annotate data_manager/bakta_build_database.py @ 18:04bee0f935a2 draft

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
author pimarin
date Wed, 09 Nov 2022 15:49:56 +0000
parents ecb7d3d41366
children c90380f8bbbc
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
1 import argparse
0
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
2 import hashlib
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
3 import json
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
4 import os
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
5 import sys
18
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
6 # import subprocess
0
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
7 import tarfile
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
8 from datetime import datetime
0
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
9 from pathlib import Path
17
ecb7d3d41366 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa-dirty
pimarin
parents: 16
diff changeset
10 # implement pip as a subprocess:
18
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
11 # subprocess.check_call([sys.executable, '-m', 'pip', 'install', 'requests'])
17
ecb7d3d41366 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa-dirty
pimarin
parents: 16
diff changeset
12
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
13 import requests
16
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
14
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
15
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
16 class GetBaktaDatabaseInfo:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
17 """
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
18 Extract bakta database information to make a json file for data_manager
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
19 """
0
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
20
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
21 def __init__(self,
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
22 data_table_name="bakta_database",
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
23 db_name=Path.cwd().joinpath("db"),
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
24 db_version="latest"):
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
25 self.bakta_table_list = None
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
26 self.db_url = None
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
27 self.data_table_entry = None
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
28 self.data_table_name = data_table_name
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
29 self.db_name = db_name
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
30 self.db_version = db_version
11
c50e8b4f74d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 572bbceeba9a22cdb591d526abda362595f8f0c4
pimarin
parents: 10
diff changeset
31 self.DB_VERSIONS_URL = 'https://raw.githubusercontent.com/oschwengers/bakta/master/db-versions.json'
0
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
32
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
33 def get_data_table_format(self):
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
34 """
18
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
35 Skeleton of a data_table format
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
36 return: a data table formated for json output
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
37 """
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
38 self.data_table_entry = {
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
39 "data_tables": {
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
40 self.data_table_name: {}
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
41 }
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
42 }
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
43 return self.data_table_entry
0
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
44
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
45 def fetch_db_versions(self, db_version="latest"):
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
46 """
18
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
47 List bakta database info depending of the db_version selected
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
48 """
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
49 try:
11
c50e8b4f74d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 572bbceeba9a22cdb591d526abda362595f8f0c4
pimarin
parents: 10
diff changeset
50 with requests.get(self.DB_VERSIONS_URL) as resp:
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
51 versions = json.loads(resp.content)
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
52 except IOError as e:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
53 print(e, file=sys.stderr)
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
54 raise e
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
55 else:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
56 if db_version == "latest":
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
57 db_date_list = []
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
58 for db_dic in versions:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
59 db_date_list.append(datetime.strptime(db_dic["date"],
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
60 '%Y-%m-%d').date())
18
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
61 filtered_version = max(versions, key=lambda x: x['date'])
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
62 elif db_version == "test":
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
63 filtered_version = {"date": "date_test",
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
64 "major": "0",
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
65 "minor": "0",
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
66 "doi": "10.5281/zenodo.7197299",
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
67 "record": "7197299",
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
68 "md5": "8b0250c17078742fc12207d4efb0fc1a",
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
69 "software-min": {"major": "0",
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
70 "minor": "0"}
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
71 }
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
72 else:
18
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
73 filtered_version = None
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
74 for item in versions:
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
75 if '{0}.{1}'.format(item["major"], item["minor"]) == db_version:
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
76 filtered_version = item
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
77 break
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
78 if filtered_version is None:
04bee0f935a2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 476cb50265f7dcb492fa2d0f03f595b1363faf7a
pimarin
parents: 17
diff changeset
79 print("No matching version detected in the list")
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
80 if filtered_version is not None:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
81 self.db_url = f"https://zenodo.org/record/" \
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
82 f"{filtered_version['record']}/files/db.tar.gz"
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
83 self.db_version = db_version
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
84 return filtered_version
0
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
85
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
86 def get_data_manager(self, bakta_database_info, output_path):
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
87 self.bakta_table_list = self.get_data_table_format()
16
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
88 bakta_value = f"V{bakta_database_info['major']}." \
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
89 f"{bakta_database_info['minor']}_" \
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
90 f"{bakta_database_info['date']}"
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
91 tool_version = str(f"{bakta_database_info['software-min']['major']}."
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
92 f"{bakta_database_info['software-min']['minor']}")
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
93 data_info = dict(value=bakta_database_info['record'],
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
94 dbkey=bakta_value,
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
95 bakta_version=tool_version,
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
96 path="db")
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
97 self.bakta_table_list["data_tables"][self.data_table_name] = [data_info]
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
98 return self.bakta_table_list
0
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
99
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
100
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
101 class InstallBaktaDatabase(GetBaktaDatabaseInfo):
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
102 """
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
103 Download the bakta database,
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
104 check md5 sum,
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
105 untar the download db and update for the amrfinderplus database
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
106 """
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
107
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
108 def __init__(self,
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
109 db_dir=Path.cwd(),
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
110 db_name="bakta",
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
111 tarball_name="db.tar.gz"):
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
112 super().__init__()
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
113 self.md5 = None
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
114 self.db_dir = db_dir
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
115 self.db_name = db_name
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
116 self.tarball_name = tarball_name
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
117 self.tarball_path = None
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
118
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
119 def download(self):
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
120 self.db_name = f'{self.db_name}_{self.db_version}'
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
121 bakta_path = Path(self.db_dir).joinpath(self.tarball_name)
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
122 try:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
123 with bakta_path.open('wb') as fh_out, \
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
124 requests.get(self.db_url, stream=True) as resp:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
125 total_length = resp.headers.get('content-length')
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
126 if total_length is None: # no content length header
15
ac4d5e1d1421 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit b082e89a0090f9c4b2a257bb06b7c6f141b42650-dirty
pimarin
parents: 14
diff changeset
127 for data in resp.iter_content(chunk_size=1024 * 1024):
ac4d5e1d1421 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit b082e89a0090f9c4b2a257bb06b7c6f141b42650-dirty
pimarin
parents: 14
diff changeset
128 fh_out.write(data)
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
129 else:
15
ac4d5e1d1421 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit b082e89a0090f9c4b2a257bb06b7c6f141b42650-dirty
pimarin
parents: 14
diff changeset
130 for data in resp.iter_content(chunk_size=1024 * 1024):
ac4d5e1d1421 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit b082e89a0090f9c4b2a257bb06b7c6f141b42650-dirty
pimarin
parents: 14
diff changeset
131 fh_out.write(data)
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
132 print(f'Download bakta database {self.db_version}')
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
133 self.tarball_path = bakta_path
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
134 except IOError:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
135 print(f'ERROR: Could not download file from Zenodo!'
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
136 f' url={self.db_url}, path={self.tarball_name}')
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
137
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
138 def untar(self):
16
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
139 db_path = Path(self.db_dir).as_posix()
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
140 try:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
141 with self.tarball_path.open('rb') as fh_in, \
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
142 tarfile.open(fileobj=fh_in, mode='r:gz') as tar_file:
16
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
143 tar_file.extractall(path=db_path)
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
144 print(f'Untar the database in {db_path}')
16
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
145 return db_path
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
146 except OSError:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
147 sys.exit(f'ERROR: Could not extract {self.tarball_name} '
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
148 f'to {self.db_name}')
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
149
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
150 def calc_md5_sum(self, buffer_size=1048576):
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
151 tarball_path = Path(self.db_dir).joinpath(self.tarball_name)
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
152 self.md5 = self.fetch_db_versions(db_version=self.db_version)["md5"]
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
153 md5 = hashlib.md5()
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
154 with tarball_path.open('rb') as fh:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
155 data = fh.read(buffer_size)
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
156 while data:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
157 md5.update(data)
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
158 data = fh.read(buffer_size)
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
159 if md5.hexdigest() == self.md5:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
160 print('\t...md5 control database OK')
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
161 else:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
162 print(f"Error: corrupt database file! "
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
163 f"calculated md5 = {md5.hexdigest()}"
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
164 f" different from {self.md5} ")
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
165
16
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
166
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
167 """
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
168 This is the method to download the amrfinderplus database need by bakta.
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
169 Deprecated to use the amrfinderplus data_manager
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
170 def update_amrfinderplus_db(self):
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
171 amrfinderplus_db_path = f"{self.db_dir}/{self.db_name}/db/amrfinderplus-db"
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
172 if self.db_version == "test":
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
173 cmd = [
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
174 'amrfinder_update',
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
175 '--database', str(amrfinderplus_db_path),
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
176 '--force_update',
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
177 '--help'
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
178 ]
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
179 else:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
180 cmd = [
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
181 'amrfinder_update',
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
182 '--database', str(amrfinderplus_db_path),
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
183 '--force_update'
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
184 ]
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
185 proc = sp.run(
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
186 cmd,
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
187 universal_newlines=True
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
188 )
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
189 if proc.returncode != 0:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
190 print(f"ERROR: AMRFinderPlus failed! "
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
191 f"command: 'amrfinder_update --force_update"
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
192 f" --database {amrfinderplus_db_path}'")
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
193 else:
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
194 print("AMRFinderPlus database download")
16
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
195 """
0
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
196
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
197
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
198 def parse_arguments():
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
199 # parse options and arguments
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
200 arg_parser = argparse.ArgumentParser()
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
201 arg_parser.add_argument("data_manager_json")
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
202 arg_parser.add_argument("-d", "--database_version",
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
203 help='Select the database version '
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
204 '(major and minor eg. 4.0),'
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
205 'default is the latest version',
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
206 default="latest",
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
207 required=True)
11
c50e8b4f74d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 572bbceeba9a22cdb591d526abda362595f8f0c4
pimarin
parents: 10
diff changeset
208 arg_parser.add_argument("-t", "--test", action='store_true',
c50e8b4f74d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 572bbceeba9a22cdb591d526abda362595f8f0c4
pimarin
parents: 10
diff changeset
209 help="option to test the script with an empty database")
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
210 return arg_parser.parse_args()
0
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
211
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
212
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
213 def main():
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
214 all_args = parse_arguments()
0
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
215
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
216 with open(all_args.data_manager_json) as fh:
16
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
217 params = json.load(fh)
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
218 target_dir = params['output_data'][0]['extra_files_path']
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
219 os.makedirs(target_dir)
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
220 # init the class to download bakta db
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
221 bakta_upload = InstallBaktaDatabase()
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
222 # extract the version
11
c50e8b4f74d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 572bbceeba9a22cdb591d526abda362595f8f0c4
pimarin
parents: 10
diff changeset
223 if all_args.test is True:
c50e8b4f74d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 572bbceeba9a22cdb591d526abda362595f8f0c4
pimarin
parents: 10
diff changeset
224 bakta_db = bakta_upload.fetch_db_versions(
c50e8b4f74d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 572bbceeba9a22cdb591d526abda362595f8f0c4
pimarin
parents: 10
diff changeset
225 db_version="test")
c50e8b4f74d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 572bbceeba9a22cdb591d526abda362595f8f0c4
pimarin
parents: 10
diff changeset
226 else:
c50e8b4f74d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 572bbceeba9a22cdb591d526abda362595f8f0c4
pimarin
parents: 10
diff changeset
227 bakta_db = bakta_upload.fetch_db_versions(
c50e8b4f74d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 572bbceeba9a22cdb591d526abda362595f8f0c4
pimarin
parents: 10
diff changeset
228 db_version=all_args.database_version)
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
229 # update the path for galaxy
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
230 bakta_upload.db_dir = target_dir
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
231 # download the database
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
232 bakta_upload.download()
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
233 # check md5 sum
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
234 bakta_upload.calc_md5_sum()
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
235 # untar db
16
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
236 bakta_upload.untar()
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
237 # make the data_manager metadata
16
718080c30a76 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
pimarin
parents: 15
diff changeset
238 bakta_data_manager = bakta_upload.get_data_manager(bakta_database_info=bakta_db, output_path=target_dir)
1
faae5d8ce0cb planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 4c11a77fb4c0bb74c0c4ae381c85971e01ed973e-dirty
pimarin
parents: 0
diff changeset
239 with open(all_args.data_manager_json, 'w') as fh:
11
c50e8b4f74d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 572bbceeba9a22cdb591d526abda362595f8f0c4
pimarin
parents: 10
diff changeset
240 json.dump(bakta_data_manager, fh, sort_keys=True)
0
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
241
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
242
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
243 if __name__ == '__main__':
9d08486abf8e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit 5f320f165a7f454193df36cc2ec77a87d50ec80f-dirty"
pimarin
parents:
diff changeset
244 main()