data-designer-config 0.4.0rc1__tar.gz → 0.4.0rc2__tar.gz

{data_designer_config-0.4.0rc1 → data_designer_config-0.4.0rc2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: data-designer-config
-Version: 0.4.0rc1
+Version: 0.4.0rc2
 Summary: Configuration layer for DataDesigner synthetic data generation
 License-Expression: Apache-2.0
 Classifier: Development Status :: 4 - Beta

{data_designer_config-0.4.0rc1 → data_designer_config-0.4.0rc2}/src/data_designer/config/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '0.4.0rc1'
-__version_tuple__ = version_tuple = (0, 4, 0, 'rc1')
+__version__ = version = '0.4.0rc2'
+__version_tuple__ = version_tuple = (0, 4, 0, 'rc2')
 
 __commit_id__ = commit_id = None

{data_designer_config-0.4.0rc1 → data_designer_config-0.4.0rc2}/src/data_designer/config/models.py

@@ -3,6 +3,7 @@
 
 from __future__ import annotations
 
+import json
 import logging
 from abc import ABC, abstractmethod
 from enum import Enum
@@ -65,7 +66,7 @@ class ModalityContext(ABC, BaseModel):
     data_type: ModalityDataType
 
     @abstractmethod
-    def get_context(self, record: dict) -> dict[str, Any]: ...
+    def get_contexts(self, record: dict) -> list[dict[str, Any]]: ...
 
 
 class ImageContext(ModalityContext):
@@ -81,25 +82,53 @@ class ImageContext(ModalityContext):
     modality: Modality = Modality.IMAGE
     image_format: ImageFormat | None = None
 
-    def get_context(self, record: dict) -> dict[str, Any]:
-        """Get the context for the image modality.
+    def get_contexts(self, record: dict) -> list[dict[str, Any]]:
+        """Get the contexts for the image modality.
 
         Args:
-            record: The record containing the image data.
+            record: The record containing the image data. The data can be:
+                - A JSON serialized list of strings
+                - A list of strings
+                - A single string
 
         Returns:
-            The context for the image modality.
+            A list of image contexts.
         """
-        context = dict(type="image_url")
-        context_value = record[self.column_name]
-        if self.data_type == ModalityDataType.URL:
-            context["image_url"] = context_value
+        raw_value = record[self.column_name]
+
+        # Normalize to list of strings
+        if isinstance(raw_value, str):
+            # Try to parse as JSON first
+            try:
+                parsed_value = json.loads(raw_value)
+                if isinstance(parsed_value, list):
+                    context_values = parsed_value
+                else:
+                    context_values = [raw_value]
+            except (json.JSONDecodeError, TypeError):
+                context_values = [raw_value]
+        elif isinstance(raw_value, list):
+            context_values = raw_value
+        elif hasattr(raw_value, "__iter__") and not isinstance(raw_value, (str, bytes, dict)):
+            # Handle array-like objects (numpy arrays, pandas Series, etc.)
+            context_values = list(raw_value)
         else:
-            context["image_url"] = {
-                "url": f"data:image/{self.image_format.value};base64,{context_value}",
-                "format": self.image_format.value,
-            }
-        return context
+            context_values = [raw_value]
+
+        # Build context list
+        contexts = []
+        for context_value in context_values:
+            context = dict(type="image_url")
+            if self.data_type == ModalityDataType.URL:
+                context["image_url"] = context_value
+            else:
+                context["image_url"] = {
+                    "url": f"data:image/{self.image_format.value};base64,{context_value}",
+                    "format": self.image_format.value,
+                }
+            contexts.append(context)
+
+        return contexts
 
     @model_validator(mode="after")
     def _validate_image_format(self) -> Self:
@@ -399,12 +428,14 @@ class ModelConfig(ConfigBase):
         inference_parameters: Inference parameters for the model (temperature, top_p, max_tokens, etc.).
             The generation_type is determined by the type of inference_parameters.
         provider: Optional model provider name if using custom providers.
+        skip_health_check: Whether to skip the health check for this model. Defaults to False.
     """
 
     alias: str
     model: str
     inference_parameters: InferenceParamsT = Field(default_factory=ChatCompletionInferenceParams)
     provider: str | None = None
+    skip_health_check: bool = False
 
     @property
     def generation_type(self) -> GenerationType:

{data_designer_config-0.4.0rc1 → data_designer_config-0.4.0rc2}/tests/config/test_models.py

@@ -4,6 +4,7 @@
 import json
 import tempfile
 from collections import Counter
+from typing import TYPE_CHECKING
 
 import pytest
 import yaml
@@ -24,22 +25,159 @@ from data_designer.config.models import (
     UniformDistributionParams,
     load_model_configs,
 )
+from data_designer.lazy_heavy_imports import np
 
+if TYPE_CHECKING:
+    import numpy as np
 
-def test_image_context_get_context():
+
+def test_image_context_get_contexts_single_string():
+    """Test get_contexts with a single string value."""
     image_context = ImageContext(
         column_name="image_base64", data_type=ModalityDataType.BASE64, image_format=ImageFormat.PNG
     )
-    assert image_context.get_context({"image_base64": "somebase64encodedimagestring"}) == {
-        "type": "image_url",
-        "image_url": {"url": "data:image/png;base64,somebase64encodedimagestring", "format": "png"},
-    }
+    assert image_context.get_contexts({"image_base64": "somebase64encodedimagestring"}) == [
+        {
+            "type": "image_url",
+            "image_url": {"url": "data:image/png;base64,somebase64encodedimagestring", "format": "png"},
+        }
+    ]
 
     image_context = ImageContext(column_name="image_url", data_type=ModalityDataType.URL)
-    assert image_context.get_context({"image_url": "https://example.com/examle_image.png"}) == {
-        "type": "image_url",
-        "image_url": "https://example.com/examle_image.png",
-    }
+    assert image_context.get_contexts({"image_url": "https://example.com/examle_image.png"}) == [
+        {
+            "type": "image_url",
+            "image_url": "https://example.com/examle_image.png",
+        }
+    ]
+
+
+def test_image_context_get_contexts_list_of_strings():
+    """Test get_contexts with a list of strings."""
+    image_context = ImageContext(
+        column_name="image_base64", data_type=ModalityDataType.BASE64, image_format=ImageFormat.PNG
+    )
+    assert image_context.get_contexts({"image_base64": ["image1base64", "image2base64", "image3base64"]}) == [
+        {
+            "type": "image_url",
+            "image_url": {"url": "data:image/png;base64,image1base64", "format": "png"},
+        },
+        {
+            "type": "image_url",
+            "image_url": {"url": "data:image/png;base64,image2base64", "format": "png"},
+        },
+        {
+            "type": "image_url",
+            "image_url": {"url": "data:image/png;base64,image3base64", "format": "png"},
+        },
+    ]
+
+    image_context = ImageContext(column_name="image_url", data_type=ModalityDataType.URL)
+    assert image_context.get_contexts(
+        {"image_url": ["https://example.com/image1.png", "https://example.com/image2.png"]}
+    ) == [
+        {
+            "type": "image_url",
+            "image_url": "https://example.com/image1.png",
+        },
+        {
+            "type": "image_url",
+            "image_url": "https://example.com/image2.png",
+        },
+    ]
+
+
+def test_image_context_get_contexts_numpy_array():
+    """Test get_contexts with numpy arrays (happens after parquet serialization)."""
+    image_context = ImageContext(
+        column_name="image_base64", data_type=ModalityDataType.BASE64, image_format=ImageFormat.PNG
+    )
+    numpy_array = np.array(["image1base64", "image2base64"])
+    assert image_context.get_contexts({"image_base64": numpy_array}) == [
+        {
+            "type": "image_url",
+            "image_url": {"url": "data:image/png;base64,image1base64", "format": "png"},
+        },
+        {
+            "type": "image_url",
+            "image_url": {"url": "data:image/png;base64,image2base64", "format": "png"},
+        },
+    ]
+
+    image_context = ImageContext(column_name="image_url", data_type=ModalityDataType.URL)
+    numpy_array = np.array(["https://example.com/image1.png", "https://example.com/image2.png"])
+    assert image_context.get_contexts({"image_url": numpy_array}) == [
+        {
+            "type": "image_url",
+            "image_url": "https://example.com/image1.png",
+        },
+        {
+            "type": "image_url",
+            "image_url": "https://example.com/image2.png",
+        },
+    ]
+
+
+def test_image_context_get_contexts_json_serialized_list():
+    """Test get_contexts with a JSON serialized list of strings."""
+    image_context = ImageContext(
+        column_name="image_base64", data_type=ModalityDataType.BASE64, image_format=ImageFormat.PNG
+    )
+    json_str = json.dumps(["image1base64", "image2base64"])
+    assert image_context.get_contexts({"image_base64": json_str}) == [
+        {
+            "type": "image_url",
+            "image_url": {"url": "data:image/png;base64,image1base64", "format": "png"},
+        },
+        {
+            "type": "image_url",
+            "image_url": {"url": "data:image/png;base64,image2base64", "format": "png"},
+        },
+    ]
+
+    image_context = ImageContext(column_name="image_url", data_type=ModalityDataType.URL)
+    json_str = json.dumps(["https://example.com/image1.png", "https://example.com/image2.png"])
+    assert image_context.get_contexts({"image_url": json_str}) == [
+        {
+            "type": "image_url",
+            "image_url": "https://example.com/image1.png",
+        },
+        {
+            "type": "image_url",
+            "image_url": "https://example.com/image2.png",
+        },
+    ]
+
+
+def test_image_context_get_contexts_json_string_not_list():
+    """Test get_contexts with a JSON string that isn't a list (should treat as single string)."""
+    image_context = ImageContext(column_name="image_url", data_type=ModalityDataType.URL)
+    json_str = json.dumps({"nested": "object"})
+    # Should treat the entire JSON string as a single image URL
+    assert image_context.get_contexts({"image_url": json_str}) == [
+        {
+            "type": "image_url",
+            "image_url": json_str,
+        }
+    ]
+
+
+def test_image_context_get_contexts_invalid_json():
+    """Test get_contexts with invalid JSON string (should treat as single string)."""
+    image_context = ImageContext(column_name="image_url", data_type=ModalityDataType.URL)
+    invalid_json = "not a valid json string"
+    assert image_context.get_contexts({"image_url": invalid_json}) == [
+        {
+            "type": "image_url",
+            "image_url": invalid_json,
+        }
+    ]
+
+
+def test_image_context_get_contexts_empty_list():
+    """Test get_contexts with an empty list."""
+    image_context = ImageContext(column_name="image_url", data_type=ModalityDataType.URL)
+    assert image_context.get_contexts({"image_url": []}) == []
 
 
 def test_image_context_validate_image_format():