computer-use-ootb-internal 0.0.171__py3-none-any.whl → 0.0.173__py3-none-any.whl

computer_use_ootb_internal/app_teachmode.py

@@ -1,611 +1,637 @@
-import argparse
-import time
-import json
-from datetime import datetime
-import threading
-import requests
-import platform  # Add platform import
-import pyautogui  # Add pyautogui import
-import webbrowser # Add webbrowser import
-import os # Import os for path joining
-import logging # Import logging
-import importlib # For dynamic imports
-import pkgutil # To find modules
-import sys # For logging setup
-import traceback # For logging setup
-from logging.handlers import RotatingFileHandler # For logging setup
-from fastapi import FastAPI, Request
-from fastapi.responses import JSONResponse
-from fastapi.middleware.cors import CORSMiddleware
-from computer_use_ootb_internal.computer_use_demo.tools.computer import get_screen_details
-from computer_use_ootb_internal.run_teachmode_ootb_args import simple_teachmode_sampling_loop
-from computer_use_ootb_internal.computer_use_demo.executor.teachmode_executor import TeachmodeExecutor
-import uvicorn # Assuming uvicorn is used to run FastAPI
-import concurrent.futures
-import asyncio
-
-# --- App Logging Setup ---
-try:
-    # Log to user's AppData directory for better accessibility
-    log_dir_base = os.environ.get('APPDATA', os.path.expanduser('~'))
-    log_dir = os.path.join(log_dir_base, 'OOTBAppLogs')
-    os.makedirs(log_dir, exist_ok=True)
-    log_file = os.path.join(log_dir, 'ootb_app.log')
-
-    log_format = '%(asctime)s - %(levelname)s - %(process)d - %(threadName)s - %(message)s'
-    log_level = logging.INFO # Or logging.DEBUG for more detail
-
-    # Use rotating file handler
-    handler = RotatingFileHandler(log_file, maxBytes=5*1024*1024, backupCount=2, encoding='utf-8')
-    handler.setFormatter(logging.Formatter(log_format))
-
-    # Configure root logger
-    logging.basicConfig(level=log_level, handlers=[handler])
-
-    # Add stream handler to see logs if running interactively (optional)
-    # logging.getLogger().addHandler(logging.StreamHandler(sys.stdout))
-
-    logging.info("="*20 + " OOTB App Starting " + "="*20)
-    logging.info(f"Running with args: {sys.argv}")
-    logging.info(f"Python Executable: {sys.executable}")
-    logging.info(f"Working Directory: {os.getcwd()}")
-    logging.info(f"User: {os.getenv('USERNAME')}")
-
-except Exception as log_setup_e:
-    print(f"FATAL: Failed to set up logging: {log_setup_e}")
-    # Fallback logging might be needed here if file logging fails
-
-# --- End App Logging Setup ---
-
-app = FastAPI()
-
-# Add CORS middleware to allow requests from the frontend
-app.add_middleware(
-    CORSMiddleware,
-    allow_origins=["*"],  
-    allow_credentials=True,
-    allow_methods=["*"],
-    allow_headers=["*"],
-)
-
-# Rate limiter for API endpoints
-class RateLimiter:
-    def __init__(self, interval_seconds=2):
-        self.interval = interval_seconds
-        self.last_request_time = {}
-        self.lock = threading.Lock()
-        
-    def allow_request(self, endpoint):
-        with self.lock:
-            current_time = time.time()
-            # Priority endpoints always allowed
-            if endpoint in ["/update_params", "/update_message"]:
-                return True
-                
-            # For other endpoints, apply rate limiting
-            if endpoint not in self.last_request_time:
-                self.last_request_time[endpoint] = current_time
-                return True
-                
-            elapsed = current_time - self.last_request_time[endpoint]
-            if elapsed < self.interval:
-                return False
-                
-            self.last_request_time[endpoint] = current_time
-            return True
-
-
-def log_ootb_request(server_url, ootb_request_type, data):
-    logging.info(f"OOTB Request: Type={ootb_request_type}, Data={data}")
-    # Keep the requests post for now if it serves a specific purpose
-    logging_data = {
-        "type": ootb_request_type,
-        "data": data,
-        "timestamp": datetime.now().strftime("%Y-%m-%d %H:%M:%S")
-    }
-    if not server_url.endswith("/update_ootb_logging"):
-        server_logging_url = server_url + "/update_ootb_logging"
-    else:
-        server_logging_url = server_url
-    try:
-        requests.post(server_logging_url, json=logging_data, timeout=5)
-    except Exception as req_log_e:
-        logging.warning(f"Could not log ootb request to server {server_logging_url}: {req_log_e}")
-
-
-class SharedState:
-    def __init__(self, args):
-        self.args = args
-        self.task_updated = False
-        self.chatbot_messages = []
-        # Store all state-related data here
-        self.model = args.model
-        self.task = getattr(args, 'task', "")
-        self.selected_screen = args.selected_screen
-        self.user_id = args.user_id
-        self.trace_id = args.trace_id
-        self.api_keys = args.api_keys
-        self.server_url = args.server_url
-        self.message_queue = []
-        self.is_processing = False
-        self.should_stop = False
-        self.is_paused = False
-        # Add a new event to better control stopping
-        self.stop_event = threading.Event()
-        # Add a reference to the processing thread
-        self.processing_thread = None
-
-shared_state = None
-rate_limiter = RateLimiter(interval_seconds=2)
-
-# Set up logging for this module
-log = logging.getLogger(__name__)
-
-def prepare_environment(state):
-    """Dynamically loads and runs preparation logic based on software name."""
-    # TODO: Replace hardcoded software name with value from shared_state when available
-    software_name = "star rail"
-    # Normalize the software name to be a valid Python module name
-    # Replace spaces/hyphens with underscores, convert to lowercase
-    module_name_base = software_name.replace(" ", "_").replace("-", "_").lower()
-    module_to_run = f"{module_name_base}_prepare"
-
-    log.info(f"Attempting preparation for software: '{software_name}' (Module: '{module_to_run}')")
-
-    try:
-        # Construct the full module path within the package
-        prep_package = "computer_use_ootb_internal.preparation"
-        full_module_path = f"{prep_package}.{module_to_run}"
-
-        # Dynamically import the module
-        # Check if module exists first using pkgutil to avoid import errors
-        # Note: pkgutil.find_loader might be deprecated, consider importlib.util.find_spec
-        loader = pkgutil.find_loader(full_module_path)
-        if loader is None:
-            log.warning(f"Preparation module '{full_module_path}' not found. Skipping preparation.")
-            return
-
-        prep_module = importlib.import_module(full_module_path)
-
-        # Check if the module has the expected function
-        if hasattr(prep_module, "run_preparation") and callable(prep_module.run_preparation):
-            log.info(f"Running preparation function from {full_module_path}...")
-            prep_module.run_preparation(state)
-            log.info(f"Preparation function from {full_module_path} completed.")
-        else:
-            log.warning(f"Module {full_module_path} found, but does not have a callable 'run_preparation' function. Skipping.")
-
-    except ModuleNotFoundError:
-        log.warning(f"Preparation module '{full_module_path}' not found. Skipping preparation.")
-    except Exception as e:
-        log.error(f"Error during dynamic preparation loading/execution for '{module_to_run}': {e}", exc_info=True)
-
-
-@app.post("/update_params")
-async def update_parameters(request: Request):
-    logging.info("Received request to /update_params")
-    try:
-        data = await request.json()
-        
-        if 'task' not in data:
-            return JSONResponse(
-                content={"status": "error", "message": "Missing required field: task"},
-                status_code=400
-            )
-        
-        # Clear message histories before updating parameters
-        shared_state.message_queue.clear()
-        shared_state.chatbot_messages.clear()
-        logging.info("Cleared message queue and chatbot messages.")
-
-        shared_state.args = argparse.Namespace(**data)
-        shared_state.task_updated = True
-        
-        # Update shared state when parameters change
-        shared_state.model = getattr(shared_state.args, 'model', "teach-mode-gpt-4o")
-        shared_state.task = getattr(shared_state.args, 'task', "Following the instructions to complete the task.")
-        shared_state.selected_screen = getattr(shared_state.args, 'selected_screen', 0)
-        shared_state.user_id = getattr(shared_state.args, 'user_id', "hero_cases")
-        shared_state.trace_id = getattr(shared_state.args, 'trace_id', "build_scroll_combat")
-        shared_state.api_keys = getattr(shared_state.args, 'api_keys', "sk-proj-1234567890")
-        shared_state.server_url = getattr(shared_state.args, 'server_url', "http://ec2-44-234-43-86.us-west-2.compute.amazonaws.com")
-        
-        log_ootb_request(shared_state.server_url, "update_params", data)
-
-        # Call the (now dynamic) preparation function here, after parameters are updated
-        prepare_environment(shared_state)
-
-        logging.info("Parameters updated successfully.")
-        return JSONResponse(
-            content={"status": "success", "message": "Parameters updated", "new_args": vars(shared_state.args)},
-            status_code=200
-        )
-    except Exception as e:
-        logging.error("Error processing /update_params:", exc_info=True)
-        return JSONResponse(content={"status": "error", "message": "Internal server error"}, status_code=500)
-
-@app.post("/update_message")
-async def update_message(request: Request):
-    data = await request.json()
-    
-    if 'message' not in data:
-        return JSONResponse(
-            content={"status": "error", "message": "Missing required field: message"},
-            status_code=400
-        )
-    
-    log_ootb_request(shared_state.server_url, "update_message", data)
-    
-    message = data['message']
-    # shared_state.chatbot_messages.append({"role": "user", "content": message, "type": "text"})
-    shared_state.task = message
-    shared_state.args.task = message
-    
-    # Reset stop event before starting
-    shared_state.stop_event.clear()
-    
-    # Start processing if not already running
-    if not shared_state.is_processing:
-        # Create and store the thread
-        shared_state.processing_thread = threading.Thread(target=process_input, daemon=True)
-        shared_state.processing_thread.start()
-    
-    return JSONResponse(
-        content={"status": "success", "message": "Message received", "task": shared_state.task},
-        status_code=200
-    )
-
-@app.get("/get_messages")
-async def get_messages(request: Request):
-    # Apply rate limiting
-    if not rate_limiter.allow_request(request.url.path):
-        return JSONResponse(
-            content={"status": "error", "message": "Rate limit exceeded. Try again after 2 seconds."},
-            status_code=429
-        )
-    
-    # log_ootb_request(shared_state.server_url, "get_messages", {})
-    
-    # Return all messages in the queue and clear it
-    messages = shared_state.message_queue.copy()
-    shared_state.message_queue = []
-    
-    return JSONResponse(
-        content={"status": "success", "messages": messages},
-        status_code=200
-    )
-
-@app.get("/get_screens")
-async def get_screens(request: Request):
-    # Apply rate limiting
-    if not rate_limiter.allow_request(request.url.path):
-        return JSONResponse(
-            content={"status": "error", "message": "Rate limit exceeded. Try again after 2 seconds."},
-            status_code=429
-        )
-    
-    log_ootb_request(shared_state.server_url, "get_screens", {})
-    
-    screen_options, primary_index = get_screen_details()
-    
-    return JSONResponse(
-        content={"status": "success", "screens": screen_options, "primary_index": primary_index},
-        status_code=200
-    )
-
-@app.post("/stop_processing")
-async def stop_processing(request: Request):
-    # Apply rate limiting
-    if not rate_limiter.allow_request(request.url.path):
-        return JSONResponse(
-            content={"status": "error", "message": "Rate limit exceeded. Try again after 2 seconds."},
-            status_code=429
-        )
-    
-    log_ootb_request(shared_state.server_url, "stop_processing", {})
-    
-    if shared_state.is_processing:
-        # Set both flags to ensure stopping the current task
-        shared_state.should_stop = True
-        shared_state.stop_event.set()
-        
-        # Clear message histories
-        shared_state.message_queue.clear()
-        shared_state.chatbot_messages.clear()
-        logging.info("Cleared message queue and chatbot messages during stop.")
-        
-        # Send an immediate message to the queue to inform the user
-        stop_initiated_msg = {"role": "assistant", "content": f"Stopping task '{shared_state.task}'...", "type": "text", "action_type": ""}
-        # Append the stop message AFTER clearing, so it's the only one left
-        shared_state.message_queue.append(stop_initiated_msg)
-        shared_state.chatbot_messages.append(stop_initiated_msg)
-        
-        return JSONResponse(
-            content={"status": "success", "message": "Task is being stopped, server will remain available for new tasks"},
-            status_code=200
-        )
-    else:
-        # Clear message histories even if not processing, to ensure clean state
-        shared_state.message_queue.clear()
-        shared_state.chatbot_messages.clear()
-        logging.info("Cleared message queue and chatbot messages (no active process to stop).")
-        return JSONResponse(
-            content={"status": "error", "message": "No active processing to stop"},
-            status_code=400
-        )
-
-@app.post("/toggle_pause")
-async def toggle_pause(request: Request):
-    # Apply rate limiting
-    if not rate_limiter.allow_request(request.url.path):
-        return JSONResponse(
-            content={"status": "error", "message": "Rate limit exceeded. Try again after 2 seconds."},
-            status_code=429
-        )
-    
-    log_ootb_request(shared_state.server_url, "toggle_pause", {})
-    
-    if not shared_state.is_processing:
-        return JSONResponse(
-            content={"status": "error", "message": "No active processing to pause/resume"},
-            status_code=400
-        )
-    
-    # Toggle the pause state
-    shared_state.is_paused = not shared_state.is_paused
-    current_state = shared_state.is_paused
-    
-    print(f"Toggled pause state to: {current_state}")
-    
-    status_message = "paused" if current_state else "resumed"
-    
-    # Add a message to the queue to inform the user
-    if current_state:
-        message = {"role": "assistant", "content": f"Task '{shared_state.task}' has been paused. Click Continue to resume.", "type": "text", "action_type": ""}
-    else:
-        message = {"role": "assistant", "content": f"Task '{shared_state.task}' has been resumed.", "type": "text", "action_type": ""}
-    
-    shared_state.chatbot_messages.append(message)
-    shared_state.message_queue.append(message)
-    
-    return JSONResponse(
-        content={
-            "status": "success", 
-            "message": f"Processing {status_message}",
-            "is_paused": current_state
-        },
-        status_code=200
-    )
-
-@app.get("/status")
-async def get_status(request: Request):
-    # Apply rate limiting
-    if not rate_limiter.allow_request(request.url.path):
-        return JSONResponse(
-            content={"status": "error", "message": "Rate limit exceeded. Try again after 2 seconds."},
-            status_code=429
-        )
-    
-    # log_ootb_request(shared_state.server_url, "get_status", {})
-    
-    print(f"Status check - Processing: {shared_state.is_processing}, Paused: {shared_state.is_paused}")
-    return JSONResponse(
-        content={
-            "status": "success",
-            "is_processing": shared_state.is_processing,
-            "is_paused": shared_state.is_paused
-        },
-        status_code=200
-    )
-
-@app.post("/exec_computer_tool")
-async def exec_computer_tool(request: Request):
-    logging.info("Received request to /exec_computer_tool")
-    try:
-        data = await request.json()
-        
-        # Extract parameters from the request
-        selected_screen = data.get('selected_screen', 0)
-        full_screen_game_mode = data.get('full_screen_game_mode', 0)
-        response = data.get('response', {})
-        
-        logging.info(f"Executing TeachmodeExecutor with: screen={selected_screen}, mode={full_screen_game_mode}, response={response}")
-        
-        # Create TeachmodeExecutor in a separate process to avoid event loop conflicts
-        # Since TeachmodeExecutor uses asyncio.run() internally, we need to run it in a way
-        # that doesn't conflict with FastAPI's event loop
-        
-        def run_executor():
-            executor = TeachmodeExecutor(
-                selected_screen=selected_screen,
-                full_screen_game_mode=full_screen_game_mode
-            )
-            
-            results = []
-            try:
-                for action_result in executor(response):
-                    results.append(action_result)
-            except Exception as exec_error:
-                logging.error(f"Error executing action: {exec_error}", exc_info=True)
-                return {"error": str(exec_error)}
-            
-            return results
-        
-        # Execute in a thread pool to avoid blocking the event loop
-        with concurrent.futures.ThreadPoolExecutor() as pool:
-            results = await asyncio.get_event_loop().run_in_executor(pool, run_executor)
-        
-        if isinstance(results, dict) and "error" in results:
-            return JSONResponse(
-                content={"status": "error", "message": results["error"]},
-                status_code=500
-            )
-            
-        logging.info(f"Action results: {results}")
-        
-        return JSONResponse(
-            content={"status": "success", "results": results},
-            status_code=200
-        )
-    except Exception as e:
-        logging.error("Error processing /exec_computer_tool:", exc_info=True)
-        return JSONResponse(
-            content={"status": "error", "message": f"Internal server error: {str(e)}"},
-            status_code=500
-        )
-
-def process_input():
-    global shared_state
-    logging.info("process_input thread started.")
-    shared_state.is_processing = True
-    shared_state.should_stop = False
-    shared_state.is_paused = False
-    shared_state.stop_event.clear()  # Ensure stop event is cleared at the start
-    
-    print(f"start sampling loop: {shared_state.chatbot_messages}")
-    print(f"shared_state.args before sampling loop: {shared_state.args}")
-
-
-    try:
-        # Get the generator for the sampling loop
-        sampling_loop = simple_teachmode_sampling_loop(
-            model=shared_state.model,
-            task=shared_state.task,
-            selected_screen=shared_state.selected_screen,
-            user_id=shared_state.user_id,
-            trace_id=shared_state.trace_id,
-            api_keys=shared_state.api_keys,
-            server_url=shared_state.server_url,
-        )
-
-        # Process messages from the sampling loop
-        for loop_msg in sampling_loop:
-            # Check stop condition more frequently
-            if shared_state.should_stop or shared_state.stop_event.is_set():
-                print("Processing stopped by user")
-                break
-                
-            # Check if paused and wait while paused
-            while shared_state.is_paused and not shared_state.should_stop and not shared_state.stop_event.is_set():
-                print(f"Processing paused at: {time.strftime('%H:%M:%S')}")
-                # Wait a short time and check stop condition regularly
-                for _ in range(5):  # Check 5 times per second
-                    if shared_state.should_stop or shared_state.stop_event.is_set():
-                        break
-                    time.sleep(0.2)
-                
-            # Check again after pause loop    
-            if shared_state.should_stop or shared_state.stop_event.is_set():
-                print("Processing stopped while paused or resuming")
-                break
-                
-            shared_state.chatbot_messages.append(loop_msg)
-            shared_state.message_queue.append(loop_msg)
-            
-            # Short sleep to allow stop signals to be processed
-            for _ in range(5):  # Check 5 times per second
-                if shared_state.should_stop or shared_state.stop_event.is_set():
-                    print("Processing stopped during sleep")
-                    break
-                time.sleep(0.1)
-                
-            if shared_state.should_stop or shared_state.stop_event.is_set():
-                break
-
-    except Exception as e:
-        # Handle any exceptions in the processing loop
-        error_msg = f"Error during task processing: {e}"
-        print(error_msg)
-        error_message = {"role": "assistant", "content": error_msg, "type": "error", "action_type": ""}
-        shared_state.message_queue.append(error_message)
-    
-    finally:
-        # Handle completion or interruption
-        if shared_state.should_stop or shared_state.stop_event.is_set():
-            stop_msg = f"Task '{shared_state.task}' was stopped. Ready for new tasks."
-            final_message = {"role": "assistant", "content": stop_msg, "type": "text", "action_type": ""}
-        else:
-            complete_msg = f"Task '{shared_state.task}' completed. Thanks for using Teachmode-OOTB."
-            final_message = {"role": "assistant", "content": complete_msg, "type": "text", "action_type": ""}
-            
-        shared_state.chatbot_messages.append(final_message)
-        shared_state.message_queue.append(final_message)
-        
-        # Reset all state flags to allow for new tasks
-        shared_state.is_processing = False
-        shared_state.should_stop = False
-        shared_state.is_paused = False
-        shared_state.stop_event.clear()
-        print("Processing completed, ready for new tasks")
-    logging.info("process_input thread finished.")
-
-def main():
-    # Logging is set up at the top level now
-    logging.info("App main() function starting setup.")
-    global app, shared_state, rate_limiter # Ensure app is global if needed by uvicorn
-    parser = argparse.ArgumentParser()
-    # Add arguments, but NOT host and port
-    parser.add_argument("--model", type=str, default="teach-mode-gpt-4o", help="Model name")
-    parser.add_argument("--task", type=str, default="Following the instructions to complete the task.", help="Initial task description")
-    parser.add_argument("--selected_screen", type=int, default=0, help="Selected screen index")
-    parser.add_argument("--user_id", type=str, default="hero_cases", help="User ID for the session")
-    parser.add_argument("--trace_id", type=str, default="build_scroll_combat", help="Trace ID for the session")
-    parser.add_argument("--api_keys", type=str, default="sk-proj-1234567890", help="API keys")
-    parser.add_argument("--server_url", type=str, default="http://ec2-44-234-43-86.us-west-2.compute.amazonaws.com", help="Server URL for the session")
-
-    args = parser.parse_args()
-
-    # Validate args or set defaults if needed (keep these)
-    if not hasattr(args, 'model'): args.model = "default_model"
-    if not hasattr(args, 'task'): args.task = "default_task"
-    if not hasattr(args, 'selected_screen'): args.selected_screen = 0
-    if not hasattr(args, 'user_id'): args.user_id = "unknown_user"
-    if not hasattr(args, 'trace_id'): args.trace_id = "unknown_trace"
-    if not hasattr(args, 'api_keys'): args.api_keys = "none"
-    if not hasattr(args, 'server_url'): args.server_url = "none"
-
-    shared_state = SharedState(args)
-    rate_limiter = RateLimiter(interval_seconds=2) # Re-initialize rate limiter
-    logging.info(f"Shared state initialized for user: {args.user_id}")
-
-    # --- Restore original port calculation logic --- 
-    port = 7888 # Default port
-    host = "0.0.0.0" # Listen on all interfaces
-    
-    if platform.system() == "Windows":
-        try:
-            username = os.environ["USERNAME"].lower()
-            logging.info(f"Determining port based on Windows username: {username}")
-            if username == "altair":
-                port = 14000
-            elif username.startswith("guest") and username[5:].isdigit():
-                num = int(username[5:])
-                if 1 <= num <= 10: # Assuming max 10 guests for this range
-                    port = 14000 + num
-                else:
-                     logging.warning(f"Guest user number {num} out of range (1-10), using default port {port}.")
-            else:
-                logging.info(f"Username '{username}' doesn't match specific rules, using default port {port}.")
-        except Exception as e:
-             logging.error(f"Error determining port from username: {e}. Using default port {port}.", exc_info=True)
-    else:
-         logging.info(f"Not running on Windows, using default port {port}.")
-    # --- End of restored port calculation --- 
-
-    logging.info(f"Final Host={host}, Port={port}")
-
-    try:
-        logging.info(f"Starting Uvicorn server on {host}:{port}")
-        # Use the calculated port and specific host
-        uvicorn.run(app, host=host, port=port)
-        logging.info("Uvicorn server stopped.")
-    except Exception as main_e:
-        logging.error("Error in main execution:", exc_info=True)
-    finally:
-        logging.info("App main() function finished.")
-
-if __name__ == "__main__":
-    main()
-
-    # Test log_ootb_request
+import argparse
+import time
+import json
+from datetime import datetime
+import threading
+import requests
+import platform  # Add platform import
+import pyautogui  # Add pyautogui import
+import webbrowser # Add webbrowser import
+import os # Import os for path joining
+import logging # Import logging
+import importlib # For dynamic imports
+import pkgutil # To find modules
+import sys # For logging setup
+import traceback # For logging setup
+from logging.handlers import RotatingFileHandler # For logging setup
+from fastapi import FastAPI, Request
+from fastapi.responses import JSONResponse
+from fastapi.middleware.cors import CORSMiddleware
+from computer_use_ootb_internal.computer_use_demo.tools.computer import get_screen_details
+from computer_use_ootb_internal.run_teachmode_ootb_args import simple_teachmode_sampling_loop
+from computer_use_ootb_internal.computer_use_demo.executor.teachmode_executor import TeachmodeExecutor
+import uvicorn # Assuming uvicorn is used to run FastAPI
+import concurrent.futures
+import asyncio
+
+# --- App Logging Setup ---
+try:
+    # Log to user's AppData directory for better accessibility
+    log_dir_base = os.environ.get('APPDATA', os.path.expanduser('~'))
+    log_dir = os.path.join(log_dir_base, 'OOTBAppLogs')
+    os.makedirs(log_dir, exist_ok=True)
+    log_file = os.path.join(log_dir, 'ootb_app.log')
+
+    log_format = '%(asctime)s - %(levelname)s - %(process)d - %(threadName)s - %(message)s'
+    log_level = logging.INFO # Or logging.DEBUG for more detail
+
+    # Use rotating file handler
+    handler = RotatingFileHandler(log_file, maxBytes=5*1024*1024, backupCount=2, encoding='utf-8')
+    handler.setFormatter(logging.Formatter(log_format))
+
+    # Configure root logger
+    logging.basicConfig(level=log_level, handlers=[handler])
+
+    # Add stream handler to see logs if running interactively (optional)
+    # logging.getLogger().addHandler(logging.StreamHandler(sys.stdout))
+
+    logging.info("="*20 + " OOTB App Starting " + "="*20)
+    logging.info(f"Running with args: {sys.argv}")
+    logging.info(f"Python Executable: {sys.executable}")
+    logging.info(f"Working Directory: {os.getcwd()}")
+    logging.info(f"User: {os.getenv('USERNAME')}")
+
+except Exception as log_setup_e:
+    print(f"FATAL: Failed to set up logging: {log_setup_e}")
+    # Fallback logging might be needed here if file logging fails
+
+# --- End App Logging Setup ---
+
+app = FastAPI()
+
+# Add CORS middleware to allow requests from the frontend
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],  
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+# Rate limiter for API endpoints
+class RateLimiter:
+    def __init__(self, interval_seconds=2):
+        self.interval = interval_seconds
+        self.last_request_time = {}
+        self.lock = threading.Lock()
+        
+    def allow_request(self, endpoint):
+        with self.lock:
+            current_time = time.time()
+            # Priority endpoints always allowed
+            if endpoint in ["/update_params", "/update_message"]:
+                return True
+                
+            # For other endpoints, apply rate limiting
+            if endpoint not in self.last_request_time:
+                self.last_request_time[endpoint] = current_time
+                return True
+                
+            elapsed = current_time - self.last_request_time[endpoint]
+            if elapsed < self.interval:
+                return False
+                
+            self.last_request_time[endpoint] = current_time
+            return True
+
+
+def log_ootb_request(server_url, ootb_request_type, data):
+    logging.info(f"OOTB Request: Type={ootb_request_type}, Data={data}")
+    # Keep the requests post for now if it serves a specific purpose
+    logging_data = {
+        "type": ootb_request_type,
+        "data": data,
+        "timestamp": datetime.now().strftime("%Y-%m-%d %H:%M:%S")
+    }
+    if not server_url.endswith("/update_ootb_logging"):
+        server_logging_url = server_url + "/update_ootb_logging"
+    else:
+        server_logging_url = server_url
+    try:
+        requests.post(server_logging_url, json=logging_data, timeout=5)
+    except Exception as req_log_e:
+        logging.warning(f"Could not log ootb request to server {server_logging_url}: {req_log_e}")
+
+
+class SharedState:
+    def __init__(self, args):
+        self.args = args
+        self.task_updated = False
+        self.chatbot_messages = []
+        # Store all state-related data here
+        self.model = args.model
+        self.task = getattr(args, 'task', "")
+        self.selected_screen = args.selected_screen
+        self.user_id = args.user_id
+        self.trace_id = args.trace_id
+        self.api_keys = args.api_keys
+        self.server_url = args.server_url
+        self.message_queue = []
+        self.is_processing = False
+        self.should_stop = False
+        self.is_paused = False
+        # Add a new event to better control stopping
+        self.stop_event = threading.Event()
+        # Add a reference to the processing thread
+        self.processing_thread = None
+
+shared_state = None
+rate_limiter = RateLimiter(interval_seconds=2)
+
+# Set up logging for this module
+log = logging.getLogger(__name__)
+
+def prepare_environment(state):
+    """Dynamically loads and runs preparation logic based on software name."""
+    # Determine software name from state (user_id, trace_id, or task)
+    software_name = ""
+    
+    # Check user_id first
+    user_id = getattr(state, 'user_id', '').lower()
+    task = getattr(state, 'task', '').lower()
+    trace_id = getattr(state, 'trace_id', '').lower()
+    
+    log.info(f"Checking for software in: user_id='{user_id}', trace_id='{trace_id}', task='{task}'")
+    
+    # Look for known software indicators
+    if "star rail" in user_id or "star rail" in trace_id:
+        software_name = "star rail"
+    elif "powerpoint" in user_id or "powerpoint" in trace_id or "powerpoint" in task:
+        software_name = "powerpoint"
+    # Add more software checks here as needed
+    
+    # If no specific software found, check task for keywords
+    if not software_name:
+        log.info("No specific software detected from IDs, checking task content")
+    
+    if not software_name:
+        log.info("No specific software preparation identified. Skipping preparation.")
+        return
+    
+    log.info(f"Identified software for preparation: '{software_name}'")
+    
+    # Normalize the software name to be a valid Python module name
+    # Replace spaces/hyphens with underscores, convert to lowercase
+    module_name_base = software_name.replace(" ", "_").replace("-", "_").lower()
+    module_to_run = f"{module_name_base}_prepare"
+
+    log.info(f"Attempting preparation for software: '{software_name}' (Module: '{module_to_run}')")
+
+    try:
+        # Construct the full module path within the package
+        prep_package = "computer_use_ootb_internal.preparation"
+        full_module_path = f"{prep_package}.{module_to_run}"
+
+        # Dynamically import the module
+        # Check if module exists first using pkgutil to avoid import errors
+        log.debug(f"Looking for preparation module: {full_module_path}")
+        loader = pkgutil.find_loader(full_module_path)
+        if loader is None:
+            log.warning(f"Preparation module '{full_module_path}' not found. Skipping preparation.")
+            return
+
+        log.debug(f"Importing preparation module: {full_module_path}")
+        prep_module = importlib.import_module(full_module_path)
+
+        # Check if the module has the expected function
+        if hasattr(prep_module, "run_preparation") and callable(prep_module.run_preparation):
+            log.info(f"Running preparation function from {full_module_path}...")
+            prep_module.run_preparation(state)
+            log.info(f"Preparation function from {full_module_path} completed.")
+        else:
+            log.warning(f"Module {full_module_path} found, but does not have a callable 'run_preparation' function. Skipping.")
+
+    except ModuleNotFoundError:
+        log.warning(f"Preparation module '{full_module_path}' not found. Skipping preparation.")
+    except Exception as e:
+        log.error(f"Error during dynamic preparation loading/execution for '{module_to_run}': {e}", exc_info=True)
+
+
+@app.post("/update_params")
+async def update_parameters(request: Request):
+    logging.info("Received request to /update_params")
+    try:
+        data = await request.json()
+        
+        if 'task' not in data:
+            return JSONResponse(
+                content={"status": "error", "message": "Missing required field: task"},
+                status_code=400
+            )
+        
+        # Clear message histories before updating parameters
+        shared_state.message_queue.clear()
+        shared_state.chatbot_messages.clear()
+        logging.info("Cleared message queue and chatbot messages.")
+
+        shared_state.args = argparse.Namespace(**data)
+        shared_state.task_updated = True
+        
+        # Update shared state when parameters change
+        shared_state.model = getattr(shared_state.args, 'model', "teach-mode-gpt-4o")
+        shared_state.task = getattr(shared_state.args, 'task', "Following the instructions to complete the task.")
+        shared_state.selected_screen = getattr(shared_state.args, 'selected_screen', 0)
+        shared_state.user_id = getattr(shared_state.args, 'user_id', "hero_cases")
+        shared_state.trace_id = getattr(shared_state.args, 'trace_id', "build_scroll_combat")
+        shared_state.api_keys = getattr(shared_state.args, 'api_keys', "sk-proj-1234567890")
+        shared_state.server_url = getattr(shared_state.args, 'server_url', "http://ec2-44-234-43-86.us-west-2.compute.amazonaws.com")
+        
+        log_ootb_request(shared_state.server_url, "update_params", data)
+
+        # Call the (now dynamic) preparation function here, after parameters are updated
+        prepare_environment(shared_state)
+
+        logging.info("Parameters updated successfully.")
+        return JSONResponse(
+            content={"status": "success", "message": "Parameters updated", "new_args": vars(shared_state.args)},
+            status_code=200
+        )
+    except Exception as e:
+        logging.error("Error processing /update_params:", exc_info=True)
+        return JSONResponse(content={"status": "error", "message": "Internal server error"}, status_code=500)
+
+@app.post("/update_message")
+async def update_message(request: Request):
+    data = await request.json()
+    
+    if 'message' not in data:
+        return JSONResponse(
+            content={"status": "error", "message": "Missing required field: message"},
+            status_code=400
+        )
+    
+    log_ootb_request(shared_state.server_url, "update_message", data)
+    
+    message = data['message']
+    # shared_state.chatbot_messages.append({"role": "user", "content": message, "type": "text"})
+    shared_state.task = message
+    shared_state.args.task = message
+    
+    # Reset stop event before starting
+    shared_state.stop_event.clear()
+    
+    # Start processing if not already running
+    if not shared_state.is_processing:
+        # Create and store the thread
+        shared_state.processing_thread = threading.Thread(target=process_input, daemon=True)
+        shared_state.processing_thread.start()
+    
+    return JSONResponse(
+        content={"status": "success", "message": "Message received", "task": shared_state.task},
+        status_code=200
+    )
+
+@app.get("/get_messages")
+async def get_messages(request: Request):
+    # Apply rate limiting
+    if not rate_limiter.allow_request(request.url.path):
+        return JSONResponse(
+            content={"status": "error", "message": "Rate limit exceeded. Try again after 2 seconds."},
+            status_code=429
+        )
+    
+    # log_ootb_request(shared_state.server_url, "get_messages", {})
+    
+    # Return all messages in the queue and clear it
+    messages = shared_state.message_queue.copy()
+    shared_state.message_queue = []
+    
+    return JSONResponse(
+        content={"status": "success", "messages": messages},
+        status_code=200
+    )
+
+@app.get("/get_screens")
+async def get_screens(request: Request):
+    # Apply rate limiting
+    if not rate_limiter.allow_request(request.url.path):
+        return JSONResponse(
+            content={"status": "error", "message": "Rate limit exceeded. Try again after 2 seconds."},
+            status_code=429
+        )
+    
+    log_ootb_request(shared_state.server_url, "get_screens", {})
+    
+    screen_options, primary_index = get_screen_details()
+    
+    return JSONResponse(
+        content={"status": "success", "screens": screen_options, "primary_index": primary_index},
+        status_code=200
+    )
+
+@app.post("/stop_processing")
+async def stop_processing(request: Request):
+    # Apply rate limiting
+    if not rate_limiter.allow_request(request.url.path):
+        return JSONResponse(
+            content={"status": "error", "message": "Rate limit exceeded. Try again after 2 seconds."},
+            status_code=429
+        )
+    
+    log_ootb_request(shared_state.server_url, "stop_processing", {})
+    
+    if shared_state.is_processing:
+        # Set both flags to ensure stopping the current task
+        shared_state.should_stop = True
+        shared_state.stop_event.set()
+        
+        # Clear message histories
+        shared_state.message_queue.clear()
+        shared_state.chatbot_messages.clear()
+        logging.info("Cleared message queue and chatbot messages during stop.")
+        
+        # Send an immediate message to the queue to inform the user
+        stop_initiated_msg = {"role": "assistant", "content": f"Stopping task '{shared_state.task}'...", "type": "text", "action_type": ""}
+        # Append the stop message AFTER clearing, so it's the only one left
+        shared_state.message_queue.append(stop_initiated_msg)
+        shared_state.chatbot_messages.append(stop_initiated_msg)
+        
+        return JSONResponse(
+            content={"status": "success", "message": "Task is being stopped, server will remain available for new tasks"},
+            status_code=200
+        )
+    else:
+        # Clear message histories even if not processing, to ensure clean state
+        shared_state.message_queue.clear()
+        shared_state.chatbot_messages.clear()
+        logging.info("Cleared message queue and chatbot messages (no active process to stop).")
+        return JSONResponse(
+            content={"status": "error", "message": "No active processing to stop"},
+            status_code=400
+        )
+
+@app.post("/toggle_pause")
+async def toggle_pause(request: Request):
+    # Apply rate limiting
+    if not rate_limiter.allow_request(request.url.path):
+        return JSONResponse(
+            content={"status": "error", "message": "Rate limit exceeded. Try again after 2 seconds."},
+            status_code=429
+        )
+    
+    log_ootb_request(shared_state.server_url, "toggle_pause", {})
+    
+    if not shared_state.is_processing:
+        return JSONResponse(
+            content={"status": "error", "message": "No active processing to pause/resume"},
+            status_code=400
+        )
+    
+    # Toggle the pause state
+    shared_state.is_paused = not shared_state.is_paused
+    current_state = shared_state.is_paused
+    
+    print(f"Toggled pause state to: {current_state}")
+    
+    status_message = "paused" if current_state else "resumed"
+    
+    # Add a message to the queue to inform the user
+    if current_state:
+        message = {"role": "assistant", "content": f"Task '{shared_state.task}' has been paused. Click Continue to resume.", "type": "text", "action_type": ""}
+    else:
+        message = {"role": "assistant", "content": f"Task '{shared_state.task}' has been resumed.", "type": "text", "action_type": ""}
+    
+    shared_state.chatbot_messages.append(message)
+    shared_state.message_queue.append(message)
+    
+    return JSONResponse(
+        content={
+            "status": "success", 
+            "message": f"Processing {status_message}",
+            "is_paused": current_state
+        },
+        status_code=200
+    )
+
+@app.get("/status")
+async def get_status(request: Request):
+    # Apply rate limiting
+    if not rate_limiter.allow_request(request.url.path):
+        return JSONResponse(
+            content={"status": "error", "message": "Rate limit exceeded. Try again after 2 seconds."},
+            status_code=429
+        )
+    
+    # log_ootb_request(shared_state.server_url, "get_status", {})
+    
+    print(f"Status check - Processing: {shared_state.is_processing}, Paused: {shared_state.is_paused}")
+    return JSONResponse(
+        content={
+            "status": "success",
+            "is_processing": shared_state.is_processing,
+            "is_paused": shared_state.is_paused
+        },
+        status_code=200
+    )
+
+@app.post("/exec_computer_tool")
+async def exec_computer_tool(request: Request):
+    logging.info("Received request to /exec_computer_tool")
+    try:
+        data = await request.json()
+        
+        # Extract parameters from the request
+        selected_screen = data.get('selected_screen', 0)
+        full_screen_game_mode = data.get('full_screen_game_mode', 0)
+        response = data.get('response', {})
+        
+        logging.info(f"Executing TeachmodeExecutor with: screen={selected_screen}, mode={full_screen_game_mode}, response={response}")
+        
+        # Create TeachmodeExecutor in a separate process to avoid event loop conflicts
+        # Since TeachmodeExecutor uses asyncio.run() internally, we need to run it in a way
+        # that doesn't conflict with FastAPI's event loop
+        
+        def run_executor():
+            executor = TeachmodeExecutor(
+                selected_screen=selected_screen,
+                full_screen_game_mode=full_screen_game_mode
+            )
+            
+            results = []
+            try:
+                for action_result in executor(response):
+                    results.append(action_result)
+            except Exception as exec_error:
+                logging.error(f"Error executing action: {exec_error}", exc_info=True)
+                return {"error": str(exec_error)}
+            
+            return results
+        
+        # Execute in a thread pool to avoid blocking the event loop
+        with concurrent.futures.ThreadPoolExecutor() as pool:
+            results = await asyncio.get_event_loop().run_in_executor(pool, run_executor)
+        
+        if isinstance(results, dict) and "error" in results:
+            return JSONResponse(
+                content={"status": "error", "message": results["error"]},
+                status_code=500
+            )
+            
+        logging.info(f"Action results: {results}")
+        
+        return JSONResponse(
+            content={"status": "success", "results": results},
+            status_code=200
+        )
+    except Exception as e:
+        logging.error("Error processing /exec_computer_tool:", exc_info=True)
+        return JSONResponse(
+            content={"status": "error", "message": f"Internal server error: {str(e)}"},
+            status_code=500
+        )
+
+def process_input():
+    global shared_state
+    logging.info("process_input thread started.")
+    shared_state.is_processing = True
+    shared_state.should_stop = False
+    shared_state.is_paused = False
+    shared_state.stop_event.clear()  # Ensure stop event is cleared at the start
+    
+    print(f"start sampling loop: {shared_state.chatbot_messages}")
+    print(f"shared_state.args before sampling loop: {shared_state.args}")
+
+
+    try:
+        # Get the generator for the sampling loop
+        sampling_loop = simple_teachmode_sampling_loop(
+            model=shared_state.model,
+            task=shared_state.task,
+            selected_screen=shared_state.selected_screen,
+            user_id=shared_state.user_id,
+            trace_id=shared_state.trace_id,
+            api_keys=shared_state.api_keys,
+            server_url=shared_state.server_url,
+        )
+
+        # Process messages from the sampling loop
+        for loop_msg in sampling_loop:
+            # Check stop condition more frequently
+            if shared_state.should_stop or shared_state.stop_event.is_set():
+                print("Processing stopped by user")
+                break
+                
+            # Check if paused and wait while paused
+            while shared_state.is_paused and not shared_state.should_stop and not shared_state.stop_event.is_set():
+                print(f"Processing paused at: {time.strftime('%H:%M:%S')}")
+                # Wait a short time and check stop condition regularly
+                for _ in range(5):  # Check 5 times per second
+                    if shared_state.should_stop or shared_state.stop_event.is_set():
+                        break
+                    time.sleep(0.2)
+                
+            # Check again after pause loop    
+            if shared_state.should_stop or shared_state.stop_event.is_set():
+                print("Processing stopped while paused or resuming")
+                break
+                
+            shared_state.chatbot_messages.append(loop_msg)
+            shared_state.message_queue.append(loop_msg)
+            
+            # Short sleep to allow stop signals to be processed
+            for _ in range(5):  # Check 5 times per second
+                if shared_state.should_stop or shared_state.stop_event.is_set():
+                    print("Processing stopped during sleep")
+                    break
+                time.sleep(0.1)
+                
+            if shared_state.should_stop or shared_state.stop_event.is_set():
+                break
+
+    except Exception as e:
+        # Handle any exceptions in the processing loop
+        error_msg = f"Error during task processing: {e}"
+        print(error_msg)
+        error_message = {"role": "assistant", "content": error_msg, "type": "error", "action_type": ""}
+        shared_state.message_queue.append(error_message)
+    
+    finally:
+        # Handle completion or interruption
+        if shared_state.should_stop or shared_state.stop_event.is_set():
+            stop_msg = f"Task '{shared_state.task}' was stopped. Ready for new tasks."
+            final_message = {"role": "assistant", "content": stop_msg, "type": "text", "action_type": ""}
+        else:
+            complete_msg = f"Task '{shared_state.task}' completed. Thanks for using Teachmode-OOTB."
+            final_message = {"role": "assistant", "content": complete_msg, "type": "text", "action_type": ""}
+            
+        shared_state.chatbot_messages.append(final_message)
+        shared_state.message_queue.append(final_message)
+        
+        # Reset all state flags to allow for new tasks
+        shared_state.is_processing = False
+        shared_state.should_stop = False
+        shared_state.is_paused = False
+        shared_state.stop_event.clear()
+        print("Processing completed, ready for new tasks")
+    logging.info("process_input thread finished.")
+
+def main():
+    # Logging is set up at the top level now
+    logging.info("App main() function starting setup.")
+    global app, shared_state, rate_limiter # Ensure app is global if needed by uvicorn
+    parser = argparse.ArgumentParser()
+    # Add arguments, but NOT host and port
+    parser.add_argument("--model", type=str, default="teach-mode-gpt-4o", help="Model name")
+    parser.add_argument("--task", type=str, default="Following the instructions to complete the task.", help="Initial task description")
+    parser.add_argument("--selected_screen", type=int, default=0, help="Selected screen index")
+    parser.add_argument("--user_id", type=str, default="hero_cases", help="User ID for the session")
+    parser.add_argument("--trace_id", type=str, default="build_scroll_combat", help="Trace ID for the session")
+    parser.add_argument("--api_keys", type=str, default="sk-proj-1234567890", help="API keys")
+    parser.add_argument("--server_url", type=str, default="http://ec2-44-234-43-86.us-west-2.compute.amazonaws.com", help="Server URL for the session")
+
+    args = parser.parse_args()
+
+    # Validate args or set defaults if needed (keep these)
+    if not hasattr(args, 'model'): args.model = "default_model"
+    if not hasattr(args, 'task'): args.task = "default_task"
+    if not hasattr(args, 'selected_screen'): args.selected_screen = 0
+    if not hasattr(args, 'user_id'): args.user_id = "unknown_user"
+    if not hasattr(args, 'trace_id'): args.trace_id = "unknown_trace"
+    if not hasattr(args, 'api_keys'): args.api_keys = "none"
+    if not hasattr(args, 'server_url'): args.server_url = "none"
+
+    shared_state = SharedState(args)
+    rate_limiter = RateLimiter(interval_seconds=2) # Re-initialize rate limiter
+    logging.info(f"Shared state initialized for user: {args.user_id}")
+
+    # --- Restore original port calculation logic --- 
+    port = 7888 # Default port
+    host = "0.0.0.0" # Listen on all interfaces
+    
+    if platform.system() == "Windows":
+        try:
+            username = os.environ["USERNAME"].lower()
+            logging.info(f"Determining port based on Windows username: {username}")
+            if username == "altair":
+                port = 14000
+            elif username.startswith("guest") and username[5:].isdigit():
+                num = int(username[5:])
+                if 1 <= num <= 10: # Assuming max 10 guests for this range
+                    port = 14000 + num
+                else:
+                     logging.warning(f"Guest user number {num} out of range (1-10), using default port {port}.")
+            else:
+                logging.info(f"Username '{username}' doesn't match specific rules, using default port {port}.")
+        except Exception as e:
+             logging.error(f"Error determining port from username: {e}. Using default port {port}.", exc_info=True)
+    else:
+         logging.info(f"Not running on Windows, using default port {port}.")
+    # --- End of restored port calculation --- 
+
+    logging.info(f"Final Host={host}, Port={port}")
+
+    try:
+        logging.info(f"Starting Uvicorn server on {host}:{port}")
+        # Use the calculated port and specific host
+        uvicorn.run(app, host=host, port=port)
+        logging.info("Uvicorn server stopped.")
+    except Exception as main_e:
+        logging.error("Error in main execution:", exc_info=True)
+    finally:
+        logging.info("App main() function finished.")
+
+if __name__ == "__main__":
+    main()
+
+    # Test log_ootb_request
     log_ootb_request("http://ec2-44-234-43-86.us-west-2.compute.amazonaws.com", "test_request", {"message": "Test message"})