Class ArxivRetriever

templates/retrieval-agent/retrieval_agent/chain.py:19–54

`Arxiv` retriever. It wraps load() to get_relevant_documents(). It uses all ArxivAPIWrapper arguments without any change.

class ArxivRetriever(BaseRetriever, ArxivAPIWrapper):
    """`Arxiv` retriever.

    It wraps load() to get_relevant_documents().
    It uses all ArxivAPIWrapper arguments without any change.
    """

    get_full_documents: bool = False

    def _get_relevant_documents(
        self, query: str, *, run_manager: CallbackManagerForRetrieverRun
    ) -> List[Document]:
        try:
            if self.is_arxiv_identifier(query):
                results = self.arxiv_search(
                    id_list=query.split(),
                    max_results=self.top_k_results,
                ).results()
            else:
                results = self.arxiv_search(  # type: ignore
                    query[: self.ARXIV_MAX_QUERY_LENGTH], max_results=self.top_k_results
                ).results()
        except self.arxiv_exceptions as ex:
            return [Document(page_content=f"Arxiv exception: {ex}")]
        docs = [
            Document(
                page_content=result.summary,
                metadata={
                    "Published": result.updated.date(),
                    "Title": result.title,
                    "Authors": ", ".join(a.name for a in result.authors),
                },
            )
            for result in results
        ]
        return docs


description = (