|
1
|
import os
|
|
2
|
import shutil
|
|
3
|
|
|
4
|
# Path to crawled data
|
|
5
|
CRAWLED_DATA_PATH = "../CrawledData/"
|
|
6
|
# Path to processed data
|
|
7
|
PROCESSED_DATA_PATH = "../ProcessedData/"
|
|
8
|
# Path to crawler logs
|
|
9
|
CRAWLER_LOGS_PATH = "../CrawlerLogs/"
|
|
10
|
# Path to dataset configuration files
|
|
11
|
CONFIG_FILES_PATH = "../DatasetConfigs"
|
|
12
|
# Path for DatasetCrawlers implementations
|
|
13
|
CRAWLER_PROGRAM_PATH = "../DatasetCrawler"
|
|
14
|
# Path for DatasetProcessors implementations
|
|
15
|
PROCESSOR_PROGRAM_PATH = "../DatasetProcessing"
|
|
16
|
|
|
17
|
|
|
18
|
def remove_dataset(dataset_name):
|
|
19
|
"""
|
|
20
|
Remove dataset
|
|
21
|
Args:
|
|
22
|
dataset_name: name of dataset that has existing configuration file
|
|
23
|
"""
|
|
24
|
shutil.rmtree(CRAWLED_DATA_PATH + dataset_name + "/")
|
|
25
|
shutil.rmtree(PROCESSED_DATA_PATH + dataset_name + "/")
|
|
26
|
shutil.rmtree(CRAWLER_LOGS_PATH + dataset_name + "/")
|
|
27
|
|
|
28
|
os.remove(CONFIG_FILES_PATH + "/" + dataset_name + ".yaml")
|
|
29
|
os.remove(CRAWLER_PROGRAM_PATH + "/" + dataset_name + "Crawler.py")
|
|
30
|
os.remove(PROCESSOR_PROGRAM_PATH + "/" + dataset_name + "Processor.py")
|
|
31
|
|
|
32
|
print("Dataset: " + dataset_name + " removed");
|
|
33
|
|
|
34
|
remove_dataset("TEST");
|