def load_temp_log_if_it_is_there(self, log_name, session, parameters=None):
"""
Load a log associated to a session if it exists in the temp logs folder
Parameters
--------------
log_name
Log name
session
Session ID
parameters
Possible parameters of the algorithm
"""
if parameters is None:
parameters = {}
force_reload = parameters["force_reload"] if "force_reload" in parameters else False
if force_reload:
return None
handler = None
file_path = self.logs_correspondence[log_name]
is_parquet = False
extension = "xes"
if file_path.endswith(".parquet") or file_path.endswith(".csv"):
is_parquet = True
extension = "parquet"
temp_log_path = os.path.join(Configuration.temp_logs_path, str(log_name)+"_"+str(session)+"."+extension)
if os.path.exists(temp_log_path):
if is_parquet:
handler = ParquetHandler(is_lazy=True)
handler.build_from_path(temp_log_path)
else:
handler = XesHandler()
handler.build_from_path(temp_log_path)
return handler
def process_quantities_test_lazy(path):
handler = ParquetHandler(is_lazy=True)
handler.build_from_path(path)
handler.get_start_activities()
handler.get_end_activities()
handler.get_variant_statistics()
cases = handler.get_case_statistics()
case_id_0 = cases[0]['caseId']
handler.get_variant_statistics()
handler.get_paths("concept:name")
handler.get_attribute_values("concept:name")
handler.get_events(case_id_0)
def test_ru_filtering(self):
handler = ParquetHandler(is_lazy=True)
handler.build_from_path("files/event_logs/running-example.parquet")
handler = handler.add_filter(
['timestamp_trace_intersecting', '1293703320@@@1294667760'],
['timestamp_trace_intersecting', '1293703320@@@1294667760'])
handler.get_start_activities()
handler.get_end_activities()
handler.get_variant_statistics()
handler = handler.add_filter(
['timestamp_trace_containing', '1293703320@@@1294667760'],
[['timestamp_trace_intersecting', '1293703320@@@1294667760'],
['timestamp_trace_containing', '1293703320@@@1294667760']])
handler.get_start_activities()
handler.get_end_activities()
handler.get_variant_statistics()
handler = handler.remove_filter(
['timestamp_trace_containing', '1293703320@@@1294667760'],
[['timestamp_trace_intersecting', '1293703320@@@1294667760']])
handler.get_start_activities()
handler.get_end_activities()
handler.get_variant_statistics()
handler = handler.remove_filter(
['timestamp_trace_intersecting', '1293703320@@@1294667760'], [])
handler.get_start_activities()
handler.get_end_activities()
handler.get_variant_statistics()
handler = handler.add_filter(['end_activities', ['reject request']],
[['end_activities', ['reject request']]])
handler.get_start_activities()
handler.get_end_activities()
handler.get_variant_statistics()
handler = handler.add_filter(
['attributes_pos_trace', ['concept:name', ['check ticket']]],
[['end_activities', ['reject request']],
['attributes_pos_trace', ['concept:name', ['check ticket']]]])
handler.get_start_activities()
handler.get_end_activities()
handler.get_variant_statistics()
handler = handler.remove_filter(
['attributes_pos_trace', ['concept:name', ['check ticket']]],
[['end_activities', ['reject request']]])
handler = handler.remove_filter(['end_activities', ['reject request']],
[])
handler.get_start_activities()
handler.get_end_activities()
handler.get_variant_statistics()
def test_parquets(self):
handler = ParquetHandler()
handler.build_from_path(os.path.join("..", "running-example.parquet"))
handler.get_schema(variant="dfg_freq")
handler.get_schema(variant="dfg_perf")
handler.get_schema(variant="alpha_freq")
handler.get_schema(variant="alpha_perf")
handler.get_schema(variant="inductive_freq")
handler.get_schema(variant="inductive_perf")
handler.get_schema(variant="tree_freq")
handler.get_schema(variant="tree_perf")
handler.get_case_duration_svg()
handler.get_events_per_time_svg()
handler.get_sna(variant="handover")
handler.get_sna(variant="working_together")
handler.get_sna(variant="subcontracting")
handler.get_sna(variant="jointactivities")
handler.get_transient(86400)
def basic_test(path):
handler = ParquetHandler()
handler.build_from_path(path)
handler.get_schema(variant="dfg_freq")
handler.get_schema(variant="dfg_perf")
handler.get_schema(variant="inductive_freq")
handler.get_schema(variant="inductive_perf")
handler.get_schema(variant="heuristics_freq")
handler.get_schema(variant="heuristics_perf")
handler.get_schema(variant="tree_freq")
handler.get_schema(variant="tree_perf")
handler.get_case_duration_svg()
handler.get_events_per_time_svg()
handler.get_sna(variant="handover")
handler.get_sna(variant="working_together")
handler.get_sna(variant="subcontracting")
handler.get_sna(variant="jointactivities")
handler.get_transient(86400)
def basic_test_lazy(path):
handler = ParquetHandler(is_lazy=True)
handler.build_from_path(path)
handler.get_schema(variant="dfg_freq")
handler.get_schema(variant="dfg_perf")
handler.get_schema(variant="inductive_freq")
handler.get_schema(variant="inductive_perf")
handler.get_schema(variant="heuristics_freq")
handler.get_schema(variant="heuristics_perf")
handler.get_schema(variant="tree_freq")
handler.get_schema(variant="tree_perf")
handler.get_case_duration_svg()
handler.get_events_per_time_svg()
handler.get_sna(variant="handover")
handler.get_sna(variant="working_together")
handler.get_sna(variant="subcontracting")
handler.get_sna(variant="jointactivities")
handler.get_transient(86400)
handler.get_events_for_dotted(
["time:timestamp", "@@case_index", "concept:name"])
def load_log_on_request(self, log_name):
"""
Loads an event log on request
Parameters
------------
log_path
Log name
"""
handler = None
file_path = self.logs_correspondence[log_name]
if file_path.endswith(".parquet"):
handler = ParquetHandler(is_lazy=True)
handler.build_from_path(file_path)
elif file_path.endswith(".csv"):
handler = ParquetHandler(is_lazy=True)
handler.build_from_csv(file_path)
elif file_path.endswith(".xes") or file_path.endswith(".xes.gz"):
handler = XesHandler()
handler.build_from_path(file_path)
elif file_path.endswith("xml"):
handler = XmlHandler()
handler.build_from_path(file_path)
return handler