def test_should_update_journals_with_document_bundle_ids(self, read_json_file_mock):
read_json_file_mock.return_value = {
"0001-3714": [
{
"id": "0034-8910-1978-v12-n2",
"number": "2",
"order": "2",
"supplement": "",
"volume": "12",
"year": "1978",
}
]
}
pipeline.import_documents_bundles_link_with_journal(
"~/json/output.json", self.session
)
_journal = self.session.journals.fetch("0001-3714")
_changes = self.session.changes.filter()
self.assertIn(
{
"id": "0034-8910-1978-v12-n2",
"number": "2",
"order": "2",
"supplement": "",
"volume": "12",
"year": "1978",
},
_journal.issues,
)
self.assertEqual(1, len(_changes))
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_not_update_journals_with_document_bundle(self, read_json_file_mock):
read_json_file_mock.return_value = {"0001-3714": []}
pipeline.import_documents_bundles_link_with_journal(
"~/json/output.json", self.session
)
_journal = self.session.journals.fetch("0001-3714")
self.assertEqual([], _journal.issues)
def test_should_not_update_journals_with_duplicated_bundles(
self, read_json_file_mock
):
read_json_file_mock.return_value = {
"0001-3714": [
{
"id": "issue-1",
"order": "0001",
"number": "01",
"volume": "01",
"year": "2019",
"supplement": "supplement",
},
{
"id": "issue-2",
"order": "0002",
"number": "02",
"volume": "02",
"year": "2019",
"supplement": "supplement",
},
{
"id": "issue-2",
"order": "0002",
"number": "02",
"volume": "02",
"year": "2019",
"supplement": "supplement",
},
]
}
pipeline.import_documents_bundles_link_with_journal(
"~/json/output.json", self.session
)
_journal = self.session.journals.fetch("0001-3714")
self.assertEqual(
[
{
"id": "issue-1",
"order": "0001",
"number": "01",
"volume": "01",
"year": "2019",
"supplement": "supplement",
},
{
"id": "issue-2",
"order": "0002",
"number": "02",
"volume": "02",
"year": "2019",
"supplement": "supplement",
},
],
_journal.issues,
)
def test_should_update_journals_with_document_bundle_ids(self, read_json_file_mock):
read_json_file_mock.return_value = {"0001-3714": ["0001-3714-1998-v29-n3"]}
pipeline.import_documents_bundles_link_with_journal(
"~/json/output.json", self.session
)
_journal = self.session.journals.fetch("0001-3714")
_changes = self.session.changes.filter()
self.assertIn("0001-3714-1998-v29-n3", _journal.issues)
self.assertEqual(1, len(_changes))
def test_should_log_dabase_exceptions(self, read_json_file_mock):
read_json_file_mock.side_effect = [
{
"0001-3714": [
{
"id": "issue-1",
"order": "0001",
"number": "01",
"volume": "01",
"year": "2019",
"supplement": "supplement",
},
{
"id": "issue-1",
"order": "0001",
"number": "01",
"volume": "01",
"year": "2019",
"supplement": "supplement",
},
{
"id": "issue-2",
"order": "0002",
"number": "02",
"volume": "02",
"year": "2019",
"supplement": "supplement",
},
]
},
{"missing-journal": []},
]
with self.assertLogs(level="DEBUG") as log:
pipeline.import_documents_bundles_link_with_journal(
"~/json/output.json", self.session
)
self.assertIn(
"Bundle issue-1 already exists in journal 0001-3714", log[-1][-1]
)
with self.assertLogs(level="DEBUG") as log:
pipeline.import_documents_bundles_link_with_journal(
"~/json/output.json", self.session
)
self.assertIn(
"Journal missing-journal does not exists, cannot link bundles.",
log[-1][-1],
)
def test_read_journal_bundle_file(self, read_json_file_mock):
read_json_file_mock.return_value = {}
pipeline.import_documents_bundles_link_with_journal(
"~/json/output.json", self.session
)
read_json_file_mock.assert_called_once_with("~/json/output.json")
