feat: Supporting invalid query response!

subquery.py

@@ -5,6 +5,7 @@
 from llama_index.core.tools import QueryEngineTool, ToolMetadata
 from llama_index.llms.openai import OpenAI
 from llama_index.question_gen.guidance import GuidanceQuestionGenerator
+from tc_hivemind_backend.db.utils.preprocess_text import BasePreprocessor
 from tc_hivemind_backend.embeddings.cohere import CohereEmbedding
 from utils.qdrant_utils import QDrantUtils
 from utils.query_engine import (
@@ -19,7 +20,7 @@
     WebsiteQueryEngine,
     prepare_discord_engine_auto_filter,
 )
-
+from utils.globals import INVALID_QUERY_RESPONSE
 
 def query_multiple_source(
     query: str,
@@ -198,6 +199,10 @@ def query_multiple_source(
                 metadata=tool_metadata,
             )
         )
+    if not BasePreprocessor().extract_main_content(text=query):
+        response = INVALID_QUERY_RESPONSE
+        source_nodes = []
+        return response, source_nodes
 
     embed_model = CohereEmbedding()
     llm = OpenAI("gpt-4o-mini")

utils/globals.py

@@ -1,2 +1,4 @@
 # the theshold to skip nodes of being included in an answer
 RETRIEVER_THRESHOLD = 0.4
+INVALID_QUERY_RESPONSE="We're unable to process your query. Please refine it and try again."
+QUERY_ERROR_MESSAGE="Sorry, we're unable to process your question at the moment. Please try again later."

worker/tasks.py

@@ -5,6 +5,7 @@
 from llama_index.core.schema import NodeWithScore
 from subquery import query_multiple_source
 from utils.data_source_selector import DataSourceSelector
+from utils.globals import QUERY_ERROR_MESSAGE
 from utils.query_engine.prepare_answer_sources import PrepareAnswerSources
 from utils.traceloop import init_tracing
 from worker.celery import app
@@ -21,7 +22,7 @@ def ask_question_auto_search(
         )
         answer_sources = PrepareAnswerSources().prepare_answer_sources(nodes=references)
     except Exception:
-        response = "Sorry, We cannot process your question at the moment."
+        response = QUERY_ERROR_MESSAGE
         answer_sources = None
         logging.error(
             f"Errors raised while processing the question for community: {community_id}!"