annotate data_manager/vsnp_genbank_fetcher.py @ 1:98babcb65125 draft default tip

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
author iuc
date Sun, 22 Nov 2020 12:55:34 +0000
parents efed56c325da
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
1 import argparse
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
2 import json
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
3 import os
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
4 import sys
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
5 try:
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
6 # For Python 3.0 and later
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
7 from urllib.request import Request, urlopen
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
8 except ImportError:
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
9 # Fall back to Python 2 imports
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
10 from urllib2 import Request, urlopen
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
11
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
12
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
13 def url_download(url, workdir):
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
14 file_path = os.path.abspath(os.path.join(workdir, os.path.basename(url)))
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
15 src = None
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
16 dst = None
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
17 try:
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
18 req = Request(url)
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
19 src = urlopen(req)
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
20 with open(file_path, 'wb') as dst:
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
21 while True:
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
22 chunk = src.read(2**10)
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
23 if chunk:
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
24 dst.write(chunk)
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
25 else:
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
26 break
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
27 except Exception as e:
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
28 sys.exit(str(e))
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
29 finally:
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
30 if src:
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
31 src.close()
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
32 return file_path
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
33
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
34
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
35 def download(dbkey, name, url, out_file):
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
36
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
37 with open(out_file) as fh:
1
98babcb65125 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents: 0
diff changeset
38 params = json.load(fh)
0
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
39
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
40 workdir = params['output_data'][0]['extra_files_path']
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
41 os.makedirs(workdir)
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
42 file_path = url_download(url, workdir)
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
43 entry_name = os.path.basename(file_path)
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
44
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
45 data_manager_json = {"data_tables": {}}
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
46 data_manager_entry = {}
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
47 data_manager_entry['value'] = dbkey
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
48 data_manager_entry['name'] = entry_name
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
49 data_manager_entry['path'] = file_path
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
50 data_manager_entry['description'] = "Genbank file for %s" % name
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
51 data_manager_json["data_tables"]["vsnp_genbank"] = data_manager_entry
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
52
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
53 with open(out_file, 'w') as fh:
1
98babcb65125 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents: 0
diff changeset
54 json.dump(data_manager_json, fh, sort_keys=True)
0
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
55
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
56
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
57 parser = argparse.ArgumentParser()
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
58
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
59 parser.add_argument('--dbkey', dest='dbkey', help='Genome reference dbkey')
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
60 parser.add_argument('--name', dest='name', help='Reference display name')
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
61 parser.add_argument('--url', dest='url', help='URL to download Genbank file')
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
62 parser.add_argument('--out_file', dest='out_file', help='JSON output file')
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
63
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
64 args = parser.parse_args()
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
65
efed56c325da "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_vsnp_genbank commit 31f4453e44da62788efd01736ed57ad483383b09"
iuc
parents:
diff changeset
66 download(args.dbkey, args.name, args.url, args.out_file)