hddlmZddlZddlmZddlmZmZmZm Z m Z m Z ddl m Z erddlZejeZe eeeeeefZdd Zdd ZGdde dZGddZdS)) annotationsN)Sequence) TYPE_CHECKINGAnyCallableLiteralOptionalUnion) TypedDictXMatrixYreturn np.ndarraycddl}t|dkst|dkr|jgS|j|}|j|}|jd|jdkr t d|jd|jd ddl}|j||j}|j||j}d|||d z }t|tr|j|gS|j|S#t$rt d |j |d }|j |d }|jd d 5|j||j|j||z }dddn #1swxYwYd||j||j|z<|cYSwxYw)z*>>? s>A2D DA.G2 )F>2 G2>G G2G )G21G24Callable[[Sequence[str]], Sequence[Sequence[float]]]c^ ddlmn#t$rtdwxYwd fd }|S) zGet the OpenAI GPT-3 encoder.r)ClientzTHe default encoder for the EmbeddingDistance class uses the OpenAI API. Please either install the openai library with `pip install openai` or provide a custom encoder function (Callable[[str], Sequence[float]]).texts Sequence[str]rSequence[Sequence[float]]c}|jt|d}d|jDS)Nztext-embedding-3-small)inputmodelcg|] }|j S) embedding).0ds r6 z<_get_openai_encoder..encode_text..Ns333 333) embeddingscreatelistdata)r;clientresponse OpenAIClients r6 encode_textz(_get_openai_encoder..encode_textIsO$++u++%=,  43X]3333rG)r;r<rr=)openair:r%)rOrNs @r6_get_openai_encoderrQ>sv 1111111     T    444444 s $c$eZdZUded<ded<dS)EmbeddingConfigz0Callable[[list[str]], Sequence[Sequence[float]]]encoderzCLiteral['cosine', 'euclidean', 'manhattan', 'chebyshev', 'hamming']rN)__name__ __module__ __qualname____annotations__rBrGr6rSrSSs*====OOOOOOrGrSF)totalceZdZ dddZdd ZddZeddZeddZeddZ eddZ eddZ dS)EmbeddingDistanceNconfigOptional[EmbeddingConfig]c|pi}|dpd|_|dp t|_dS)NrrrT)getdistancerQrT)selfr\s r6__init__zEmbeddingDistance.__init__YsH2 8,,8 zz),,E0C0E0E rG predictionstr referencerr$c ddl}n#t$rtdwxYw|||g}|j|}||d|dS)NrzWThe EmbeddingDistance class requires NumPy. Please install it with `pip install numpy`.r)rr%rTr_compute_distanceitem)rarcrer0rHvectors r6evaluatezEmbeddingDistance.evaluateas          '   \\:y"9:: *%%%%fQi;;@@BBBs!arb np.floatingcz|jdkr|||S|jdkr|||S|jdkr|||S|jdkr|||S|jdkr|||St d|j)Nr euclidean manhattan chebyshevhammingzInvalid distance metric: )r`_cosine_distance_euclidean_distance_manhattan_distance_chebyshev_distance_hamming_distancer)rarkrls r6rgz#EmbeddingDistance._compute_distanceqs =H $ $((A.. . ]k ) )++Aq11 1 ]k ) )++Aq11 1 ]k ) )++Aq11 1 ]i ' '))!Q// /HHHII IrGc,dt|g|gz S)zCompute the cosine distance between two vectors. Args: a (np.ndarray): The first vector. b (np.ndarray): The second vector. Returns: np.ndarray: The cosine distance. g?)r7rkrls r6rsz"EmbeddingDistance._cosine_distances&sQC0000rGcFtj||z S)zCompute the Euclidean distance between two vectors. Args: a (np.ndarray): The first vector. b (np.ndarray): The second vector. Returns: np.floating: The Euclidean distance. )r0r(r)rys r6rtz%EmbeddingDistance._euclidean_distancesy~~a!e$$$rGcTtjtj||z S)zCompute the Manhattan distance between two vectors. Args: a (np.ndarray): The first vector. b (np.ndarray): The second vector. Returns: np.floating: The Manhattan distance. )r0sumabsrys r6ruz%EmbeddingDistance._manhattan_distance vbfQUmm$$$rGcTtjtj||z S)zCompute the Chebyshev distance between two vectors. Args: a (np.ndarray): The first vector. b (np.ndarray): The second vector. Returns: np.floating: The Chebyshev distance. )r0maxr}rys r6rvz%EmbeddingDistance._chebyshev_distancer~rGc2tj||kS)zCompute the Hamming distance between two vectors. Args: a (np.ndarray): The first vector. b (np.ndarray): The second vector. Returns: np.floating: The Hamming distance. )r0meanrys r6rwz#EmbeddingDistance._hamming_distanceswqAvrG)N)r\r])rcrdrerdrr$)rkrrlrrrm)rkrrlrrr) rUrVrWrbrjrg staticmethodrsrtrurvrwrBrGr6r[r[Xs-1FFFFFCCCC J J J J 1 1 1\ 1 % % %\ % % % %\ % % % %\ %   \   rGr[)r r rr rr)rr8) __future__rloggingcollections.abcrtypingrrrrr r typing_extensionsr rr0 getLoggerrUr&rJr$r r7rQrSr[rBrGr6rs{""""""$$$$$$('''''  8 $ $ tDK $s)S0 1""""J*PPPPPiuPPPP ffffffffffrG