config.py

"""
Configuration management for Quran Transcription API
"""

import os
from typing import Optional
from pathlib import Path

# Handle both pydantic v1 and v2
try:
    from pydantic_settings import BaseSettings
except ImportError:
    from pydantic import BaseSettings


class Settings(BaseSettings):
    """Application settings loaded from environment variables and .env file"""
    
    # Server configuration
    host: str = "0.0.0.0"
    port: int = 8888
    reload: bool = False
    workers: int = 1
    
    # API configuration
    title: str = "Quran Recitation Transcription API"
    description: str = "Arabic/Quran speech-to-text service using Faster-Whisper"
    version: str = "1.0.0"
    
    # CORS configuration
    cors_origins: str = "http://localhost:3000,http://localhost:5173"
    
    # Model configuration
    whisper_model: str = "OdyAsh/faster-whisper-base-ar-quran"
    compute_type: str = "float32"  # float32, float16, int8
    device: Optional[str] = None  # auto-detect if None
    
    # GPU configuration
    cuda_visible_devices: Optional[str] = "0"
    
    # File configuration
    max_file_size_mb: int = 100
    allowed_audio_formats: list[str] = ["mp3", "wav", "flac", "m4a", "aac", "ogg", "opus", "webm"]
    
    # Logging configuration
    log_level: str = "INFO"
    
    # Transcription parameters
    beam_size: int = 1
    vad_filter: bool = True
    vad_min_silence_duration_ms: int = 500
    language: str = "ar"
    
    class Config:
        env_file = ".env"
        env_file_encoding = "utf-8"
        case_sensitive = False
        
        # Example values for documentation
        json_schema_extra = {
            "example": {
                "host": "0.0.0.0",
                "port": 8888,
                "whisper_model": "OdyAsh/faster-whisper-base-ar-quran",
                "compute_type": "float32"
            }
        }


def get_settings() -> Settings:
    """Get settings instance with cached results"""
    return Settings()


def get_device() -> str:
    """Determine device based on CUDA availability and environment"""
    import torch
    
    settings = get_settings()
    
    if settings.device:
        return settings.device
    
    # Auto-detect
    if settings.cuda_visible_devices and torch.cuda.is_available():
        return "cuda"
    
    return "cpu"


def get_cors_origins() -> list[str]:
    """Parse CORS origins from settings"""
    settings = get_settings()
    return [origin.strip() for origin in settings.cors_origins.split(",")]


# Export settings instance
settings = get_settings()