arm-core/api/arm_statements/model.py

"""Типы входа (кандидат формы) и выхода (нормализованный документ) пакета."""

from dataclasses import dataclass
from datetime import date
from decimal import Decimal
from enum import StrEnum


class Source(StrEnum):
    OCR = "ocr"
    EXCEL = "excel"


class DocumentType(StrEnum):
    BALANCE = "BALANCE"
    INCOME = "INCOME"
    DISCLOSURE = "DISCLOSURE"


class FormVariant(StrEnum):
    FULL = "full"
    SIMPLIFIED = "simplified"


class Edition(StrEnum):
    RAS_66N = "66n"
    FSBU4 = "fsbu4"


class PeriodKind(StrEnum):
    ABSOLUTE = "absolute"
    CURRENT = "relative:current"
    PREV_YEAR_END = "relative:prev_year_end"
    PREV_PREV_YEAR_END = "relative:prev_prev_year_end"
    PREV_PERIOD = "relative:prev_period"
    NONE = "none"


class LineStatus(StrEnum):
    OK = "ok"
    UNREADABLE = "unreadable"
    MISSING = "missing"
    DETAIL = "detail"


class Severity(StrEnum):
    ERROR = "error"
    INFO = "info"


class DocStatus(StrEnum):
    DONE = "done"
    NEEDS_CHECK = "needs_check"


class EnvelopeStatus(StrEnum):
    DONE = "done"
    NEEDS_CHECK = "needs_check"
    ERROR = "error"


class SignForm(StrEnum):
    PARENTHESES = "parentheses"
    MINUS = "minus"
    UNSIGNED = "unsigned"


class RowRole(StrEnum):
    TITLE = "title"
    HEADER = "header"
    NUMBERING = "numbering"
    DATA = "data"


CellValue = Decimal | date | bool | None


@dataclass(frozen=True, slots=True)
class Cell:
    text: str
    value: CellValue
    confidence: float
    ref: str


@dataclass(frozen=True, slots=True)
class Evidence:
    okud: Cell | None = None
    title: Cell | None = None


@dataclass(frozen=True, slots=True)
class HeaderCells:
    organization_name: Cell | None = None
    inn: Cell | None = None
    report_date: Cell | None = None
    units: Cell | None = None
    okud: Cell | None = None
    okpo: Cell | None = None
    okopf: Cell | None = None
    okfs: Cell | None = None
    okei: Cell | None = None
    address: Cell | None = None


HEADER_FIELDS: tuple[str, ...] = (
    "organization_name",
    "inn",
    "report_date",
    "units",
    "okud",
    "okpo",
    "okopf",
    "okfs",
    "okei",
    "address",
)


@dataclass(frozen=True, slots=True)
class StatementWarning:
    code: str
    severity: Severity
    message: str
    line_refs: tuple[str, ...] = ()
    column_index: int | None = None
    field: str | None = None
    locator: str | None = None


@dataclass(frozen=True, slots=True)
class FormCandidate:
    source: Source
    locator: str
    type_hint: DocumentType | None
    evidence: Evidence
    header: HeaderCells
    rows: tuple[tuple[Cell | None, ...], ...]
    segment_breaks: tuple[int, ...] = ()
    warnings: tuple[StatementWarning, ...] = ()


@dataclass(frozen=True, slots=True)
class HeaderField:
    text: str | None
    confidence: float | None
    value: str | None


@dataclass(frozen=True, slots=True)
class DateField:
    text: str | None
    confidence: float | None
    value: date | None


@dataclass(frozen=True, slots=True)
class StatementHeader:
    organization_name: HeaderField
    inn: HeaderField
    report_date: DateField
    units: HeaderField
    okud: HeaderField
    okpo: HeaderField
    okopf: HeaderField
    okfs: HeaderField
    okei: HeaderField
    address: HeaderField
    form_name: str | None = None
    needs_review_fields: tuple[str, ...] = ()


@dataclass(frozen=True, slots=True)
class StatementColumn:
    column_index: int
    label: str
    label_confidence: float | None
    period_date: date | None
    period_kind: PeriodKind


@dataclass(frozen=True, slots=True)
class StatementValue:
    column_index: int
    amount: Decimal | None
    string_value: str | None
    text: str | None
    confidence: float
    is_recognized: bool
    needs_review: bool = False


@dataclass(frozen=True, slots=True)
class StatementLine:
    row_index: int
    ref: str
    line_status: LineStatus
    line_code: str | None
    line_code_text: str | None
    line_code_confidence: float | None
    line_name: str | None
    line_name_confidence: float | None
    parent_ref: str | None
    needs_review: bool
    values: tuple[StatementValue, ...]


@dataclass(frozen=True, slots=True)
class FailedRelation:
    column_index: int | None
    column_key: str
    relation: str
    codes: tuple[str, ...]
    expected: Decimal | None
    actual: Decimal | None


@dataclass(frozen=True, slots=True)
class StatementChecks:
    relations_failed: tuple[FailedRelation, ...]
    soft_relations_failed: tuple[FailedRelation, ...]


@dataclass(frozen=True, slots=True)
class StatementDocument:
    document_type: DocumentType
    form_variant: FormVariant | None
    locator: str
    header: StatementHeader
    columns: tuple[StatementColumn, ...]
    lines: tuple[StatementLine, ...]
    checks: StatementChecks | None
    status: DocStatus
    warnings: tuple[StatementWarning, ...]


@dataclass(frozen=True, slots=True)
class CandidateResult:
    document: StatementDocument | None
    file_warnings: tuple[StatementWarning, ...]


@dataclass(frozen=True, slots=True)
class NormalizationResult:
    documents: tuple[StatementDocument, ...]
    file_warnings: tuple[StatementWarning, ...]