Update app/rag.py

app/rag.py

@@ -8,7 +8,8 @@ from pympler import asizeof
 ayat_arabic = pd.read_csv("app/data/ayas.csv", encoding="utf-8")
 ayat_arabic_list = ayat_arabic['arabic'].tolist()
 ayat_english_list = ayat_arabic['answers'].tolist()  # assumes you have an 'english' column
-
+def normalize(s: str):
+    return re.sub(r'[^a-zA-Z0-9 ]', '', s).lower().strip()
 def find_top_5_ayahs_qdrant(question: str):
     q_emb = model.encode(question).tolist()
     search_result = client.search(
@@ -24,7 +25,11 @@ def find_top_5_ayahs_qdrant(question: str):
 
         # Try to find the Arabic equivalent by matching the English verse
         print(f"{english_ayah.strip}")
-        match = ayat_arabic[ayat_arabic['answers'].str.strip() == english_ayah.strip()]
+        normalized_english_ayah = normalize(english_ayah)
+        normalized_df = ayat_arabic.copy()
+        normalized_df['normalized'] = ayat_arabic['answers'].apply(normalize)
+
+        match = normalized_df[normalized_df['normalized'] == normalized_english_ayah]
         arabic_ayah = match.iloc[0]['arabic'] if not match.empty else "❌ Not found"
 
         results.append({