Update app.py

app.py

@@ -1,74 +1,60 @@
-import pandas as pd
-from sentence_transformers import SentenceTransformer
-import faiss
 import streamlit as st
+import pandas as pd
+from sentence_transformers import SentenceTransformer, util
 from gtts import gTTS
-import base64
-from io import BytesIO
-
-@st.cache_resource
-def load_data_and_model():
-    df = pd.read_csv("Bhagwad_Gita.csv")
-
-    documents = []
-    for _, row in df.iterrows():
-        doc = {
-            "verse": f"Chapter {row['Chapter']}, Verse {row['Verse']}",
-            "shloka": row['Shloka'],
-            "eng": row['EngMeaning'],
-            "hindi": row.get('HindiMeaning', 'हिंदी अनुवाद उपलब्ध नहीं है।')
-        }
-        documents.append(doc)
-
-    texts = [f"{doc['verse']}\n{doc['shloka']}\n{doc['eng']}" for doc in documents]
-    model = SentenceTransformer('all-MiniLM-L6-v2')
-    embeddings = model.encode(texts, show_progress_bar=False)
-    index = faiss.IndexFlatL2(embeddings[0].shape[0])
-    index.add(embeddings)
-
-    return model, index, documents
-
-def generate_audio(text):
-    tts = gTTS(text)
-    buf = BytesIO()
-    tts.write_to_fp(buf)
-    return buf.getvalue()
+import tempfile
+import os
 
-def render_audio(audio_bytes):
-    b64 = base64.b64encode(audio_bytes).decode()
-    md = f"""
-        <audio controls>
-        <source src="data:audio/mp3;base64,{b64}" type="audio/mp3">
-        </audio>
-    """
-    st.markdown(md, unsafe_allow_html=True)
+# Load the Gita dataset
+@st.cache_data
+def load_data():
+    return pd.read_csv("Bhagwad_Gita.csv")
 
-# Load model & data
-model, index, documents = load_data_and_model()
+data = load_data()
 
-# UI Design
-st.set_page_config(page_title="🕉️ Geeta GPT", layout="centered")
-st.image("https://upload.wikimedia.org/wikipedia/commons/thumb/d/d6/Lord_krishna_with_flute.jpg/640px-Lord_krishna_with_flute.jpg", use_column_width=True)
-st.title("🕉️ Geeta GPT - Ask Bhagavad Gita")
-st.markdown("🙏 Ask questions about life, karma, fear, love, or devotion... Let Shri Krishna guide you!")
-
-query = st.text_input("🔎 Your Question:")
-
-if query:
-    st.markdown("🙏 **Jai Shri Krishna!** Let us seek wisdom from the Gita...")
-
-    query_embedding = model.encode([query])
-    D, I = index.search(query_embedding, k=3)
-
-    for i in I[0]:
-        doc = documents[i]
-        st.markdown(f"### 📜 {doc['verse']}")
-        st.markdown(f"**🔸 Shloka:**\n{doc['shloka']}")
-        st.markdown(f"**📖 English Meaning:**\n{doc['eng']}")
-        st.markdown(f"**🇮🇳 Hindi Meaning:**\n{doc['hindi']}")
-
-        # Audio
-        audio_bytes = generate_audio(doc['eng'])
-        render_audio(audio_bytes)
-
-        st.markdown("---")
+# Load the embedding model
+@st.cache_resource
+def load_model():
+    return SentenceTransformer('all-MiniLM-L6-v2')
+
+model = load_model()
+
+# Preprocess verses and create embeddings
+@st.cache_data
+def get_embeddings(data):
+    verses = data['Verse'].astype(str).tolist()
+    embeddings = model.encode(verses, convert_to_tensor=True)
+    return verses, embeddings
+
+verses, verse_embeddings = get_embeddings(data)
+
+# App Title
+st.title("🙏 GeetaGPT – Divine Wisdom from the Bhagavad Gita")
+
+# User Input
+user_question = st.text_input("Ask your question to Lord Krishna:")
+
+if user_question:
+    # Embed the question
+    question_embedding = model.encode(user_question, convert_to_tensor=True)
+
+    # Compute cosine similarities
+    scores = util.pytorch_cos_sim(question_embedding, verse_embeddings)[0]
+    best_idx = scores.argmax().item()
+
+    # Get the best matching verse
+    matched_verse = verses[best_idx]
+    chapter = data.iloc[best_idx]['Chapter']
+    verse_number = data.iloc[best_idx]['Verse Number']
+
+    # Greet and show result
+    greeting = "🕉️ Jai Shri Krishna!\n\n"
+    st.markdown(f"{greeting}**Chapter {chapter}, Verse {verse_number}:**\n\n*{matched_verse}*")
+
+    # Optional: Generate audio
+    if st.checkbox("🔊 Hear it aloud"):
+        tts = gTTS(text=matched_verse, lang='en')
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as fp:
+            tts.save(fp.name)
+            st.audio(fp.name, format="audio/mp3")
+            os.remove(fp.name)