app.py

import gradio as gr
import subprocess
import sys
import os
import threading
import time

class Logger:
    def __init__(self, filename):
        self.terminal = sys.stdout
        self.log = open(filename, "w")

    def write(self, message):
        self.terminal.write(message)
        self.log.write(message)
        self.log.flush()

    def flush(self):
        self.terminal.flush()
        self.log.flush()

    def isatty(self):
        return False

log_file = "bigcodebench_output.log"
sys.stdout = Logger(log_file)

default_command = "bigcodebench.evaluate"
is_running = False

def generate_command(
    jsonl_file, split, subset, save_pass_rate, parallel,
    min_time_limit, max_as_limit, max_data_limit, max_stack_limit,
    check_gt_only, no_gt
):
    command = [default_command]
    
    if jsonl_file is not None:
        samples = os.path.basename(jsonl_file.name)
        command.extend(["--samples", samples])
    
    command.extend(["--split", split, "--subset", subset])
    
    if save_pass_rate:
        command.append("--save_pass_rate")
    
    if parallel is not None and parallel != 0:
        command.extend(["--parallel", str(int(parallel))])
    
    command.extend([
        "--min-time-limit", str(min_time_limit),
        "--max-as-limit", str(int(max_as_limit)),
        "--max-data-limit", str(int(max_data_limit)),
        "--max-stack-limit", str(int(max_stack_limit))
    ])
    
    if check_gt_only:
        command.append("--check-gt-only")
    
    if no_gt:
        command.append("--no-gt")
    
    return " ".join(command)

def run_bigcodebench(command):
    global is_running
    is_running = True
    print(f"Executing command: {command}")
    
    process = subprocess.Popen(command.split(), stdout=subprocess.PIPE, stderr=subprocess.STDOUT, universal_newlines=True)
    
    for line in process.stdout:
        print(line, end='')
    
    process.wait()
    
    if process.returncode != 0:
        print(f"Error: Command exited with status {process.returncode}")
    
    cleanup_command = "pids=$(ps -u $(id -u) -o pid,comm | grep 'bigcodebench' | awk '{print $1}'); if [ -n \"$pids\" ]; then echo $pids | xargs -r kill; fi; rm -rf /tmp/*"
    subprocess.run(cleanup_command, shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
    
    is_running = False

def read_logs():
    with open(log_file, "r") as f:
        return f.read()

with gr.Blocks() as demo:
    gr.Markdown("# BigCodeBench Evaluator")
    
    with gr.Row():
        jsonl_file = gr.File(label="Upload JSONL file", file_types=[".jsonl"])
        split = gr.Dropdown(choices=["complete", "instruct"], label="Split", value="complete")
        subset = gr.Dropdown(choices=["full", "hard"], label="Subset", value="full")
    
    with gr.Row():
        save_pass_rate = gr.Checkbox(label="Save Pass Rate")
        parallel = gr.Number(label="Parallel (optional)", precision=0)
        min_time_limit = gr.Number(label="Min Time Limit", value=1, precision=1)
        max_as_limit = gr.Number(label="Max AS Limit", value=128*1024, precision=0)
    
    with gr.Row():
        max_data_limit = gr.Number(label="Max Data Limit", value=4*1024, precision=0)
        max_stack_limit = gr.Number(label="Max Stack Limit", value=5, precision=0)
        check_gt_only = gr.Checkbox(label="Check GT Only")
        no_gt = gr.Checkbox(label="No GT")
    
    command_output = gr.Textbox(label="Command", lines=2, value=default_command, interactive=False)
    submit_btn = gr.Button("Run Evaluation")
    log_output = gr.Textbox(label="Execution Logs", lines=10)
    
    def update_command(*args):
        return generate_command(*args)
    
    input_components = [
        jsonl_file, split, subset, save_pass_rate, parallel,
        min_time_limit, max_as_limit, max_data_limit, max_stack_limit,
        check_gt_only, no_gt
    ]
    
    for component in input_components:
        component.change(update_command, inputs=input_components, outputs=command_output)
    
    def on_submit(command):
        global is_running
        if is_running:
            return "A command is already running. Please wait for it to finish."
        threading.Thread(target=run_bigcodebench, args=(command,), daemon=True).start()
        return "Evaluation started. Please wait for the logs to update..."
    
    submit_btn.click(on_submit, inputs=[command_output], outputs=[log_output])
    
    demo.load(read_logs, None, log_output, every=1)

if __name__ == "__main__":
    demo.queue(max_size=300).launch(server_name="0.0.0.0", server_port=7860)