annotate data_manager/gtdbtk_database_installer.py @ 0:3ab83cb7e2d2 draft

Uploaded
author greg
date Tue, 15 Mar 2022 15:32:31 +0000
parents
children 7093598fa300
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
1 #!/usr/bin/env python
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
2
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
3 import argparse
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
4 import json
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
5 import os
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
6 import sys
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
7 import tarfile
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
8 from urllib.request import Request, urlopen
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
9 from urllib.parse import urlparse
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
10
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
11
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
12 def url_download(url, work_dir):
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
13 url_parts = urlparse(url)
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
14 file_path = os.path.abspath(os.path.join(work_dir, os.path.basename(url_parts.path)))
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
15 src = None
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
16 dst = None
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
17 try:
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
18 req = Request(url)
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
19 src = urlopen(req)
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
20 with open(file_path, 'wb') as dst:
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
21 while True:
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
22 chunk = src.read(2**10)
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
23 if chunk:
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
24 dst.write(chunk)
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
25 else:
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
26 break
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
27 except Exception as e:
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
28 sys.exit(str(e))
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
29 finally:
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
30 if src:
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
31 src.close()
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
32 if tarfile.is_tarfile(file_path):
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
33 fh = tarfile.open(file_path, 'r:*')
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
34 else:
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
35 return file_path
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
36 fh.extractall(work_dir)
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
37 os.remove(file_path)
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
38 return work_dir
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
39
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
40
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
41 def download(database_id, database_name, url, out_file):
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
42
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
43 with open(out_file) as fh:
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
44 params = json.load(fh)
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
45
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
46 work_dir = params['output_data'][0]['extra_files_path']
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
47 os.makedirs(work_dir)
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
48 file_path = url_download(url, work_dir)
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
49
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
50 data_manager_json = {"data_tables": {}}
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
51 data_manager_entry = {}
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
52 data_manager_entry['value'] = database_id
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
53 data_manager_entry['name'] = database_name
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
54 data_manager_entry['path'] = file_path
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
55 data_manager_json["data_tables"]["gtdbtk_database"] = data_manager_entry
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
56
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
57 with open(out_file, 'w') as fh:
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
58 json.dump(data_manager_json, fh, sort_keys=True)
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
59
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
60
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
61 parser = argparse.ArgumentParser()
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
62
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
63 parser.add_argument('--database_name', dest='database_name', help='GTDB-Tk database display name')
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
64 parser.add_argument('--database_id', dest='database_id', help='Unique GTDB-Tk database id')
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
65 parser.add_argument('--url', dest='url', help='URL to download GTDB-Tk databse version')
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
66 parser.add_argument('--out_file', dest='out_file', help='JSON output file')
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
67
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
68 args = parser.parse_args()
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
69
3ab83cb7e2d2 Uploaded
greg
parents:
diff changeset
70 download(args.database_id, args.database_name, args.url, args.out_file)