format check for the uploaded event log

backend/eventlog.py

@@ -95,6 +95,36 @@ def convert_to_dataframe(name):
     return event_log_df
 
 
+def format_check(name):
+    event_log = convert_to_dataframe(name)
+
+    allowed_columns = {'CaseID', 'Activity', 'StartTimestamp', 'EndTimestamp'}
+    allowed_activities = {'place order', 'arrange standard order', 'arrange custom order', 'pick from stock A', 'pick from stock B', 'pick from stock C', 'manufacture A', 'manufacture B', 'pack A', 'pack B', 'pack C', 'attempt delivery A', 'attempt delivery B', 'attempt delivery C', 'order completed'}
+
+    for column_name in event_log.columns:
+        if column_name not in allowed_columns:
+            return False
+
+    for index, event in event_log.iterrows():
+        case_id = event['CaseID']
+        if not isinstance(case_id, int):
+            return False
+
+        activity = event['Activity']
+        if activity not in allowed_activities:
+            return False
+        
+        start = event['StartTimestamp']
+        if not isinstance(start, float):
+            return False
+        
+        end = event['EndTimestamp']
+        if not isinstance(end, float):
+            return False
+        
+    return True
+
+
 def get_active_cases(name):
     event_log_df = convert_to_dataframe(name)
     
@@ -280,7 +310,8 @@ def show_active_cases(name):
 
 def main():
     # generate_event_log(10000)
-    print(get_state(5, "eventlog.csv"))
+    # print(get_state(5, "eventlog.xes"))
+    print(format_check("eventlog.xes"))
 
 
 if __name__ == "__main__":