a
    •˜ëjî.  ã                   @  sV  d Z ddlmZ ddlZddlmZ ddlmZmZ ddl	m
Z
 ddlmZmZmZ ddlmZ dd	lmZ dd
lmZ ddlmZ ddlmZ ddlmZmZmZ ddlmZ ddlm Z  ddl!m"Z" ddl#m$Z$ ddl%m&Z& ddl'm(Z( ddl)m*Z* e
dddd�G dd„ de ƒƒZ+e
dddd�G dd„ de+ƒƒZ,e
dddd�G dd„ de+ƒƒZ-dS ) ú7Chain for question-answering against a vector database.é    )ÚannotationsN)Úabstractmethod)ÚAnyÚOptional)Ú
deprecated)ÚAsyncCallbackManagerForChainRunÚCallbackManagerForChainRunÚ	Callbacks)ÚDocument)ÚBaseLanguageModel)ÚPromptTemplate)ÚBaseRetriever)ÚVectorStore)Ú
ConfigDictÚFieldÚmodel_validator)Úoverride)ÚChain)ÚBaseCombineDocumentsChain)ÚStuffDocumentsChain)ÚLLMChain©Úload_qa_chain)ÚPROMPT_SELECTORz0.2.13z1.0z³This class is deprecated. Use the `create_retrieval_chain` constructor instead. See migration guide here: https://python.langchain.com/docs/versions/migrating_chains/retrieval_qa/)ZsinceZremovalÚmessagec                	   @  sþ   e Zd ZU dZded< dZded< dZded< d	Zd
ed< edddd�Z	e
ddœdd„ƒZe
ddœdd„ƒZed2dddddd dœdd„ƒZed3ddddd dœd d!„ƒZedd"d#d$œd%d&„ƒZd4d'd(d'd)œd*d+„Zedd,d#d$œd-d.„ƒZd5d'd/d'd)œd0d1„ZdS )6ÚBaseRetrievalQAz)Base class for question-answering chains.r   Úcombine_documents_chainÚqueryÚstrÚ	input_keyÚresultÚ
output_keyFÚboolÚreturn_source_documentsTZforbid)Zpopulate_by_nameZarbitrary_types_allowedÚextraz	list[str]©Úreturnc                 C  s   | j gS )z,Input keys.

        :meta private:
        )r    ©Úself© r*   úa/mnt/bigstor/www/usecases/.venv/lib/python3.9/site-packages/langchain/chains/retrieval_qa/base.pyÚ
input_keys8   s    zBaseRetrievalQA.input_keysc                 C  s   | j g}| jrg |¢d‘}|S )z-Output keys.

        :meta private:
        Úsource_documents)r"   r$   )r)   Z_output_keysr*   r*   r+   Úoutput_keys@   s    zBaseRetrievalQA.output_keysNr   zOptional[PromptTemplate]r
   zOptional[dict]r   )ÚllmÚpromptÚ	callbacksÚllm_chain_kwargsÚkwargsr'   c           
      K  sZ   |pt  |¡}tf |||dœ|p"i ¤Ž}tdgdd�}t|d||d�}	| f |	|dœ|¤ŽS )zInitialize from LLM.)r/   r0   r1   Zpage_contentzContext:
{page_content})Zinput_variablesÚtemplateÚcontext)Ú	llm_chainZdocument_variable_nameÚdocument_promptr1   )r   r1   )r   Z
get_promptr   r   r   )
Úclsr/   r0   r1   r2   r3   Z_promptr6   r7   r   r*   r*   r+   Úfrom_llmK   s0    
ýüþüþýzBaseRetrievalQA.from_llmÚstuff)r/   Ú
chain_typeÚchain_type_kwargsr3   r'   c                 K  s.   |pi }t |fd|i|¤Ž}| f d|i|¤ŽS )zLoad chain from chain type.r;   r   r   )r8   r/   r;   r<   r3   Z_chain_type_kwargsr   r*   r*   r+   Úfrom_chain_typem   s    	ÿþýzBaseRetrievalQA.from_chain_typer	   úlist[Document]©ÚquestionÚrun_managerr'   c                C  s   dS ©z,Get documents to do question answering over.Nr*   ©r)   r@   rA   r*   r*   r+   Ú	_get_docs~   s    zBaseRetrievalQA._get_docsúdict[str, Any]z$Optional[CallbackManagerForChainRun])ÚinputsrA   r'   c                 C  sz   |p
t  ¡ }|| j }dt | j¡jv }|r<| j||d�}n
|  |¡}| jj||| 	¡ d�}| j
rp| j|d|iS | j|iS )ái  Run get_relevant_text and llm on input query.

        If chain has 'return_source_documents' as 'True', returns
        the retrieved documents as well under the key 'source_documents'.

        Example:
        .. code-block:: python

        res = indexqa({'query': 'This is my query'})
        answer, docs = res['result'], res['source_documents']

        rA   ©rA   ©Zinput_documentsr@   r1   r-   )r	   Úget_noop_managerr    ÚinspectÚ	signaturerD   Ú
parametersr   ÚrunÚ	get_childr$   r"   ©r)   rF   rA   Z_run_managerr@   Zaccepts_run_managerÚdocsZanswerr*   r*   r+   Ú_call‡   s    
ÿ
ýzBaseRetrievalQA._callr   c                Ã  s   dS rB   r*   rC   r*   r*   r+   Ú
_aget_docs«   s    zBaseRetrievalQA._aget_docsz)Optional[AsyncCallbackManagerForChainRun]c                 Ã  sŒ   |p
t  ¡ }|| j }dt | j¡jv }|rB| j||d�I dH }n|  |¡I dH }| jj||| 	¡ d�I dH }| j
r‚| j|d|iS | j|iS )rG   rA   rH   NrI   r-   )r   rJ   r    rK   rL   rS   rM   r   ZarunrO   r$   r"   rP   r*   r*   r+   Ú_acall´   s    
ÿýzBaseRetrievalQA._acall)NNN)r:   N)N)N)Ú__name__Ú
__module__Ú__qualname__Ú__doc__Ú__annotations__r    r"   r$   r   Zmodel_configÚpropertyr,   r.   Úclassmethodr9   r=   r   rD   rR   rS   rT   r*   r*   r*   r+   r      s>   

ý
   û!  ü ý$ ýr   z0.1.17c                   @  sZ   e Zd ZU dZedd�Zded< dddd	œd
d„Zdddd	œdd„Ze	ddœdd„ƒZ
dS )ÚRetrievalQAa„  Chain for question-answering against an index.

    This class is deprecated. See below for an example implementation using
    `create_retrieval_chain`:

        .. code-block:: python

            from langchain.chains import create_retrieval_chain
            from langchain.chains.combine_documents import create_stuff_documents_chain
            from langchain_core.prompts import ChatPromptTemplate
            from langchain_openai import ChatOpenAI


            retriever = ...  # Your retriever
            llm = ChatOpenAI()

            system_prompt = (
                "Use the given context to answer the question. "
                "If you don't know the answer, say you don't know. "
                "Use three sentence maximum and keep the answer concise. "
                "Context: {context}"
            )
            prompt = ChatPromptTemplate.from_messages(
                [
                    ("system", system_prompt),
                    ("human", "{input}"),
                ]
            )
            question_answer_chain = create_stuff_documents_chain(llm, prompt)
            chain = create_retrieval_chain(retriever, question_answer_chain)

            chain.invoke({"input": query})

    Example:
        .. code-block:: python

            from langchain_community.llms import OpenAI
            from langchain.chains import RetrievalQA
            from langchain_community.vectorstores import FAISS
            from langchain_core.vectorstores import VectorStoreRetriever

            retriever = VectorStoreRetriever(vectorstore=FAISS(...))
            retrievalQA = RetrievalQA.from_llm(llm=OpenAI(), retriever=retriever)

    T)Úexcluder   Ú	retrieverr   r	   r>   r?   c                C  s   | j j|d| ¡ id�S )ú	Get docs.r1   ©Úconfig)r^   ZinvokerO   rC   r*   r*   r+   rD     s    
þzRetrievalQA._get_docsr   c                Ã  s   | j j|d| ¡ id�I dH S )r_   r1   r`   N)r^   ZainvokerO   rC   r*   r*   r+   rS     s    
þzRetrievalQA._aget_docsr&   c                 C  s   dS )úReturn the chain type.Zretrieval_qar*   r(   r*   r*   r+   Ú_chain_type+  s    zRetrievalQA._chain_typeN)rU   rV   rW   rX   r   r^   rY   rD   rS   rZ   rc   r*   r*   r*   r+   r\   Ù   s   

.r\   c                   @  s¨   e Zd ZU dZeddd�Zded< dZded< d	Zd
ed< ee	d�Z
ded< edd�edddœdd„ƒƒZed
dddœdd„ƒZd
dddœdd„Zed
dœdd „ƒZd!S )"Ú
VectorDBQAr   TÚvectorstore)r]   Úaliasr   é   ÚintÚkÚ
similarityr   Úsearch_type)Údefault_factoryrE   Úsearch_kwargsÚbefore)ÚmodeÚdictr   )Úvaluesr'   c                 C  s0   d|v r,|d }|dvr,d|› d�}t |ƒ‚|S )zValidate search type.rk   )rj   Úmmrúsearch_type of ú not allowed.)Ú
ValueError)r8   rq   rk   Úmsgr*   r*   r+   Úvalidate_search_typeF  s    zVectorDBQA.validate_search_typer	   r>   r?   c                C  sj   | j dkr(| jj|fd| ji| j¤Ž}n>| j dkrP| jj|fd| ji| j¤Ž}nd| j › d�}t|ƒ‚|S )r_   rj   ri   rr   rs   rt   )rk   re   Zsimilarity_searchri   rm   Zmax_marginal_relevance_searchru   )r)   r@   rA   rQ   rv   r*   r*   r+   rD   Q  s&    
ÿþý
ÿþýzVectorDBQA._get_docsr   c                Ã  s   d}t |ƒ‚dS )r_   z!VectorDBQA does not support asyncN)ÚNotImplementedError)r)   r@   rA   rv   r*   r*   r+   rS   j  s    zVectorDBQA._aget_docsr&   c                 C  s   dS )rb   Zvector_db_qar*   r(   r*   r*   r+   rc   t  s    zVectorDBQA._chain_typeN)rU   rV   rW   rX   r   re   rY   ri   rk   rp   rm   r   r[   rw   r   rD   rS   rZ   rc   r*   r*   r*   r+   rd   1  s   

	
rd   ).rX   Ú
__future__r   rK   Úabcr   Útypingr   r   Zlangchain_core._apir   Zlangchain_core.callbacksr   r	   r
   Zlangchain_core.documentsr   Zlangchain_core.language_modelsr   Zlangchain_core.promptsr   Zlangchain_core.retrieversr   Zlangchain_core.vectorstoresr   Zpydanticr   r   r   Ztyping_extensionsr   Zlangchain.chains.baser   Z'langchain.chains.combine_documents.baser   Z(langchain.chains.combine_documents.stuffr   Zlangchain.chains.llmr   Z#langchain.chains.question_answeringr   Z0langchain.chains.question_answering.stuff_promptr   r   r\   rd   r*   r*   r*   r+   Ú<module>   sL   ü	 2ü	Oü	