comparison data_manager/gtdbtk_database_installer.py @ 0:629464b96c2e draft

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gtdbtk_database_installer commit 69da0d34ea6e72c0427ec2d3e882c86cbb90833f
author iuc
date Thu, 08 Dec 2022 19:26:49 +0000
parents
children 6ab422fba1a3
comparison
equal deleted inserted replaced
-1:000000000000 0:629464b96c2e
1 #!/usr/bin/env python
2
3 import argparse
4 import json
5 import os
6 import shutil
7 import sys
8 import tarfile
9 from urllib.parse import urlparse
10 from urllib.request import Request
11 from urllib.request import urlopen
12
13
14 def url_download(url, target_directory):
15 url_parts = urlparse(url)
16 tarball = os.path.abspath(os.path.join(target_directory, os.path.basename(url_parts.path)))
17 src = None
18 dst = None
19 try:
20 req = Request(url)
21 src = urlopen(req)
22 with open(tarball, 'wb') as dst:
23 while True:
24 chunk = src.read(2**10)
25 if chunk:
26 dst.write(chunk)
27 else:
28 break
29 except Exception as e:
30 sys.exit(str(e))
31 finally:
32 if src is not None:
33 src.close()
34 if tarfile.is_tarfile(tarball):
35 fh = tarfile.open(tarball, 'r:*')
36 else:
37 return tarball
38 fh.extractall(target_directory)
39 fh.close()
40 os.remove(tarball)
41 # The tarball extraction will create a directory named
42 # something like release202 in the target_directory, so
43 # we need to move the items in that directory to the
44 # target directory.
45 subdir = next(os.walk(target_directory))[1][0]
46 subdir_path = os.path.join(target_directory, subdir)
47 items = os.listdir(subdir_path)
48 for item in items:
49 item_path = os.path.join(subdir_path, item)
50 shutil.move(item_path, target_directory)
51 os.rmdir(subdir_path)
52 return target_directory
53
54
55 def download(database_id, database_name, url, out_file):
56
57 with open(out_file) as fh:
58 params = json.load(fh)
59
60 target_directory = params['output_data'][0]['extra_files_path']
61 os.makedirs(target_directory)
62 file_path = url_download(url, target_directory)
63
64 data_manager_json = {"data_tables": {}}
65 data_manager_entry = {}
66 data_manager_entry['value'] = database_id
67 data_manager_entry['name'] = database_name
68 data_manager_entry['path'] = file_path
69 data_manager_json["data_tables"]["gtdbtk_database"] = data_manager_entry
70
71 with open(out_file, 'w') as fh:
72 json.dump(data_manager_json, fh, sort_keys=True)
73
74
75 parser = argparse.ArgumentParser()
76
77 parser.add_argument('--database_name', dest='database_name', help='GTDB-Tk database display name')
78 parser.add_argument('--database_id', dest='database_id', help='Unique GTDB-Tk database id')
79 parser.add_argument('--url', dest='url', help='URL to download GTDB-Tk databse version')
80 parser.add_argument('--out_file', dest='out_file', help='JSON output file')
81
82 args = parser.parse_args()
83
84 download(args.database_id, args.database_name, args.url, args.out_file)