import gradio as gr
import subprocess
import os

def run_sample_trial():
    # Sanity check to make sure the sample data exists
    if not os.path.exists("candidates.jsonl"):
        return "Error: Sample dataset 'candidates.jsonl' is missing from the container."
        
    yield "Initializing Docker Container Sandbox Env...\nSpawning pipeline subprocess...\n\n"
    
    # Trigger your actual optimized script file via terminal sub-process
        # Trigger your actual optimized script file via terminal sub-process
    process = subprocess.Popen(
        [
            "python", "run_pipeline.py", 
            "--candidates", "candidates.jsonl", 
            "--ranker", "regressor_no_prescore.pkl"
        ],
        stdout=subprocess.PIPE,
        stderr=subprocess.STDOUT,
        text=True
    )

    
    # Stream the console print blocks to the web text box line-by-line in real-time
    output_log = ""
    for line in process.stdout:
        output_log += line
        yield output_log

# Build the layout with a text output console block and a single Run button
with gr.Blocks(title="Redrob Ranker Sandbox") as demo:
    gr.Markdown("# 🚀 Redrob Candidate Ranking Pipeline Sandbox")
    gr.Markdown("Click the button below to execute the end-to-end processing pipeline on a sample data subset inside this container terminal.")
    
    run_btn = gr.Button("▶ Run Pipeline Trial", variant="primary")
    console_output = gr.Textbox(label="Terminal Live Output Logs", lines=22, max_lines=30, interactive=False)
    
    run_btn.click(fn=run_sample_trial, outputs=console_output)

# Launch on port 7860 to clear the Hugging Face health checks
demo.queue().launch(server_name="0.0.0.0", server_port=7860)