Convert a mht file for Microsoft Teams Wiki into markdown format

convert_teams_wiki.py

import argparse
import logging
import sys

from bs4 import Tag, NavigableString, BeautifulSoup

logger = logging.getLogger(__name__)


class Context:
    def __init__(self):
        self.tags = []
        self.indent_level = 0
        self.table_shape = (0, 0)
        self.table_index = (0, 0)

    @property
    def tag_depth(self):
        return len(self.tags)

    def down_tag(self, tag: str):
        self.tags.append(tag)

    def up_tag(self):
        return self.tags.pop()

    def inc_indent(self):
        self.indent_level += 1
        return self.indent_level

    def dec_indent(self):
        self.indent_level -= 1
        return self.indent_level

    def bullet_prefix(self):
        """
        Return a bullet sign for listing.
        """
        indent = '    ' * (self.indent_level - 1)
        for tag in reversed(self.tags):
            if tag == 'ul':
                return indent + '-'
            elif tag == 'ol':
                return indent + '1.'
        raise ValueError('Not inside "ul" or "ol".')

    def inside_table(self):
        return 'table' in self.tags

    def is_second_row(self):
        row, _ = self.table_index
        return row == 2

    def next_row_index(self):
        row, col = self.table_index
        self.table_index = (row + 1, col)

    def next_column_index(self):
        row, col = self.table_index
        self.table_index = (row, col + 1)

    def clear_table_index(self):
        self.table_index = (0, 0)


class MarkdownConverter:
    def __init__(self):
        self._context: Context = None

    def convert(self, content: Tag):
        self._context = Context()
        return self._convert(content)

    def _convert(self, content):
        if isinstance(content, Tag):
            logger.info('%s => %s', type(content), content.name)
            try:
                self._context.down_tag(content.name)
                return self._convert_tag(content)
            finally:
                self._context.up_tag()
        elif isinstance(content, NavigableString):
            logger.info('%s => %s', type(content), content)
            return str(content).strip()

    def _convert_tag(self, tag: Tag):
        if tag.name == 'html':
            return self._convert_html(tag)
        elif tag.name == 'h1':
            return f'# {self._convert_contents(tag)}\n\n'
        elif tag.name == 'h3':
            return self._convert_h3(tag)
        elif tag.name == 'div':
            return self._convert_div(tag)
        elif tag.name == 'p':
            return f'{self._convert_contents(tag)}\n'
        elif tag.name == 'blockquote':
            return f'>{self._convert_contents(tag)}\n'
        elif tag.name == 'pre':
            # pre配下のタグは読み取らずテキストとして解釈する
            return f'```\n{tag.text.strip()}\n```\n'
        elif tag.name == 'code':
            return f'`{self._convert_contents(tag)}`'
        elif tag.name == 'a':
            return self._convert_a(tag)
        elif tag.name == 'span':
            return f'{self._convert_contents(tag)}'
        elif tag.name == 'u':
            return f'{self._convert_contents(tag)}'
        elif tag.name == 'b':
            return f'**{self._convert_contents(tag)}**'
        elif tag.name == 'em':
            return f'*{self._convert_contents(tag)}*'
        elif tag.name == 'strong':
            return f'__{self._convert_contents(tag)}__'
        elif tag.name == 'br':
            return self._convert_br()
        elif tag.name == 'ul':
            return self._convert_list(tag)
        elif tag.name == 'ol':
            return self._convert_list(tag)
        elif tag.name == 'li':
            return self._convert_li(tag)
        elif tag.name == 'table':
            return self._convert_table(tag)
        elif tag.name == 'colgroup':
            return ''  # ignore
        elif tag.name == 'tbody':
            return f'{self._convert_contents(tag)}\n'
        elif tag.name == 'tr':
            self._context.next_row_index()
            return self._convert_tr(tag)
        elif tag.name == 'td':
            self._context.next_column_index()
            return f'{self._convert_contents(tag).strip()} | '
        elif tag.name == 'img':
            return self._convert_img(tag)
        else:
            return f'<<<CANNOT CONVERT {tag.name}>>>'

    def _convert_contents(self, tag: Tag):
        return ''.join([self._convert(c) for c in tag.children])

    def _convert_a(self, a: Tag):
        href = a.attrs['href']
        text = self._convert_contents(a)
        if href == text:
            # 表示テキストとURLが一致する場合はURLのみ出力
            return href
        else:
            return f'[{text}]({href})'

    def _convert_list(self, tag: Tag):
        try:
            self._context.inc_indent()
            return f'\n{self._convert_contents(tag)}'
        finally:
            self._context.dec_indent()

    def _convert_li(self, li: Tag):
        inner_text = self._convert_contents(li)
        prefix = self._context.bullet_prefix()
        return f'{prefix} {inner_text}\n'

    def _convert_html(self, html: Tag):
        # Embedded metadata into HTML comment
        metadata = f'''\
<!--
page: {html.attrs["data-page"]}
canvas: {html.attrs["data-canvas"]}
site: {html.attrs["data-site"]}
list: {html.attrs["data-list"]}
tabId: {html.attrs["data-tabid"]}
slug: {html.attrs["data-slug"]}
threadId: {html.attrs["data-threadid"]}
-->

'''
        return metadata + self._convert_contents(html)

    def _convert_h3(self, tag: Tag):
        if 'wiki-mht-note' in tag.attrs.get('class', []):
            # wiki-mht-noteクラスはTeams特有の非表示クラスなので
            # 特別扱いしてMarkdownに変換しない
            return ''
        return f'### {self._convert_contents(tag)}\n\n'

    def _convert_br(self):
        if self._context.inside_table():
            return '<br />'
        else:
            return ' \n'

    def _convert_div(self, tag):
        if self._context.inside_table():
            return f'{self._convert_contents(tag)}<br />'
        else:
            return f'{self._convert_contents(tag)}\n'

    def _convert_table(self, table: Tag):
        try:
            rows = len(table.find_all('tr'))
            cols = len(table.find_all('col'))
            self._context.table_shape = (rows, cols)
            return f'\n{self._convert_contents(table)}\n'
        finally:
            self._context.clear_table_index()

    def _convert_tr(self, tr: Tag):
        if self._context.is_second_row():
            # 2行目はヘッダーとボディのセパレーターを入れる
            separator = '|' + (' --- |' * self._context.table_shape[1]) + '\n'
        else:
            separator = ''
        return separator + '| ' + self._convert_contents(tr) + '\n'

    def _convert_img(self, img: Tag):
        if 'data-preview-src' in img.attrs:
            src = img.attrs['data-preview-src']
            # "data-preview-src" attribute is a string like below.
            # "/sites/msteams_1b3d5f/Teams Wiki Data/General/img-123-fe3028d38dc34d1e94b6cd350d0f9941.png"
            # Last entry "img-123-fe3028d38dc34d1e94b6cd350d0f9941.png" is a image file stored in
            # "Teams Wiki Data" folder, so link it as embedded image.
            url = src.split('/')[-1]
        else:
            # Preserve original "src" attrribute
            url = img.attrs['src']
        return f'![]({url})'

def cli_main():
    parser = argparse.ArgumentParser()
    parser.add_argument("--verbose", action="store_true", help="output logging message")
    parser.add_argument("file", type=str, help="mhtml file to convert into markdown")
    args = parser.parse_args(sys.argv[1:])

    logging.basicConfig(level=logging.INFO if args.verbose else logging.WARNING, stream=sys.stderr)
    with open(args.file, mode='r', encoding='utf-8') as f:
        html_text = f.read()
    soup = BeautifulSoup(html_text, "html.parser")  # HTMLを解析する

    # print(soup.prettify())
    html: Tag = soup.html
    converter = MarkdownConverter()
    markdown = converter.convert(html)
    print(markdown)


if __name__ == '__main__':
    cli_main()

pyproject.toml

[project]
name = "convert-teams-wiki"
version = "0.1.1"
description = "Convert a mht file for Microsoft Teams Wiki into markdown format"
requires-python = ">=3.6"
classifiers = [
    "Programming Language :: Python :: 3",
]
dependencies = [
    "beautifulsoup4>=4.5",
]

[project.scripts]
convert-teams-wiki = "convert_teams_wiki:cli_main"

[build-system]
requires = ["setuptools>=40.8.0", "wheel"]
build-backend = "setuptools.build_meta:__legacy__"

Thank you for this! This was super helpful and worked beautifully for me.