Merge pull request #10 from DSD-DBS/raillabel_4

validate_schema()

pyproject.toml

@@ -40,6 +40,7 @@ dependencies = [
   "raillabel>=4.0.0",
   "pyyaml>=6.0.0",
   "numpy>=1.24.4",
+  "pydantic<3.0.0",
 ]
 
 [project.urls]

raillabel_providerkit/validation/__init__.py

@@ -3,5 +3,6 @@
 """Package for validating raillabel data regarding the format requirements."""
 
 from .validate_onthology.validate_onthology import validate_onthology
+from .validate_schema import validate_schema
 
-__all__ = ["validate_onthology"]
+__all__ = ["validate_onthology", "validate_schema"]

raillabel_providerkit/validation/validate.py

@@ -3,24 +3,16 @@
 
 from __future__ import annotations
 
-from pathlib import Path
+from . import validate_schema
 
-import raillabel
 
-from . import validate_onthology
-
-
-def validate(scene: raillabel.Scene, onthology: dict | Path) -> list[str]:
+def validate(scene_dict: dict) -> list[str]:
     """Validate a scene based on the Deutsche Bahn Requirements.
 
     Parameters
     ----------
-    scene : raillabel.Scene
-        The scene containing the annotations.
-    onthology : dict or Path
-        Onthology YAML-data or file containing a information about all classes and their
-        attributes. The onthology must adhere to the onthology_schema. If a path is provided, the
-        file is loaded as a YAML.
+    scene_dict : dict
+        The scene as a dictionary directly from `json.load()` in the raillabel format.
 
     Returns
     -------
@@ -31,6 +23,9 @@ def validate(scene: raillabel.Scene, onthology: dict | Path) -> list[str]:
     """
     errors = []
 
-    errors += validate_onthology(scene, onthology)
+    errors.extend(validate_schema(scene_dict))
+
+    if len(errors) > 0:
+        return errors
 
     return errors

raillabel_providerkit/validation/validate_schema/__init__.py

@@ -0,0 +1,8 @@
+# Copyright DB InfraGO AG and contributors
+# SPDX-License-Identifier: Apache-2.0
+
+"""Code for schema validation."""
+
+from .validate_schema import validate_schema
+
+__all__ = ["validate_schema"]

raillabel_providerkit/validation/validate_schema/validate_schema.py

@@ -0,0 +1,86 @@
+# Copyright DB InfraGO AG and contributors
+# SPDX-License-Identifier: Apache-2.0
+
+from __future__ import annotations
+
+import json
+
+from pydantic_core import ValidationError
+from raillabel.json_format import JSONScene
+
+
+def validate_schema(data: dict) -> list[str]:
+    """Validate a scene for adherence to the raillabel schema."""
+    try:
+        JSONScene(**data)
+    except ValidationError as errors:
+        return _make_errors_readable(errors)
+    else:
+        return []
+
+
+def _make_errors_readable(errors: ValidationError) -> list[str]:  # noqa: C901
+    readable_errors = []
+    for error in json.loads(errors.json()):
+        if error["type"] == "missing":
+            readable_errors.append(_convert_missing_error_to_string(error))
+        elif error["type"] == "extra_forbidden":
+            readable_errors.append(_convert_unexpected_field_error_to_string(error))
+        elif error["type"] == "literal_error":
+            readable_errors.append(_convert_literal_error_to_string(error))
+        elif error["type"] in ["bool_type", "bool_parsing"]:
+            readable_errors.append(_convert_false_type_error_to_string(error, "bool"))
+        elif error["type"] in ["int_type", "int_parsing", "int_from_float"]:
+            readable_errors.append(_convert_false_type_error_to_string(error, "int"))
+        elif error["type"] in ["decimal_type", "decimal_parsing"]:
+            readable_errors.append(_convert_false_type_error_to_string(error, "Decimal"))
+        elif error["type"] in ["string_type", "string_parsing"]:
+            readable_errors.append(_convert_false_type_error_to_string(error, "str"))
+        elif error["type"] in ["float_type", "float_parsing"]:
+            readable_errors.append(_convert_false_type_error_to_string(error, "float"))
+        elif error["type"] in ["uuid_type", "uuid_parsing"]:
+            readable_errors.append(_convert_false_type_error_to_string(error, "UUID"))
+        elif error["type"] == "too_long":
+            readable_errors.append(_convert_too_long_error_to_string(error))
+        else:
+            readable_errors.append(str(error))
+
+    return readable_errors
+
+
+def _build_error_path(loc: list[str]) -> str:
+    path = "$"
+    for part in loc:
+        path += f".{part}"
+    return path
+
+
+def _convert_missing_error_to_string(error: dict) -> str:
+    return f"{_build_error_path(error["loc"][:-1])}: required field '{error["loc"][-1]}' is missing."
+
+
+def _convert_unexpected_field_error_to_string(error: dict) -> str:
+    return f"{_build_error_path(error["loc"][:-1])}: found unexpected field '{error["loc"][-1]}'."
+
+
+def _convert_literal_error_to_string(error: dict) -> str:
+    return (
+        f"{_build_error_path(error["loc"])}: value '{error["input"]}' does not match allowed values "
+        f"({error["ctx"]["expected"]})."
+    )
+
+
+def _convert_false_type_error_to_string(error: dict, target_type: str) -> str:
+    if "[key]" in error["loc"]:
+        error_path = _build_error_path(error["loc"][:-2])
+    else:
+        error_path = _build_error_path(error["loc"])
+
+    return f"{error_path}: value '{error["input"]}' could not be interpreted " f"as {target_type}."
+
+
+def _convert_too_long_error_to_string(error: dict) -> str:
+    return (
+        f"{_build_error_path(error["loc"])}: should have length of {error["ctx"]["actual_length"]} "
+        f"but  has length of {error["ctx"]["max_length"]}."
+    )

tests/test_raillabel_providerkit/format/understand_ai/test_uai_bounding_box_2d.py

@@ -4,8 +4,8 @@
 from uuid import UUID
 
 import pytest
-import raillabel.format.understand_ai as uai_format
-from raillabel.format.understand_ai._translation import translate_class_id
+import raillabel_providerkit.format.understand_ai as uai_format
+from raillabel_providerkit.format.understand_ai._translation import translate_class_id
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/format/understand_ai/test_uai_bounding_box_3d.py

@@ -4,8 +4,8 @@
 from uuid import UUID
 
 import pytest
-import raillabel.format.understand_ai as uai_format
-from raillabel.format.understand_ai._translation import translate_class_id
+import raillabel_providerkit.format.understand_ai as uai_format
+from raillabel_providerkit.format.understand_ai._translation import translate_class_id
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/format/understand_ai/test_uai_coordinate_system.py

@@ -2,7 +2,7 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import pytest
-import raillabel.format.understand_ai as uai_format
+import raillabel_providerkit.format.understand_ai as uai_format
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/format/understand_ai/test_uai_frame.py

@@ -2,8 +2,8 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import pytest
-import raillabel.format.understand_ai as uai_format
-from raillabel._util._warning import _WarningsLogger
+import raillabel_providerkit.format.understand_ai as uai_format
+from raillabel_providerkit._util._warning import _WarningsLogger
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/format/understand_ai/test_uai_metadata.py

@@ -2,7 +2,7 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import pytest
-import raillabel.format.understand_ai as uai_format
+import raillabel_providerkit.format.understand_ai as uai_format
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/format/understand_ai/test_uai_point_3d.py

@@ -2,7 +2,7 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import pytest
-import raillabel.format.understand_ai as uai_format
+import raillabel_providerkit.format.understand_ai as uai_format
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/format/understand_ai/test_uai_polygon_2d.py

@@ -4,8 +4,8 @@
 from uuid import UUID
 
 import pytest
-import raillabel.format.understand_ai as uai_format
-from raillabel.format.understand_ai._translation import translate_class_id
+import raillabel_providerkit.format.understand_ai as uai_format
+from raillabel_providerkit.format.understand_ai._translation import translate_class_id
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/format/understand_ai/test_uai_polyline_2d.py

@@ -4,8 +4,8 @@
 from uuid import UUID
 
 import pytest
-import raillabel.format.understand_ai as uai_format
-from raillabel.format.understand_ai._translation import translate_class_id
+import raillabel_providerkit.format.understand_ai as uai_format
+from raillabel_providerkit.format.understand_ai._translation import translate_class_id
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/format/understand_ai/test_uai_quaternion.py

@@ -2,7 +2,7 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import pytest
-import raillabel.format.understand_ai as uai_format
+import raillabel_providerkit.format.understand_ai as uai_format
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/format/understand_ai/test_uai_scene.py

@@ -2,8 +2,8 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import pytest
-import raillabel.format.understand_ai as uai_format
-from raillabel._util._warning import _WarningsLogger
+import raillabel_providerkit.format.understand_ai as uai_format
+from raillabel_providerkit._util._warning import _WarningsLogger
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/format/understand_ai/test_uai_segmentation_3d.py

@@ -4,8 +4,8 @@
 from uuid import UUID
 
 import pytest
-import raillabel.format.understand_ai as uai_format
-from raillabel.format.understand_ai._translation import translate_class_id
+import raillabel_providerkit.format.understand_ai as uai_format
+from raillabel_providerkit.format.understand_ai._translation import translate_class_id
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/format/understand_ai/test_uai_sensor_reference.py

@@ -4,7 +4,7 @@
 from decimal import Decimal
 
 import pytest
-import raillabel.format.understand_ai as uai_format
+import raillabel_providerkit.format.understand_ai as uai_format
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/format/understand_ai/test_uai_size_3d.py

@@ -2,7 +2,7 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import pytest
-import raillabel.format.understand_ai as uai_format
+import raillabel_providerkit.format.understand_ai as uai_format
 
 # == Fixtures =========================
 

tests/test_raillabel_providerkit/validation/test_validate.py

@@ -5,15 +5,17 @@
 
 from raillabel_providerkit import validate
 
-# == Tests ============================
 
+def test_no_errors_in_empty_scene():
+    scene_dict = {"openlabel": {"metadata": {"schema_version": "1.0.0"}}}
+    actual = validate(scene_dict)
+    assert len(actual) == 0
 
-def test_no_errors(demo_onthology, valid_onthology_scene):
-    assert validate(valid_onthology_scene, demo_onthology) == []
 
-
-def test_onthology_errors(demo_onthology, invalid_onthology_scene):
-    assert len(validate(invalid_onthology_scene, demo_onthology)) == 1
+def test_schema_errors():
+    scene_dict = {"openlabel": {}}
+    actual = validate(scene_dict)
+    assert len(actual) == 1
 
 
 if __name__ == "__main__":