Team Ai
Apppublic

documentExtractionag051/ExtractDocument

sourceHugging Faceupdated 9mo agoView on Hugging Face
0likes
filter_utils.py30 linesDownload Raw Back to utils
1from enum import Enum2 3from .geometry_utils import TextRect4 5 6class FilterType(Enum):7    PAGE = 'page'8    BLOCK_TYPE = 'block_type'9    TEXT_CONTENT = 'text_content'10 11 12class FilterMethod(Enum):13    EQUALS = 'equals'14    CONTAINS = 'contains'15 16 17class FilterRects:18    19    @staticmethod20    def filter_by_page(rects: list[TextRect], page_num: int) -> list[TextRect]:21        return [r for r in rects if r.page_num == page_num]22    23    @staticmethod24    def filter_by_block_type(rects: list[TextRect], block_type: str) -> list[TextRect]:25        return [r for r in rects if r.block_type == block_type]26    27    @staticmethod28    def filter_by_text_content(rects: list[TextRect], text_substring: str) -> list[TextRect]:29        return [r for r in rects if r.text and text_substring in r.text]30