Update app/rag.py

app/rag.py

@@ -1,38 +1,36 @@
-from app.data_loader import model
-from sklearn.metrics.pairwise import cosine_similarity
-from app.qdrant_client import client
-
-from pympler import asizeof
-"""
-def find_top_5_ayahs(question: str):
-    q_emb = model.encode(question)
-    sims = cosine_similarity([q_emb], embeddings)[0]
-    top_indices = sims.argsort()[-5:][::-1]
-
-    results = []
-    for i in top_indices:
-        ayah_text = ayat[i]  # single column with the ayah text
-        results.append({
-            "question": question,   # the question from the request
-            "answer": ayah_text     # the ayah text
-        })
-    return results
-"""
-# find top 5 ayahs from Qdrant (alternative method)
-def find_top_5_ayahs_qdrant(question: str):
-    q_emb = model.encode(question).tolist()
-    search_result = client.search(
-        collection_name="ayahs_collection",
-        query_vector=q_emb,
-        limit=5
-    )
-
-    results = []
-    for point in search_result:
-        ayah_text = point.payload['text']
-        results.append({
-            "question": question,
-            "answer": ayah_text
-        })
-    print("results size : ",asizeof.asizeof(results))
-    return results
+from app.data_loader import model
+from sklearn.metrics.pairwise import cosine_similarity
+from app.qdrant_client import client
+import pandas as pd
+from pympler import asizeof
+
+# Load Arabic ayahs
+ayat_arabic = pd.read_csv("app/data/ayas.csv", encoding="utf-8")
+ayat_arabic_list = ayat_arabic['arabic'].tolist()
+ayat_english_list = ayat_arabic['english'].tolist()  # assumes you have an 'english' column
+
+def find_top_5_ayahs_qdrant(question: str):
+    q_emb = model.encode(question).tolist()
+    search_result = client.search(
+        collection_name="ayahs_collection",
+        query_vector=q_emb,
+        limit=5
+    )
+
+    results = []
+
+    for point in search_result:
+        english_ayah = point.payload['text']
+
+        # Try to find the Arabic equivalent by matching the English verse
+        match = ayat_arabic[ayat_arabic['english'].str.strip() == english_ayah.strip()]
+        arabic_ayah = match.iloc[0]['arabic'] if not match.empty else "❌ Not found"
+
+        results.append({
+            "question": question,
+            "answer": english_ayah,
+            "arabic": arabic_ayah
+        })
+
+    print("results size:", asizeof.asizeof(results))
+    return results