data_explorer.py

import gradio as gr
import pandas as pd
from datasets import load_dataset
import os

# --- Backend Functions ---

def preview_dataset(dataset_id, config_name, split_name, num_rows):
    """
    Streams the first N rows of a dataset without downloading the whole thing.
    """
    if not dataset_id:
        return pd.DataFrame(), "⚠️ Please enter a Dataset ID."
    
    try:
        # If config is empty, assume None or 'default'
        config_name = config_name if config_name.strip() else None
        
        # Stream the dataset
        ds_stream = load_dataset(
            dataset_id, 
            name=config_name, 
            split=split_name, 
            streaming=True
        )
        
        # Get first N rows
        data_head = list(ds_stream.take(int(num_rows)))
        df = pd.DataFrame(data_head)
        
        return df, f"✅ Successfully loaded first {num_rows} rows from {dataset_id}"
        
    except Exception as e:
        return pd.DataFrame(), f"❌ Error: {str(e)}"

def upload_to_hub(files, repo_name, is_private):
    """
    Placeholder function for development.
    The Token field was removed, so this functionality is currently disabled.
    """
    if not files:
        return "⚠️ No files selected."
    
    # Just return a message for now since we removed the token logic
    return f"🛠️ Dev Mode: Upload to '{repo_name}' triggered (functionality disabled)."

# --- Frontend Interface ---

def render_explorer():
    """
    Returns the layout (Column) and the back button.
    """
    with gr.Column(visible=False) as explorer_view:
        
        gr.Markdown("# 🤗 Hugging Face Dataset Manager")
        
        with gr.Tabs():
            
            # TAB 1: EXPLORE
            with gr.TabItem("🔍 Explore Datasets"):
                gr.Markdown("Preview any dataset on the Hub without downloading it.")
                
                with gr.Row():
                    in_id = gr.Textbox(label="Dataset ID", placeholder="e.g. imdb, glue, common_voice", value="imdb")
                    in_config = gr.Textbox(label="Config / Subset", placeholder="e.g. default", value="")
                    in_split = gr.Textbox(label="Split", value="train")
                    in_rows = gr.Slider(minimum=5, maximum=100, value=10, step=5, label="Rows to Preview")
                
                btn_preview = gr.Button("Load Preview", variant="primary")
                out_status = gr.Markdown()
                out_df = gr.Dataframe(label="Dataset Preview", wrap=True)
                
                btn_preview.click(
                    fn=preview_dataset, 
                    inputs=[in_id, in_config, in_split, in_rows], 
                    outputs=[out_df, out_status]
                )
    
            # TAB 2: UPLOADER
            with gr.TabItem("☁️ Upload New Dataset"):
                gr.Markdown("Upload CSV, JSON, or Parquet files to create a new dataset.")
                
                # --- TOKEN FIELD REMOVED ---
                
                with gr.Row():
                    in_repo = gr.Textbox(label="New Repo Name", placeholder="username/my-new-dataset")
                    in_private = gr.Checkbox(label="Private Dataset", value=True)
                
                in_files = gr.File(label="Drag & Drop Files", file_count="multiple")
                btn_upload = gr.Button("Push to Hub", variant="primary")
                out_upload_status = gr.Markdown()
                
                # Updated inputs to exclude the token
                btn_upload.click(
                    fn=upload_to_hub,
                    inputs=[in_files, in_repo, in_private],
                    outputs=[out_upload_status]
                )

        gr.Markdown("---")
        
        btn_back = gr.Button("⬅️ Back to Home", variant="secondary")

    return explorer_view, btn_back

if __name__ == "__main__":
    with gr.Blocks() as demo:
        view, _ = render_explorer()
        view.visible = True 
    demo.launch()