def test_new_file():
""" Test that new file are created correctly """
os.environ['DATASERVICE_API'] = 'http://api.com/'
mock = patch('service.requests')
req = mock.start()
mock_resp = MagicMock()
mock_resp.status_code = 201
mock_resp.json.return_value = {'results': {'kf_id': 'GF_00000000'}}
req.post.return_value = mock_resp
importer = service.FileImporter('http://api.com/', 'abc123')
res = importer.new_file(BUCKET, OBJECT, 'd41d8cd98f00b204e9800998ecf8427e',
1024)
expected = {
'file_name': '60d33dec-98db-446c-ac64-f4d027588f26.cram',
'file_format': 'cram',
'data_type': 'Aligned Reads',
'controlled_access': True,
'availability': 'Immediate Download',
'is_harmonized': True,
'reference_genome': 'GRCh38',
'hashes': {
'etag': 'd41d8cd98f00b204e9800998ecf8427e'
},
'size': 1024,
'urls': ['s3://{}/{}'.format(BUCKET, OBJECT)],
'acl': []
}
req.post.assert_called_with('http://api.com/genomic-files', json=expected)
assert req.post.call_count == 1
def test_no_biospecimen(event, obj):
""" Test that nothing is done if the biospecimen does not exist """
obj()
s3 = boto3.client('s3')
mock = patch('service.requests')
req = mock.start()
mock_resp = MagicMock()
mock_resp.status_code = 404
req.get.return_value = mock_resp
importer = service.FileImporter('http://api.com/', 'abc123')
with pytest.raises(service.ImportException):
importer.import_harmonized(event['Records'][0])
req.get.assert_any_call('http://api.com/biospecimens/BS_QV3Z0DZM')
mock.stop()
def test_new_file_error():
""" Test that new file with predefined kf_id """
os.environ['DATASERVICE_API'] = 'http://api.com/'
mock = patch('service.requests')
req = mock.start()
mock_resp = MagicMock()
mock_resp.status_code = 400
mock_resp.json.return_value = {'results': {'kf_id': 'GF_00000002'}}
req.post.return_value = mock_resp
importer = service.FileImporter('http://api.com/', 'abc123')
with pytest.raises(service.DataServiceException):
res = importer.new_file(BUCKET, OBJECT,
'd41d8cd98f00b204e9800998ecf8427e', 1024)
assert req.post.call_count == 1
def test_get_gf_id_tag():
""" Test that gf_id is returned from tags correctly """
mock = patch('service.requests')
req = mock.start()
mock_resp = MagicMock()
mock_resp.status_code = 404
mock_resp.json.return_value = {'_status': {'code': 404}}
req.post.return_value = mock_resp
importer = service.FileImporter('http://api.com/', 'abc123')
tags = {'gf_id': 'GF_00000003'}
res = importer.get_gf_id_tag(tags)
assert res == 'GF_00000003'
assert req.get.call_count == 1
mock.stop()
def test_get_gf_id_tag_exists():
""" Test getting gf_id tag if it already exists """
mock = patch('service.requests')
req = mock.start()
mock_resp = MagicMock()
mock_resp.status_code = 200
mock_resp.json.return_value = {'results': {'kf_id': 'GF_00000003'}}
req.get.return_value = mock_resp
importer = service.FileImporter('http://api.com/', 'abc123')
tags = {'gf_id': 'GF_00000003'}
with pytest.raises(service.ImportException):
res = importer.get_gf_id_tag(tags)
assert req.get.call_count == 1
mock.stop()
def test_req_tags(event, obj):
""" Test that error is raised for missing tags """
obj()
s3 = boto3.client('s3')
# Add a gf_id
tags = TAGS.copy()
tags['TagSet'] = tags['TagSet'][:4]
response = s3.put_object_tagging(Bucket=BUCKET, Key=OBJECT, Tagging=tags)
mock = patch('service.requests')
req = mock.start()
mock_resp = MagicMock()
mock_resp.status_code = 404
req.get.return_value = mock_resp
importer = service.FileImporter('http://api.com/', 'abc123')
with pytest.raises(service.ImportException):
importer.import_harmonized(event['Records'][0])
mock.stop()
def test_file_formats(filename, file_format, data_type):
""" Test that file formats and data types are discovered correctly """
mock = patch('service.requests')
req = mock.start()
mock_resp = MagicMock()
mock_resp.status_code = 201
mock_resp.json.return_value = {'results': {'kf_id': 'GF_00000001'}}
req.post.return_value = mock_resp
importer = service.FileImporter('http://api.com/', 'abc123')
res = importer.new_file(SOURCE_BUCKET,
filename,
'd41d8cd98f00b204e9800998ecf8427e',
1024,
gf_id='GF_00000001',
bs_id='BS_00000000',
study_id='SD_00000000')
expected = {
'kf_id': 'GF_00000001',
'file_name': filename,
'file_format': file_format,
'acl': ['SD_00000000'],
'data_type': data_type,
'controlled_access': True,
'availability': 'Immediate Download',
'is_harmonized': False,
'biospecimen_id': 'BS_00000000',
'hashes': {
'etag': 'd41d8cd98f00b204e9800998ecf8427e'
},
'size': 1024,
'urls': ['s3://{}/{}'.format(SOURCE_BUCKET, SOURCE_OBJECT)]
}
assert req.post.call_count == 1
called_with = req.post.call_args_list[0][1]['json']
assert called_with['file_format'] == file_format
assert called_with['data_type'] == data_type
def test_new_file_bs_id():
""" Test that new source file registers with a biospecimen """
mock = patch('service.requests')
req = mock.start()
mock_resp = MagicMock()
mock_resp.status_code = 201
mock_resp.json.return_value = {'results': {'kf_id': 'GF_00000001'}}
req.post.return_value = mock_resp
importer = service.FileImporter('http://api.com/', 'abc123')
res = importer.new_file(SOURCE_BUCKET,
SOURCE_OBJECT,
'd41d8cd98f00b204e9800998ecf8427e',
1024,
gf_id='GF_00000001',
bs_id='BS_00000000',
study_id='SD_00000000')
expected = {
'kf_id': 'GF_00000001',
'file_name': 'bd042b24ae844a57ace28cf70cb3c852.bam.bai',
'file_format': 'bai',
'acl': ['SD_00000000'],
'data_type': 'Aligned Reads Index',
'controlled_access': True,
'availability': 'Immediate Download',
'is_harmonized': False,
'biospecimen_id': 'BS_00000000',
'hashes': {
'etag': 'd41d8cd98f00b204e9800998ecf8427e'
},
'size': 1024,
'urls': ['s3://{}/{}'.format(SOURCE_BUCKET, SOURCE_OBJECT)]
}
req.post.assert_called_with('http://api.com/genomic-files', json=expected)
assert req.post.call_count == 1
def test_get_gf_id_tag_no_tag():
""" Test getting gf_id tag if its not in the tags """
importer = service.FileImporter('http://api.com/', 'abc123')
tags = {'fake': 'tag'}
res = importer.get_gf_id_tag(tags)
assert res == None