o +i@sddlmZddlZddlmZddlmZmZmZm Z m Z m Z ddl m Z er,ddlZeeZe eeeeeefZdd d ZdddZGddde ddZGdddZdS)) annotationsN)Sequence) TYPE_CHECKINGAnyCallableLiteralOptionalUnion) TypedDictXMatrixYreturn np.ndarrayc Csnddl}t|dkst|dkr|gS||}||}|jd|jdkr6td|jd|jdz0ddl}|j||jd}|j||jd}d|j||dd }t|t ra||gWS||WSt yt d |j j|dd }|j j|dd }|jd d d |||j|||}Wdn1swYd|||||B<|YSw)z       r64Callable[[Sequence[str]], Sequence[Sequence[float]]]cs:zddlmWn tytdwd fdd }|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]]cs*}|jjt|dd}dd|jDS)Nztext-embedding-3-small)inputmodelcSsg|]}|jqSr4) embedding).0dr4r4r5 Nsz<_get_openai_encoder..encode_text..) embeddingscreatelistdata)r9clientresponse OpenAIClientr4r5 encode_textIs z(_get_openai_encoder..encode_textN)r9r:rr;)openair8r#)rJr4rHr5_get_openai_encoder>s rLc@seZdZUded<ded<dS)EmbeddingConfigz0Callable[[list[str]], Sequence[Sequence[float]]]encoderzCLiteral['cosine', 'euclidean', 'manhattan', 'chebyshev', 'hamming']rN)__name__ __module__ __qualname____annotations__r4r4r4r5rMSs  rMF)totalc@steZdZ ddddZdd d Zd ddZed!ddZed ddZed ddZ ed ddZ ed ddZ dS)"EmbeddingDistanceNconfigOptional[EmbeddingConfig]cCs.|pi}|dp d|_|dpt|_dS)NrrrN)getdistancerLrN)selfrUr4r4r5__init__YszEmbeddingDistance.__init__ predictionstr referencerr"cCsTzddl}Wn tytdw|||g}||}||d|dS)NrzWThe EmbeddingDistance class requires NumPy. Please install it with `pip install numpy`.r)rr#rNr_compute_distanceitem)rYr[r]r.rBvectorr4r4r5evaluateas   zEmbeddingDistance.evaluatearb np.floatingcCs~|jdkr |||S|jdkr|||S|jdkr!|||S|jdkr,|||S|jdkr7|||Std|j)Nr euclidean manhattan chebyshevhammingzInvalid distance metric: )rX_cosine_distance_euclidean_distance_manhattan_distance_chebyshev_distance_hamming_distancer)rYrbrcr4r4r5r^qs          z#EmbeddingDistance._compute_distancecCsdt|g|gS)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?)r6rbrcr4r4r5ris z"EmbeddingDistance._cosine_distancecCstj||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. )r.r&r'rnr4r4r5rjs z%EmbeddingDistance._euclidean_distancecCtt||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. )r.sumabsrnr4r4r5rk z%EmbeddingDistance._manhattan_distancecCro)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. )r.maxrqrnr4r4r5rlrrz%EmbeddingDistance._chebyshev_distancecCst||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. )r.meanrnr4r4r5rms z#EmbeddingDistance._hamming_distance)N)rUrV)r[r\r]r\rr")rbrrcrrrd)rbrrcrrr) rOrPrQrZrar^ staticmethodrirjrkrlrmr4r4r4r5rTXs       rT)r r r r rr)rr7) __future__rloggingcollections.abcrtypingrrrrrr typing_extensionsr rr. getLoggerrOr$rDr"r r6rLrMrTr4r4r4r5s       %