src/streamlit_app.py

import streamlit as st
import torch
import json
import os
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import numpy as np

# ==========================
# LABEL MAPPING
# ==========================
LABEL_MAPPING = {
    0: "51-100",
    1: "101-150", 
    2: "151-200",
    3: "201-250",
    4: "251-500",
    5: "501-1000",
    6: "1001-2000",
    7: "2001-5000",
    8: "5001-10000",
    9: "10001-100000",
    10: "100001-200000",
    11: "200001+"
}

# ==========================
# PAGE CONFIG
# ==========================
st.set_page_config(
    page_title="KVE",
    page_icon="📊",
    layout="centered"
)

hide_streamlit_style = """
<style>
#MainMenu {visibility: hidden;}
</style>
"""
st.markdown(hide_streamlit_style, unsafe_allow_html=True)

st.logo(
    image="https://dejan.ai/wp-content/uploads/2024/02/dejan-300x103.png",
    link="https://dejan.ai/",
    size="large"
)

# ==========================
# LOAD MODEL
# ==========================
@st.cache_resource
def load_model():
    """Load model and tokenizer from Hugging Face"""
    model_name = "dejanseo/qve-klook"
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = AutoModelForSequenceClassification.from_pretrained(model_name)
    model.eval()
    return model, tokenizer

# ==========================
# PREDICTION
# ==========================
def predict(query, model, tokenizer):
    inputs = tokenizer(query, padding="max_length", truncation=True, max_length=20, return_tensors="pt")
    with torch.no_grad():
        outputs = model(**inputs)
        probabilities = torch.softmax(outputs.logits, dim=-1).squeeze().numpy()
    predicted_class = np.argmax(probabilities)
    confidence = probabilities[predicted_class]
    return predicted_class, confidence, probabilities

# ==========================
# CONFIDENCE INDICATOR
# ==========================
def get_confidence_indicator(confidence):
    if confidence > 0.8:
        return "🟢", "High Confidence", "#4CAF50"
    elif confidence > 0.5:
        return "🟡", "Moderate Confidence", "#FFC107"
    else:
        return "🔴", "Low Confidence", "#f44336"

# ==========================
# MAIN APP
# ==========================
def main():
    try:
        with st.spinner("Loading model..."):
            model, tokenizer = load_model()
        
        query = st.text_input(
            "Enter search query to classify:",
            placeholder="Type your search query here...",
            key="input_text"
        )
        
        col1, col2, col3 = st.columns([1, 1, 2])
        with col1:
            classify_btn = st.button("🔍 Classify", type="primary", use_container_width=True)
        with col2:
            clear_btn = st.button("🗑️ Clear", use_container_width=True)
        
        if clear_btn:
            st.rerun()
        
        if classify_btn and query:
            with st.spinner("Analyzing..."):
                predicted_class, confidence, probabilities = predict(query, model, tokenizer)
            
            predicted_label = LABEL_MAPPING.get(predicted_class, f"Class {predicted_class}")
            emoji, conf_text, color = get_confidence_indicator(confidence)
            
            result_html = f"""
            <div style="
                background: linear-gradient(135deg, {color}22 0%, {color}11 100%);
                border-left: 4px solid {color};
                padding: 20px;
                border-radius: 10px;
                margin: 10px 0;
            ">
                <h2 style="margin: 0; color: {color};">
                    {emoji} {predicted_label}
                </h2>
            </div>
            """
            st.markdown(result_html, unsafe_allow_html=True)
            
            st.markdown("Search Volume Probabilities")
            
            top_10_idx = np.argsort(probabilities)[::-1][:10]
            for rank, idx in enumerate(top_10_idx, 1):
                label_name = LABEL_MAPPING.get(idx, f"Class {idx}")
                prob = probabilities[idx]
                bar_width = int(prob * 100)
                bar_color = color if idx == predicted_class else "#B0BEC5"
                
                st.markdown(f"""
                <div style="margin-bottom: 8px;">
                    <div style="display: flex; align-items: center;">
                        <span style="width: 30px;">#{rank}</span>
                        <span style="flex: 1; font-weight: {'bold' if idx == predicted_class else 'normal'};">
                            {label_name}
                        </span>
                        <span style="width: 60px; text-align: right;">{prob:.1%}</span>
                    </div>
                    <div style="background: #E0E0E0; height: 4px; border-radius: 2px; margin-top: 2px;">
                        <div style="background: {bar_color}; width: {bar_width}%; height: 100%; border-radius: 2px;"></div>
                    </div>
                </div>
                """, unsafe_allow_html=True)
        
        elif classify_btn and not query:
            st.warning("⚠️ Please enter some text to classify")
    
    except Exception as e:
        st.error(f"❌ Error: {str(e)}")

if __name__ == "__main__":
    main()