app.py

import streamlit as st
import joblib

# ✅ MUST be first Streamlit command
st.set_page_config(page_title="Stack Overflow Tag Predictor", layout="centered")

# ✅ Define your custom tokenizer BEFORE loading the vectorizer
def customs_tokenization(text):
    return text.split()

# --- Load model and vectorizers ---
@st.cache_resource
def load_pickle(path):
    return joblib.load(path)

model = load_pickle("compressed_logistic_reg1 (1).pkl")
vectorizer = load_pickle("tfidf_vectorizer (2).pkl")
mlb = load_pickle("multilabel_binarizer (1).pkl")

# --- Prediction Function ---
def predict_tags(title, description):
    title, description = title.strip(), description.strip()

    if not title or not description:
        return "⚠️ Please enter both title and description."

    full_text = title + " " + description
    vectorized = vectorizer.transform([full_text])
    prediction = model.predict(vectorized)
    tags = mlb.inverse_transform(prediction)

    if tags and tags[0]:
        return "✅ Predicted Tags: " + ", ".join(tags[0])
    else:
        return "ℹ️ No tags predicted. Try refining your question."


# --- Streamlit UI ---
st.title("🔖 Stack Overflow Tag Predictor")
st.markdown("Enter a question title and description to receive relevant tag suggestions.")

# Input fields
title_input = st.text_input("Question Title", placeholder="e.g. How to merge dictionaries in Python?")
desc_input = st.text_area("Question Description", placeholder="e.g. I have two dictionaries and I want to merge them...")

# Predict button
if st.button("Predict Tags"):
    result = predict_tags(title_input, desc_input)
    st.markdown(f"### Result:\n{result}")