def test_data_type_builder_doc_bad():
a = DocumentProto()
a.id = 'a236cbb0eda62d58'
with pytest.raises(BadDocType):
_new_doc_from_data(b'BREAKIT!' + a.SerializeToString(), DataInputType.DOCUMENT)
with pytest.raises(BadDocType):
_new_doc_from_data(MessageToJson(a) + '�', DataInputType.DOCUMENT)
def test_data_type_builder_auto(input_type):
if 'JINA_ARRAY_QUANT' in os.environ:
print(f'quant is on: {os.environ["JINA_ARRAY_QUANT"]}')
del os.environ['JINA_ARRAY_QUANT']
d, t = _new_doc_from_data('123', input_type)
assert d.text == '123'
assert t == DataInputType.CONTENT
d, t = _new_doc_from_data(b'123', input_type)
assert t == DataInputType.CONTENT
assert d.blob == b'123'
c = np.random.random([10, 10])
d, t = _new_doc_from_data(c, input_type)
np.testing.assert_equal(d.tensor, c)
assert t == DataInputType.CONTENT
def test_data_type_builder_doc(builder, input_data_type, output_data_type):
a = Document()
a.id = 'a236cbb0eda62d58'
a.text = 'text test'
d, t = _new_doc_from_data(builder(a), input_data_type)
if input_data_type != DataInputType.CONTENT:
assert d.id == a.id
assert d.text == a.text
assert t == output_data_type
def test_data_type_builder_doc(builder):
a = DocumentProto()
a.id = 'a236cbb0eda62d58'
d, t = _new_doc_from_data(builder(a), DataInputType.DOCUMENT)
assert d.id == a.id
assert t == DataInputType.DOCUMENT
