Mercurial > repos > pimarin > data_manager_bakta
changeset 16:718080c30a76 draft
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_bakta commit e7c35e529ae95a2c9f2ecd87a3716e6866feabfa
| author | pimarin |
|---|---|
| date | Fri, 04 Nov 2022 15:46:06 +0000 |
| parents | ac4d5e1d1421 |
| children | ecb7d3d41366 |
| files | data_manager/bakta_build_database.py data_manager/bakta_build_database.xml data_manager/macro.xml data_manager_conf.xml test-data/bakta_test.loc test-data/bakta_test_data_manager.json tool-data/bakta_database.loc tool_data_table_conf.xml.sample tool_data_table_conf.xml.test |
| diffstat | 9 files changed, 37 insertions(+), 38 deletions(-) [+] |
line wrap: on
line diff
--- a/data_manager/bakta_build_database.py Wed Nov 02 16:21:11 2022 +0000 +++ b/data_manager/bakta_build_database.py Fri Nov 04 15:46:06 2022 +0000 @@ -2,13 +2,14 @@ import hashlib import json import os -import subprocess as sp import sys import tarfile from datetime import datetime from pathlib import Path import requests + + class GetBaktaDatabaseInfo: """ Extract bakta database information to make a json file for data_manager @@ -64,7 +65,6 @@ for db_dic in versions: db_date_list.append(datetime.strptime(db_dic["date"], '%Y-%m-%d').date()) - max(db_date_list) filtered_version = next(item for item in versions if max(db_date_list)) elif db_version == "test": @@ -96,17 +96,16 @@ def get_data_manager(self, bakta_database_info, output_path): self.bakta_table_list = self.get_data_table_format() - bakta_value = f"bakta_{bakta_database_info['major']}." \ - f"{bakta_database_info['minor']}" - data_info = dict(value=bakta_value, - dbkey=bakta_database_info['date'], - database_record=bakta_database_info['record'], - bakta_version=str( - f"{bakta_database_info['software-min']['major']}." - f"{bakta_database_info['software-min']['minor']}" - ), - path=output_path) - self.bakta_table_list["data_tables"][self.data_table_name] = data_info + bakta_value = f"V{bakta_database_info['major']}." \ + f"{bakta_database_info['minor']}_" \ + f"{bakta_database_info['date']}" + tool_version = str(f"{bakta_database_info['software-min']['major']}." + f"{bakta_database_info['software-min']['minor']}") + data_info = dict(value=bakta_database_info['record'], + dbkey=bakta_value, + bakta_version=tool_version, + path="db") + self.bakta_table_list["data_tables"][self.data_table_name] = [data_info] return self.bakta_table_list @@ -148,13 +147,13 @@ f' url={self.db_url}, path={self.tarball_name}') def untar(self): - db_path = Path(self.db_dir).joinpath(self.db_name) + db_path = Path(self.db_dir).as_posix() try: with self.tarball_path.open('rb') as fh_in, \ tarfile.open(fileobj=fh_in, mode='r:gz') as tar_file: - tar_file.extractall(path=str(db_path)) + tar_file.extractall(path=db_path) print(f'Untar the database in {db_path}') - return db_path.as_posix() + return db_path except OSError: sys.exit(f'ERROR: Could not extract {self.tarball_name} ' f'to {self.db_name}') @@ -175,6 +174,10 @@ f"calculated md5 = {md5.hexdigest()}" f" different from {self.md5} ") + +""" +This is the method to download the amrfinderplus database need by bakta. +Deprecated to use the amrfinderplus data_manager def update_amrfinderplus_db(self): amrfinderplus_db_path = f"{self.db_dir}/{self.db_name}/db/amrfinderplus-db" if self.db_version == "test": @@ -200,6 +203,7 @@ f" --database {amrfinderplus_db_path}'") else: print("AMRFinderPlus database download") +""" def parse_arguments(): @@ -221,11 +225,9 @@ all_args = parse_arguments() with open(all_args.data_manager_json) as fh: - data_manager_input = json.load(fh) - - target_dir = data_manager_input['output_data'][0]['extra_files_path'] + params = json.load(fh) + target_dir = params['output_data'][0]['extra_files_path'] os.makedirs(target_dir) - # init the class to download bakta db bakta_upload = InstallBaktaDatabase() # extract the version @@ -242,11 +244,9 @@ # check md5 sum bakta_upload.calc_md5_sum() # untar db - bakta_extracted_path = bakta_upload.untar() - # update for amrfinderplus - bakta_upload.update_amrfinderplus_db() + bakta_upload.untar() # make the data_manager metadata - bakta_data_manager = bakta_upload.get_data_manager(bakta_database_info=bakta_db, output_path=bakta_extracted_path) + bakta_data_manager = bakta_upload.get_data_manager(bakta_database_info=bakta_db, output_path=target_dir) with open(all_args.data_manager_json, 'w') as fh: json.dump(bakta_data_manager, fh, sort_keys=True)
--- a/data_manager/bakta_build_database.xml Wed Nov 02 16:21:11 2022 +0000 +++ b/data_manager/bakta_build_database.xml Fri Nov 04 15:46:06 2022 +0000 @@ -33,7 +33,7 @@ <!-- Test download --> <test expect_num_outputs="1"> <param name="test_data_manager" value="--test"/> - <output name="output_file" value="bakta_test_data_manager.json" compare="contains"/> + <output name="output_file" value="bakta_test_data_manager.json" /> </test> </tests> <help><![CDATA[
--- a/data_manager/macro.xml Wed Nov 02 16:21:11 2022 +0000 +++ b/data_manager/macro.xml Fri Nov 04 15:46:06 2022 +0000 @@ -5,9 +5,9 @@ <token name="@VERSION_SUFFIX@">0</token> <token name="@PROFILE@">21.05</token> <xml name="requirements"> - <requirements> + <requirements>w <requirement type="package" version="@PYTHON_VERSION@">python</requirement> - <requirement type="package" version="@REQUEST_VERSION@">requests</requirement> + <requirement type="package" version="@REQUESTS_VERSION@">requests</requirement> </requirements> </xml> </macros>
--- a/data_manager_conf.xml Wed Nov 02 16:21:11 2022 +0000 +++ b/data_manager_conf.xml Fri Nov 04 15:46:06 2022 +0000 @@ -4,15 +4,14 @@ <output> <column name="value"/> <column name="dbkey"/> - <column name="database_record"/> <column name="bakta_version"/> <column name="path" output_ref="output_file"> <move type="directory" relativize_symlinks="True"> <source>${path}</source> - <target base="${GALAXY_DATA_MANAGER_DATA_PATH}">bakta_database/${dbkey}/${path}</target> + <target base="${GALAXY_DATA_MANAGER_DATA_PATH}">bakta_database/${value}</target> </move> - <value_translation>${GALAXY_DATA_MANAGER_DATA_PATH}/bakta_database/${dbkey}/${path}</value_translation> - <value_translation type="function">abspath</value_translation> + <value_translation>${GALAXY_DATA_MANAGER_DATA_PATH}/bakta_database/${value}</value_translation> + <value_translation type="function">abspath</value_translation> </column> </output> </data_table>
--- a/test-data/bakta_test.loc Wed Nov 02 16:21:11 2022 +0000 +++ b/test-data/bakta_test.loc Fri Nov 04 15:46:06 2022 +0000 @@ -1,7 +1,7 @@ # this is a tab separated file describing the location of bakta database # # the columns are: -# value, dbkey, database_record, bakta_version, path +# value, dbkey, bakta_version, path # # for example -bakta_0.0 date_test 7197299 0.0 ${__HERE__} +7197299 V0.0_date_test 0.0 ${__HERE__}
--- a/test-data/bakta_test_data_manager.json Wed Nov 02 16:21:11 2022 +0000 +++ b/test-data/bakta_test_data_manager.json Fri Nov 04 15:46:06 2022 +0000 @@ -1,1 +1,1 @@ -{"data_tables":{"bakta_database": [{"bakta_version": "0.0","database_record": "7197299","dbkey": "date_test","path": "/tmp/tmpu_ccwfs9/job_working_directory/000/1/working/dataset_fb48f154-bf1b-4974-8759-e4f0e230e917_files/bakta_test","value": "bakta_0.0"}]}} +{"data_tables": {"bakta_database": [{"bakta_version": "0.0", "dbkey": "V0.0_date_test", "path": "db", "value": "7197299"}]}}
--- a/tool-data/bakta_database.loc Wed Nov 02 16:21:11 2022 +0000 +++ b/tool-data/bakta_database.loc Fri Nov 04 15:46:06 2022 +0000 @@ -1,7 +1,7 @@ # this is a tab separated file describing the location of bakta database # # the columns are: -# value, dbkey, database_record, bakta_version, path +# value, dbkey, bakta_version, path # # for example -#bakta_0.0 date_test 7197299 0.0 path/to/db +#7197299 V0.0_date_test 0.0 ${__HERE__}
--- a/tool_data_table_conf.xml.sample Wed Nov 02 16:21:11 2022 +0000 +++ b/tool_data_table_conf.xml.sample Fri Nov 04 15:46:06 2022 +0000 @@ -1,7 +1,7 @@ <tables> <!-- Locations of bakta database in the required format --> <table name="bakta_database" comment_char="#"> - <columns>value, dbkey, database_record, bakta_version, path</columns> + <columns>value, dbkey, bakta_version, path</columns> <file path="tool-data/bakta_database.loc" /> </table> </tables>
--- a/tool_data_table_conf.xml.test Wed Nov 02 16:21:11 2022 +0000 +++ b/tool_data_table_conf.xml.test Fri Nov 04 15:46:06 2022 +0000 @@ -1,7 +1,7 @@ <tables> <!-- Locations of bakta database in the required format --> <table name="bakta_database" comment_char="#" allow_duplicate_entries="False"> - <columns>value, dbkey, database_record, bakta_version, path</columns> + <columns>value, dbkey, bakta_version, path</columns> <file path="${__HERE__}/test-data/bakta_test.loc" /> </table> </tables>