fix(search-service): Add structured_answer support to SearchContext and SearchOutput

- Add structured_answer field to SearchOutput schema with StructuredAnswer import
- Add structured_answer field to SearchContext dataclass for pipeline data flow
- Fix quote style in search_service.py debug logging (double quotes)
- Apply Ruff formatting to search_service.py

This ensures structured output with citations generated by generation_stage.py
flows through to the SearchOutput response and reaches the frontend.

Related to PR #626 (Structured Output schema)
Enables PR #630 (Frontend Citations UI)

Signed-off-by: Claude <[email protected]>
Signed-off-by: manavgup <[email protected]>

Author: manavgup

backend/rag_solution/schemas/search_schema.py

@@ -5,6 +5,7 @@
 from pydantic import UUID4, BaseModel, ConfigDict
 
 from rag_solution.schemas.llm_usage_schema import TokenWarning
+from rag_solution.schemas.structured_output_schema import StructuredAnswer
 from vectordbs.data_types import DocumentMetadata, QueryResult
 
 
@@ -44,6 +45,7 @@ class SearchOutput(BaseModel):
         query_results: List of QueryResult
         rewritten_query: Optional rewritten version of the original query
         evaluation: Optional evaluation metrics and results
+        structured_answer: Optional structured answer with citations (when requested)
     """
 
     answer: str
@@ -55,5 +57,6 @@ class SearchOutput(BaseModel):
     cot_output: dict[str, Any] | None = None  # Chain of Thought reasoning steps when requested
     metadata: dict[str, Any] | None = None  # Additional metadata including conversation context
     token_warning: TokenWarning | None = None  # Token usage warning if approaching limits
+    structured_answer: StructuredAnswer | None = None  # Structured output with citations when requested
 
     model_config = ConfigDict(from_attributes=True)

backend/rag_solution/services/pipeline/search_context.py

@@ -14,6 +14,7 @@
 from rag_solution.schemas.chain_of_thought_schema import ChainOfThoughtOutput
 from rag_solution.schemas.llm_usage_schema import TokenWarning
 from rag_solution.schemas.search_schema import SearchInput
+from rag_solution.schemas.structured_output_schema import StructuredAnswer
 from vectordbs.data_types import DocumentMetadata, QueryResult
 
 
@@ -46,6 +47,7 @@ class SearchContext:  # pylint: disable=too-many-instance-attributes
         evaluation: Answer quality evaluation
         cot_output: Chain of Thought reasoning steps
         token_warning: Token usage warnings
+        structured_answer: Structured answer with citations
 
         # Execution Metadata
         start_time: When search started
@@ -73,6 +75,7 @@ class SearchContext:  # pylint: disable=too-many-instance-attributes
     evaluation: dict[str, Any] | None = None
     cot_output: ChainOfThoughtOutput | None = None
     token_warning: TokenWarning | None = None
+    structured_answer: StructuredAnswer | None = None
 
     # Execution Metadata
     start_time: float = field(default_factory=time.time)

backend/rag_solution/services/search_service.py

@@ -570,9 +570,16 @@ async def _search_with_pipeline(self, search_input: SearchInput) -> SearchOutput
         cot_output_dict = result_context.cot_output.model_dump() if result_context.cot_output else None
 
         # Debug: Log document_metadata before creating SearchOutput
-        logger.info("📊 SEARCH_SERVICE: result_context.document_metadata has %d items", len(result_context.document_metadata))
+        logger.info(
+            "📊 SEARCH_SERVICE: result_context.document_metadata has %d items", len(result_context.document_metadata)
+        )
         if result_context.document_metadata:
-            logger.info("📊 SEARCH_SERVICE: First doc_metadata = %s", result_context.document_metadata[0].document_name if hasattr(result_context.document_metadata[0], 'document_name') else 'NO DOCUMENT_NAME')
+            logger.info(
+                "📊 SEARCH_SERVICE: First doc_metadata = %s",
+                result_context.document_metadata[0].document_name
+                if hasattr(result_context.document_metadata[0], "document_name")
+                else "NO DOCUMENT_NAME",
+            )
 
         search_output = SearchOutput(
             answer=cleaned_answer,