MrSimple01 commited on
Commit
b03f5ea
·
verified ·
1 Parent(s): e20de14

Update utils.py

Browse files
Files changed (1) hide show
  1. utils.py +14 -5
utils.py CHANGED
@@ -195,14 +195,23 @@ def debug_search_tables(vector_index, search_term="С-25"):
195
 
196
  return matching
197
 
 
 
 
 
198
  def answer_question(question, query_engine, reranker, current_model, chunks_df=None):
 
 
 
199
  if query_engine is None:
200
  return "<div style='background-color: #e53e3e; color: white; padding: 20px; border-radius: 10px;'>Система не инициализирована</div>", "", ""
201
 
202
  try:
203
  start_time = time.time()
204
- retrieved_nodes = query_engine.retriever.retrieve(question)
 
205
  log_message(f"user query: {question}")
 
206
 
207
 
208
  log_message(f"RETRIEVED: {len(retrieved_nodes)} nodes")
@@ -218,11 +227,11 @@ def answer_question(question, query_engine, reranker, current_model, chunks_df=N
218
  log_message(f" [{i+1}] {doc_id} - Table {table_num}: {table_title[:50]}")
219
  log_message(f"UNIQUE NODES: {len(unique_retrieved)} nodes")
220
 
221
- # Simple reranking
222
- reranked_nodes = rerank_nodes(question, unique_retrieved, reranker, top_k=20)
223
 
224
- # Direct query without formatting
225
- response = query_engine.query(question)
226
 
227
  end_time = time.time()
228
  processing_time = end_time - start_time
 
195
 
196
  return matching
197
 
198
+ # Add this import at the top of utils.py
199
+ from documents_prep import normalize_text
200
+
201
+ # MODIFIED: Update answer_question function
202
  def answer_question(question, query_engine, reranker, current_model, chunks_df=None):
203
+ # NORMALIZE the question to convert C to С
204
+ normalized_question = normalize_text(question)
205
+
206
  if query_engine is None:
207
  return "<div style='background-color: #e53e3e; color: white; padding: 20px; border-radius: 10px;'>Система не инициализирована</div>", "", ""
208
 
209
  try:
210
  start_time = time.time()
211
+ # Use NORMALIZED question for retrieval
212
+ retrieved_nodes = query_engine.retriever.retrieve(normalized_question)
213
  log_message(f"user query: {question}")
214
+ log_message(f"normalized query: {normalized_question}")
215
 
216
 
217
  log_message(f"RETRIEVED: {len(retrieved_nodes)} nodes")
 
227
  log_message(f" [{i+1}] {doc_id} - Table {table_num}: {table_title[:50]}")
228
  log_message(f"UNIQUE NODES: {len(unique_retrieved)} nodes")
229
 
230
+ # Simple reranking with NORMALIZED question
231
+ reranked_nodes = rerank_nodes(normalized_question, unique_retrieved, reranker, top_k=20)
232
 
233
+ # Direct query without formatting - use normalized question
234
+ response = query_engine.query(normalized_question)
235
 
236
  end_time = time.time()
237
  processing_time = end_time - start_time