data-designer-engine 0.9.2__tar.gz → 0.9.3__tar.gz

{data_designer_engine-0.9.2 → data_designer_engine-0.9.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: data-designer-engine
-Version: 0.9.2
+Version: 0.9.3
 Summary: Generation engine for DataDesigner synthetic data generation
 License-Expression: Apache-2.0
 Classifier: Development Status :: 4 - Beta
@@ -18,7 +18,7 @@ Requires-Dist: anyascii<1,>=0.3.3
 Requires-Dist: asciichartpy<2,>=1.5.25
 Requires-Dist: chardet<6,>=3.0.2
 Requires-Dist: cryptography<=49,>=48.0.1
-Requires-Dist: data-designer-config==0.9.2
+Requires-Dist: data-designer-config==0.9.3
 Requires-Dist: duckdb<2,>=1.5.0
 Requires-Dist: faker<21,>=20.1.0
 Requires-Dist: fsspec<2026,>=2025.3.0

{data_designer_engine-0.9.2 → data_designer_engine-0.9.3}/src/data_designer/engine/compiler.py

@@ -11,7 +11,12 @@ from data_designer.config.errors import InvalidConfigError
 from data_designer.config.sampler_params import UUIDSamplerParams
 from data_designer.engine.resources.resource_provider import ResourceProvider
 from data_designer.engine.resources.seed_reader import SeedReader, SeedReaderConfigError
-from data_designer.engine.validation import ViolationLevel, rich_print_violations, validate_data_designer_config
+from data_designer.engine.validation import (
+    ViolationLevel,
+    resolve_processor_dropped_columns,
+    rich_print_violations,
+    validate_data_designer_config,
+)
 
 logger = logging.getLogger(__name__)
 
@@ -43,7 +48,26 @@ def _resolve_and_add_seed_columns(config: DataDesignerConfig, seed_reader: SeedR
             "Please remove the conflicting columns or use a seed dataset with different column names."
         )
 
-    config.columns.extend([SeedDatasetColumnConfig(name=col_name) for col_name in seed_col_names])
+    seed_columns = [SeedDatasetColumnConfig(name=col_name) for col_name in seed_col_names]
+    config.columns.extend(seed_columns)
+    _mark_processor_dropped_seed_columns(config, seed_columns)
+
+
+def _mark_processor_dropped_seed_columns(
+    config: DataDesignerConfig, seed_columns: list[SeedDatasetColumnConfig]
+) -> None:
+    """Marks seed columns targeted by a drop processor with `drop=True`.
+
+    `DataDesignerConfigBuilder.add_processor` does this for configured columns, but seed
+    columns don't exist yet at that point — they're resolved here. Without this, a
+    processor-dropped seed column is still handed to the dataset profiler, which either
+    reports a column the emitted dataset no longer has or fails outright when
+    `preserve_dropped_columns=False`.
+    """
+    dropped_column_names = resolve_processor_dropped_columns(config.columns, config.processors or [])
+    for column in seed_columns:
+        if column.name in dropped_column_names:
+            column.drop = True
 
 
 def _add_internal_row_id_column_if_needed(config: DataDesignerConfig) -> None:

{data_designer_engine-0.9.2 → data_designer_engine-0.9.3}/src/data_designer/engine/models/clients/errors.py

@@ -15,6 +15,7 @@ class ProviderErrorKind(str, Enum):
     API_ERROR = "api_error"
     API_CONNECTION = "api_connection"
     AUTHENTICATION = "authentication"
+    CLIENT_ERROR = "client_error"
     CONTEXT_WINDOW_EXCEEDED = "context_window_exceeded"
     QUOTA_EXCEEDED = "quota_exceeded"
     UNSUPPORTED_PARAMS = "unsupported_params"
@@ -101,6 +102,8 @@ def map_http_status_to_provider_error_kind(status_code: int, body_text: str = ""
         return ProviderErrorKind.BAD_REQUEST
     if 500 <= status_code <= 599:
         return ProviderErrorKind.INTERNAL_SERVER
+    if 400 <= status_code <= 499:
+        return ProviderErrorKind.CLIENT_ERROR
     return ProviderErrorKind.API_ERROR
 
 

{data_designer_engine-0.9.2 → data_designer_engine-0.9.3}/src/data_designer/engine/models/errors.py

@@ -98,6 +98,9 @@ class ModelUnsupportedParamsError(DataDesignerError): ...
 class ModelBadRequestError(DataDesignerError): ...
 
 
+class ModelRequestRejectedError(DataDesignerError): ...
+
+
 class ModelInternalServerError(DataDesignerError): ...
 
 
@@ -172,7 +175,7 @@ def _attach_provider_message(
     formatted_message: FormattedLLMErrorMessage,
     exception: ProviderError,
 ) -> FormattedLLMErrorMessage:
-    if exception.status_code != 400:
+    if exception.status_code != 400 and exception.kind != ProviderErrorKind.CLIENT_ERROR:
         return formatted_message
     normalized = _normalize_error_detail(exception.message)
     if normalized is None:
@@ -320,6 +323,7 @@ def _raise_from_provider_error(
         ProviderErrorKind.NOT_FOUND: ModelNotFoundError,
         ProviderErrorKind.PERMISSION_DENIED: ModelPermissionDeniedError,
         ProviderErrorKind.UNSUPPORTED_PARAMS: ModelUnsupportedParamsError,
+        ProviderErrorKind.CLIENT_ERROR: ModelRequestRejectedError,
         ProviderErrorKind.INTERNAL_SERVER: ModelInternalServerError,
         ProviderErrorKind.UNPROCESSABLE_ENTITY: ModelUnprocessableEntityError,
         ProviderErrorKind.API_CONNECTION: ModelAPIConnectionError,
@@ -350,6 +354,12 @@ def _raise_from_provider_error(
             f"One or more of the parameters you provided were found to be unsupported by model {model_name!r} while {purpose}.",
             f"Review the documentation for model provider {model_provider_name!r} and adjust your request.",
         ),
+        ProviderErrorKind.CLIENT_ERROR: (
+            f"Model provider {model_provider_name!r} rejected the request for model {model_name!r} "
+            f"with HTTP status {exception.status_code} while {purpose}.",
+            "Verify the model name, credentials, and model configuration. If the provider is behind a gateway or "
+            "proxy, contact its administrator and provide the HTTP status.",
+        ),
         ProviderErrorKind.INTERNAL_SERVER: (
             f"Model {model_name!r} is currently experiencing internal server issues while {purpose}.",
             f"Try again in a few moments. Check with your model provider {model_provider_name!r} if the issue persists.",

{data_designer_engine-0.9.2 → data_designer_engine-0.9.3}/src/data_designer/engine/processing/ginja/environment.py

@@ -180,6 +180,11 @@ class UserTemplateSandboxEnvironment(ImmutableSandboxedEnvironment):
             or may be malicious in nature. If not specified, defaults to
             MAX_AST_DEPTH set by this module.
 
+        prefer_dict_key_access (optional, bool): When enabled, prefer dict key
+            lookup over attribute access so that keys like "items" resolve to
+            dict["items"] instead of dict.items. If not specified, defaults to
+            False.
+
         **kwargs: Additional kwargs passed to ImmutableSandboxedEnvironment.
         """
         super().__init__(autoescape=False, **kwargs)

{data_designer_engine-0.9.2 → data_designer_engine-0.9.3}/src/data_designer/engine/validation.py

@@ -260,7 +260,7 @@ def validate_columns_not_all_dropped(
     processor_configs: list[ProcessorConfigT] | None = None,
 ) -> list[Violation]:
     processor_configs = processor_configs or []
-    processor_dropped_columns = _resolve_processor_dropped_columns(columns, processor_configs)
+    processor_dropped_columns = resolve_processor_dropped_columns(columns, processor_configs)
     remaining_cols = [
         c
         for c in columns
@@ -287,10 +287,11 @@ def validate_columns_not_all_dropped(
     return []
 
 
-def _resolve_processor_dropped_columns(
+def resolve_processor_dropped_columns(
     columns: list[ColumnConfigT],
     processor_configs: list[ProcessorConfigT],
 ) -> set[str]:
+    """Return the column names targeted by drop processors, expanding globs against `columns`."""
     column_names = {c.name for c in columns}
     dropped_columns: set[str] = set()
     for processor_config in processor_configs:

{data_designer_engine-0.9.2 → data_designer_engine-0.9.3}/tests/engine/models/clients/test_client_errors.py

@@ -59,7 +59,9 @@ class StubHttpResponse:
         (400, "maximum context length exceeded", ProviderErrorKind.CONTEXT_WINDOW_EXCEEDED),
         (500, "", ProviderErrorKind.INTERNAL_SERVER),
         (503, "", ProviderErrorKind.INTERNAL_SERVER),
-        (418, "", ProviderErrorKind.API_ERROR),
+        (418, "", ProviderErrorKind.CLIENT_ERROR),
+        (424, "", ProviderErrorKind.CLIENT_ERROR),
+        (302, "", ProviderErrorKind.API_ERROR),
     ],
 )
 def test_map_http_status_to_provider_error_kind(

{data_designer_engine-0.9.2 → data_designer_engine-0.9.3}/tests/engine/models/test_model_errors.py

@@ -25,6 +25,7 @@ from data_designer.engine.models.errors import (
     ModelQuotaExceededError,
     ModelRateLimitError,
     ModelRequestAdmissionTimeoutError,
+    ModelRequestRejectedError,
     ModelTimeoutError,
     ModelUnprocessableEntityError,
     ModelUnsupportedCapabilityError,
@@ -51,7 +52,7 @@ stub_purpose = "running generation for column 'test'"
             ),
             ModelBadRequestError,
             (
-                f"Provider message: Unexpected field 'foo' in request payload.\n  | Cause: The request for model "
+                "Provider message: Unexpected field 'foo' in request payload.\n  | Cause: The request for model "
                 f"'{stub_model_name}' was found to be malformed or missing required parameters while {stub_purpose}."
             ),
         ),
@@ -167,11 +168,25 @@ stub_purpose = "running generation for column 'test'"
             ModelUnprocessableEntityError,
             f"Cause: The request to model '{stub_model_name}' failed despite correct request format while {stub_purpose}.",
         ),
+        (
+            ProviderError(
+                kind=ProviderErrorKind.CLIENT_ERROR,
+                message="Backend returned 404: 404 page not found",
+                status_code=424,
+            ),
+            ModelRequestRejectedError,
+            (
+                "Provider message: Backend returned 404: 404 page not found\n  | Cause: Model provider "
+                f"'{stub_model_provider_name}' rejected the request for model "
+                f"'{stub_model_name}' with HTTP status 424 while {stub_purpose}.\n  | Solution: Verify the model name, "
+                "credentials, and model configuration."
+            ),
+        ),
         (
             ProviderError(
                 kind=ProviderErrorKind.API_ERROR,
                 message="Unknown API error",
-                status_code=418,
+                status_code=302,
             ),
             ModelAPIError,
             f"Cause: An unexpected API error occurred with model '{stub_model_name}' while {stub_purpose}.",
@@ -218,6 +233,7 @@ stub_purpose = "running generation for column 'test'"
         "not_found",
         "internal_server",
         "unprocessable_entity",
+        "client_error",
         "api_error",
         "bad_request_multimodal",
         "generation_validation_failure",
@@ -232,6 +248,19 @@ def test_handle_llm_exceptions(
         handle_llm_exceptions(exception, stub_model_name, stub_model_provider_name, stub_purpose)
 
 
+def test_handle_llm_exceptions_does_not_attach_provider_message_for_server_error() -> None:
+    exception = ProviderError(
+        kind=ProviderErrorKind.INTERNAL_SERVER,
+        message="Gateway temporarily unavailable",
+        status_code=502,
+    )
+
+    with pytest.raises(ModelInternalServerError) as exc_info:
+        handle_llm_exceptions(exception, stub_model_name, stub_model_provider_name, stub_purpose)
+
+    assert "Provider message:" not in str(exc_info.value)
+
+
 def test_generation_validation_failure_error_stores_truncation_reason() -> None:
     exception = GenerationValidationFailureError(
         " Generation validation failure ",

{data_designer_engine-0.9.2 → data_designer_engine-0.9.3}/tests/engine/test_compiler.py

@@ -8,6 +8,7 @@ import pytest
 from data_designer.config.column_configs import ExpressionColumnConfig, SamplerColumnConfig
 from data_designer.config.config_builder import DataDesignerConfigBuilder
 from data_designer.config.errors import InvalidConfigError
+from data_designer.config.processors import DropColumnsProcessorConfig
 from data_designer.config.sampler_params import CategorySamplerParams, SamplerType, UUIDSamplerParams
 from data_designer.config.seed_source import FileContentsSeedSource, HuggingFaceSeedSource
 from data_designer.engine.compiler import compile_data_designer_config
@@ -174,3 +175,43 @@ def test_does_not_add_id_column_when_seed_dataset_exists(resource_provider: Reso
     assert len(config.columns) == 3
     assert config.columns[0].name == "derived_value"
     assert not any(col.name == "_internal_row_id" for col in config.columns)
+
+
+def _seed_column(config, name: str):
+    return next(col for col in config.columns if col.name == name)
+
+
+def test_seed_columns_are_not_dropped_without_a_drop_processor(resource_provider: ResourceProvider):
+    builder = DataDesignerConfigBuilder()
+    builder.add_column(SamplerColumnConfig(name="row_id", sampler_type=SamplerType.UUID, params=UUIDSamplerParams()))
+    builder.with_seed_dataset(HuggingFaceSeedSource(path="hf://datasets/test/data.csv"))
+
+    config = compile_data_designer_config(builder.build(), resource_provider)
+
+    assert _seed_column(config, "age").drop is False
+    assert _seed_column(config, "city").drop is False
+
+
+def test_marks_seed_columns_named_by_drop_processor(resource_provider: ResourceProvider):
+    """A drop processor naming a seed column must mark it dropped so the profiler skips it."""
+    builder = DataDesignerConfigBuilder()
+    builder.add_column(SamplerColumnConfig(name="row_id", sampler_type=SamplerType.UUID, params=UUIDSamplerParams()))
+    builder.with_seed_dataset(HuggingFaceSeedSource(path="hf://datasets/test/data.csv"))
+    builder.add_processor(DropColumnsProcessorConfig(name="drop_age", column_names=["age"]))
+
+    config = compile_data_designer_config(builder.build(), resource_provider)
+
+    assert _seed_column(config, "age").drop is True
+    assert _seed_column(config, "city").drop is False
+
+
+def test_marks_seed_columns_matched_by_drop_processor_glob(resource_provider: ResourceProvider):
+    builder = DataDesignerConfigBuilder()
+    builder.add_column(SamplerColumnConfig(name="row_id", sampler_type=SamplerType.UUID, params=UUIDSamplerParams()))
+    builder.with_seed_dataset(HuggingFaceSeedSource(path="hf://datasets/test/data.csv"))
+    builder.add_processor(DropColumnsProcessorConfig(name="drop_c", column_names=["c*"]))
+
+    config = compile_data_designer_config(builder.build(), resource_provider)
+
+    assert _seed_column(config, "city").drop is True
+    assert _seed_column(config, "age").drop is False