def test_generate_dataframe_when_aggregation_nodepoint_with_many_entries_same_subkey(
monkeypatch):
content_list = [
'''[{ "id": "chain_1", "shops": [
{"id": "shop_1", "name": "shopname_1"}
] }]''', '''[
{ "sales": [
{ "billing": 111 },
{ "billing": 222 },
{ "billing": 333 }
] }
]'''
]
nodepoint_specs = [{
'name': 'test',
'type': 'aggregation',
'aggregation_key': 'billing',
'column_suffix': 'billing',
'subkey': 'sales'
}]
mock_request = build_mock_request(content_list)
monkeypatch.setattr(requests, 'request', mock_request)
expected = pd.DataFrame([
['chain_1', 'shop_1', 'shopname_1', 3, 666, 0],
],
columns=[
'chain_id', 'shop_id', 'shop_name',
'test_count', 'test_billing', 'test_malformed'
])
found = generate_dataframe(nodepoint_specs)
pd.testing.assert_frame_equal(expected, found, check_dtype=False)
def test_generate_dataframe_when_one_shop_one_nodepoint(monkeypatch):
content_list = [
'[{ "id": "chain_1", "shops": [ {"id": "shop_id_1", "name":"shop_name_1"} ] }]',
'''[
{ "originalId": "oid1", "name":"name1" },
{ "originalId": "oid1", "name":"name1" },
{ "id": "badentry" },
{ "originalId": "oid2", "name":"name2" }
]'''
]
nodepoint_specs = [
{
"name": "sellers",
"type": "raw",
"column_suffix": "distinct",
"equality_key": "originalId"
},
]
mock_request = build_mock_request(content_list)
monkeypatch.setattr(requests, 'request', mock_request)
expected = pd.DataFrame([['chain_1', 'shop_id_1', 'shop_name_1', 3, 2, 1]],
columns=[
'chain_id',
'shop_id',
'shop_name',
'sellers_count',
'sellers_distinct',
'sellers_malformed',
])
found = generate_dataframe(nodepoint_specs)
pd.testing.assert_frame_equal(expected, found, check_dtype=False)
def test_generate_dataframe_when_no_shops(monkeypatch):
content_list = ['[{ "id": "anychain", "shops": [] }]']
nodepoint_specs = [
{
"name": "products",
"type": "raw",
"column_suffix": "distinct",
"equality_key": "originalId"
},
{
"name": "sellers",
"type": "raw",
"column_suffix": "distinct",
"equality_key": "originalId"
},
]
mock_request = build_mock_request(content_list)
monkeypatch.setattr(requests, 'request', mock_request)
expected = pd.DataFrame(columns=[
'chain_id',
'shop_id',
'shop_name',
'products_count',
'products_distinct',
'products_malformed',
'sellers_count',
'sellers_distinct',
'sellers_malformed',
])
found = generate_dataframe(nodepoint_specs)
pd.testing.assert_frame_equal(expected, found, check_dtype=False)
def test_generate_dataframe_when_response_error(monkeypatch):
def mock_request(*args, **kwargs):
return MockResponse(status_code=400)
monkeypatch.setattr(requests, 'request', mock_request)
nodepoint_specs = [
{
"name": "products",
"type": "raw",
"column_suffix": "distinct",
"equality_key": "originalId"
},
{
"name": "sellers",
"type": "raw",
"column_suffix": "distinct",
"equality_key": "originalId"
},
]
expected = pd.DataFrame(columns=[
'chain_id',
'shop_id',
'shop_name',
'products_count',
'products_distinct',
'products_malformed',
'sellers_count',
'sellers_distinct',
'sellers_malformed',
])
found = generate_dataframe(nodepoint_specs)
pd.testing.assert_frame_equal(expected, found, check_dtype=False)
def test_generate_dataframe_when_raw_and_aggregation_many_nodepoints_many_shops_and_chains(
monkeypatch):
content_list = [
'''[{ "id": "chain_1", "shops": [
{"id": "shop_1", "name": "shopname_1"},
{"id": "shop_2", "name": "shopname_2"}
] },
{ "id": "chain_2", "shops": [
{"id": "shop_3", "name": "shopname_3"}
]}
]''',
'''[
{ "whatis": "for chain_1 shop_1 test_1", "sales": [
{ "billing": 1 }
] },
{ "whatis": "for chain_1 shop_1 test_1", "sales": [
{ "billing": 10 }
] },
{ "whatis": "for chain_1 shop_1 test_1", "sales": [
{ "billing": 100 }
] }
]''',
'''[
{ "whatis": "for chain_1 shop_1 test_2", "originalId": "oid1"},
{ "whatis": "for chain_1 shop_1 test_2", "originalId": "oid2"},
{ "whatis": "for chain_1 shop_1 test_2", "originalId": "oid2"},
{ "whatis": "for chain_1 shop_1 test_2", "unexpected key": "oid3"}
]''',
'''[
{ "whatis": "for chain_1 shop_1 test_3","billing": 111 },
{ "whatis": "for chain_1 shop_1 test_3","billing": 222 },
{ "whatis": "for chain_1 shop_1 test_3","non expected": 555 }
]''',
'''[
{ "whatis": "for chain_1 shop_2 test_1", "sales": [
{ "billing": 1000 }
] }
]''',
'''[
{ "whatis": "for chain_1 shop_2 test_2", "originalId": "oid122"}
]''',
'''[
{ "whatis": "for chain_1 shop_2 test_3",
"billing": 3000 }
]''',
'''[
{ "whatis": "for chain_2 shop_3 test_1", "sales": [
{ "billing": 10000 }
] }
]''',
'''[
{ "whatis": "for chain_1 shop_3 test_2", "originalId": "oid232"}
]''',
'''[
{ "whatis": "for chain_2 shop_3 test_3",
"billing": 30000 }
]''',
]
nodepoint_specs = [{
'name': 'test_1',
'type': 'aggregation',
'aggregation_key': 'billing',
'column_suffix': 'billing',
'subkey': 'sales'
}, {
'name': 'test_2',
'type': 'raw',
'equality_key': 'originalId',
'column_suffix': 'distinct'
}, {
'name': 'test_3',
'type': 'aggregation',
'aggregation_key': 'billing',
'column_suffix': 'billing',
'subkey': None
}]
mock_request = build_mock_request(content_list)
monkeypatch.setattr(requests, 'request', mock_request)
expected = pd.DataFrame([[
'chain_1', 'shop_1', 'shopname_1', 3, 111, 0, 3, 2, 1, 2, 333, 1
], [
'chain_1', 'shop_2', 'shopname_2', 1, 1000, 0, 1, 1, 0, 1, 3000, 0
], ['chain_2', 'shop_3', 'shopname_3', 1, 10000, 0, 1, 1, 0, 1, 30000, 0]],
columns=[
'chain_id', 'shop_id', 'shop_name',
'test_1_count', 'test_1_billing',
'test_1_malformed', 'test_2_count',
'test_2_distinct', 'test_2_malformed',
'test_3_count', 'test_3_billing',
'test_3_malformed'
])
found = generate_dataframe(nodepoint_specs)
pd.testing.assert_frame_equal(expected, found, check_dtype=False)
def test_generate_dataframe_when_aggregation_mani_nodepoints_same_shop(
monkeypatch):
content_list = [
'''[{ "id": "chain_1", "shops": [
{"id": "shop_1", "name": "shopname_1"}
] }]''', '''[
{ "sales": [
{ "billing": 1 }
] },
{ "sales": [
{ "billing": 10 }
] },
{ "sales": [
{ "billing": 100 }
] }
]''', '''[
{ "sales": [
{ "billing": 2 }
] },
{ "sales": [
{ "billing": 200 }
] }
]''', '''[
{ "billing": 111 },
{ "billing": 222 },
{ "non expected": 555 }
]'''
]
nodepoint_specs = [{
'name': 'test_1',
'type': 'aggregation',
'aggregation_key': 'billing',
'column_suffix': 'billing',
'subkey': 'sales'
}, {
'name': 'test_2',
'type': 'aggregation',
'aggregation_key': 'billing',
'column_suffix': 'billing',
'subkey': 'sales'
}, {
'name': 'test_3',
'type': 'aggregation',
'aggregation_key': 'billing',
'column_suffix': 'billing',
'subkey': None
}]
mock_request = build_mock_request(content_list)
monkeypatch.setattr(requests, 'request', mock_request)
expected = pd.DataFrame([
['chain_1', 'shop_1', 'shopname_1', 3, 111, 0, 2, 202, 0, 2, 333, 1],
],
columns=[
'chain_id', 'shop_id', 'shop_name',
'test_1_count', 'test_1_billing',
'test_1_malformed', 'test_2_count',
'test_2_billing', 'test_2_malformed',
'test_3_count', 'test_3_billing',
'test_3_malformed'
])
found = generate_dataframe(nodepoint_specs)
pd.testing.assert_frame_equal(expected, found, check_dtype=False)
