Function _extract_pdf_url

agents/reading-agent/main.py:1745–1781

(paper: Dict[str, Any])

def _extract_pdf_url(paper: Dict[str, Any]) -> str:
    candidates = [
        paper.get("pdf_url"),
        paper.get("url"),
        paper.get("paper_url"),
        paper.get("doi_url"),
        paper.get("openreview_url"),
        (paper.get("metadata") or {}).get("pdf_url"),
        (paper.get("metadata") or {}).get("url"),
        (paper.get("metadata") or {}).get("link"),
        (paper.get("metadata") or {}).get("paper_url"),
        (paper.get("metadata") or {}).get("doi_url"),
        (paper.get("metadata") or {}).get("openreview_url"),
    ]

    for candidate in candidates:
        text = _clean_text(candidate)
        if _looks_like_pdf_url(text):
            return text

    for source_url in _collect_source_page_urls(paper):
        try:
            resolved = _resolve_pdf_url_from_source_page(source_url)
        except Exception as exc:
            print(f"  Failed to resolve PDF from source page {source_url[:120]}: {exc}")
            heuristic_candidates = _build_pdf_url_candidates_from_source_url(source_url)
            if heuristic_candidates:
                print(f"  Falling back to heuristic PDF URL: {heuristic_candidates[0][:120]}")
                return heuristic_candidates[0]
            continue
        if _clean_text(resolved):
            return resolved

    arxiv_id = _clean_text(paper.get("arxiv_id"))
    if arxiv_id:
        return f"https://arxiv.org/pdf/{arxiv_id}.pdf"
    return ""


def _get_pdf_enrichment_mode() -> str: