annotate data_manager/bakta_build_database.py @ 19:c90380f8bbbc draft

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
author pimarin
date Fri, 13 Jan 2023 14:11:30 +0000
parents 04bee0f935a2
children ddeedb302cf1
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
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
6 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
7 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
8 from pathlib import Path
19
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
9
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
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
11 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
12
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
13
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
14 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
15 """
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 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
17 """
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
18
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
19 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
20 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
21 db_name=Path.cwd().joinpath("db"),
19
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
22 db_version="latest",
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
23 test_mode=False):
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
24 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
25 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
26 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
27 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
28 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
29 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
30 self.DB_VERSIONS_URL = 'https://raw.githubusercontent.com/oschwengers/bakta/master/db-versions.json'
19
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
31 self.DB_TEST_URL = 'https://zenodo.org/record/7360542/files/db-versions.json'
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
32 self.test_mode = test_mode
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
33
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
34 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
35 """
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
36 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
37 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
38 """
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 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
40 "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
41 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
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 }
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
44 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
45
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
46 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
47 """
19
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
48 List bakta database info related to 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
49 """
19
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
50 if self.test_mode is True:
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
51 self.DB_VERSIONS_URL = self.DB_TEST_URL
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
52 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
53 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
54 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
55 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
56 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
57 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
58 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
59 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
60 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
61 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
62 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
63 '%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
64 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
65 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
66 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
67 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
68 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
69 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
70 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
71 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
72 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
73 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
74 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
75 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
76 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
77 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
78
19
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
79 def get_data_manager(self, bakta_database_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
80 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
81 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
82 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
83 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
84 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
85 f"{bakta_database_info['software-min']['minor']}")
19
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
86 data_info = dict(value=bakta_value,
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
87 dbkey=bakta_database_info['record'],
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_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
89 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
90 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
91 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
92
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
93
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
94 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
95 """
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
96 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
97 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
98 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
99 """
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
100
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 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
102 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
103 db_name="bakta",
19
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
104 tarball_name="db.tar.gz",
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
105 test_mode=False):
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
106 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
107 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
108 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
109 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
110 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
111 self.tarball_path = None
19
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
112 self.test_mode = test_mode
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
113
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 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
115 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
116 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
117 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
118 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
119 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
120 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
121 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
122 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
123 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
124 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
125 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
126 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
127 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
128 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
129 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
130 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
131 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
132
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 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
134 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
135 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
136 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
137 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
138 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
139 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
140 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
141 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
142 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
143 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
144
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
145 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
146 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
147 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
148 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
149 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
150 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
151 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
152 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
153 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
154 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
155 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
156 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
157 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
158 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
159 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
160
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
161
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
162 """
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
163 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
164 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
165 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
166 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
167 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
168 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
169 '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
170 '--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
171 '--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
172 '--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
173 ]
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 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
175 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
176 '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
177 '--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
178 '--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
179 ]
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 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
181 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
182 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
183 )
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 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
185 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
186 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
187 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
188 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
189 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
190 """
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
191
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
192
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
193 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
194 # 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
195 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
196 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
197 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
198 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
199 '(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
200 '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
201 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
202 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
203 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
204 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
205 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
206
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
207
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
208 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
209 all_args = 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
210 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
211 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
212 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
213 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
214 # init the class to download bakta db
19
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
215 bakta_upload = InstallBaktaDatabase(test_mode=all_args.test)
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
216 bakta_db = bakta_upload.fetch_db_versions(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
217 # 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
218 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
219 # 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
220 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
221 # 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
222 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
223 # 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
224 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
225 # make the data_manager metadata
19
c90380f8bbbc planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_bakta_database commit 8eb28a93e6f2688bb2f3f85aea0389e1b1148816-dirty
pimarin
parents: 18
diff changeset
226 bakta_data_manager = bakta_upload.get_data_manager(bakta_database_info=bakta_db)
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
227 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
228 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
229
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
230
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
231 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
232 main()