Vision-module-auto/claude_vision_auto/config.py

"""
Configuration for Claude Vision Auto
"""

import os
import yaml
from pathlib import Path
from typing import Dict, Any


def get_config_dir() -> Path:
    """Get configuration directory"""
    config_dir = Path.home() / ".config" / "claude-vision-auto"
    config_dir.mkdir(parents=True, exist_ok=True)
    return config_dir


def get_cache_dir() -> Path:
    """Get cache directory for screenshots"""
    cache_dir = Path.home() / ".cache" / "claude-vision-auto"
    cache_dir.mkdir(parents=True, exist_ok=True)
    return cache_dir


def load_config() -> Dict[str, Any]:
    """
    Load configuration from YAML files with priority:
    1. User config (~/.config/claude-vision-auto/config.yaml)
    2. Default config (package default_config.yaml)
    3. Environment variables (highest priority)
    """
    # Load default config
    default_config_path = Path(__file__).parent / "default_config.yaml"
    with open(default_config_path, 'r') as f:
        config = yaml.safe_load(f)

    # Load user config if exists
    user_config_path = get_config_dir() / "config.yaml"
    if user_config_path.exists():
        with open(user_config_path, 'r') as f:
            user_config = yaml.safe_load(f)
            # Deep merge user config
            if user_config:
                config = deep_merge(config, user_config)

    # Override with environment variables
    if os.getenv("OLLAMA_URL"):
        config['ollama']['url'] = os.getenv("OLLAMA_URL")
    if os.getenv("VISION_MODEL"):
        config['ollama']['model'] = os.getenv("VISION_MODEL")
    if os.getenv("IDLE_THRESHOLD"):
        config['timing']['idle_threshold'] = float(os.getenv("IDLE_THRESHOLD"))
    if os.getenv("RESPONSE_DELAY"):
        config['timing']['response_delay'] = float(os.getenv("RESPONSE_DELAY"))
    if os.getenv("DEBUG"):
        config['debug'] = os.getenv("DEBUG", "false").lower() in ("true", "1", "yes")

    return config


def deep_merge(base: Dict, override: Dict) -> Dict:
    """Deep merge two dictionaries"""
    result = base.copy()
    for key, value in override.items():
        if key in result and isinstance(result[key], dict) and isinstance(value, dict):
            result[key] = deep_merge(result[key], value)
        else:
            result[key] = value
    return result


def create_user_config():
    """Create user config file with defaults"""
    config_dir = get_config_dir()
    user_config_path = config_dir / "config.yaml"

    if not user_config_path.exists():
        # Copy default config to user config
        default_config_path = Path(__file__).parent / "default_config.yaml"
        with open(default_config_path, 'r') as f:
            default_content = f.read()

        with open(user_config_path, 'w') as f:
            f.write(default_content)

        return user_config_path
    return None


def create_user_config_cli():
    """CLI command to create user configuration file"""
    import sys

    config_path = create_user_config()

    if config_path:
        print(f"✅ Created user configuration file:")
        print(f"   {config_path}")
        print()
        print("Edit this file to customize:")
        print(f"   - Vision model (minicpm-v, llama3.2-vision, llava)")
        print(f"   - Vision prompt for better responses")
        print(f"   - Timing settings (idle threshold, response delay)")
        print(f"   - Approval keywords")
        print()
        print(f"Edit with: nano {config_path}")
    else:
        config_dir = get_config_dir()
        config_path = config_dir / "config.yaml"
        print(f"ℹ️  Configuration file already exists:")
        print(f"   {config_path}")
        print()
        print(f"Edit with: nano {config_path}")

    sys.exit(0)


# Load configuration
_config = load_config()

# Export commonly used values
OLLAMA_URL = _config['ollama']['url']
VISION_MODEL = _config['ollama']['model']
VISION_TIMEOUT = _config['ollama']['timeout']

IDLE_THRESHOLD = _config['timing']['idle_threshold']
RESPONSE_DELAY = _config['timing']['response_delay']
SCREENSHOT_TIMEOUT = _config['timing']['screenshot_timeout']

VISION_PROMPT = _config['vision_prompt']
RESPONSE_MAPPING = _config['response_mapping']
APPROVAL_KEYWORDS = _config['approval_keywords']

OUTPUT_BUFFER_SIZE = _config['buffer']['size']
SCREENSHOT_TOOLS = _config['screenshot']['tools']
SCREENSHOT_CACHE_CLEANUP = _config['screenshot']['cache_cleanup_seconds']

DEBUG = _config['debug']


def get_config() -> Dict[str, Any]:
    """Get full configuration dict"""
    return _config.copy()


def reload_config():
    """Reload configuration from files"""
    global _config, OLLAMA_URL, VISION_MODEL, VISION_TIMEOUT
    global IDLE_THRESHOLD, RESPONSE_DELAY, SCREENSHOT_TIMEOUT
    global VISION_PROMPT, RESPONSE_MAPPING, APPROVAL_KEYWORDS
    global OUTPUT_BUFFER_SIZE, SCREENSHOT_TOOLS, SCREENSHOT_CACHE_CLEANUP, DEBUG

    _config = load_config()

    OLLAMA_URL = _config['ollama']['url']
    VISION_MODEL = _config['ollama']['model']
    VISION_TIMEOUT = _config['ollama']['timeout']

    IDLE_THRESHOLD = _config['timing']['idle_threshold']
    RESPONSE_DELAY = _config['timing']['response_delay']
    SCREENSHOT_TIMEOUT = _config['timing']['screenshot_timeout']

    VISION_PROMPT = _config['vision_prompt']
    RESPONSE_MAPPING = _config['response_mapping']
    APPROVAL_KEYWORDS = _config['approval_keywords']

    OUTPUT_BUFFER_SIZE = _config['buffer']['size']
    SCREENSHOT_TOOLS = _config['screenshot']['tools']
    SCREENSHOT_CACHE_CLEANUP = _config['screenshot']['cache_cleanup_seconds']

    DEBUG = _config['debug']