def test_should_insert_issue_in_changes(self, read_json_file_mock):
read_json_file_mock.return_value = SAMPLE_ISSUES_JSON
pipeline.import_issues("~/json/issues.json", self.session)
_changes = self.session.changes.filter()
self.assertEqual(1, len(_changes))
self.assertEqual("0001-3714-1998-v29-n3", _changes[0]["id"])
def test_pipeline_should_insert_issue_in_database_and_register_change(
self, read_json_file_mock
):
read_json_file_mock.return_value = SAMPLE_ISSUES_JSON
pipeline.import_issues("~/json/issues.json", self.session)
expected = SAMPLE_ISSUES_KERNEL[0]["_id"]
self.assertEqual(expected, self.session.documents_bundles.fetch(expected).id())
def migrate_isis_parser(sargs):
parser = argparse.ArgumentParser(description="ISIS database migration tool")
subparsers = parser.add_subparsers(title="Commands", metavar="", dest="command")
extract_parser = subparsers.add_parser("extract", help="Extract mst files to json")
extract_parser.add_argument(
"mst_file_path", metavar="file", help="Path to MST file that will be extracted"
)
extract_parser.add_argument("--output", required=True, help="The output file path")
import_parser = subparsers.add_parser(
"import",
parents=[mongodb_parser(sargs)],
help="Process JSON files then import into Kernel database",
)
import_parser.add_argument(
"import_file",
metavar="file",
help="JSON file path that contains mst extraction result, e.g: collection-title.json",
)
import_parser.add_argument(
"--type",
help="Type of JSON file that will load into Kernel database",
choices=["journal", "issue", "documents-bundles-link"],
required=True,
)
link_parser = subparsers.add_parser(
"link",
help="Generate JSON file of journals' ids and their issues linked by ISSN",
)
link_parser.add_argument(
"issues",
help="JSON file path that contains mst extraction result, e.g: ~/json/collection-issues.json",
)
link_parser.add_argument("--output", required=True, help="The output file path")
args = parser.parse_args(sargs)
if args.command == "extract":
extract_isis.create_output_dir(args.output)
extract_isis.run(args.mst_file_path, args.output)
elif args.command == "import":
mongo = ds_adapters.MongoDB(uri=args.uri, dbname=args.db)
Session = ds_adapters.Session.partial(mongo)
if args.type == "journal":
pipeline.import_journals(args.import_file, session=Session())
elif args.type == "issue":
pipeline.import_issues(args.import_file, session=Session())
elif args.type == "documents-bundles-link":
pipeline.import_documents_bundles_link_with_journal(
args.import_file, session=Session()
)
elif args.command == "link":
pipeline.link_documents_bundles_with_journals(args.issues, args.output)
else:
parser.print_help()
def test_should_raise_already_exception_if_try_to_insert_same_id_twice(
self, read_json_file_mock
):
read_json_file_mock.return_value = SAMPLE_ISSUES_JSON
with self.assertLogs(level="INFO") as log:
pipeline.import_issues("~/json/issues.json", self.session)
pipeline.import_issues("~/json/issues.json", self.session)
self.assertIn("pipeline", log[1][0])
def test_pipeline_should_read_correct_json_file(self, read_json_file_mock):
pipeline.import_issues("~/json/issues.json", self.session)
read_json_file_mock.assert_called_once_with("~/json/issues.json")