|
7
|
1 import os
|
|
|
2 import sys
|
|
|
3 import uuid
|
|
|
4 import json
|
|
0
|
5 import argparse
|
|
|
6 import datetime
|
|
7
|
7 import requests
|
|
|
8 from requests.exceptions import ContentDecodingError
|
|
0
|
9
|
|
|
10 parser = argparse.ArgumentParser(description='Create data manager json.')
|
|
7
|
11 parser.add_argument('--out',
|
|
|
12 dest='output',
|
|
|
13 action='store',
|
|
|
14 help='JSON filename',
|
|
|
15 )
|
|
|
16 parser.add_argument('--name',
|
|
|
17 dest='name',
|
|
|
18 action='store',
|
|
|
19 default=uuid.uuid4(),
|
|
|
20 help='Data table entry unique ID'
|
|
|
21 )
|
|
|
22 parser.add_argument('--url',
|
|
|
23 dest='url',
|
|
|
24 action='store',
|
|
8
|
25 help='Download URL'
|
|
|
26 )
|
|
0
|
27
|
|
|
28 args = parser.parse_args()
|
|
|
29
|
|
7
|
30
|
|
19
|
31 def url_download(url, name):
|
|
7
|
32 response = requests.get(url=url, stream=True)
|
|
|
33
|
|
22
|
34 # Generate file_name
|
|
23
|
35 file_name = response.url.split("/")[-1]
|
|
7
|
36
|
|
|
37 block_size = 10 * 1024 * 1024 # 10MB chunked download
|
|
23
|
38 with open(file_name, 'w+') as f:
|
|
7
|
39 try:
|
|
24
|
40 # Good to note here that requests' iter_content() will
|
|
|
41 # automatically handle decoding "gzip" and "deflate" formats
|
|
7
|
42 for buf in response.iter_content(block_size):
|
|
|
43 f.write(buf)
|
|
|
44 except (ContentDecodingError, IOError) as e:
|
|
|
45 sys.stderr.write("Error occured downloading reference file: %s"
|
|
|
46 % e)
|
|
23
|
47 os.remove(file_name)
|
|
7
|
48
|
|
24
|
49 return os.path.join(os.getcwd(), file_name)
|
|
0
|
50
|
|
|
51
|
|
|
52 def main(args):
|
|
7
|
53
|
|
|
54 # Attempt to download gene annotation file from given url
|
|
24
|
55 gene_annotation_file_path = url_download(
|
|
23
|
56 "http://www.scott-ouellette.com/gene_annotations/chr1-hg19_genes.gtf", args.name)
|
|
7
|
57
|
|
22
|
58 # Update Data Manager JSON and write to file
|
|
7
|
59 data_manager_entry = {
|
|
|
60 'data_tables': {
|
|
|
61 'gene_annotation': {
|
|
22
|
62 'value': str(datetime.datetime.now()),
|
|
10
|
63 'dbkey': str(args.name),
|
|
25
|
64 'name': gene_annotation_file_path.split(".")[0]
|
|
7
|
65 }
|
|
|
66 }
|
|
|
67 }
|
|
|
68
|
|
19
|
69 with open(os.path.join(args.output), "w+") as f:
|
|
7
|
70 f.write(json.dumps(data_manager_entry))
|
|
0
|
71
|
|
|
72 if __name__ == '__main__':
|
|
7
|
73 main(args)
|
