tri_matrix.py

from dataclasses import dataclass

@dataclass
class TriBool:
    """Holds two boolean values.

    Attributes:
        tr_in_ocr: if condition is true for transcription word in ocr word
        ocr_in_tr: if condition is true for ocr word in transcription word
    """
    tr_in_ocr : bool = False
    ocr_in_tr : bool = False

    def __str__(self):
        return f"{self.tr_in_ocr=}, {self.ocr_in_tr=}"

class IsHeadBodyTail:
    """Holds substring checking values for two words.

    Attributes:
        is_head (TriBool): if either word is the head of the other
            (one word is the start of the other word, but
            is not completely the same as the other word)
        is_body (TriBool): if either word is the body of the other
            (one word is a substring of the other word, but
            is not completely the same as the other word)
        is_tail (TriBool): if either word is the tail of the other
            (one word is the end of the other word, but
            is not completely the same as the other word)
    """

    def __init__(self, is_head: TriBool = None, is_body: TriBool = None, is_tail: TriBool = None) -> None:
        if is_head is None:
            is_head = TriBool()
        if is_body is None:
            is_body = TriBool()
        if is_tail is None:
            is_tail = TriBool()
        self.is_head = is_head
        self.is_body = is_body
        self.is_tail = is_tail
    
    def __str__(self):
        return f"{self.is_head=}, {self.is_body=}, {self.is_tail=}"
    
    def set_headbodytail(self, head: bool, tail: bool, tr_len: int, ocr_len: int, body: bool = None) -> None:
        """Sets the head body and tail values, dependent on the length of the words."""
        self.is_head.tr_in_ocr = head and tr_len < ocr_len
        self.is_head.ocr_in_tr = head and ocr_len < tr_len
        self.is_tail.tr_in_ocr = tail and tr_len < ocr_len
        self.is_tail.ocr_in_tr = tail and ocr_len < tr_len
        self.is_body.tr_in_ocr = body and tr_len < ocr_len
        self.is_body.ocr_in_tr = body and ocr_len < tr_len