interfaces.py

"""Common interfaces for the MRZ reader components"""
import abc
from dataclasses import dataclass
from typing import Optional


@dataclass
class PreProcessors:
    """A configuration of preprocessors"""

    grayscale: Optional[bool] = None
    threshold: Optional[int] = None
    variable_threshold: Optional[bool] = None


@dataclass
class PostProcessors:
    """A configuration of postprocessors"""

    character_height: Optional[bool] = None
    mrz_fields: Optional[bool] = None
    line_lengths: Optional[bool] = None


@dataclass
class PostProcessorMetadata:
    """Metadata generated by the engine for the postprocessors"""

    box_heights: Optional[list[float]] = None


class Engine(abc.ABC):
    """The engines capable of doing the main OCR task"""

    @abc.abstractmethod
    def get_mrz_text(
        self, image, verbose=False
    ) -> Optional[tuple[str, PostProcessorMetadata]]:
        """Read the MRZ text from an image, and generate metadata for postprocessing"""