app.py

import gradio as gr
from datasets import load_dataset
import pandas as pd
import matplotlib.pyplot as plt

# ---------- CONFIGURATION ---------- #
DATASET_NAME = "ag_news"
LABEL_MAPPING = {
    0: "World",
    1: "Sports",
    2: "Business",
    3: "Science/Technology"
}
SPLIT = "train[:1000]"  # For demo purposes, only use a subset

# ---------- DATA LOADING ---------- #
def load_data():
    dataset = load_dataset(DATASET_NAME, split=SPLIT)
    df = pd.DataFrame(dataset)
    df["label_name"] = df["label"].map(LABEL_MAPPING)
    return df

df = load_data()

# ---------- APP FUNCTIONALITY ---------- #
def get_data_preview(n_rows: int = 5) -> pd.DataFrame:
    """Return the top n rows of the dataset."""
    return df.head(n_rows)

def get_label_distribution_plot():
    """Return a bar chart of label distribution."""
    counts = df["label_name"].value_counts().sort_index()
    fig, ax = plt.subplots()
    counts.plot(kind="bar", ax=ax, color="#4C72B0")
    ax.set_title("Label Distribution")
    ax.set_ylabel("Number of Samples")
    ax.set_xlabel("Category")
    ax.grid(axis="y", linestyle="--", alpha=0.7)
    plt.tight_layout()
    return fig

# ---------- UI LAYOUT ---------- #
with gr.Blocks(title="AG News Dataset Explorer") as demo:
    gr.Markdown("""
    # 🧠 AG News Dataset Explorer  
    Explore a cleaned and labeled version of the AG News dataset using Hugging Face `datasets`.

    - View sample records  
    - Understand label distribution  
    - Great for data engineers & NLP practitioners  
    """)

    with gr.Tab("📄 Data Preview"):
        with gr.Row():
            n_slider = gr.Slider(1, 20, value=5, label="Number of rows to preview")
            preview_btn = gr.Button("Show Preview")
        preview_table = gr.Dataframe(label="Sample Rows")
        preview_btn.click(get_data_preview, inputs=n_slider, outputs=preview_table)

    with gr.Tab("📊 Label Distribution"):
        dist_btn = gr.Button("Generate Distribution Plot")
        dist_plot = gr.Plot(label="Label Distribution")
        dist_btn.click(get_label_distribution_plot, outputs=dist_plot)

    gr.Markdown("""
    ---
    🔗 **Dataset**: [AG News on Hugging Face](https://huggingface.co/datasets/ag_news)  
    ✨
    """)

# ---------- LAUNCH ---------- #
demo.launch()