def setup_data(model_path: Path, log_path: Path):
settings = Configuration(model_path=Path(model_path),
log_path=Path(log_path))
settings.fill_in_derived_fields()
log = LogReader(log_path)
graph = BPMNGraph.from_bpmn_path(model_path)
return graph, log, settings
def simulate(settings: Configuration, log_data, evaluate_fn: Callable = None):
"""General simulation function that takes in different simulators and evaluators."""
if evaluate_fn is None:
evaluate_fn = evaluate_logs
if isinstance(settings, dict):
settings = Configuration(**settings)
# Simulator choice based on configuration
if settings.simulator is SimulatorKind.BIMP:
raise NotImplementedError('BIMP simulator is not implemented')
elif settings.simulator is SimulatorKind.CUSTOM:
simulate_fn = diffresbp_simulator
settings.read_options.column_names = {
'CaseID': 'caseid',
'Activity': 'task',
'EnableTimestamp': 'enabled_timestamp',
'StartTimestamp': 'start_timestamp',
'EndTimestamp': 'end_timestamp',
'Resource': 'user'
}
else:
raise ValueError(f'Unknown simulator {settings.simulator}')
# Number of cases to simulate
n_cases = len(log_data.caseid.unique())
settings.simulation_cases = n_cases
reps = settings.repetitions
cpu_count = multiprocessing.cpu_count()
w_count = reps if reps <= cpu_count else cpu_count
pool = multiprocessing.Pool(processes=w_count)
# Simulate
args = [(settings, rep) for rep in range(reps)]
p = pool.map_async(simulate_fn, args)
progress_bar_async(p, 'simulating', reps)
# Read simulated logs
p = pool.map_async(_read_stats_alt, args)
progress_bar_async(p, 'reading simulated logs', reps)
# Evaluate
args = [(settings, log_data, log) for log in p.get()]
if n_cases > 1000:
pool.close()
results = [
evaluate_fn(arg) for arg in tqdm(args, 'evaluating results')
]
sim_values = list(itertools.chain(*results))
else:
p = pool.map_async(evaluate_fn, args)
progress_bar_async(p, 'evaluating results', reps)
pool.close()
sim_values = list(itertools.chain(*p.get()))
return sim_values
def discover(ctx, config_path):
repository_dir = get_project_dir()
ctx.params['config_path'] = repository_dir.joinpath(config_path)
config_data = config_data_from_file(config_path)
config_data.update(ctx.params)
config = Configuration(**config_data)
config.fill_in_derived_fields()
discoverer = Discoverer(config)
discoverer.run()
def test_AndPriorORemove_default(entry_point):
config_path = entry_point / 'optimize_debug_config_2.yml'
config = config_data_from_file(config_path)
assert config['strc'] is not None
structure_config = Configuration(**config['strc'])
assert structure_config.and_prior == [AndPriorORemove.FALSE]
assert structure_config.or_rep == [AndPriorORemove.FALSE]
def test_remove_outliers(args):
for arg in args:
settings = Configuration()
log_path = arg['log_path']
log = LogReader(log_path)
print(f'Running test for {log_path}')
result = remove_outliers(log.df)
assert result is not None
assert 'caseid' in result.keys()
assert 'duration_seconds' not in result.keys()
def test_splitminer(entry_point, arg, tmp_path):
log_path = entry_point / arg['log_path']
read_options = arg['read_options']
log = LogReader(log_path)
assert len(log.data) != 0
config = Configuration()
config.project_name = os.path.basename(log_path).split('.')[0]
config.read_options = read_options
config.output = tmp_path
output_path = config.output / (config.project_name + '.xes')
write_xes(log, output_path)
print(tmp_path.absolute())
config.epsilon = 0.5
config.eta = 0.5
config.and_prior = AndPriorORemove.FALSE
config.or_rep = AndPriorORemove.FALSE
exit_code = StructureMiner._sm3_miner(
log_path, config) # TODO: make _sm3_miner to return exit code
def test_diffresbp_simulator(entry_point, arg):
qbp_path = entry_point / arg['qbp_path']
config = Configuration()
config.output = qbp_path.parent
config.project_name, _ = os.path.splitext(qbp_path.name)
config.repetitions = 1
config.simulation_cases = get_number_of_cases(qbp_path)
diffresbp_simulator((config, config.repetitions))
json_path = qbp_path.with_suffix('.json')
assert json_path.exists()
output_dir = config.output / 'sim_data'
shutil.rmtree(output_dir)
def optimize(ctx, config_path):
repository_dir = get_project_dir()
ctx.params['config_path'] = repository_dir.joinpath(config_path)
config_data = config_data_from_file(config_path)
config_data.update(ctx.params)
strc_data = config_data.pop('strc')
tm_data = config_data.pop('tm')
global_data = config_data
global_config = Configuration(**global_data)
global_config.fill_in_derived_fields()
strc_data.update(global_data)
structure_optimizer_config = Configuration(**strc_data)
structure_optimizer_config.fill_in_derived_fields()
tm_data.update(global_data)
time_optimizer_config = Configuration(**tm_data)
time_optimizer_config.fill_in_derived_fields()
optimizer = Optimizer({'gl': global_config, 'strc': structure_optimizer_config, 'tm': time_optimizer_config})
optimizer.run(discover_model=global_config.model_path is None)