import os
from dataclasses import dataclass, field
from typing import Dict


@dataclass
class ModelConfig:
    
    HF_TOKEN: str = field(default_factory=lambda: os.environ.get("HF_TOKEN", "HF_TOKEN"))

    
    LID_MODEL: str = "drrobot9/wav2vec2-nigerian-language-identifier-v8"

   
    ASR_MODELS: Dict[str, str] = field(default_factory=lambda: {
        "yoruba":  "NCAIR1/Yoruba-ASR",
        "igbo":    "NCAIR1/Igbo-ASR",
        "hausa":   "NCAIR1/Hausa-ASR",
        "english": "NCAIR1/NigerianAccentedEnglish",
    })

    # LLM
    LLM_MODEL: str = "NCAIR1/N-ATLaS"

    # Audio
    SAMPLING_RATE: int = 16_000

    # LLM generation
    LLM_MAX_NEW_TOKENS:     int   = 1000
    LLM_TEMPERATURE:        float = 0.1
    LLM_REPETITION_PENALTY: float = 1.12

    # Memory
    MAX_HISTORY_TURNS: int = 10


config = ModelConfig()