src/gaia/config/env.py

"""
Environment variable handler for GAIA configuration.

This module handles the loading of environment variables for GAIA configuration,
providing a structured way to override default configuration values with
environment variables.
"""

import os
from typing import Dict, Any, Optional

# Try to load dotenv for development environments
try:
    from dotenv import load_dotenv
    load_dotenv(verbose=False)
except ImportError:
    # No-op implementation when dotenv is not available
    def load_dotenv(*args, **kwargs):
        pass

# Environment variable mapping to configuration paths
ENV_CONFIG_MAPPING = {
    # API Keys
    "OPENAI_API_KEY": "api.openai.api_key",
    "OPENAI_API_BASE": "api.openai.api_base",
    "SERPER_API_KEY": "api.serper.api_key",
    "SERPER_API_URL": "api.serper.api_url",
    "PERPLEXITY_API_KEY": "api.perplexity.api_key",
    "PERPLEXITY_API_URL": "api.perplexity.api_url",
    "SUPABASE_KEY": "api.supabase.key",
    "SUPABASE_URL": "api.supabase.url",
    "HF_TOKEN": "api.huggingface.token",
    
    # Model Configuration
    "GAIA_DEFAULT_MODEL": "models.default_model",
    "GAIA_EMBEDDING_MODEL": "models.embedding_model",
    "GAIA_FALLBACK_MODEL": "models.fallback_model",
    "GAIA_VISION_MODEL": "models.vision_model",
    "GAIA_REASONING_MODEL": "models.reasoning_model",
    "GAIA_MODEL_TEMPERATURE": "models.temperature",
    "GAIA_MODEL_MAX_TOKENS": "models.max_tokens",
    
    # Tool Configuration
    "GAIA_WEB_SEARCH_RESULT_COUNT": "tools.web_search.result_count",
    "GAIA_WEB_SEARCH_TIMEOUT": "tools.web_search.timeout",
    "GAIA_ARXIV_MAX_RESULTS": "tools.arxiv.max_results",
    "GAIA_DUCKDUCKGO_TIMEOUT": "tools.duckduckgo.timeout",
    "GAIA_DUCKDUCKGO_MAX_RESULTS": "tools.duckduckgo.max_results",
    "GAIA_PERPLEXITY_TIMEOUT": "tools.perplexity.timeout",
    "GAIA_PERPLEXITY_MODEL": "tools.perplexity.model",
    
    # Memory Settings
    "GAIA_MEMORY_ENABLED": "memory.enabled",
    "GAIA_MEMORY_TABLE_NAME": "memory.table_name",
    "GAIA_MEMORY_TTL": "memory.ttl",
    "GAIA_MEMORY_CACHE_SIZE": "memory.cache_size",
    "GAIA_MEMORY_MAX_ENTRIES": "memory.max_entries",
    
    # Agent Settings
    "GAIA_MAX_ITERATIONS": "agent.max_iterations",
    "GAIA_VERBOSE": "agent.verbose",
    "GAIA_TIMEOUT": "agent.timeout",
    "GAIA_MAX_RETRIES": "agent.max_retries",
    "GAIA_RETRY_DELAY": "agent.retry_delay",
    
    # Logging Configuration
    "GAIA_LOG_LEVEL": "logging.level",
    "GAIA_LOG_FILE": "logging.file",
    "GAIA_LOG_FORMAT": "logging.format",
}

def get_env_value(env_var: str, default=None) -> Any:
    """
    Get an environment variable value with type conversion.
    
    Args:
        env_var: The environment variable name
        default: Default value if environment variable is not set
        
    Returns:
        The environment variable value with appropriate type conversion
    """
    value = os.getenv(env_var)
    
    if value is None:
        return default
    
    # Convert boolean strings
    if value.lower() in ("true", "yes", "1", "on"):
        return True
    if value.lower() in ("false", "no", "0", "off"):
        return False
    
    # Convert numeric strings
    try:
        if "." in value:
            return float(value)
        return int(value)
    except ValueError:
        # If conversion fails, return as string
        return value

def load_env_config() -> Dict[str, Any]:
    """
    Load configuration from environment variables.
    
    Returns:
        Dict containing configuration values from environment variables
    """
    env_config = {}
    
    # Process each environment variable in the mapping
    for env_var, config_path in ENV_CONFIG_MAPPING.items():
        value = get_env_value(env_var)
        
        if value is not None:
            # Split the config path and build the nested dictionary
            parts = config_path.split(".")
            current = env_config
            
            # Navigate to the correct nested level
            for i, part in enumerate(parts):
                if i == len(parts) - 1:
                    # Last part is the key to set
                    current[part] = value
                else:
                    # Create the nested dict if it doesn't exist
                    if part not in current:
                        current[part] = {}
                    current = current[part]
    
    return env_config