pyaltt 0.0.7

Various tools for functions, looped and queued workers etc.

pyaltt

Various tools for functions, looped and queued workers etc.

License: Apache License 2.0

FunctionCollection

Collect and run a pack of functions

Usage example:

from pyaltt import FunctionCollecton

funcs = FunctionCollecton()

@funcs
def f1():
    print('I am function 1')

@funcs
def f2():
    print('I am function 2')

funcs.run()

Real life example: define shutdown function collection and call shutdown.run() to stop all background threads.

Parameters:

• on_error function which is called if any function throws an exeption (with e=Exception argument), or if remove method is called and function doesn't exist in collection.
• on_error_kwargs additional arguments for on_error

Background workers

Looped worker decorator

Usage example:

from pyaltt import background_worker

#transforms function into background worker which run in a loop
@background_worker
def myworker(**kwargs):
    print('I\'m a worker ' + kwargs.get('worker_name'))

myworker.start()
# ................
myworker.stop()

Parameters for @background_worker:

• name alternative worker name (default is function name)
• daemon set background worker as daemon (default is True)
• o any object which worker can get with kwargs.get('o')
• on_error function which is called if worker throws an exeption (with e=Exception argument)
• on_error_kwargs additional arguments for on_error
• delay_before delay before each worker function execution
• delay_after (or simply delay) delay after each worker function execution
• interval same as delay_after but it will try to keep execution interval exactly
• poll_delay sleep precision (lower is better but uses CPU, default is 0.1 sec)
• realtime Use different algorythm for sleeping to be even more exact

Note: if running in virtual machine, unset "sync guest time with host" otherwise real time sleep may work unpredictable.

Poll delay should be lower or equal to delays. If you set very short delays, don't forget to decrease poll delay as well.

Real time loop execution in this library is not 100% exact and can't be used e.g. in heavy industry. Real time loops require dedicated coding for the particular task and for small delays can't be coded with Python at all or require special tricks/hardware.

Parameters for start():

• _daemon override initial params and set worker as daemon or not

• delay, interval and realtime parameters can be overriden in start (use kwargs with _: _delay, _interval etc.)

• All other parameters are passed to worker function (both args and kwargs)

Worker gets all parameters used in start() plus additionally, worker_name and o in kwargs (can be overriden in start()).

Paramter wait=False can be used for stop() to send a signal to worker and continue (default is wait until worker finish)

Queue worker decorator

from pyaltt import background_worker

#transforms function into background worker which run on task in queue
@background_worker(q=True)
def myworker(task, **kwargs):
    print('Got a task %s' % task)

myworker.start()
# ................
myworker.put('task 1')
# ................
myworker.stop()

Parameters:

• name, daemon, o, on_error, on_error_kwargs same as for looped worker
• queue_class use alternative queue class (e.g. queue.PriorityQueue, default is queue,Queue). If this parameter is set, q=True is not required.

Task can be any object (obvious). Worker has task always as first parameter.

Parameters for start() / stop() are the same as for looped worker.

Stopping loop from worker

This can be done in 2 ways: calling myworker.terminate() from worker function or simply return False.

Working directly with classes

Classes BackgroundWorker and BackgroundQueueWorker can be used without a wrapper.

You can override loop method to have own function executed when worker starts.

Loop calls run function which's actually a worker function (the same decorated/transformed in the examples above).

(c) 2018-2019 Altertech Group, https://www.altertech.com/