"""Шапка документа из полей `data_ocr` (§11.1 п. 8): ИНН — только цифры, единицы — очищенный текст, дата отчёта —
`parse_date` после замены двойников; «распознано» — `confidence == 100` и сырой текст без мусора."""

from collections.abc import Mapping
from dataclasses import dataclass
from datetime import date

from src.normalizer.ocr.dates import is_date_sized, parse_date
from src.normalizer.ocr.text import clamp_confidence, clean_value, collapse_whitespace, digits_only, fix_lookalikes

__all__ = ["FileHeader", "parse_file_header"]

FULL_CONFIDENCE = 100.0
CAPPED_CONFIDENCE = 99.0


@dataclass(frozen=True)
class FileHeader:
    organization_name: str | None
    organization_confidence: float | None
    inn: str | None
    inn_confidence: float | None
    units: str | None
    report_date_text: str | None
    report_date: date | None
    report_date_recognized: bool


def _field(data_ocr: Mapping[str, object], name: str) -> tuple[str, float | None]:
    raw = data_ocr.get(name)
    if not isinstance(raw, Mapping):
        return "", None
    text = raw.get("form_name")
    text_str = str(text) if isinstance(text, str) else ""
    return (text_str, clamp_confidence(raw.get("confidence")))


def parse_file_header(data_ocr: Mapping[str, object], lookalikes: Mapping[str, str]) -> FileHeader:
    organization, organization_confidence = _field(data_ocr, "organization")
    inn_raw, inn_confidence = _field(data_ocr, "inn")
    units_raw, _ = _field(data_ocr, "units")
    date_raw, date_confidence = _field(data_ocr, "date")

    organization_name = collapse_whitespace(organization) or None
    inn = digits_only(inn_raw) or None
    if inn is None:
        inn_confidence = None
    elif inn_confidence is not None and inn_raw.strip() != inn:
        inn_confidence = min(inn_confidence, CAPPED_CONFIDENCE)
    # решение №31: длинный текст не разбирается и не проходит `fix_lookalikes`
    report_date = (
        parse_date(fix_lookalikes(date_raw, lookalikes)) if date_raw.strip() and is_date_sized(date_raw) else None
    )
    return FileHeader(
        organization_name=organization_name,
        organization_confidence=organization_confidence if organization_name is not None else None,
        inn=inn,
        inn_confidence=inn_confidence,
        units=clean_value(units_raw) or None,
        report_date_text=date_raw.strip() or None,
        report_date=report_date,
        report_date_recognized=date_confidence == FULL_CONFIDENCE and report_date is not None,
    )