INNER CODE UNIT · Python

extract_paper_data

DmitryRyumin/ICCV-2023-25-Papers · code/markdown_to_json_parser.py:356

def extract_paper_data(paper_section, columns):
    title_column = columns[0]
    # title = title_column.get_text(strip=True)
    title = (
        title_column.a.encode_contents().decode("utf-8")
        if title_column.a is not None
        else (
            title_column.encode_contents().decode("utf-8")
            if title_column.get_text(strip=True) is not None
            else None
        )
    )

    title = re.sub(r"<(?:br\s*/?>|img[^>]*>)", "", title)
    title = title.strip()

    html_entities = {
        "&amp;": "&",

View source record →