scale-nucleus 0.1.24__py3-none-any.whl → 0.6.4__py3-none-any.whl

nucleus/utils.py

@@ -1,10 +1,10 @@
 """Shared stateless utility function library"""
 
-from collections import defaultdict
 import io
-import uuid
 import json
-from typing import IO, Dict, List, Sequence, Union
+import uuid
+from collections import defaultdict
+from typing import IO, Dict, List, Sequence, Type, Union
 
 import requests
 from requests.models import HTTPError
@@ -12,9 +12,10 @@ from requests.models import HTTPError
 from nucleus.annotation import (
     Annotation,
     BoxAnnotation,
+    CategoryAnnotation,
     CuboidAnnotation,
+    MultiCategoryAnnotation,
     PolygonAnnotation,
-    CategoryAnnotation,
     SegmentationAnnotation,
 )
 
@@ -22,55 +23,126 @@ from .constants import (
     ANNOTATION_TYPES,
     ANNOTATIONS_KEY,
     BOX_TYPE,
-    CUBOID_TYPE,
     CATEGORY_TYPE,
+    CUBOID_TYPE,
     ITEM_KEY,
+    MULTICATEGORY_TYPE,
     POLYGON_TYPE,
     REFERENCE_ID_KEY,
     SEGMENTATION_TYPE,
 )
 from .dataset_item import DatasetItem
-from .prediction import BoxPrediction, CuboidPrediction, PolygonPrediction
+from .prediction import (
+    BoxPrediction,
+    CategoryPrediction,
+    CuboidPrediction,
+    PolygonPrediction,
+    SegmentationPrediction,
+)
 from .scene import LidarScene
 
+STRING_REPLACEMENTS = {
+    "\\\\n": "\n",
+    "\\\\t": "\t",
+    '\\\\"': '"',
+}
 
-def _get_all_field_values(metadata_list: List[dict], key: str):
-    return {metadata[key] for metadata in metadata_list if key in metadata}
 
+class KeyErrorDict(dict):
+    """Wrapper for response dicts with deprecated keys.
 
-def suggest_metadata_schema(
-    data: Union[
-        List[DatasetItem],
-        List[BoxPrediction],
-        List[PolygonPrediction],
-        List[CuboidPrediction],
-    ]
-):
-    metadata_list: List[dict] = [
-        d.metadata for d in data if d.metadata is not None
-    ]
-    schema = {}
-    all_keys = {k for metadata in metadata_list for k in metadata.keys()}
-
-    all_key_values: Dict[str, set] = {
-        k: _get_all_field_values(metadata_list, k) for k in all_keys
-    }
+    Parameters:
+        **kwargs: Mapping from the deprecated key to a warning message.
+    """
+
+    def __init__(self, **kwargs):
+        self._deprecated = {}
+
+        for key, msg in kwargs.items():
+            if not isinstance(key, str):
+                raise TypeError(
+                    f"All keys must be strings! Received non-string '{key}'"
+                )
+            if not isinstance(msg, str):
+                raise TypeError(
+                    f"All warning messages must be strings! Received non-string '{msg}'"
+                )
+
+            self._deprecated[key] = msg
 
-    for key, values in all_key_values.items():
-        entry: dict = {}
-        if all(isinstance(x, (float, int)) for x in values):
-            entry["type"] = "number"
-        elif len(values) <= 50:
-            entry["type"] = "category"
-            entry["choices"] = list(values)
-        else:
-            entry["type"] = "text"
-        schema[key] = entry
-    return schema
+        super().__init__()
+
+    def __missing__(self, key):
+        """Raises KeyError for deprecated keys, otherwise uses base dict logic."""
+        if key in self._deprecated:
+            raise KeyError(self._deprecated[key])
+        try:
+            super().__missing__(key)
+        except AttributeError as e:
+            raise KeyError(key) from e
+
+
+def format_prediction_response(
+    response: dict,
+) -> Union[
+    dict,
+    List[
+        Union[
+            BoxPrediction,
+            PolygonPrediction,
+            CuboidPrediction,
+            CategoryPrediction,
+            SegmentationPrediction,
+        ]
+    ],
+]:
+    """Helper function to convert JSON response from endpoints to python objects
+
+    Args:
+        response: JSON dictionary response from REST endpoint.
+    Returns:
+        annotation_response: Dictionary containing a list of annotations for each type,
+            keyed by the type name.
+    """
+    annotation_payload = response.get(ANNOTATIONS_KEY, None)
+    if not annotation_payload:
+        # An error occurred
+        return response
+    annotation_response = {}
+    type_key_to_class: Dict[
+        str,
+        Union[
+            Type[BoxPrediction],
+            Type[PolygonPrediction],
+            Type[CuboidPrediction],
+            Type[CategoryPrediction],
+            Type[SegmentationPrediction],
+        ],
+    ] = {
+        BOX_TYPE: BoxPrediction,
+        POLYGON_TYPE: PolygonPrediction,
+        CUBOID_TYPE: CuboidPrediction,
+        CATEGORY_TYPE: CategoryPrediction,
+        SEGMENTATION_TYPE: SegmentationPrediction,
+    }
+    for type_key in annotation_payload:
+        type_class = type_key_to_class[type_key]
+        annotation_response[type_key] = [
+            type_class.from_json(annotation)
+            for annotation in annotation_payload[type_key]
+        ]
+    return annotation_response
 
 
 def format_dataset_item_response(response: dict) -> dict:
-    """Format the raw client response into api objects."""
+    """Format the raw client response into api objects.
+
+    Args:
+      response: JSON dictionary response from REST endpoint
+    Returns:
+      item_dict: A dictionary with two entries, one for the dataset item, and annother
+        for all of the associated annotations.
+    """
     if ANNOTATIONS_KEY not in response:
         raise ValueError(
             f"Server response was missing the annotation key: {response}"
@@ -96,6 +168,15 @@ def format_dataset_item_response(response: dict) -> dict:
 
 
 def convert_export_payload(api_payload):
+    """Helper function to convert raw JSON to API objects
+
+    Args:
+        api_payload: JSON dictionary response from REST endpoint
+    Returns:
+        return_payload: A list of dictionaries for each dataset item. Each dictionary
+            is in the same format as format_dataset_item_response: one key for the
+            dataset item, another for the annotations.
+    """
     return_payload = []
     for row in api_payload:
         return_payload_row = {}
@@ -123,6 +204,11 @@ def convert_export_payload(api_payload):
             annotations[CATEGORY_TYPE].append(
                 CategoryAnnotation.from_json(category)
             )
+        for multicategory in row[MULTICATEGORY_TYPE]:
+            multicategory[REFERENCE_ID_KEY] = row[ITEM_KEY][REFERENCE_ID_KEY]
+            annotations[MULTICATEGORY_TYPE].append(
+                MultiCategoryAnnotation.from_json(multicategory)
+            )
         return_payload_row[ANNOTATIONS_KEY] = annotations
         return_payload.append(return_payload_row)
     return return_payload
@@ -132,6 +218,10 @@ def serialize_and_write(
     upload_units: Sequence[Union[DatasetItem, Annotation, LidarScene]],
     file_pointer,
 ):
+    if len(upload_units) == 0:
+        raise ValueError(
+            "Expecting at least one object when serializing objects to upload, but got zero.  Please try again."
+        )
     for unit in upload_units:
         try:
             if isinstance(unit, (DatasetItem, Annotation, LidarScene)):
@@ -168,6 +258,7 @@ def serialize_and_write_to_presigned_url(
     dataset_id: str,
     client,
 ):
+    """This helper function can be used to serialize a list of API objects to NDJSON."""
     request_id = uuid.uuid4().hex
     response = client.make_request(
         payload={},
@@ -180,3 +271,9 @@ def serialize_and_write_to_presigned_url(
     strio.seek(0)
     upload_to_presigned_url(response["signed_url"], strio)
     return request_id
+
+
+def replace_double_slashes(s: str) -> str:
+    for key, val in STRING_REPLACEMENTS.items():
+        s = s.replace(key, val)
+    return s

nucleus/validate/__init__.py

@@ -0,0 +1,24 @@
+"""Model CI Python Library."""
+
+__all__ = [
+    "Validate",
+    "ScenarioTest",
+    "EvaluationCriterion",
+]
+
+from .client import Validate
+from .constants import ThresholdComparison
+from .data_transfer_objects.eval_function import (
+    EvalFunctionEntry,
+    EvaluationCriterion,
+    GetEvalFunctions,
+)
+from .data_transfer_objects.scenario_test import CreateScenarioTestRequest
+from .errors import CreateScenarioTestError
+from .eval_functions.available_eval_functions import AvailableEvalFunctions
+from .scenario_test import ScenarioTest
+from .scenario_test_evaluation import (
+    ScenarioTestEvaluation,
+    ScenarioTestItemEvaluation,
+)
+from .scenario_test_metric import ScenarioTestMetric

nucleus/validate/client.py

@@ -0,0 +1,168 @@
+from typing import List
+
+from nucleus.connection import Connection
+from nucleus.job import AsyncJob
+
+from .constants import SCENARIO_TEST_ID_KEY
+from .data_transfer_objects.eval_function import (
+    EvaluationCriterion,
+    GetEvalFunctions,
+)
+from .data_transfer_objects.scenario_test import CreateScenarioTestRequest
+from .errors import CreateScenarioTestError
+from .eval_functions.available_eval_functions import AvailableEvalFunctions
+from .scenario_test import ScenarioTest
+
+SUCCESS_KEY = "success"
+EVAL_FUNCTIONS_KEY = "eval_functions"
+
+
+class Validate:
+    """Model CI Python Client extension."""
+
+    def __init__(self, api_key: str, endpoint: str):
+        self.connection = Connection(api_key, endpoint)
+
+    def __repr__(self):
+        return f"Validate(connection='{self.connection}')"
+
+    def __eq__(self, other):
+        return self.connection == other.connection
+
+    @property
+    def eval_functions(self) -> AvailableEvalFunctions:
+        """List all available evaluation functions which can be used to set up evaluation criteria.::
+
+            import nucleus
+            client = nucleus.NucleusClient("YOUR_SCALE_API_KEY")
+
+            scenario_test_criterion = client.validate.eval_functions.bbox_iou() > 0.5  # Creates an EvaluationCriterion by comparison
+
+        Returns:
+            :class:`AvailableEvalFunctions`: A container for all the available eval functions
+        """
+        response = self.connection.get(
+            "validate/eval_fn",
+        )
+        payload = GetEvalFunctions.parse_obj(response)
+        return AvailableEvalFunctions(payload.eval_functions)
+
+    def create_scenario_test(
+        self,
+        name: str,
+        slice_id: str,
+        evaluation_criteria: List[EvaluationCriterion],
+    ) -> ScenarioTest:
+        """Creates a new Scenario Test from an existing Nucleus :class:`Slice`:. ::
+
+            import nucleus
+            client = nucleus.NucleusClient("YOUR_SCALE_API_KEY")
+
+            scenario_test = client.validate.create_scenario_test(
+                name="sample_scenario_test",
+                slice_id="YOUR_SLICE_ID",
+                evaluation_criteria=[client.validate.eval_functions.bbox_iou() > 0.5]
+            )
+
+        Args:
+            name: unique name of test
+            slice_id: id of (pre-defined) slice of items to evaluate test on.
+            evaluation_criteria: :class:`EvaluationCriterion` defines a pass/fail criteria for the test. Created with a
+                comparison with an eval functions. See :class:`eval_functions`.
+
+        Returns:
+            Created ScenarioTest object.
+        """
+        if not evaluation_criteria:
+            raise CreateScenarioTestError(
+                "Must pass an evaluation_criteria to the scenario test! I.e. "
+                "evaluation_criteria = [client.validate.eval_functions.bbox_iou() > 0.5]"
+            )
+        response = self.connection.post(
+            CreateScenarioTestRequest(
+                name=name,
+                slice_id=slice_id,
+                evaluation_criteria=evaluation_criteria,
+            ).dict(),
+            "validate/scenario_test",
+        )
+        return ScenarioTest(response[SCENARIO_TEST_ID_KEY], self.connection)
+
+    def get_scenario_test(self, scenario_test_id: str) -> ScenarioTest:
+        response = self.connection.get(
+            f"validate/scenario_test/{scenario_test_id}",
+        )
+        return ScenarioTest(response["id"], self.connection)
+
+    def list_scenario_tests(self) -> List[ScenarioTest]:
+        """Lists all Scenario Tests of the current user. ::
+
+            import nucleus
+            client = nucleus.NucleusClient("YOUR_SCALE_API_KEY")
+            scenario_test = client.validate.create_scenario_test(
+                "sample_scenario_test", "slc_bx86ea222a6g057x4380"
+            )
+
+            client.validate.list_scenario_tests()
+
+        Returns:
+            A list of ScenarioTest objects.
+        """
+        response = self.connection.get(
+            "validate/scenario_test",
+        )
+        return [
+            ScenarioTest(test_id, self.connection)
+            for test_id in response["scenario_test_ids"]
+        ]
+
+    def delete_scenario_test(self, scenario_test_id: str) -> bool:
+        """Deletes a Scenario Test. ::
+
+            import nucleus
+            client = nucleus.NucleusClient("YOUR_SCALE_API_KEY")
+            scenario_test = client.validate.list_scenario_tests()[0]
+
+            success = client.validate.delete_scenario_test(scenario_test.id)
+
+        Args:
+            scenario_test_id: unique ID of scenario test
+
+        Returns:
+            Whether deletion was successful.
+        """
+        response = self.connection.delete(
+            f"validate/scenario_test/{scenario_test_id}",
+        )
+        return response[SUCCESS_KEY]
+
+    def evaluate_model_on_scenario_tests(
+        self, model_id: str, scenario_test_names: List[str]
+    ) -> AsyncJob:
+        """Evaluates the given model on the specified Scenario Tests. ::
+
+            import nucleus
+            client = nucleus.NucleusClient("YOUR_SCALE_API_KEY")
+            model = client.list_models()[0]
+            scenario_test = client.validate.create_scenario_test(
+                "sample_scenario_test", "slc_bx86ea222a6g057x4380"
+            )
+
+            job = client.validate.evaluate_model_on_scenario_tests(
+                model_id=model.id,
+                scenario_test_names=["sample_scenario_test"],
+            )
+            job.sleep_until_complete() # Not required. Will block and update on status of the job.
+
+        Args:
+            model_id: ID of model to evaluate
+            scenario_test_names: list of scenario test names of test to evaluate
+
+        Returns:
+            AsyncJob object of evaluation job
+        """
+        response = self.connection.post(
+            {"test_names": scenario_test_names},
+            f"validate/{model_id}/evaluate",
+        )
+        return AsyncJob.from_json(response, self.connection)

nucleus/validate/constants.py

@@ -0,0 +1,20 @@
+from enum import Enum
+
+EVALUATION_ID_KEY = "evaluation_id"
+EVAL_FUNCTION_ID_KEY = "eval_function_id"
+ID_KEY = "id"
+PASS_KEY = "pass"
+RESULT_KEY = "result"
+THRESHOLD_COMPARISON_KEY = "threshold_comparison"
+THRESHOLD_KEY = "threshold"
+SCENARIO_TEST_ID_KEY = "scenario_test_id"
+SCENARIO_TEST_NAME_KEY = "scenario_test_name"
+
+
+class ThresholdComparison(str, Enum):
+    """Comparator between the result and the threshold."""
+
+    GREATER_THAN = "greater_than"
+    GREATER_THAN_EQUAL_TO = "greater_than_equal_to"
+    LESS_THAN = "less_than"
+    LESS_THAN_EQUAL_TO = "less_than_equal_to"

nucleus/validate/data_transfer_objects/eval_function.py

@@ -0,0 +1,81 @@
+from typing import List, Optional
+
+from pydantic import validator
+
+from ...pydantic_base import ImmutableModel
+from ..constants import ThresholdComparison
+
+
+class EvaluationCriterion(ImmutableModel):
+    """
+    An Evaluation Criterion is defined as an evaluation function, threshold, and comparator.
+    It describes how to apply an evaluation function
+
+    Notes:
+        To define the evaluation criteria for a scenario test we've created some syntactic sugar to make it look closer to an
+        actual function call, and we also hide away implementation details related to our data model that simply are not clear,
+        UX-wise.
+
+        Instead of defining criteria like this::
+
+            from nucleus.validate.data_transfer_objects.eval_function import (
+                EvaluationCriterion,
+                ThresholdComparison,
+            )
+
+            criteria = [
+                EvaluationCriterion(
+                    eval_function_id="ef_c6m1khygqk400918ays0",  # bbox_recall
+                    threshold_comparison=ThresholdComparison.GREATER_THAN,
+                    threshold=0.5,
+                ),
+            ]
+
+        we define it like this::
+
+            bbox_recall = client.validate.eval_functions.bbox_recall
+            criteria = [
+                bbox_recall() > 0.5
+            ]
+
+        The chosen method allows us to document the available evaluation functions in an IDE friendly fashion and hides away
+        details like internal IDs (`"ef_...."`).
+
+        The actual `EvaluationCriterion` is created by overloading the comparison operators for the base class of an evaluation
+        function. Instead of the comparison returning a bool, we've made it create an `EvaluationCriterion` with the correct
+        signature to send over the wire to our API.
+
+
+    Parameters:
+        eval_function_id (str): ID of evaluation function
+        threshold_comparison (:class:`ThresholdComparison`): comparator for evaluation. i.e. threshold=0.5 and threshold_comparator > implies that a test only passes if score > 0.5.
+        threshold (float): numerical threshold that together with threshold comparison, defines success criteria for test evaluation.
+    """
+
+    # TODO: Having only eval_function_id hurts readability -> Add function name
+    eval_function_id: str
+    threshold_comparison: ThresholdComparison
+    threshold: float
+
+    @validator("eval_function_id")
+    def valid_eval_function_id(cls, v):  # pylint: disable=no-self-argument
+        if not v.startswith("ef_"):
+            raise ValueError(f"Expected field to start with 'ef_', got '{v}'")
+        return v
+
+
+class EvalFunctionEntry(ImmutableModel):
+    """Encapsulates information about an evaluation function for Model CI."""
+
+    id: str
+    name: str
+    is_public: bool
+    user_id: str
+    serialized_fn: Optional[str] = None
+    raw_source: Optional[str] = None
+
+
+class GetEvalFunctions(ImmutableModel):
+    """Expected format from GET validate/eval_fn"""
+
+    eval_functions: List[EvalFunctionEntry]

nucleus/validate/data_transfer_objects/scenario_test.py

@@ -0,0 +1,19 @@
+from typing import List
+
+from pydantic import validator
+
+from nucleus.pydantic_base import ImmutableModel
+
+from .eval_function import EvaluationCriterion
+
+
+class CreateScenarioTestRequest(ImmutableModel):
+    name: str
+    slice_id: str
+    evaluation_criteria: List[EvaluationCriterion]
+
+    @validator("slice_id")
+    def startswith_slice_indicator(cls, v):  # pylint: disable=no-self-argument
+        if not v.startswith("slc_"):
+            raise ValueError(f"Expected field to start with 'slc_', got '{v}'")
+        return v

nucleus/validate/data_transfer_objects/scenario_test_evaluations.py

@@ -0,0 +1,11 @@
+from typing import List
+
+from nucleus.pydantic_base import ImmutableModel
+
+
+class EvalDetail(ImmutableModel):
+    id: str
+
+
+class GetEvalHistory(ImmutableModel):
+    evaluations: List[EvalDetail]

nucleus/validate/data_transfer_objects/scenario_test_metric.py

@@ -0,0 +1,12 @@
+from nucleus.pydantic_base import ImmutableModel
+
+from ..constants import ThresholdComparison
+
+
+class AddScenarioTestMetric(ImmutableModel):
+    """Data transfer object to add a scenario test."""
+
+    scenario_test_name: str
+    eval_function_id: str
+    threshold: float
+    threshold_comparison: ThresholdComparison

nucleus/validate/errors.py

@@ -0,0 +1,6 @@
+class CreateScenarioTestError(Exception):
+    pass
+
+
+class EvalFunctionNotAvailableError(Exception):
+    pass