Create quran_search.py

tools/quran_search.py

@@ -0,0 +1,40 @@
+import pandas as pd
+from sentence_transformers import SentenceTransformer
+from sklearn.metrics.pairwise import cosine_similarity
+import numpy as np
+
+class QuranSearchEngine:
+    def __init__(self):
+        self.data_loaded = False
+        
+    def load_data(self):
+        """Lazy load data and model"""
+        if not self.data_loaded:
+            # Load Quran data
+            self.quran_df = pd.read_csv("https://raw.githubusercontent.com/mafahim/quran-json/main/quran_clean.csv")
+            
+            # Load model
+            self.model = SentenceTransformer(
+                'sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2'
+            )
+            
+            # Pre-compute embeddings
+            self.verse_embeddings = self.model.encode(self.quran_df['text'].tolist())
+            self.data_loaded = True
+    
+    def search(self, query, top_k=5):
+        self.load_data()
+        query_embedding = self.model.encode([query])
+        similarities = cosine_similarity(query_embedding, self.verse_embeddings)[0]
+        top_indices = np.argsort(similarities)[-top_k:][::-1]
+        
+        results = []
+        for idx in top_indices:
+            verse = self.quran_df.iloc[idx]
+            results.append({
+                "surah": verse['surah'],
+                "ayah": verse['ayah'],
+                "text": verse['text'],
+                "similarity": f"{similarities[idx]:.2f}"
+            })
+        return results