1 |
d6ca840d
|
petrh
|
import os
|
2 |
|
|
from Utilities import folder_processor
|
3 |
|
|
from Utilities.Database import database_loader
|
4 |
|
|
from Utilities import configure_functions
|
5 |
|
|
|
6 |
|
|
# Path to crawled data
|
7 |
|
|
CRAWLED_DATA_PATH = "CrawledData/"
|
8 |
|
|
# Path to processed data
|
9 |
|
|
PROCESSED_DATA_PATH = "ProcessedData/"
|
10 |
|
|
# Path to crawler logs
|
11 |
|
|
CRAWLER_LOGS_PATH = "CrawlerLogs/"
|
12 |
|
|
# Path to dataset configuration files
|
13 |
|
|
CONFIG_FILES_PATH = "DatasetConfigs"
|
14 |
|
|
|
15 |
|
|
|
16 |
|
|
def reset_dataset(dataset_name):
|
17 |
|
|
"""
|
18 |
|
|
Resets all saved data in dataset except config and implementation
|
19 |
|
|
Args:
|
20 |
|
|
dataset_name: name of dataset that has existing configuration file
|
21 |
|
|
"""
|
22 |
|
|
|
23 |
|
|
path = CRAWLED_DATA_PATH + dataset_name + "/"
|
24 |
|
|
folder_processor.clean_folder(path)
|
25 |
|
|
|
26 |
|
|
path = PROCESSED_DATA_PATH + dataset_name + "/"
|
27 |
|
|
folder_processor.clean_folder(path)
|
28 |
|
|
|
29 |
|
|
database_loader.remove_dataset_database(dataset_name)
|
30 |
|
|
|
31 |
|
|
|
32 |
|
|
def reset_all_datasets():
|
33 |
|
|
"""
|
34 |
|
|
Resets all saved data in all datasets with config file except configs and implementation
|
35 |
|
|
"""
|
36 |
|
|
datasets = os.listdir(CONFIG_FILES_PATH)
|
37 |
|
|
|
38 |
|
|
for dataset in datasets:
|
39 |
|
|
reset_dataset(dataset.split('.')[0])
|
40 |
|
|
|
41 |
|
|
|
42 |
|
|
|
43 |
|
|
print("Zadejte jméno Datasetu který chcete resetovat (pokud všechny zadejte '-ALL'):\n")
|
44 |
|
|
|
45 |
|
|
dataset_name = input().upper()
|
46 |
|
|
|
47 |
|
|
if dataset_name == '-ALL':
|
48 |
|
|
reset_all_datasets()
|
49 |
|
|
else:
|
50 |
|
|
test = configure_functions.check_if_there_is_a_config_file(dataset_name)
|
51 |
|
|
if test == True:
|
52 |
|
|
reset_dataset(dataset_name)
|
53 |
|
|
else:
|
54 |
|
|
print("Tento dataset v architektuře neexistuje")
|