vision_engine.py

import os
import base64
import requests
from dotenv import load_dotenv

# Load environment variables from .env file
load_dotenv()

# CapSolver API configuration
CAPSOLVER_API_KEY = os.getenv("CAPSOLVER_API_KEY")
CREATE_TASK_URL = "https://api.capsolver.com/createTask"

# Vision Engine modules
MODULES = {
    "slider_1": "Slider captcha - requires image and imageBackground",
    "rotate_1": "Rotation detection - requires image and imageBackground",
    "rotate_2": "Standalone rotation - requires only image",
    "shein": "Multi-object selection with coordinates - requires image and question"
}


def solve_slider(image_base64, background_base64):
    """
    Solve slider captcha.

    Args:
        image_base64: Base64 encoded slider piece image
        background_base64: Base64 encoded background image

    Returns:
        Solution containing distance to slide
    """
    payload = {
        "clientKey": CAPSOLVER_API_KEY,
        "task": {
            "type": "VisionEngine",
            "module": "slider_1",
            "image": image_base64,
            "imageBackground": background_base64
        }
    }

    response = requests.post(CREATE_TASK_URL, json=payload)
    result = response.json()

    if result.get("errorId") != 0:
        raise Exception(f"Failed to solve: {result.get('errorDescription')}")

    return result.get("solution", {})


def solve_rotate(image_base64, background_base64=None, module="rotate_1"):
    """
    Solve rotation captcha.

    Args:
        image_base64: Base64 encoded image to rotate
        background_base64: Base64 encoded background (required for rotate_1)
        module: "rotate_1" (with background) or "rotate_2" (standalone)

    Returns:
        Solution containing angle to rotate
    """
    payload = {
        "clientKey": CAPSOLVER_API_KEY,
        "task": {
            "type": "VisionEngine",
            "module": module,
            "image": image_base64
        }
    }

    if background_base64 and module == "rotate_1":
        payload["task"]["imageBackground"] = background_base64

    response = requests.post(CREATE_TASK_URL, json=payload)
    result = response.json()

    if result.get("errorId") != 0:
        raise Exception(f"Failed to solve: {result.get('errorDescription')}")

    return result.get("solution", {})


def solve_shein(image_base64, question):
    """
    Solve Shein-style multi-object selection captcha.

    Args:
        image_base64: Base64 encoded image
        question: The object type to identify

    Returns:
        Solution containing rects (coordinate arrays)
    """
    payload = {
        "clientKey": CAPSOLVER_API_KEY,
        "task": {
            "type": "VisionEngine",
            "module": "shein",
            "image": image_base64,
            "question": question
        }
    }

    response = requests.post(CREATE_TASK_URL, json=payload)
    result = response.json()

    if result.get("errorId") != 0:
        raise Exception(f"Failed to solve: {result.get('errorDescription')}")

    return result.get("solution", {})


def load_image_as_base64(image_path):
    """Helper to load image as base64."""
    with open(image_path, "rb") as f:
        return base64.b64encode(f.read()).decode("utf-8")


def main():
    if not CAPSOLVER_API_KEY:
        print("Error: CAPSOLVER_API_KEY not found in .env file")
        print("Please create a .env file with your API key:")
        print("CAPSOLVER_API_KEY=your_api_key_here")
        return

    print("Vision Engine - Slider/Rotate Captcha Solver")

    print("\nAvailable modules:")
    for module, desc in MODULES.items():
        print(f"  {module}: {desc}")

    print("\nExample usage:")
    print("\n  # Slider captcha:")
    print('  slider_img = load_image_as_base64("slider.png")')
    print('  bg_img = load_image_as_base64("background.png")')
    print("  solution = solve_slider(slider_img, bg_img)")
    print("  # Returns: {'distance': 150}  # pixels to slide")

    print("\n  # Rotation captcha (with background):")
    print("  solution = solve_rotate(image_base64, background_base64, module='rotate_1')")
    print("  # Returns: {'angle': 45}  # degrees to rotate")

    print("\n  # Rotation captcha (standalone):")
    print("  solution = solve_rotate(image_base64, module='rotate_2')")
    print("  # Returns: {'angle': 90}")

    print("\n  # Shein multi-object selection:")
    print('  solution = solve_shein(image_base64, "Select all shoes")')
    print("  # Returns: {'rects': [[x1,y1,x2,y2], [x1,y1,x2,y2]]}  # object coordinates")


if __name__ == "__main__":
    main()