admin管理员组

文章数量:1289509

I'm implementing a self-query retriever using langchain with OpenSearch as the target vectore store, so far everything is good but we need to capture the generated query in DSL, for debugging and auditing purposes, after some testing I cannot find how to do it, I found how to return thye StructuredQuery, and how to use the StructuredQuery and OpenSearchTranslator to get a step closer to the final query, however it is not the final query sent to OpenSearch. Question is, how to get the query? This is my current code(that returns something close to it but not the final version):

opensearch_translator = OpenSearchTranslator()
def show_translated_query(query):
    chain_structured_query = retriever.llm_chain.invoke(query)
    print("langchain structured query:")
    print(chain_structured_query)
    os_structured_query = opensearch_translator.visit_structured_query(chain_structured_query)
    print("OS query(semantic, filter):")
    print(os_structured_query)

show_translated_query("a fire ocurring before 2023")
>>langchain structured query:
>>query='fire' filter=Comparison(comparator=<Comparator.LT: 'lt'>, attribute='year', value=2023) limit=None
>>OS query(semantic, filter):
>>('fire', {'filter': {'range': {'metadata.year': {'lt': 2023}}}})

本文标签： pythonhow to get the target generated query on a selfquery retriever(langchain)Stack Overflow

版权声明：本文标题：python - how to get the target generated query on a self-query retriever(langchain) - Stack Overflow 内容由网友自发贡献，该文观点仅代表作者本人， 转载请联系作者并注明出处：http://www.betaflare.com/web/1741405468a2376918.html， 本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌抄袭侵权/违法违规的内容，一经查实，本站将立刻删除。