from pydantic_settings import BaseSettings


# Each entry: full metadata shown in the model picker UI
# First entry = default model loaded at startup
AVAILABLE_MODELS: list[dict] = [
    {
        "id": "dslim/bert-base-NER",
        "label": "Standard",
        "quality": "good",
        "quality_score": 2,
        "speed": "fast",
        "is_default": False,
        "entity_types": ["PER", "ORG", "LOC"],
        "description": "The recommended starting point. Reliably finds names, companies and places — and also catches emails, phone numbers, IBANs and tax codes. Fast and accurate, with very few false alarms. Good for most everyday documents.",
    },
    {
        "id": "Jean-Baptiste/roberta-large-ner-english",
        "label": "Thorough",
        "quality": "best",
        "quality_score": 3,
        "speed": "slow",
        "is_default": True,
        "entity_types": ["PER", "ORG", "LOC", "MISC"],
        "description": "A more powerful mode that catches sensitive data even in complex or messy documents — think CVs, contracts, mixed-language content. It finds more than Standard, but takes a bit longer to process. Use this when thoroughness matters more than speed.",
    },
    {
        "id": "iiiorg/piiranha-v1-detect-personal-information",
        "label": "Light",
        "quality": "limited",
        "quality_score": 1,
        "speed": "fast",
        "is_default": False,
        "entity_types": ["PER", "EMAIL", "PHONE", "ADDRESS", "ORG", "LOC", "DATE", "ID"],
        "description": "A fast, lightweight scan covering a wide range of data types including dates and IDs. It may occasionally miss part of a name or produce false positives. Best for quick, informal checks rather than critical documents.",
    },
]

DEFAULT_MODEL_ID: str = AVAILABLE_MODELS[1]["id"]


class Settings(BaseSettings):
    # Environment: "local" | "staging" | "production"
    app_env: str = "local"

    default_model_name: str = DEFAULT_MODEL_ID
    host: str = "0.0.0.0"
    port: int = 8000

    # Set FRONTEND_URL in HF Space secrets for production (e.g. https://sanitizemydata.com)
    frontend_url: str = "https://sanitizemydata.com"

    # Base origins always allowed; production frontend URL added from env or default
    @property
    def allowed_origins(self) -> list[str]:
        origins = ["http://localhost:3000", "http://127.0.0.1:3000"]
        if self.frontend_url:
            origins.append(self.frontend_url)
        return origins

    model_config = {"env_file": ".env", "env_file_encoding": "utf-8"}


settings = Settings()