config.py

import os
from typing import Optional

class Config:
    """Application configuration for working with local GGUF models"""
    
    # Model settings - using Hugging Face downloaded model
    MODEL_REPO: str = os.getenv("MODEL_REPO", "lmstudio-community/gemma-3n-E4B-it-text-GGUF")
    MODEL_FILENAME: str = os.getenv("MODEL_FILENAME", "gemma-3n-E4B-it-Q8_0.gguf")
    MODEL_PATH: str = os.getenv("MODEL_PATH", "/app/models/gemma-3n-E4B-it-Q8_0.gguf")
    HUGGINGFACE_TOKEN: str = os.getenv("HUGGINGFACE_TOKEN", "")
    
    # Model loading settings - optimized for Docker container
    N_CTX: int = int(os.getenv("N_CTX", "4096"))  # Reduced context window for Docker
    N_GPU_LAYERS: int = int(os.getenv("N_GPU_LAYERS", "0"))  # CPU-only for Docker by default
    N_THREADS: int = int(os.getenv("N_THREADS", "4"))  # Conservative thread count
    N_BATCH: int = int(os.getenv("N_BATCH", "512"))  # Smaller batch size for Docker
    USE_MLOCK: bool = os.getenv("USE_MLOCK", "false").lower() == "true"  # Disabled for Docker
    USE_MMAP: bool = os.getenv("USE_MMAP", "true").lower() == "true"  # Keep memory mapping
    F16_KV: bool = os.getenv("F16_KV", "true").lower() == "true"  # Use 16-bit keys and values
    SEED: int = int(os.getenv("SEED", "42"))  # Random seed for reproducibility
    
    # Server settings - Docker compatible
    HOST: str = os.getenv("HOST", "0.0.0.0")
    GRADIO_PORT: int = int(os.getenv("GRADIO_PORT", "7860"))  # Standard HuggingFace Spaces port
    API_PORT: int = int(os.getenv("API_PORT", "8000"))
    
    # Generation settings - optimized for Docker
    MAX_NEW_TOKENS: int = int(os.getenv("MAX_NEW_TOKENS", "256"))  # Reduced for faster response
    TEMPERATURE: float = float(os.getenv("TEMPERATURE", "0.1"))
    
    # File upload settings
    MAX_FILE_SIZE: int = int(os.getenv("MAX_FILE_SIZE", "10485760"))  # 10MB
    ALLOWED_IMAGE_EXTENSIONS: set = {".jpg", ".jpeg", ".png", ".gif", ".bmp", ".webp"}
    
    @classmethod
    def is_model_available(cls) -> bool:
        """Check if local model file exists"""
        return os.path.exists(cls.MODEL_PATH)
    
    @classmethod
    def get_model_path(cls) -> str:
        """Get absolute path to model file"""
        return os.path.abspath(cls.MODEL_PATH)
    
    @classmethod
    def get_models_dir(cls) -> str:
        """Get models directory path"""
        return os.path.dirname(cls.MODEL_PATH)
    
    @classmethod
    def load_from_env_file(cls, env_file: str = ".env") -> None:
        """Load configuration from .env file"""
        if os.path.exists(env_file):
            with open(env_file, 'r') as f:
                for line in f:
                    line = line.strip()
                    if line and not line.startswith('#') and '=' in line:
                        key, value = line.split('=', 1)
                        os.environ[key.strip()] = value.strip()

# Automatically load from .env file on import
Config.load_from_env_file()