Update news_category_similar_news_prediction.py

news_category_similar_news_prediction.py

@@ -32,7 +32,7 @@ def date_time_parser(dt):
 def delete_outdated_news(final_df: pd.DataFrame):
     logger.warning("Entering delete_outdated_news()")
     final_df = final_df.copy()
-    final_df["parsed_date"] = final_df["parsed_date"].map(correct_date)
+    # final_df["parsed_date"] = final_df["parsed_date"].map(correct_date)
     final_df["parsed_date"] = final_df["parsed_date"].map(parser.parse)
     final_df["elapsed_time"] = final_df["parsed_date"].apply(date_time_parser)
     final_df = final_df.loc[final_df["elapsed_time"] <= 720, :].copy() # 1440=24 hrs and 720=12 hrs
@@ -94,6 +94,11 @@ def predict_news_category_similar_news(old_news: pd.DataFrame, new_news: pd.Data
             new_news = delete_outdated_news(new_news)
             logger.warning(f'{len(new_news)} new news items found after deleting outdated news')
             logger.warning(f'new news columns: {[*new_news.columns]}')
+            if len(new_news) <= 1:
+                new_news = None
+
+        if not isinstance(new_news, pd.DataFrame):
+            raise Exception('No New News Found after deleting outdated news')
             
         
         if isinstance(old_news, pd.DataFrame):
@@ -105,9 +110,15 @@ def predict_news_category_similar_news(old_news: pd.DataFrame, new_news: pd.Data
             
             old_news.drop(columns='_id', inplace=True)
             logger.warning('Dropped _id column from old news data frame.')
+            if len(old_news) <= 1:
+                old_news = None
         else:
             logger.warning('No old news is found')
             old_news = new_news.copy()
+
+        if not isinstance(old_news, pd.DataFrame):
+            logger.warning('No old news is found after deleting outdate news')
+            old_news = new_news.copy()
       
         if 'category' not in [*old_news.columns]:
             logger.warning('No prior predictions found in old news')