comparison data_manager/vsnp_excel_fetcher.py @ 0:84a356e0950a draft

Uploaded
author greg
date Fri, 07 Feb 2020 13:02:25 -0500
parents
children d1bf63483e33
comparison
equal deleted inserted replaced
-1:000000000000 0:84a356e0950a
1 import argparse
2 import json
3 import os
4 try:
5 # For Python 3.0 and later
6 from urllib.request import Request, urlopen
7 except ImportError:
8 # Fall back to Python 2 imports
9 from urllib2 import Request, urlopen
10
11
12 def url_download(url, target_directory):
13 file_path = os.path.abspath(os.path.join(target_directory, os.path.basename(url)))
14 src = None
15 dst = None
16 try:
17 req = Request(url)
18 src = urlopen(req)
19 with open(file_path, 'wb') as dst:
20 while True:
21 chunk = src.read(2**10)
22 if chunk:
23 dst.write(chunk)
24 else:
25 break
26 finally:
27 if src:
28 src.close()
29 return file_path
30
31
32 def download(dbkey, display_name, url, out_file):
33
34 with open(out_file) as fh:
35 params = json.loads(fh.read())
36
37 work_directory = params['output_data'][0]['extra_files_path']
38 target_directory = os.path.abspath(os.path.join(work_directory, "vsnp", dbkey))
39 os.mkdir(target_directory)
40 file_path = url_download(url, target_directory)
41 entry_name = "%s" % os.path.basename(file_path)
42
43 data_manager_json = {"data_tables": {}}
44 data_manager_entry = {}
45 data_manager_entry['value'] = dbkey
46 data_manager_entry['name'] = entry_name
47 data_manager_entry['path'] = file_path
48 data_manager_entry['description'] = "Excel file for %s" % display_name
49 data_manager_json["data_tables"]["vsnp_excel"] = data_manager_entry
50
51 with open(out_file, 'w') as fh:
52 fh.write(json.dumps(data_manager_json, sort_keys=True))
53
54
55 parser = argparse.ArgumentParser()
56
57 parser.add_argument('--dbkey', dest='dbkey', help='Genome reference dbkey')
58 parser.add_argument('--display_name', dest='display_name', help='Reference display name')
59 parser.add_argument('--url', dest='url', help='URL to download Excel file')
60 parser.add_argument('--out_file', dest='out_file', help='JSON output file')
61
62 args = parser.parse_args()
63
64 download(args.dbkey, args.display_name, args.url, args.out_file)