(html: str, *, strip_boilerplate: bool)
| 11 | |
| 12 | |
| 13 | def _extract_sync(html: str, *, strip_boilerplate: bool) -> str: |
| 14 | extracted = trafilatura.extract( |
| 15 | html, |
| 16 | output_format="markdown", |
| 17 | include_links=True, |
| 18 | include_images=False, |
| 19 | include_tables=True, |
| 20 | favor_recall=not strip_boilerplate, |
| 21 | ) |
| 22 | return extracted or "" |
| 23 | |
| 24 | |
| 25 | async def html_to_markdown( |
nothing calls this directly
no outgoing calls
no test coverage detected