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 default="gene_annotation.json"
|
|
16 )
|
|
17 parser.add_argument('--name',
|
|
18 dest='name',
|
|
19 action='store',
|
|
20 default=uuid.uuid4(),
|
|
21 help='Data table entry unique ID'
|
|
22 )
|
|
23 parser.add_argument('--url',
|
|
24 dest='url',
|
|
25 action='store',
|
|
26 help='Download URL',
|
|
27 default="http://www.scott-ouellette.com/gene_annotations/chr1-hg19_genes.gtf")
|
0
|
28
|
|
29 args = parser.parse_args()
|
|
30
|
7
|
31
|
|
32 def url_download(url, name, workdir):
|
|
33 # Create path if it doesn't exist
|
0
|
34 if not os.path.exists(workdir):
|
|
35 os.makedirs(workdir)
|
7
|
36
|
|
37 response = requests.get(url=url, stream=True)
|
|
38
|
|
39 # Create path that we will write the file to
|
|
40 file_path = os.path.join(workdir, 'download_{}.dat'.format(name))
|
|
41
|
|
42 block_size = 10 * 1024 * 1024 # 10MB chunked download
|
|
43 with open(file_path, 'w+') as f:
|
|
44 try:
|
|
45 for buf in response.iter_content(block_size):
|
|
46 f.write(buf)
|
|
47 except (ContentDecodingError, IOError) as e:
|
|
48 sys.stderr.write("Error occured downloading reference file: %s"
|
|
49 % e)
|
|
50 os.remove(file_path)
|
|
51
|
|
52 return file_path
|
0
|
53
|
|
54
|
|
55 def main(args):
|
|
56 workdir = os.path.join(os.getcwd(), 'gene_annotation')
|
7
|
57
|
|
58 # Attempt to download gene annotation file from given url
|
|
59 gene_annotation_file_path = url_download(args.url, args.name, workdir)
|
|
60
|
|
61 # Update Data Manager Json and write out
|
|
62 data_manager_entry = {
|
|
63 'data_tables': {
|
|
64 'gene_annotation': {
|
|
65 'date': str(datetime.datetime.now()),
|
|
66 'name': str(args.name),
|
|
67 'path': gene_annotation_file_path
|
|
68 }
|
|
69 }
|
|
70 }
|
|
71
|
|
72 with open(os.path.join(workdir, args.output), "w+") as f:
|
|
73 f.write(json.dumps(data_manager_entry))
|
0
|
74
|
|
75 if __name__ == '__main__':
|
7
|
76 main(args)
|