BIDINLTK/dev/bert/3.py

from docx import Document
from html2docx import html2docx

def convert_html_to_docx(html_content, output_path):
    # 将 HTML 内容转换为 DOCX
    _io = html2docx(html_content, "1234")

    with open(output_path,"wb") as f:
        f.write(_io.getvalue())

    print(f"DOCX file saved to: {output_path}")

if __name__ == "__main__":
    # 示例：将 HTML 内容转换为 DOCX
    html_content = """
<div>
<div>

  <b>Hello</b>, <i>this is</i> <font color='red'>rich text</font>.
</div>
<div>
  <img src="D:/Workspace2016/BIDINLTK/dev/bert/img.png"/>
  </div>

</div>
    """

    output_docx_path = "output_document.docx"

    convert_html_to_docx(html_content, output_docx_path)