main.py

#!/usr/bin/env python3
import time
import json
from camera import Camera
from openai_client import send_image_and_get_instructions, parse_response
from audio import SpeechToText, TextToSpeech
from motor_control import init_gpio, cleanup_gpio, MotorDriver, mecanum_drive

def parse_command(command):
    if not command:
        return ("waiting", None)
    command_lower = command.lower()
    if "look for" in command_lower:
        target = command_lower.split("look for", 1)[1].strip()
        return ("search", target)
    elif "find" in command_lower:
        target = command_lower.split("find", 1)[1].strip()
        return ("search", target)
    elif ("what" in command_lower or "see" in command_lower or "describe" in command_lower) and ("front" in command_lower or "ahead" in command_lower):
        return ("see", None)
    else:
        return ("waiting", None)

def search_for_target(target, camera, tts, simulate, lf, lr, rf, rr):
    # Append "book" if not already there
    if "book" not in target:
        target += " book"
    
    tts.speak(f"Searching for {target}.")
    attempts = 0
    max_attempts = 36  # roughly one full rotation (360°)
    rotation_speed = 40
    rotation_duration = 0.3

    while attempts < max_attempts:
        if attempts > 0:
            tts.speak("Stabilizing position.")
            time.sleep(5)
        tts.speak("Capturing image.")
        b64_img = camera.capture_frame()
        prompt = (
            f"Analyze the image and determine if the book '{target}' is visible. "
            "If it is, respond with a JSON object: {\"found\": true, \"description\": <brief description>}. "
            "If it is not visible, respond with {\"found\": false}."
        )
        tts.speak("Analyzing.")
        response_text = send_image_and_get_instructions(b64_img, prompt, simulate=simulate)
        print("\n--- RAW API RESPONSE ---")
        print(response_text)
        
        # If no valid response, say "Thinking" and retry after waiting.
        if not response_text:
            tts.speak("Thinking.")
            time.sleep(5)
            continue
        
        try:
            data = json.loads(response_text)
            found = data.get("found", False)
        except Exception as e:
            print("JSON parsing error:", e)
            try:
                if target.lower() in response_text.lower():
                    found = True
                else:
                    found = False
            except Exception as inner_e:
                print("Inner error:", inner_e)
                found = False

        if found:
            description = data.get("description", "it looks interesting") if isinstance(data, dict) else "it looks interesting"
            tts.speak(f"Found {target}: {description}.")
            return True
        else:
            tts.speak("Not found here. Waiting 5 seconds.")
            time.sleep(5)
        
        tts.speak("Rotating slowly.")
        # Rotate left at defined speed and duration.
        mecanum_drive(lf, lr, rf, rr, "rotate_left", rotation_speed)
        time.sleep(rotation_duration)
        mecanum_drive(lf, lr, rf, rr, "stop", 0)
        attempts += 1

    tts.speak(f"Could not locate {target} after a full rotation.")
    return False

def main():
    # Initialize GPIO and motors.
    init_gpio()
    lf = MotorDriver(5, 6, 12)   # Left Front Motor
    lr = MotorDriver(27, 22, 18)  # Left Rear Motor
    rf = MotorDriver(24, 23, 19)  # Right Front Motor
    rr = MotorDriver(26, 17, 13)  # Right Rear Motor

    # Use a lower resolution to reduce token load.
    camera = Camera(capture_resolution=(640, 480), output_resolution=(320, 240))
    stt = SpeechToText()
    tts = TextToSpeech()
    SIMULATE_GPT = False  # Make sure your API key is properly set in your environment

    detailed_prompt = (
        "Analyze the provided camera image and give a short description of what you see. "
        "If there is a book or notebook visible, include a brief description of its cover. "
        "Then answer: What do you see?"
    )

    tts.speak("Voice Interactive Visual Assistant started. Please say a command.")
    print("I am waiting for instructions.")

    try:
        while True:
            command_text = stt.listen_command()
            print("Recognized command:", command_text)
            mode, target = parse_command(command_text)
            
            if mode == "search" and target:
                if search_for_target(target, camera, tts, SIMULATE_GPT, lf, lr, rf, rr):
                    tts.speak("Consider it done!")
                    tts.speak("Moving forward at normal speed for one and a half seconds.")
                    forward_speed = 60
                    forward_duration = 1.5
                    mecanum_drive(lf, lr, rf, rr, "forward", forward_speed)
                    time.sleep(forward_duration)
                    mecanum_drive(lf, lr, rf, rr, "stop", 0)
                else:
                    tts.speak("I couldn't find the target. Please try again.")
            elif mode == "see":
                b64_img = camera.capture_frame()
                response_text = send_image_and_get_instructions(b64_img, detailed_prompt, simulate=SIMULATE_GPT)
                if response_text:
                    final_answer = parse_response(response_text)
                    tts.speak(final_answer)
                else:
                    tts.speak("Sorry, I didn't get a response.")
            else:
                tts.speak("I am waiting for instructions.")
            time.sleep(1)
    except KeyboardInterrupt:
        tts.speak("Exiting. Goodbye!")
        print("Exiting due to keyboard interrupt.")
    finally:
        camera.close()
        cleanup_gpio()

if __name__ == "__main__":
    main()