Revize 753d424e

Predelany komentáře upravena kontrola datasetu

modules/crawler/reset_datasets.py

@@ -13 +13 @@
 CONFIG_FILES_PATH = "DatasetConfigs"
 
 
-def reset_dataset(dataset_name):
+def hard_reset_dataset(dataset_name):
     """
     Resets all saved data in dataset except config and implementation
     Args:
@@ -29 +29 @@
     database_loader.remove_dataset_database(dataset_name)
 
 
-def reset_all_datasets():
+def soft_reset_dataset(dataset_name):
+    """
+    Resets all saved data in dataset except config and implementation
+    Args:
+        dataset_name: name of dataset that has existing configuration file
+    """
+    path = PROCESSED_DATA_PATH + dataset_name + "/"
+    folder_processor.clean_folder(path)
+
+    database_loader.remove_dataset_database(dataset_name)
+
+
+
+def soft_reset_all_datasets():
     """
     Resets all saved data in all datasets with config file except configs and implementation
     """
     datasets = os.listdir(CONFIG_FILES_PATH)
 
     for dataset in datasets:
-        reset_dataset(dataset.split('.')[0])
+        soft_reset_dataset(dataset.split('.')[0])
 
 
+def hard_reset_all_datasets():
+    """
+    Resets all saved data in all datasets with config file except configs and implementation
+    """
+    datasets = os.listdir(CONFIG_FILES_PATH)
+
+    for dataset in datasets:
+        hard_reset_dataset(dataset.split('.')[0])
+
 
 print("Zadejte jméno Datasetu který chcete resetovat (pokud všechny zadejte '-ALL'):\n")
 
 dataset_name = input().upper()
 
+print("Chcete smazat i stažené stažená data ? (ANO/NE) \n")
+
+vstup = input().upper()
+
 if dataset_name == '-ALL':
-    reset_all_datasets()
+    if vstup == 'ANO':
+        hard_reset_all_datasets()
+    elif vstup == 'NE':
+        soft_reset_all_datasets()
+    else:
+        print('Neplatný vstup (ANO/NE)')
 else:
     test = configure_functions.check_if_there_is_a_config_file(dataset_name)
     if test == True:
-        reset_dataset(dataset_name)
+        if vstup == 'ANO':
+            hard_reset_dataset(dataset_name)
+        elif vstup == 'NE':
+            soft_reset_dataset(dataset_name)
+        else:
+            print('Neplatný vstup (ANO/NE)')
     else:
         print("Tento dataset v architektuře neexistuje")
+