(candidate: Any)
| 1559 | |
| 1560 | |
| 1561 | def _looks_like_pdf_url(candidate: Any) -> bool: |
| 1562 | text = _clean_text(candidate).lower() |
| 1563 | if not (text.startswith("http://") or text.startswith("https://")): |
| 1564 | return False |
| 1565 | return ( |
| 1566 | ".pdf" in text |
| 1567 | or "/pdf?" in text |
| 1568 | or "/pdf/" in text |
| 1569 | or "arxiv.org/pdf/" in text |
| 1570 | or "openreview.net/pdf" in text |
| 1571 | ) |
| 1572 | |
| 1573 | |
| 1574 | def _collect_source_page_urls(paper: Dict[str, Any]) -> List[str]: |
no test coverage detected