initialize each process in the multiprocessing pool with a global variable ctrl_c_entered that is set initially to False. I have completed your class TestClass with a method foo which will be the worker function invoked. It must when invoked:

1. Test the global flag ctrl_c_entered and if True, immediately return.
2. Have its own KeyboardInterrupt handler and on such an interrupt it must set the global ctrl_c_entered flag to True and return.
3. Update: However, the CTRL-C could be entered when the pool process has not transferred control to the worker function. For instance, it could be in the process of grabbing from the input queue the next task to run. In this case there would otherwise be no try/catch for KeyboardInterrupt exceptions in effect. So we needs to set an interrupt handler for the SIGINT interrupt for each process in the pool that will set the ctrl_c_entered flag to True. But this now means that the original, default SIGINT interrupt handler must be temporarily restored in Step 2 above in order to catch KeyboardInterrupt exceptions.

You also have to let all the submitted asyncio tasks complete. So we set a signal.SIGINT interrupt handler that sets a global ctrl_c_entered flags for the main process to True if CTRL-C has been entered (we do not break out of the asyncio.run(main() statement. Our long-running asyncio tasks must check this ctrl_c_entered flag and terminate if it is set to True.