from __future__ import annotations

import re
from dataclasses import dataclass, asdict


ORDER_RE = re.compile(r"\b(?P<order>\d{4}/\d{4}/(?:[1-9]|[1-9]\d))\b")


@dataclass
class ParsedLabel:
    order_number: str | None
    color_code: str | None
    product_model: str | None
    raw_text: str

    def to_dict(self) -> dict[str, str | None]:
        return asdict(self)


def normalize_ocr_text(text: str) -> str:
    return " ".join(text.replace("\n", " ").replace("\r", " ").split())


def parse_label_text(text: str, known_colors: list[str], known_models: list[str]) -> ParsedLabel:
    normalized = normalize_ocr_text(text)
    order_match = ORDER_RE.search(normalized)

    normalized_upper = normalized.upper()
    color_code = next(
        (color for color in known_colors if color.upper() in normalized_upper),
        None,
    )
    product_model = next(
        (model for model in known_models if re.search(rf"\b{re.escape(model)}\b", normalized, re.I)),
        None,
    )

    return ParsedLabel(
        order_number=order_match.group("order") if order_match else None,
        color_code=color_code,
        product_model=product_model,
        raw_text=normalized,
    )