(soup: BeautifulSoup, document_id: str)
| 29 | |
| 30 | # Inside documents.py |
| 31 | def extract_title(soup: BeautifulSoup, document_id: str) -> str: |
| 32 | title = soup.title.string if soup.title else None |
| 33 | if not title: |
| 34 | title = f"Untitled Document {document_id}" |
| 35 | meta_title = soup.find('meta', attrs={'property': 'og:title'}) |
| 36 | if meta_title: |
| 37 | title = meta_title.get('content', title) |
| 38 | return title |
| 39 | |
| 40 | |
| 41 | def extract_primary_image(soup: BeautifulSoup) -> str: |