def test_migrate_record_from_mirror_invalidates_local_file_cache_if_no_local_file(
inspire_app, s3, redis, datadir
):
expected_key = "f43f40833edfd8227c4deb9ad05b321e"
create_s3_bucket(expected_key)
with patch.dict(
current_app.config, {"LABS_AFS_HTTP_SERVICE": "http://inspire-afs-web.cern.ch/"}
):
redis.delete("afs_file_locations")
# populate cache with invalid file path
redis.hset(
"afs_file_locations",
"http://inspire-afs-web.cern.ch/var/data/files/g97/1940001/content.pdf%3B2",
"/api/files/ddb1a354-1d2a-40b6-9cc4-2e823b6bef81/0000000000000000000000000000000000000000",
)
raw_record_path = (datadir / "1313624.xml").as_posix()
migrate_from_file(raw_record_path)
record = LiteratureRecord.get_record_by_pid_value("1313624")
assert redis.hlen("afs_file_locations") > 0
assert (
record["documents"][0]["original_url"]
== "http://inspire-afs-web.cern.ch/var/data/files/g97/1940001/content.pdf%3B2"
)
def test_orcid_push_disabled_on_migrate_from_mirror(
inspire_app, enable_orcid_push_feature
):
record_fixture_path = pkg_resources.resource_filename(
__name__, os.path.join("fixtures", "dummy.xml")
)
with patch(
"inspirehep.orcid.domain_models.OrcidPusher"
) as mock_orcid_pusher, patch(
"inspirehep.orcid.push_access_tokens"
) as mock_push_access_tokens:
mock_push_access_tokens.get_access_tokens.return_value.remote_account.extra_data[
"orcid"
] = "0000-0002-1825-0097"
mock_push_access_tokens.get_access_tokens.return_value.access_token = "mytoken"
migrate_from_file(record_fixture_path)
mock_orcid_pusher.assert_not_called()
prod_record = LegacyRecordsMirror.query.filter(
LegacyRecordsMirror.recid == 12345
).one()
assert prod_record.valid
assert current_app.config["FEATURE_FLAG_ENABLE_ORCID_PUSH"]
def test_migrate_from_mirror_doesnt_raise_on_job_records(inspire_app):
record_fixture_path = pkg_resources.resource_filename(
__name__, os.path.join("fixtures", "1695012.xml"))
migrate_from_file(record_fixture_path)
job_rec = LegacyRecordsMirror.query.filter(
LegacyRecordsMirror.recid == 1695012).one()
assert job_rec.valid
def test_migrate_record_from_mirror_with_download_file_error_not_caused_by_invalid_cache(
inspire_app, s3, redis, datadir):
with patch.dict(
current_app.config,
{"LABS_AFS_HTTP_SERVICE": "http://inspire-afs-web.cern.ch/"}):
redis.delete("afs_file_locations")
raw_record_path = (datadir / "1313624.xml").as_posix()
with requests_mock.Mocker() as mocker:
mocker.get(
"http://inspire-afs-web.cern.ch/var/data/files/g97/1940001/content.pdf%3B2",
status_code=404,
)
migrate_from_file(raw_record_path)
record_mirror = LegacyRecordsMirror.query.filter(
LegacyRecordsMirror.recid == 1313624).one()
assert record_mirror.error.startswith("DownloadFileError")
def test_migrate_record_from_mirror_uses_local_cache_for_afs_files(
inspire_app, s3, redis, datadir):
expected_key = "f43f40833edfd8227c4deb9ad05b321e"
create_s3_bucket(expected_key)
with patch.dict(
current_app.config,
{"LABS_AFS_HTTP_SERVICE": "http://inspire-afs-web.cern.ch/"}):
redis.delete("afs_file_locations")
raw_record_path = (datadir / "1313624.xml").as_posix()
migrate_from_file(raw_record_path)
assert redis.hlen("afs_file_locations") > 0
migrate_from_file(raw_record_path)
record = LiteratureRecord.get_record_by_pid_value("1313624")
# No original_url as source is local file
assert "original_url" not in record["documents"][0]
def test_migrate_from_mirror_doesnt_index_deleted_records(inspire_app):
record_fixture_path = pkg_resources.resource_filename(
__name__, os.path.join("fixtures", "dummy.xml"))
record_fixture_path_deleted = pkg_resources.resource_filename(
__name__, os.path.join("fixtures", "deleted_record.xml"))
migrate_from_file(record_fixture_path)
migrate_from_file(record_fixture_path_deleted)
current_search.flush_and_refresh("records-hep")
expected_record_lit_es_len = 1
record_lit_uuid = LiteratureRecord.get_uuid_from_pid_value(12345)
record_lit_es = LiteratureSearch().get_record(
str(record_lit_uuid)).execute().hits
record_lit_es_len = len(record_lit_es)
assert expected_record_lit_es_len == record_lit_es_len
def test_migrate_from_mirror_doesnt_index_deleted_records(
base_app, db, es_clear):
record_fixture_path = pkg_resources.resource_filename(
__name__, os.path.join("fixtures", "dummy.xml"))
record_fixture_path_deleted = pkg_resources.resource_filename(
__name__, os.path.join("fixtures", "deleted_record.xml"))
migrate_from_file(record_fixture_path)
migrate_from_file(record_fixture_path_deleted)
es_clear.indices.refresh("records-hep")
expected_record_lit_es_len = 1
record_lit_uuid = LiteratureRecord.get_uuid_from_pid_value(12345)
with pytest.raises(PIDDoesNotExistError):
LiteratureRecord.get_uuid_from_pid_value(1234)
record_lit_es = LiteratureSearch().get_record(
str(record_lit_uuid)).execute().hits
record_lit_es_len = len(record_lit_es)
assert expected_record_lit_es_len == record_lit_es_len
def test_migrate_from_mirror_removes_record_from_es(inspire_app, datadir):
data = orjson.loads((datadir / "dummy_record.json").read_text())
create_record("lit", data=data)
expected_record_lit_es_len = 1
record_lit_uuid = LiteratureRecord.get_uuid_from_pid_value(12345)
record_lit_es = LiteratureSearch().get_record(str(record_lit_uuid)).execute().hits
record_lit_es_len = len(record_lit_es)
assert expected_record_lit_es_len == record_lit_es_len
record_deleted_fixture_path = pkg_resources.resource_filename(
__name__, os.path.join("fixtures", "dummy_deleted.xml")
)
migrate_from_file(record_deleted_fixture_path)
current_search.flush_and_refresh("records-hep")
expected_record_lit_es_len = 0
record_lit_uuid = LiteratureRecord.get_uuid_from_pid_value(12345)
record_lit_es = LiteratureSearch().get_record(str(record_lit_uuid)).execute().hits
record_lit_es_len = len(record_lit_es)
assert expected_record_lit_es_len == record_lit_es_len
