src/streamlit_app.py

import os
import joblib
import pandas as pd
import streamlit as st

st.set_page_config(
    page_title="Toxic Comment Classifier",
    page_icon="☣️",
    layout="centered"
)

BASE_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
MODEL_PATH = os.path.join(BASE_DIR, "toxic_model.pkl")
VECTORIZER_PATH = os.path.join(BASE_DIR, "toxic_vectorizer.pkl")
COLUMNS_PATH = os.path.join(BASE_DIR, "toxic_columns.pkl")


@st.cache_resource
def load_artifacts():
    model = joblib.load(MODEL_PATH)
    vectorizer = joblib.load(VECTORIZER_PATH)
    label_columns = joblib.load(COLUMNS_PATH)
    return model, vectorizer, label_columns


model, vectorizer, label_columns = load_artifacts()

st.title("☣️ Toxic Comment Classifier")
st.write(
    "Enter a comment and predict whether it belongs to one or more toxicity categories."
)

comment = st.text_area(
    "Comment Text",
    placeholder="Type or paste a comment here..."
)

if st.button("Predict"):
    if not comment.strip():
        st.warning("Please enter a comment.")
    else:
        text_vec = vectorizer.transform([comment])
        pred = model.predict(text_vec)[0]

        result_df = pd.DataFrame({
            "Label": label_columns,
            "Prediction": pred
        })

        st.subheader("Prediction Results")
        st.dataframe(result_df, use_container_width=True)

        positive_labels = result_df.loc[result_df["Prediction"] == 1, "Label"].tolist()

        if positive_labels:
            st.error("Detected labels: " + ", ".join(positive_labels))
        else:
            st.success("No toxic label detected.")

        st.subheader("Quick Summary")
        st.write(
            {
                "toxic_labels_count": int(result_df["Prediction"].sum()),
                "labels_detected": positive_labels
            }
        )