Handle saved search docs in eval flow (#2075)

backend/danswer/chat/process_message.py

@@ -61,7 +61,7 @@ from danswer.search.retrieval.search_runner import inference_sections_from_ids
 from danswer.search.utils import chunks_or_sections_to_search_docs
 from danswer.search.utils import dedupe_documents
 from danswer.search.utils import drop_llm_indices
-from danswer.search.utils import relevant_documents_to_indices
+from danswer.search.utils import relevant_sections_to_indices
 from danswer.server.query_and_chat.models import ChatMessageDetail
 from danswer.server.query_and_chat.models import CreateChatMessageRequest
 from danswer.server.utils import get_json_line
@@ -637,9 +637,9 @@ def stream_chat_message_objects(
                     relevance_sections = packet.response
 
                     if reference_db_search_docs is not None:
-                        llm_indices = relevant_documents_to_indices(
+                        llm_indices = relevant_sections_to_indices(
                             relevance_sections=relevance_sections,
-                            search_docs=[
+                            items=[
                                 translate_db_search_doc_to_server_search_doc(doc)
                                 for doc in reference_db_search_docs
                             ],

backend/danswer/search/pipeline.py

@@ -402,6 +402,6 @@ class SearchPipeline:
     def section_relevance_list(self) -> list[bool]:
         llm_indices = relevant_sections_to_indices(
             relevance_sections=self.section_relevance,
-            inference_sections=self.final_context_sections,
+            items=self.final_context_sections,
         )
         return [ind in llm_indices for ind in range(len(self.final_context_sections))]

backend/danswer/search/utils.py

@@ -19,6 +19,14 @@ T = TypeVar(
     SavedSearchDocWithContent,
 )
 
+TSection = TypeVar(
+    "TSection",
+    InferenceSection,
+    SearchDoc,
+    SavedSearchDoc,
+    SavedSearchDocWithContent,
+)
+
 
 def dedupe_documents(items: list[T]) -> tuple[list[T], list[int]]:
     seen_ids = set()
@@ -39,30 +47,9 @@ def dedupe_documents(items: list[T]) -> tuple[list[T], list[int]]:
 
 
 def relevant_sections_to_indices(
-    relevance_sections: list[SectionRelevancePiece] | None,
+    relevance_sections: list[SectionRelevancePiece] | None, items: list[TSection]
-    inference_sections: list[InferenceSection],
 ) -> list[int]:
-    if relevance_sections is None:
+    if not relevance_sections:
-        return []
-
-    relevant_set = {
-        (chunk.document_id, chunk.chunk_id)
-        for chunk in relevance_sections
-        if chunk.relevant
-    }
-    relevant_indices = [
-        index
-        for index, section in enumerate(inference_sections)
-        if (section.center_chunk.document_id, section.center_chunk.chunk_id)
-        in relevant_set
-    ]
-    return relevant_indices
-
-
-def relevant_documents_to_indices(
-    relevance_sections: list[SectionRelevancePiece] | None, search_docs: list[SearchDoc]
-) -> list[int]:
-    if relevance_sections is None:
         return []
 
     relevant_set = {
@@ -73,8 +60,18 @@ def relevant_documents_to_indices(
 
     return [
         index
-        for index, section in enumerate(search_docs)
+        for index, item in enumerate(items)
-        if (section.document_id, section.chunk_ind) in relevant_set
+        if (
+            (
+                isinstance(item, InferenceSection)
+                and (item.center_chunk.document_id, item.center_chunk.chunk_id)
+                in relevant_set
+            )
+            or (
+                not isinstance(item, (InferenceSection))
+                and (item.document_id, item.chunk_ind) in relevant_set
+            )
+        )
     ]
 
 

backend/ee/danswer/server/query_and_chat/query_backend.py

@@ -1,5 +1,3 @@
-from typing import cast
-
 from fastapi import APIRouter
 from fastapi import Depends
 from fastapi import HTTPException
@@ -11,9 +9,7 @@ from danswer.configs.danswerbot_configs import DANSWER_BOT_TARGET_CHUNK_PERCENTA
 from danswer.danswerbot.slack.handlers.handle_standard_answers import (
     oneoff_standard_answers,
 )
-from danswer.db.chat import translate_db_search_doc_to_server_search_doc
 from danswer.db.engine import get_session
-from danswer.db.models import SearchDoc
 from danswer.db.models import User
 from danswer.db.persona import get_persona_by_id
 from danswer.llm.answering.prompts.citations_prompt import (
@@ -31,7 +27,7 @@ from danswer.search.models import SearchRequest
 from danswer.search.pipeline import SearchPipeline
 from danswer.search.utils import dedupe_documents
 from danswer.search.utils import drop_llm_indices
-from danswer.search.utils import relevant_documents_to_indices
+from danswer.search.utils import relevant_sections_to_indices
 from danswer.utils.logger import setup_logger
 from ee.danswer.server.query_and_chat.models import DocumentSearchRequest
 from ee.danswer.server.query_and_chat.models import StandardAnswerRequest
@@ -113,12 +109,8 @@ def handle_search_request(
     if search_request.retrieval_options.dedupe_docs:
         deduped_docs, dropped_inds = dedupe_documents(top_docs)
 
-    llm_indices = relevant_documents_to_indices(
+    llm_indices = relevant_sections_to_indices(
-        relevance_sections=relevance_sections,
+        relevance_sections=relevance_sections, items=deduped_docs
-        search_docs=[
-            translate_db_search_doc_to_server_search_doc(cast(SearchDoc, doc))
-            for doc in deduped_docs
-        ],
     )
 
     if dropped_inds: