Mercurial > repos > devteam > data_manager_fetch_ncbi_taxonomy

annotate data_manager/data_manager.py @ 0:078e803565cf draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
author devteam
date Fri, 06 Nov 2015 14:16:28 -0500
parents
children 8dd4076813ef
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
1 import argparse
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
2 import datetime
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
3 import json
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
4 import os
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
5 import shutil
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
6 import sys
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
7 import tarfile
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
8 import urllib2
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
9 import zipfile
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
10
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
11 parser = argparse.ArgumentParser(description='Create data manager json.')
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
12 parser.add_argument('--out', dest='output', action='store', help='JSON filename')
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
13 parser.add_argument('--name', dest='name', action='store', default=str(datetime.date.today()), help='Data table entry unique ID')
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
14 parser.add_argument('--url', dest='url', action='store', default='ftp://ftp.ncbi.nih.gov/pub/taxonomy/taxdump.tar.gz', help='Download URL')
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
15
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
16 args = parser.parse_args()
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
17
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
18 def url_download(url, workdir):
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
19 file_path = os.path.join(workdir, 'download.dat')
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
20 if not os.path.exists(workdir):
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
21 os.makedirs(workdir)
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
22 src = None
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
23 dst = None
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
24 try:
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
25 req = urllib2.Request(url)
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
26 src = urllib2.urlopen(req)
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
27 dst = open(file_path, 'wb')
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
28 while True:
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
29 chunk = src.read(2**10)
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
30 if chunk:
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
31 dst.write(chunk)
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
32 else:
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
33 break
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
34 except Exception, e:
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
35 print >>sys.stderr, str(e)
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
36 finally:
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
37 if src:
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
38 src.close()
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
39 if dst:
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
40 dst.close()
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
41 if tarfile.is_tarfile(file_path):
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
42 fh = tarfile.open(file_path, 'r:*')
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
43 elif zipfile.is_zipfile(file_path):
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
44 fh = zipfile.ZipFile(file_path, 'r')
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
45 else:
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
46 return
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
47 fh.extractall(workdir)
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
48 os.remove(file_path)
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
49
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
50
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
51 def main(args):
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
52 workdir = os.path.join(os.getcwd(), 'taxonomy')
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
53 url_download(args.url, workdir)
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
54 data_manager_entry = {}
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
55 data_manager_entry['value'] = args.name.lower()
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
56 data_manager_entry['name'] = args.name
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
57 data_manager_entry['path'] = '.'
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
58 data_manager_json = dict(data_tables=dict(ncbi_taxonomy=data_manager_entry))
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
59 params = json.loads(open(args.output).read())
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
60 target_directory = params['output_data'][0]['extra_files_path']
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
61 os.mkdir(target_directory)
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
62 output_path = os.path.abspath(os.path.join(os.getcwd(), 'taxonomy'))
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
63 for filename in os.listdir(workdir):
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
64 shutil.move(os.path.join(output_path, filename), target_directory)
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
65 file(args.output, 'w').write(json.dumps(data_manager_json))
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
66
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
67 if __name__ == '__main__':
078e803565cf planemo upload for repository https://github.com/galaxyproject/tools-devteam/blob/master/data_managers/data_manager_fetch_ncbi_taxonomy/ commit 86cf90107482cab1cb47fc0d42d6705f8077daa7
devteam
parents:
diff changeset
68 main(args)