Fix Azure OpenAI embeddings by validating deployment_id parameter (#2320)

Co-Authored-By: Joe Moura <joao@crewai.com>

src/crewai/utilities/embedding_configurator.py

@@ -75,10 +75,36 @@ class EmbeddingConfigurator:
 
     @staticmethod
     def _configure_azure(config, model_name):
+        """
+        Configure an Azure OpenAI embedding function.
+        
+        Args:
+            config: A dictionary containing Azure OpenAI configuration parameters.
+                Required parameters:
+                - api_key: Azure OpenAI API key
+                - api_base: Azure OpenAI API base URL
+                - api_version: Azure OpenAI API version
+                - deployment_id: Azure OpenAI deployment ID for the embedding model
+            model_name: The name of the embedding model
+            
+        Returns:
+            An OpenAIEmbeddingFunction configured for Azure OpenAI
+            
+        Raises:
+            ValueError: If required parameters are missing
+        """
         from chromadb.utils.embedding_functions.openai_embedding_function import (
             OpenAIEmbeddingFunction,
         )
-
+        
+        # Check if deployment_id is provided for Azure OpenAI
+        deployment_id = config.get("deployment_id")
+        if not deployment_id:
+            raise ValueError(
+                "Missing required parameter 'deployment_id' for Azure OpenAI embeddings. "
+                "Please provide a deployment_id in your Azure embedder configuration."
+            )
+        
         return OpenAIEmbeddingFunction(
             api_key=config.get("api_key"),
             api_base=config.get("api_base"),
@@ -87,7 +113,7 @@ class EmbeddingConfigurator:
             model_name=model_name,
             default_headers=config.get("default_headers"),
             dimensions=config.get("dimensions"),
-            deployment_id=config.get("deployment_id"),
+            deployment_id=deployment_id,
             organization_id=config.get("organization_id"),
         )
 

tests/memory/azure_memory_test.py

@@ -0,0 +1,53 @@
+import pytest
+from unittest.mock import patch, MagicMock
+
+from crewai.utilities.embedding_configurator import EmbeddingConfigurator
+
+
+def test_azure_embedder_missing_deployment_id():
+    """Test that Azure embedder raises an error when deployment_id is missing"""
+    embedder_config = {
+        "provider": "azure",
+        "config": {
+            "model": "text-embedding-ada-002",
+            "api_key": "test-key",
+            "api_base": "https://test.openai.azure.com",
+            "api_version": "2023-05-15",
+            "api_type": "azure",
+        }
+    }
+    
+    configurator = EmbeddingConfigurator()
+    
+    with pytest.raises(ValueError) as excinfo:
+        configurator.configure_embedder(embedder_config)
+        
+    assert "Missing required parameter 'deployment_id'" in str(excinfo.value)
+
+
+@patch("chromadb.utils.embedding_functions.openai_embedding_function.OpenAIEmbeddingFunction")
+def test_azure_embedder_with_deployment_id(mock_openai_embedding):
+    """Test that Azure embedder works when deployment_id is provided"""
+    mock_instance = MagicMock()
+    mock_openai_embedding.return_value = mock_instance
+    
+    embedder_config = {
+        "provider": "azure",
+        "config": {
+            "model": "text-embedding-ada-002",
+            "api_key": "test-key",
+            "api_base": "https://test.openai.azure.com",
+            "api_version": "2023-05-15",
+            "api_type": "azure",
+            "deployment_id": "text-embedding-ada-002",
+        }
+    }
+    
+    configurator = EmbeddingConfigurator()
+    result = configurator.configure_embedder(embedder_config)
+    
+    assert result == mock_instance
+    mock_openai_embedding.assert_called_once()
+    # Verify deployment_id was passed correctly
+    call_kwargs = mock_openai_embedding.call_args.kwargs
+    assert call_kwargs["deployment_id"] == "text-embedding-ada-002"