def model_run(CLIENT):
ds = CLIENT.create_dataset(TEST_DATASET_NAME)
ds_items = []
for url in TEST_IMG_URLS[:2]:
ds_items.append(
DatasetItem(
image_location=url,
reference_id=reference_id_from_url(url),
))
response = ds.append(ds_items)
assert ERROR_PAYLOAD not in response.json()
model = CLIENT.add_model(name=TEST_MODEL_NAME,
reference_id="model_" + str(time.time()))
run = model.create_run(name=TEST_MODEL_RUN, dataset=ds, predictions=[])
yield run
response = CLIENT.delete_dataset(ds.id)
assert response == {"message": "Beginning dataset deletion..."}
response = CLIENT.delete_model(model.id)
assert response == {}
def test_slice_append(dataset):
# Dataset upload
ds_items = []
for url in TEST_IMG_URLS:
ds_items.append(
DatasetItem(
image_location=url,
reference_id=reference_id_from_url(url),
))
response = dataset.append(ds_items)
assert ERROR_PAYLOAD not in response.json()
# Slice creation
slc = dataset.create_slice(
name=TEST_SLICE_NAME,
reference_ids=[ds_items[0].reference_id],
)
# Insert duplicate first item
slc.append(reference_ids=[item.reference_id for item in ds_items[:3]])
response = slc.info()
assert len(response["dataset_items"]) == 3
for item in ds_items[:3]:
assert (item.reference_id == response["dataset_items"][0]["ref_id"]
or item.reference_id == response["dataset_items"][1]["ref_id"]
or item.reference_id == response["dataset_items"][2]["ref_id"])
def test_slice_create_and_delete_and_list(dataset):
# Dataset upload
ds_items = []
for url in TEST_IMG_URLS:
ds_items.append(
DatasetItem(
image_location=url,
reference_id=reference_id_from_url(url),
))
response = dataset.append(ds_items)
assert ERROR_PAYLOAD not in response.json()
# Slice creation
slc = dataset.create_slice(
name=TEST_SLICE_NAME,
reference_ids=[item.reference_id for item in ds_items[:2]],
)
dataset_slices = dataset.slices
assert len(dataset_slices) == 1
assert slc.slice_id == dataset_slices[0]
response = slc.info()
assert response["name"] == TEST_SLICE_NAME
assert response["dataset_id"] == dataset.id
assert len(response["dataset_items"]) == 2
for item in ds_items[:2]:
assert (item.reference_id == response["dataset_items"][0]["ref_id"]
or item.reference_id == response["dataset_items"][1]["ref_id"])
def test_dataset_append_local(CLIENT, dataset):
ds_items_local = [DatasetItem(image_location=LOCAL_FILENAME)]
response = dataset.append(ds_items_local)
assert isinstance(response, UploadResponse)
resp_json = response.json()
assert resp_json[DATASET_ID_KEY] == dataset.id
assert resp_json[NEW_ITEMS] == 1
assert resp_json[UPDATED_ITEMS] == 0
assert resp_json[IGNORED_ITEMS] == 0
assert resp_json[ERROR_ITEMS] == 0
assert ERROR_PAYLOAD not in resp_json
def test_append_with_special_chars(dataset):
url = TEST_IMG_URLS[0]
ref_id = "test/reference/id"
ds_items = [
DatasetItem(
image_location=url,
reference_id=ref_id,
metadata={"test": "metadata"},
),
]
dataset.append(ds_items)
dataset.refloc(ref_id)
def test_dataset_update_metadata(dataset):
dataset.append([
DatasetItem(
image_location=TEST_IMG_URLS[0],
metadata={"snake_field": 0},
reference_id="test_image",
)
])
dataset.append(
[
DatasetItem(
image_location=TEST_IMG_URLS[0],
metadata={"snake_field": 1},
reference_id="test_image",
)
],
update=True,
)
resulting_item = dataset.iloc(0)["item"]
print(resulting_item)
assert resulting_item.metadata["snake_field"] == 1
def test_dataset_append_local(CLIENT, dataset):
ds_items_local_error = [
DatasetItem(image_location=LOCAL_FILENAME, metadata={"test": math.nan})
]
with pytest.raises(ValueError) as e:
dataset.append(ds_items_local_error)
assert "Out of range float values are not JSON compliant" in str(
e.value)
ds_items_local = [
DatasetItem(image_location=LOCAL_FILENAME, metadata={"test": 0})
]
response = dataset.append(ds_items_local)
assert isinstance(response, UploadResponse)
resp_json = response.json()
assert resp_json[DATASET_ID_KEY] == dataset.id
assert resp_json[NEW_ITEMS] == 1
assert resp_json[UPDATED_ITEMS] == 0
assert resp_json[IGNORED_ITEMS] == 0
assert resp_json[ERROR_ITEMS] == 0
assert ERROR_PAYLOAD not in resp_json
def test_dataset_append(dataset):
def check_is_expected_response(response):
assert isinstance(response, UploadResponse)
resp_json = response.json()
assert resp_json[DATASET_ID_KEY] == dataset.id
assert resp_json[NEW_ITEMS] == len(TEST_IMG_URLS)
assert resp_json[UPDATED_ITEMS] == 0
assert resp_json[IGNORED_ITEMS] == 0
assert resp_json[ERROR_ITEMS] == 0
assert ERROR_PAYLOAD not in resp_json
# Plain image upload
ds_items_plain = []
for url in TEST_IMG_URLS:
ds_items_plain.append(DatasetItem(image_location=url))
response = dataset.append(ds_items_plain)
check_is_expected_response(response)
# With reference ids and metadata:
ds_items_with_metadata = []
for i, url in enumerate(TEST_IMG_URLS):
ds_items_with_metadata.append(
DatasetItem(
image_location=url,
reference_id=reference_id_from_url(url),
metadata={
"made_with_pytest": True,
"example_int": i,
"example_str": "hello",
"example_float": 0.5,
"example_dict": {
"nested": True,
},
"example_list": ["hello", i, False],
},
))
response = dataset.append(ds_items_with_metadata)
check_is_expected_response(response)
def dataset(CLIENT):
ds = CLIENT.create_dataset(TEST_DATASET_NAME)
ds_items = []
for url in TEST_IMG_URLS:
ds_items.append(
DatasetItem(
image_location=url,
reference_id=reference_id_from_url(url),
))
response = ds.append(ds_items)
assert ERROR_PAYLOAD not in response.json()
yield ds
response = CLIENT.delete_dataset(ds.id)
assert response == {"message": "Beginning dataset deletion..."}
def test_append_and_export(dataset):
# Dataset upload
url = TEST_IMG_URLS[0]
box_annotation = BoxAnnotation(**TEST_BOX_ANNOTATIONS[0])
segmentation_annotation = SegmentationAnnotation.from_json(
TEST_SEGMENTATION_ANNOTATIONS[0])
polygon_annotation = PolygonAnnotation.from_json(
TEST_POLYGON_ANNOTATIONS[0])
ds_items = [
DatasetItem(
image_location=url,
reference_id=reference_id_from_url(url),
metadata={"test": "metadata"},
),
]
response = dataset.append(ds_items)
assert ERROR_PAYLOAD not in response.json()
dataset.annotate(annotations=[
box_annotation,
polygon_annotation,
segmentation_annotation,
])
# We don't export everything on the annotations in order to speed up export.
def clear_fields(annotation):
cleared_annotation = copy.deepcopy(annotation)
cleared_annotation.annotation_id = None
cleared_annotation.metadata = {}
return cleared_annotation
def sort_labelmap(segmentation_annotation):
segmentation_annotation.annotations = sorted(
segmentation_annotation.annotations, key=lambda x: x.index)
exported = dataset.items_and_annotations()
assert exported[0][ITEM_KEY] == ds_items[0]
assert exported[0][ANNOTATIONS_KEY][BOX_TYPE][0] == clear_fields(
box_annotation)
assert sort_labelmap(
exported[0][ANNOTATIONS_KEY][SEGMENTATION_TYPE]) == sort_labelmap(
clear_fields(segmentation_annotation))
assert exported[0][ANNOTATIONS_KEY][POLYGON_TYPE][0] == clear_fields(
polygon_annotation)
def make_dataset_items():
ds_items_with_metadata = []
for i, url in enumerate(TEST_IMG_URLS):
ds_items_with_metadata.append(
DatasetItem(
image_location=url,
reference_id=reference_id_from_url(url),
metadata={
"made_with_pytest": True,
"example_int": i,
"example_str": "hello",
"example_float": 0.5,
"example_dict": {
"nested": True,
},
"example_list": ["hello", i, False],
},
))
return ds_items_with_metadata
def test_slice_append(dataset):
# Dataset upload
ds_items = []
for url in TEST_IMG_URLS:
ds_items.append(
DatasetItem(
image_location=url,
reference_id=reference_id_from_url(url),
))
response = dataset.append(ds_items)
assert ERROR_PAYLOAD not in response.json()
# Slice creation
slc = dataset.create_slice(
name=TEST_SLICE_NAME,
reference_ids=[ds_items[0].reference_id],
)
# Insert duplicate first item
slc.append(reference_ids=[item.reference_id for item in ds_items[:3]])
response = slc.info()
assert len(response["dataset_items"]) == 3
for item in ds_items[:3]:
assert (item.reference_id == response["dataset_items"][0]["ref_id"]
or item.reference_id == response["dataset_items"][1]["ref_id"]
or item.reference_id == response["dataset_items"][2]["ref_id"])
all_stored_items = [_[ITEM_KEY] for _ in slc.items_and_annotations()]
def sort_by_reference_id(items):
# Remove the generated item_ids and standardize
# empty metadata so we can do an equality check.
for item in items:
item.item_id = None
if item.metadata == {}:
item.metadata = None
return sorted(items, key=lambda x: x.reference_id)
assert sort_by_reference_id(all_stored_items) == sort_by_reference_id(
ds_items[:3])
def test_reprs():
# Have to define here in order to have access to all relevant objects
def test_repr(test_object: any):
assert eval(str(test_object)) == test_object
test_repr(
DatasetItem(
image_location="test_url",
reference_id="test_reference_id",
metadata={
"made_with_pytest": True,
"example_int": 0,
"example_str": "hello",
"example_float": 0.5,
"example_dict": {
"nested": True,
},
"example_list": ["hello", 1, False],
},
))
test_repr(Dataset("test_dataset", NucleusClient(api_key="fake_key")))
def test_dataset_append(dataset):
def check_is_expected_response(response):
assert isinstance(response, UploadResponse)
resp_json = response.json()
assert resp_json[DATASET_ID_KEY] == dataset.id
assert resp_json[NEW_ITEMS] == len(TEST_IMG_URLS)
assert resp_json[UPDATED_ITEMS] == 0
assert resp_json[IGNORED_ITEMS] == 0
assert resp_json[ERROR_ITEMS] == 0
assert ERROR_PAYLOAD not in resp_json
# Plain image upload
ds_items_plain = []
for url in TEST_IMG_URLS:
ds_items_plain.append(DatasetItem(image_location=url))
response = dataset.append(ds_items_plain)
check_is_expected_response(response)
# With reference ids and metadata:
response = dataset.append(make_dataset_items())
check_is_expected_response(response)
def test_slice_send_to_labeling(dataset):
# Dataset upload
ds_items = []
for url in TEST_IMG_URLS:
ds_items.append(
DatasetItem(
image_location=url,
reference_id=reference_id_from_url(url),
))
response = dataset.append(ds_items)
assert ERROR_PAYLOAD not in response.json()
# Slice creation
slc = dataset.create_slice(
name=(TEST_SLICE_NAME + str(uuid.uuid4())[-10:]),
reference_ids=[ds_items[0].reference_id, ds_items[1].reference_id],
)
response = slc.info()
assert len(response["dataset_items"]) == 2
response = slc.send_to_labeling(TEST_PROJECT_ID)
assert isinstance(response, AsyncJob)
TEST_SLICE_NAME = "[PyTest] Test Slice"
TEST_MODEL_NAME = "[PyTest] Test Model Name"
TEST_MODEL_REFERENCE = "[PyTest] Test Model Reference"
TEST_MODEL_RUN = "[PyTest] Test Model Run Reference"
TEST_DATASET_NAME = "[PyTest] Test Dataset"
TEST_SLICE_NAME = "[PyTest] Test Slice"
TEST_IMG_URLS = [
"s3://scaleapi-cust-lidar/Hesai/raw_data/2019-5-11/hesai_data_1557540003/undistorted/front_camera/1557540143.650423lf.jpg",
"s3://scaleapi-cust-lidar/Hesai/raw_data/2019-5-11/hesai_data_1557540003/undistorted/back_camera/1557540143.600352lf.jpg",
"s3://scaleapi-cust-lidar/Hesai/raw_data/2019-5-11/hesai_data_1557540003/undistorted/right_camera/1557540143.681730lf.jpg",
"s3://scaleapi-cust-lidar/Hesai/raw_data/2019-5-11/hesai_data_1557540003/undistorted/front_left_camera/1557540143.639619lf.jpg",
"s3://scaleapi-cust-lidar/Hesai/raw_data/2019-5-11/hesai_data_1557540003/undistorted/front_right_camera/1557540143.661212lf.jpg",
]
TEST_DATASET_ITEMS = [
DatasetItem(TEST_IMG_URLS[0], "1"),
DatasetItem(TEST_IMG_URLS[1], "2"),
DatasetItem(TEST_IMG_URLS[2], "3"),
DatasetItem(TEST_IMG_URLS[3], "4"),
]
LOCAL_FILENAME = "tests/test_img.jpg"
TEST_PREDS = [
BoxPrediction("[Pytest Box Prediction 1]", 0, 0, 100, 100, "1"),
BoxPrediction("[Pytest Box Prediction 2]", 0, 0, 100, 100, "2"),
BoxPrediction("[Pytest Box Prediction 3]", 0, 0, 100, 100, "3"),
BoxPrediction("[Pytest Box Prediction 4]", 0, 0, 100, 100, "4"),
]
def get_signed_url(url):
