Revert "Fix Azure Search validation errors" - restore required fields

Co-authored-by: owndev <69784886+owndev@users.noreply.github.com>

Author: Copilot

docs/azure-ai-integration.md

@@ -55,6 +55,9 @@ AZURE_SEARCH_ENDPOINT="https://your-search-service.search.windows.net"
 # Azure Search index name containing the documents
 AZURE_SEARCH_INDEX_NAME="your-index-name"
 
+# Azure Search project resource ID (optional)
+AZURE_SEARCH_PROJECT_RESOURCE_ID="your-project-resource-id"
+
 # Azure Search API key (if using api_key authentication)
 AZURE_SEARCH_KEY="your-search-api-key"
 
@@ -64,6 +67,12 @@ AZURE_SEARCH_AUTHENTICATION_TYPE="system_assigned_managed_identity"
 # Semantic configuration name for Azure Search
 AZURE_SEARCH_SEMANTIC_CONFIGURATION="azureml-default"
 
+# Azure Search embedding endpoint (optional)
+AZURE_SEARCH_EMBEDDING_ENDPOINT="your-embedding-endpoint"
+
+# Azure Search embedding API key (optional)
+AZURE_SEARCH_EMBEDDING_KEY="your-embedding-key"
+
 # Query type for Azure Search
 AZURE_SEARCH_QUERY_TYPE="vectorSimpleHybrid"
 
@@ -95,43 +104,15 @@ Configure Azure Search by setting the following environment variables:
 
 #### Optional Settings:
 
+- **AZURE_SEARCH_PROJECT_RESOURCE_ID**: Project resource ID
 - **AZURE_SEARCH_SEMANTIC_CONFIGURATION**: Semantic configuration name
+- **AZURE_SEARCH_EMBEDDING_ENDPOINT**: Embedding service endpoint
+- **AZURE_SEARCH_EMBEDDING_KEY**: Embedding service API key
 - **AZURE_SEARCH_QUERY_TYPE**: Query type (`vectorSimpleHybrid`, `vector`, `semantic`)
 - **AZURE_SEARCH_IN_SCOPE**: Limit to indexed documents only
 - **AZURE_SEARCH_ROLE_INFORMATION**: Role information for responses
 - **AZURE_SEARCH_STRICTNESS**: Strictness level (1-5)
-- **AZURE_SEARCH_TOP_N_DOCUMENTS**: Number of documents to retrieve
-
-#### Example Generated Data Sources Configuration:
-
-When Azure Search is configured, requests to Azure AI will automatically include:
-
-```json
-{
-  "model": "gpt-4o",
-  "messages": [...],
-  "data_sources": [
-    {
-      "type": "azure_search",
-      "parameters": {
-        "filter": null,
-        "endpoint": "https://your-search-service.search.windows.net",
-        "index_name": "your-index-name",
-        "authentication": {
-          "type": "system_assigned_managed_identity",
-          "key": null
-        },
-        "query_type": "vectorSimpleHybrid",
-        "in_scope": false,
-        "role_information": "You are an AI assistant.",
-        "strictness": 5,
-        "top_n_documents": 20,
-        "semantic_configuration": "azureml-default"
-      }
-    }
-  ]
-}
-```
+- **AZURE_SEARCH_TOP_N_DOCUMENTS**: Number of documents to retrieve
 
 > [!TIP]  
 > To use **Azure OpenAI** and other **Azure AI** models **simultaneously**, you can use the following URL: `https://<your project>.services.ai.azure.com/models/chat/completions?api-version=2024-05-01-preview`

pipelines/azure/azure_ai_foundry.py

@@ -187,6 +187,11 @@ class Valves(BaseModel):
             description="Azure Search index name for RAG",
         )
 
+        AZURE_SEARCH_PROJECT_RESOURCE_ID: str = Field(
+            default=os.getenv("AZURE_SEARCH_PROJECT_RESOURCE_ID", ""),
+            description="Azure Search project resource ID",
+        )
+
         AZURE_SEARCH_KEY: EncryptedStr = Field(
             default=os.getenv("AZURE_SEARCH_KEY", ""),
             description="Azure Search API key",
@@ -204,6 +209,16 @@ class Valves(BaseModel):
             description="Azure Search semantic configuration name",
         )
 
+        AZURE_SEARCH_EMBEDDING_ENDPOINT: str = Field(
+            default=os.getenv("AZURE_SEARCH_EMBEDDING_ENDPOINT", ""),
+            description="Azure Search embedding endpoint URL",
+        )
+
+        AZURE_SEARCH_EMBEDDING_KEY: EncryptedStr = Field(
+            default=os.getenv("AZURE_SEARCH_EMBEDDING_KEY", ""),
+            description="Azure Search embedding API key",
+        )
+
         AZURE_SEARCH_QUERY_TYPE: str = Field(
             default=os.getenv("AZURE_SEARCH_QUERY_TYPE", "vectorSimpleHybrid"),
             description="Azure Search query type (vectorSimpleHybrid, vector, semantic)",
@@ -324,7 +339,7 @@ def get_azure_search_data_sources(self) -> Optional[List[Dict[str, Any]]]:
         ):
             auth_config["key"] = self.valves.AZURE_SEARCH_KEY.get_decrypted()
 
-        # Build the data source configuration with only officially supported parameters
+        # Build the data source configuration
         data_source = {
             "type": "azure_search",
             "parameters": {
@@ -340,12 +355,40 @@ def get_azure_search_data_sources(self) -> Optional[List[Dict[str, Any]]]:
             },
         }
 
+        # Add optional project resource ID if configured
+        if self.valves.AZURE_SEARCH_PROJECT_RESOURCE_ID:
+            data_source["parameters"]["project_resource_id"] = (
+                self.valves.AZURE_SEARCH_PROJECT_RESOURCE_ID
+            )
+
         # Add semantic configuration if configured
         if self.valves.AZURE_SEARCH_SEMANTIC_CONFIGURATION:
             data_source["parameters"]["semantic_configuration"] = (
                 self.valves.AZURE_SEARCH_SEMANTIC_CONFIGURATION
             )
 
+        # Add embedding configuration if configured
+        if self.valves.AZURE_SEARCH_EMBEDDING_ENDPOINT:
+            data_source["parameters"]["embeddingEndpoint"] = (
+                self.valves.AZURE_SEARCH_EMBEDDING_ENDPOINT
+            )
+            data_source["parameters"]["embedding_dependency"] = None
+
+        if self.valves.AZURE_SEARCH_EMBEDDING_KEY:
+            data_source["parameters"]["embeddingKey"] = (
+                self.valves.AZURE_SEARCH_EMBEDDING_KEY.get_decrypted()
+            )
+
+        # Add additional Azure Search parameters if using API key
+        if (
+            self.valves.AZURE_SEARCH_AUTHENTICATION_TYPE == "api_key"
+            and self.valves.AZURE_SEARCH_KEY
+        ):
+            data_source["parameters"]["key"] = (
+                self.valves.AZURE_SEARCH_KEY.get_decrypted()
+            )
+            data_source["parameters"]["indexName"] = self.valves.AZURE_SEARCH_INDEX_NAME
+
         return [data_source]
 
     def parse_models(self, models_str: str) -> List[str]: