Update news_category_similar_news_prediction.py

news_category_similar_news_prediction.py

@@ -15,6 +15,7 @@ logger = get_logger()
 from dateutil import parser
 def correct_date(x):
     if (not isinstance(x, str)) or (str(x).find(":") == -1):
+        logger.warning(f'correct_date() error: {x} is not the right date format')
         return "2020-11-07 00:36:44+05:30"
     return x
 
@@ -27,16 +28,17 @@ def date_time_parser(dt):
     try:
         return int(np.round((dt.now(dt.tz) - dt).total_seconds() / 60, 0))
     except:
+        logger.warning(f'date_time_parser() error: {dt} is not the right date format')
         return 100000
 
 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(parser.parse)
-    final_df["elapsed_time"] = final_df["parsed_date"].apply(date_time_parser)
+    final_df["parsed_date_1"] = final_df["parsed_date"].map(correct_date)
+    final_df["parsed_date_1"] = final_df["parsed_date_1"].map(parser.parse)
+    final_df["elapsed_time"] = final_df["parsed_date_1"].apply(date_time_parser)
     final_df = final_df.loc[final_df["elapsed_time"] <= 720, :].copy() # 1440=24 hrs and 720=12 hrs
-    final_df.drop(columns='elapsed_time', inplace=True)
+    final_df.drop(columns=['elapsed_time', 'parsed_date_1'], inplace=True)
     final_df.reset_index(drop=True, inplace=True)
     logger.warning("Exiting delete_outdated_news()")
     return final_df