Function _normalize_paper

agents/reading-agent/main.py:1181–1226

(paper: Dict[str, Any])

def _normalize_paper(paper: Dict[str, Any]) -> Dict[str, Any]:
    normalized = dict(paper or {})
    normalized["authors"] = _parse_jsonish_list(normalized.get("authors"))

    metadata = normalized.get("metadata")
    if isinstance(metadata, str):
        try:
            metadata = json.loads(metadata)
        except json.JSONDecodeError:
            metadata = {}
    if not isinstance(metadata, dict):
        metadata = {}
    normalized["metadata"] = metadata

    for field in (
        "title",
        "institution",
        "venue",
        "publish_date",
        "arxiv_id",
        "doi",
        "pdf_url",
        "url",
        "paper_url",
        "doi_url",
        "openreview_url",
        "cvf_url",
        "ecva_url",
        "dblp_url",
        "journal",
        "source",
        "category",
    ):
        if field in normalized:
            normalized[field] = _clean_text(normalized.get(field))

    if "abstract" in normalized:
        normalized["abstract"] = _clean_abstract_text(normalized.get("abstract"))

    score = normalized.get("score")
    try:
        normalized["score"] = float(score)
    except (TypeError, ValueError):
        normalized["score"] = 0.0

    return normalized


def _is_direct_upload_request(request_metadata: Dict[str, Any]) -> bool:
    source_type = _clean_text((request_metadata or {}).get("report_source_type"))