app.py

import streamlit as st
from datetime import datetime
from pymongo import MongoClient
import pandas as pd
from transformers import pipeline

# =============================================================================
#  AI CLASSIFIER LOGIC
# =============================================================================

# Load the pipeline once using caching for efficiency
@st.cache_resource
def load_classifier():
    """Loads the zero-shot classification model."""
    return pipeline(
        "zero-shot-classification",
        model="joeddav/xlm-roberta-large-xnli",
        cache_dir="./hf_cache"  # Prevents permission errors
    )

classifier = load_classifier()

# Define the candidate labels
CATEGORIES = {
    "Family": "కుటుంబం", "Friendship": "స్నేహం", "Morality": "నీతి",
    "Hard Work": "శ్రమ", "Knowledge": "జ్ఞానం", "Devotion": "భక్తి",
    "Culture": "సంస్కృతి", "Literature": "సాహిత్యం", "Humility": "వినయం",
    "Patience": "సహనం", "Courage": "ధైర్యం", "Arrogance": "అహంకారం",
    "Love": "ప్రేమ", "Greed": "దురాశ", "Wisdom": "ఆలోచన",
    "Responsibility": "బాధ్యత", "Satire": "వ్యంగ్యం", "Politics": "రాజకీయం",
    "Wealth": "ధనము", "Time": "సమయం"
}

def classify_proverb(text):
    """Classifies the proverb and returns the Telugu label."""
    result = classifier(text, list(CATEGORIES.keys()))
    top_label = result["labels"][0]
    return CATEGORIES[top_label]

# =============================================================================
#  DATABASE LOGIC
# =============================================================================

# Use @st.cache_resource to establish the database connection only once.
@st.cache_resource
def init_connection():
    """Initializes a connection to the MongoDB database."""
    return MongoClient(st.secrets["MONGO_URI"])

client = init_connection()
collection = client["proverbs_and_meanings"]["proverbs"]

# Use @st.cache_data to load data only when it changes.
@st.cache_data
def load_proverbs():
    """Retrieves all proverbs, sorted by newest first."""
    data = list(collection.find({}, {'_id': 0}).sort("Timestamp", -1))
    return pd.DataFrame(data) if data else pd.DataFrame()

def save_proverb(proverb, meaning):
    """Saves a new proverb to the database and clears the data cache."""
    # This now calls the classify_proverb function defined above
    category = classify_proverb(proverb + " " + meaning)
    new_entry = {
        "Proverb": proverb.strip(),
        "Meaning": meaning.strip(),
        "Category": category,
        "Timestamp": datetime.now()
    }
    collection.insert_one(new_entry)
    load_proverbs.clear()

# =============================================================================
#  STREAMLIT UI & APP LOGIC
# =============================================================================

# --- Page Configuration ---
st.set_page_config(
    page_title="తెలుగు సామెతల ఖజానా", page_icon="📜", layout="wide"
)

# --- Title & Introduction ---
st.title("📜 తెలుగు సామెతల ఖజానా")
st.markdown("### మీకు తెలిసిన సామెతను పంచుకోండి ✍️")
st.markdown("వెనుక తరాల నుంచి ముందుకు, మనం భద్రపరిద్దాం మన తెలుగు జ్ఞానం.")
st.markdown("<hr style='border:1px solid #555'>", unsafe_allow_html=True)

# --- Main Layout (Two Columns) ---
col1, col2 = st.columns([1, 2], gap="large")

# --- Column 1: Submission Form ---
with col1:
    st.subheader("📥 కొత్త సామెతను జత చేయండి")
    with st.form("proverb_form", clear_on_submit=True):
        proverb = st.text_area("**సామెత (Proverb)**", height=100, max_chars=150)
        meaning = st.text_area("**భావం (Meaning)**", height=100, max_chars=500)
        
        submitted = st.form_submit_button("✅ సామెతను జత చేయండి", use_container_width=True, type="primary")

        if submitted:
            if proverb and meaning:
                save_proverb(proverb, meaning)
                st.success("ధన్యవాదాలు! మీ సామెత జత చేయబడింది.")
                st.balloons()
            else:
                st.error("దయచేసి సామెత మరియు భావం రెండూ నమోదు చేయండి.")

# --- Column 2: Gallery Section ---
with col2:
    st.subheader("🗂️ సామెతల గ్యాలరీ")
    df = load_proverbs()
    st.markdown(f"🔢 ఇప్పటివరకు మొత్తం **{len(df)}** సామెతలు జత చేయబడ్డాయి.")

    if df.empty:
        st.info("ఇప్పటి వరకు ఎలాంటి సామెతలు జత చేయలేదు. మీరే మొదటిది జత చేయండి!")
    else:
        for _, row in df.iterrows():
            # This block safely handles both datetime objects and old string formats
            timestamp_str = "N/A"
            timestamp_val = row.get('Timestamp')
            if isinstance(timestamp_val, datetime):
                timestamp_str = timestamp_val.strftime('%d %B %Y, %I:%M %p')
            elif timestamp_val:
                try:
                    # Try to convert old string format to datetime
                    timestamp_obj = datetime.strptime(str(timestamp_val), '%Y-%m-%d %H:%M:%S')
                    timestamp_str = timestamp_obj.strftime('%d %B %Y, %I:%M %p')
                except (ValueError, TypeError):
                    # If conversion fails, just show the original string
                    timestamp_str = str(timestamp_val)

            st.markdown(f"""
            <div style='border: 1px solid #444; padding: 15px; border-radius: 10px; margin-bottom: 15px; background-color: #0e1117;'>
                <h5>📝 {row['Proverb']}</h5>
                <b>భావం:</b> {row['Meaning']}<br>
                <span style='color:#bbb;'>🏷️ శ్రేణి: {row.get('Category', 'N/A')}</span><br>
                <span style='font-size: 0.8em; color: gray;'>🕒 {timestamp_str}</span>
            </div>
            """, unsafe_allow_html=True)

# --- Footer ---
st.markdown("---")
st.markdown("<center><small>© తెలుగు సామెతల ఖజానా – viswam.ai</small></center>", unsafe_allow_html=True)