Text dropped in table after setting `include_formatting=True`

[SnowFox4004]

content in table was dropped if it contains a format tag and enabled include_formatting
example:

<table>
    <tr>
        <td>
            <p><b>GPT-5.4</b></p>
        </td>
    </tr>
</table>

will become

<table>
  <row>
    <cell> <p/> </cell>
  </row>
</table>

I don't have any idea why this happenedso I ask GLM-5 and it said this:

---

When include_formatting=True, trafilatura keeps format tags such as <b><strong><i>  and converts them to the internal> <hi rend="#b">
  format. The problem is its strip_tags() process:

1. Normal Flow (include_formatting=False):
      - <td><p><b>GPT-5.4</b></p></td> → <cell><p>GPT-5.4</p></cell> ✅

2. Question flow (include_formatting=True):
      - <td><p><b>GPT-5.4</b></p></td>
      - → Convert to <cell><p><hi rend="#b">GPT-5.4</hi></p></cell>
      - → In a cleanup/merge step,<hi> the text inside was incorrectly handled
      - → results become <cell><p></p></cell> ❌

---

Since I hope to use it to extract webpage content and made them an ebook to read, I hope I could keep formatting and corect table structure.
Now I can set include_formatting=False to fix the problem but it's not perfect.
I would like to help if anyone could tell me how to fix it.