Update withlogging.py

withlogging.py

@@ -12,37 +12,49 @@ def date_dict(df_list, pif_key):
     else:
         return ''
 
-def report_date_check(dict_list, df_list, pif_keys_with_report_date, pif_keys_without_report_date):
-    col_names = {col['attribute_name'] for col in dict_list}
-    pif_key = next((col['attribute_prediction'] for col in dict_list if col['attribute_name'] == 'pif_key'), None)
-    if pif_key is not None:
-        if 'report_date' in col_names:
-            pif_keys_with_report_date.add(pif_key)
-        else:
-            pif_keys_without_report_date.add(pif_key)
-            date_insert_dict = date_dict(df_list, pif_key)
-            if date_insert_dict:
-                dict_list.append(date_insert_dict)
-    return dict_list, pif_keys_with_report_date, pif_keys_without_report_date
-
-def json_report_date_insertion(json_data, df_list):
-    pif_keys = set()
+def report_date_check(dict_list, df_list, logging_df):
     pif_keys_with_report_date = set()
     pif_keys_without_report_date = set()
 
+    for col in dict_list:
+        pif_key = col.get('attribute_prediction', None)
+        if pif_key is not None:
+            if col['attribute_name'] == 'report_date':
+                pif_keys_with_report_date.add(pif_key)
+            else:
+                pif_keys_without_report_date.add(pif_key)
+                latest_date = date_dict(df_list, pif_key)
+                if latest_date:
+                    logging_df = logging_df.append({'pif_key': pif_key,
+                                                    'report_date_exists': False,
+                                                    'report_date_missing': True,
+                                                    'encounter_dates': None,
+                                                    'latest_date': latest_date}, ignore_index=True)
+                else:
+                    logging_df = logging_df.append({'pif_key': pif_key,
+                                                    'report_date_exists': False,
+                                                    'report_date_missing': True,
+                                                    'encounter_dates': None,
+                                                    'latest_date': ''}, ignore_index=True)
+
+    for pif_key in pif_keys_with_report_date:
+        logging_df = logging_df.append({'pif_key': pif_key,
+                                        'report_date_exists': True,
+                                        'report_date_missing': False,
+                                        'encounter_dates': None,
+                                        'latest_date': ''}, ignore_index=True)
+
+    return logging_df
+
+def json_report_date_insertion(json_data, df_list):
+    logging_df = pd.DataFrame(columns=['pif_key', 'report_date_exists', 'report_date_missing', 'encounter_dates', 'latest_date'])
+
     for biomarker_detail in json_data['patient_level']['biomarkers']['details']:
         for attribute in biomarker_detail['attribute']:
             attribute_details = attribute['attribute_details']
-            attribute_details, pif_keys_with_report_date, pif_keys_without_report_date = report_date_check(
-                attribute_details, df_list, pif_keys_with_report_date, pif_keys_without_report_date)
-            pif_keys.add(attribute_details['pif_key'])
-    
-    # Logging
-    print("1. No of unique pif_key present in a json:", len(pif_keys))
-    print("2. Pif_keys with report_date in json:", pif_keys_with_report_date)
-    print("3. Pif_keys without report_date in json:", pif_keys_without_report_date)
+            logging_df = report_date_check(attribute_details, df_list, logging_df)
 
-    return json_data
+    return logging_df
 
 # Usage
 # Load dataframes df2022, df2023, df2024
@@ -52,4 +64,5 @@ def json_report_date_insertion(json_data, df_list):
 
 # json_data = {}  # Load JSON data
 
-# json_report_date_insertion(json_data, [df2022, df2023, df2024])
+# logging_df = json_report_date_insertion(json_data, [df2022, df2023, df2024])
+# print(logging_df)