Merge pull request #14 from rwth-acis/release/0.1.10

Add deserialize_remarks parameter to generateXESfile and generate_eventlog functions

app.py

@@ -177,7 +177,7 @@ def send_xml_file_for_bot(botName):
     app.run(port=port, debug=True)
 
 
-def generateXESfile(db_connection, start_date=None, end_date=None, resource_ids=None, include_bot_messages=False, include_life_cycle_start=False):
+def generateXESfile(db_connection, start_date=None, end_date=None, resource_ids=None, include_bot_messages=False, include_life_cycle_start=False,deserialize_remarks=False):
     """
     This function generates an XES file from the events in the database
     
@@ -205,7 +205,7 @@ def generateXESfile(db_connection, start_date=None, end_date=None, resource_ids=
     if not os.path.exists(os.path.join(current_dir, 'event_logs')):
         os.makedirs(os.path.join(current_dir, 'event_logs'))
     event_log = generate_eventlog(db_connection, start_date, end_date, resource_ids,
-                                  include_bot_messages=include_bot_messages, include_life_cycle_start=include_life_cycle_start)
+                                  include_bot_messages=include_bot_messages, include_life_cycle_start=include_life_cycle_start,deserialize_remarks=deserialize_remarks)
     if event_log is None or event_log.empty:
         logger.info('No events found for resource ids: '+str(resource_ids))
         return None

event_log_generator/event_reader.py

@@ -4,7 +4,7 @@
 # pip install dogpile.cache for caching the sql results
 
 
-def generate_eventlog(db_connection, start_date=None, end_date=None, resource_ids=None, include_bot_messages=False, include_life_cycle_start=False):
+def generate_eventlog(db_connection, start_date=None, end_date=None, resource_ids=None, include_bot_messages=False, include_life_cycle_start=False, deserialize_remarks=False):
     df = _read_events_into_df(db_connection, start_date, end_date, resource_ids,
                               include_bot_messages=include_bot_messages, include_life_cycle_start=include_life_cycle_start)
 
@@ -29,7 +29,8 @@ def generate_eventlog(db_connection, start_date=None, end_date=None, resource_id
         end_date = df['time:timestamp'].max().strftime('%Y-%m-%d')
 
     # extract fields from remarks column
-    df = df.apply(_extract_remarks, axis=1)
+    if deserialize_remarks:
+        df = df.apply(_extract_remarks, axis=1)
     if ('lifecycle:transition' in df.columns):
         df.loc[:, ['lifecycle:transition']] = df[[
             'lifecycle:transition']].fillna('complete')
@@ -69,6 +70,7 @@ def _read_events_into_df(db_connection, start_date=None, end_date=None, resource
 
     df = pd.read_sql(statement, con=db_connection,
                      params=params)
+    print(f'Read {len(df)} events from database')
     df.rename(columns={'CASE_ID': 'case:concept:name', 'ACTIVITY_NAME': 'concept:name',
               'TIME_STAMP': 'time:timestamp', 'LIFECYCLE_PHASE': 'lifecycle:transition'}, inplace=True)
     # convert time:timestamp to datetime

event_log_generator/setup.py

@@ -2,7 +2,7 @@
 
 setup(
     name='event_reader',
-    version='0.1.9',
+    version='0.1.10',
     description='A package to read events from a database using SQLAlchemy and Pandas',
     author='Ben Lakhoune',
     author_email='[email protected]',