async def test_make_entity_links():
schema = make_pfb_schema([person_entity_def, pet_entity_def])
file = make_avro_file(schema, [
{'name': 'person', 'id': '123', 'object': {'first_name': 'Brian'}},
{'name': 'pet', 'id': '456', 'object': {'pet_name': 'Asa'}, 'relations': [
{'dst_name': 'person', 'dst_id': '123'}
]}
])
result = await translate(fastavro.reader(file))
assert result == [
{
'name': '123',
'entityType': 'person',
'operations': [
add_update_attribute('first_name', 'Brian')
]
},
{
'name': '456',
'entityType': 'pet',
'operations': [
add_update_attribute('pet_name', 'Asa'),
add_update_attribute('person', {
'entityType': 'person',
'entityName': '123'
})
]
}
]
async def test_decode_enum_value():
use_base64_encoded_enums = True
schema = make_pfb_schema([make_sample_entity_def(use_base64_encoded_enums)])
file = make_avro_file(schema, [
{'name': 'sample', 'id': '123', 'object': {'tissue_type': tissue_type_b64_values['Tumor']}},
{'name': 'sample', 'id': '456', 'object': {'tissue_type': tissue_type_b64_values['Normal']}},
{'name': 'sample', 'id': '789', 'object': {'tissue_type': tissue_type_b64_values['Unknown']}}
])
result = await translate(fastavro.reader(file), {'b64-decode-enums': use_base64_encoded_enums})
assert result == [
{
'name': '123',
'entityType': 'sample',
'operations': [
add_update_attribute('tissue_type', 'Tumor')
]
},
{
'name': '456',
'entityType': 'sample',
'operations': [
add_update_attribute('tissue_type', 'Normal')
]
},
{
'name': '789',
'entityType': 'sample',
'operations': [
add_update_attribute('tissue_type', 'Unknown')
]
}
]
async def test_make_file_links():
schema = make_pfb_schema([file_def])
file = make_avro_file(schema, [
{'name': 'file', 'id': '123', 'object': {'object_id': 'abc.de/12345'}}
])
result = await translate(fastavro.reader(file))
assert result == [
{
'name': '123',
'entityType': 'file',
'operations': [
add_update_attribute('object_id', 'abc.de/12345')
]
}
]
file.seek(0)
result = await translate(fastavro.reader(file), {'prefix-object-ids': True})
assert result == [
{
'name': '123',
'entityType': 'file',
'operations': [
add_update_attribute('object_id', 'drs://abc.de/12345')
]
}
]
async def test_disambiguate_enum_and_non_enum():
"""
Verify enum value decoding when different entities have fields with the same names
"""
schema = make_pfb_schema([make_sample_entity_def(), other_sample_entity_def])
file = make_avro_file(schema, [
{'name': 'sample', 'id': '123', 'object': {'tissue_type': tissue_type_b64_values['Tumor']}},
{'name': 'other_sample', 'id': '456', 'object': {'tissue_type': 'not a tumor'}},
])
result = await translate(fastavro.reader(file), {'b64-decode-enums': True})
assert result == [
{
'name': '123',
'entityType': 'sample',
'operations': [
add_update_attribute('tissue_type', 'Tumor')
]
},
{
'name': '456',
'entityType': 'other_sample',
'operations': [
add_update_attribute('tissue_type', 'not a tumor')
]
}
]
async def test_ignore_metadata():
schema = make_pfb_schema()
file = make_avro_file(schema, [
{'name': 'Metadata', 'object': {}}
])
result = await translate(fastavro.reader(file))
assert result == []
async def test_filter_none_values():
schema = make_pfb_schema([person_entity_def])
file = make_avro_file(schema, [
{'name': 'person', 'id': '123', 'object': {'first_name': 'Test'}}
])
result = await translate(fastavro.reader(file))
assert result == [
{
'name': '123',
'entityType': 'person',
'operations': [
add_update_attribute('first_name', 'Test')
]
}
]
async def test_rename_name_attributes():
schema = make_pfb_schema([named_entity_def])
file = make_avro_file(schema, [
{'name': 'thing', 'id': '1', 'object': {'name': 'Thing 1'}}
])
result = await translate(fastavro.reader(file))
assert result == [
{
'name': '1',
'entityType': 'thing',
'operations': [
add_update_attribute('thing_name', 'Thing 1')
]
}
]
async def test_translate_record():
schema = make_pfb_schema([person_entity_def])
file = make_avro_file(schema, [
{'name': 'person', 'id': '123', 'object': {
'first_name': 'Test', 'last_name': 'Dummy', 'eye_color': 'gray'
}}
])
result = await translate(fastavro.reader(file))
assert result == [
{
'name': '123',
'entityType': 'person',
'operations': [
add_update_attribute('first_name', 'Test'),
add_update_attribute('last_name', 'Dummy'),
add_update_attribute('eye_color', 'gray')
]
}
]
async def test_translate_nothing():
schema = make_pfb_schema()
file = make_avro_file(schema)
result = await translate(fastavro.reader(file))
assert result == []