o piP@sddlZddlZddlZddlZddlZddlZddlZddlm Z ddl m Z m Z m Z mZddlZddlmZddlmZmZejdZGdddeZGd d d eZGd d d ejZGd ddZdS)N)Path)CallableMappingOptionalUnion)Response)ConnectionError HTTPErrorzpyannoteai-sdkc@eZdZdZdS)PyannoteAIFailedJobz2Raised when a job failed on the pyannoteAI web APIN__name__ __module__ __qualname____doc__rrI/home/ubuntu/.local/lib/python3.10/site-packages/pyannoteai/sdk/client.pyr (r c@r )PyannoteAICanceledJobz8Raised when a job was canceled on the pyannoteAI web APINr rrrrr.rrcsBeZdZdZdededeffdd Zd deffd d ZZ S) _UploadingCallbackBytesIOaKBytesIO subclass that calls a callback during the upload process Parameters ---------- callback : Callable Callback called during upload as `callback(total_in_bytes, completed_in_bytes)` total_size : int Total size to upload (in bytes) initial_bytes : bytes Initial bytes to upload callback total_size initial_bytescs"||_d|_||_t|dS)Nr)r_completed_size _callbacksuper__init__)selfrrr __class__rrrAsz"_UploadingCallbackBytesIO.__init__returncs:t|}|jt|7_|jr|j|j|jd|S)N)total completed)rreadrlenrr)rsizedatarrrr$Ls z_UploadingCallbackBytesIO.read)r ) r rrrrintbytesrr$ __classcell__rrrrr4s  rc@seZdZdZdZd:deefddZdefdd Z d ed efd d Z d:d edee d efddZ e ddZejded dfddZded efddZdeeefd efddZ  d;deeBe eeeBfBdeedeed efddZ        dded0e eefd1ed2ed!edBd"edBd#edBd$ed%ed&ed'ed efd3d4Zd?d6ed7ed e fd8d9ZdS)@ClientaOfficial client for pyannoteAI web API Parameters ---------- token : str, optional pyannoteAI API key created from https://dashboard.pyannote.ai. Defaults to using `PYANNOTEAI_API_KEY` environment variable. Usage ----- # instantiate client for pyannoteAI web API >>> from pyannoteai.sdk import Client >>> client = Client(token="{PYANNOTEAI_API_KEY}") # upload your audio file to the pyannoteAI web API # and store it for a few hours for later re-use. >>> media_url = client.upload("/path/to/your/audio.wav") # initiate a diarization job on the pyannoteAI web API >>> job_id = client.diarize(media_url) # retrieve prediction from pyannoteAI web API >>> prediction = client.retrieve(job_id) zhttps://api.pyannote.ai/v1NtokencKs||_|p tjdd|_dS)NPYANNOTEAI_API_KEY)r,osenvirongetapi_key)rr,kwargsrrrrtszClient.__init__responsecCs|jdkr td|dS)z9Raise an exception if the response status code is not 2xxiz Failed to authenticate to pyannoteAI API. Please create an API key on https://dashboard.pyannote.ai/ and provide it either via `PYANNOTEAI_API_KEY` environment variable or with `token` parameter.N) status_coder raise_for_status)rr4rrr_raise_for_statusxs  zClient._raise_for_statusrouter!cCsDztj|j||jd}Wn tytdw|||S)a,Send GET authenticated request to pyannoteAI API Parameters ---------- route : str API route to send the GET request to. Returns ------- response : Response Raises ------ ConnectionError HTTPError )headers Failed to connect to pyannoteAI web API. Please check your internet connection or visit https://pyannote.openstatus.dev/ to check the status of the pyannoteAI web API.)requestsr1API_URL_headersrr7)rr8r4rrr_authenticated_gets  zClient._authenticated_getjsoncCsFztj|j|||jd}Wn tytdw|||S)aSend POST authenticated request to pyannoteAI web API Parameters ---------- route : str API route to send the GET request to. json : dict, optional Request body to send with the POST request. Returns ------- response : Response Raises ------ ConnectionError HTTPError )r?r9r:)r;postr<r=rr7)rr8r?r4rrr_authenticated_posts  zClient._authenticated_postcCs|jSN)_api_key)rrrrr2szClient.api_keyr2cCs<|std||_d|jdtdd|_|ddS)Nz Failed to authenticate to pyannoteAI web API. Please create an API key on https://dashboard.pyannote.ai/ and provide it either via `PYANNOTEAI_API_KEY` environment variable or with `token` parameter.zBearer zpyannoteAI-sdk-python/zapplication/json) Authorizationz User-Agentz Content-Typez/test) ValueErrorrC __version__r=r>)rr2rrrr2s  media_urlcCs|jdd|id}|dS)aZCreate a presigned URL to upload audio file to pyannoteAI platform Parameters ---------- media_url : str Unique identifier used to retrieve the uploaded audio file on the pyannoteAI platform. Any combination of letters (a-z, A-Z), digits (0-9), and the characters -./ prefixed with media:// is allowed. One would usually use a string akin to a path on filesystem (e.g. "media://path/to/audio.wav"). Returns ------- url : str Presigned URL to upload audio file to pyannoteAI platform z /media/inputurlr?)rAr?)rrGr4rrr_create_presigned_urls zClient._create_presigned_urlfilecsbt}t|dtfdddD]}||qWd|S1s(wY|S)zACompute MD5 hash of a file (used for media_url when not provided)rbcs dS)Ni)r$rfrrs z"Client._hash_md5..N)hashlibmd5openiterupdate hexdigest)rrKhash_md5chunkrrMr _hash_md5s   zClient._hash_md5audiorc Csd}t|trnd|vr|d}n^d|vrjd}zddl}Wn ty'tdw|d}|d}t|dd r?|djdd }tjd dd }|jj |j || | |j }Wdn1sdwYntd tj|} |durd||}t|d}t|| |} Wdn1swYt|tr|dstd|d||} z(z tj| | d} Wntytd| dwW|rtj|rt|n|rtj|rt|www| t d|S)aUpload audio file to pyannoteAI platform Parameters ---------- audio : str or Path or dict Path to audio file to be uploaded. Can be a "str" or "Path" instance, or a dict with an "audio" key (e.g. {"audio": "/path/to/audio.wav"}). media_url : str, optional Unique identifier used to retrieve the uploaded audio file on the pyannoteAI platform. Any combination of letters {a-z, A-Z}, digits {0-9}, and {-./} characters prefixed with 'media://' is allowed. One would usually use a string akin to a path on filesystem (e.g. "media://path/to/audio.wav"). Defaults to media://{md5-hash-of-audio-file}. callback : Callable, optional When provided, `callback` is called during the uploading process with the following signature: callback(total=..., # number of bytes to upload completed=...) # number of bytes uploaded) Returns ------- media_url : str Same as the input `media_url` parameter when provided, or "media://{md5-hash-of-audio-file}" otherwise. FrZwaveformTrNz>To process the waveform directly, you need to install `scipy`. sample_ratenumpy)forcez.wav)suffixdeletezRWhen `audio` is a dict, it must provide the path to the audio file in 'audio' key.zmedia://rLz Invalid media URI: zp. Any combination of letters {a-z, A-Z}, digits {0-9}, and {-./} characters prefixed with 'media://' is allowed.)r'z) Failed to upload audio to presigned URL z|. Please check your internet connection or visit https://pyannote.openstatus.dev/ to check the status of the pyannoteAI API.aR You are using pyannoteAI's temporary storage solution. Your file will be permanently deleted from our servers within 24hs. If you are running in production, we highly recommend to use your own storage to reduce network latency and obtain results faster. Please check our documentation at https://docs.pyannote.ai/ for more information.)! isinstancerscipy.io ImportErrorhasattrr]tempfileNamedTemporaryFileiowavfilewritenamesqueezeflushrEr/pathgetsizerYrSrr$str startswithrJr;putrexistsremover6warningswarn) rrZrGrr`scipyr\r[rNrr' presigned_urlr4rrruploadst           z Client.uploadF precision-2 num_speakers min_speakers max_speakers confidenceturn_level_confidence exclusivemodel transcriptiontranscription_configc  KsP||||||||| d } | dur| | d<| | |jd| d} | }|dS)a,Initiate a diarization job on the pyannoteAI web API Parameters ---------- media_url : str media://{...} URL created with the `upload` method or any other public URL pointing to an audio file. num_speakers : int, optional Force number of speakers to diarize. If not provided, the number of speakers will be determined automatically. min_speakers : int, optional Minimum number of speakers. max_speakers : int, optional Maximum number of speakers. confidence : bool, optional Enable confidence scores. turn_level_confidence: bool, optional Enable turn-based confidence scores. exclusive: bool, optional Enable exclusive speaker diarization. model : str, optional Defaults to "precision-2" transcription : bool, optional Enable STT orchestration. transcription_config : dict, optional STT configuration parameters, including model selection. **kwargs : optional Extra arguments to send in the body of the request. Returns ------- job_id: str Raises ------ HTTPError If something else went wrong ) rHr numSpeakers minSpeakers maxSpeakersr}turnLevelConfidencerrNtranscriptionConfigz/diarizerIjobIdrUrAr?)rrGrzr{r|r}r~rrrrr3r?r4r'rrrdiarizeks 6  zClient.diarizecKs2||d}|||jd|d}|}|dS)a)Initiate a voiceprint job on the pyannoteAI web API Parameters ---------- media_url : str media://{...} URL created with the `upload` method or any other public URL pointing to an audio file. model : str, optional Defaults to "precision-2". **kwargs : optional Extra arguments to send in the body of the request. Returns ------- job_id: str Raises ------ HTTPError If something else went wrong )rHrz /voiceprintrIrr)rrGrr3r?r4r'rrr voiceprints  zClient.voiceprintT voiceprintsexclusive_matchingmatching_thresholdc  KsV|| ||||| | dd|D||dd } | | |jd| d}|}|dS)aInitiate an identification job on the pyannoteAI web API Parameters ---------- media_url : str media://{...} URL created with the `upload` method or any other public URL pointing to an audio file. voiceprints : dict Voiceprints. exclusive_matching : bool, optional Prevent multiple speakers from being matched to the same voiceprint. Defaults to True. matching_threshold : float, optional Prevent matching if confidence score is below this threshold. Value is between 0 and 100. Default is 0, meaning all voiceprints are matched. num_speakers : int, optional Force number of speakers to diarize. If not provided, the number of speakers will be determined automatically. min_speakers : int, optional Minimum number of speakers. max_speakers : int, optional Maximum number of speakers. confidence : bool, optional Enable confidence scores. turn_level_confidence: bool, optional Enable turn-based confidence scores. exclusive: bool, optional Enable exclusive speaker diarization. model : str, optional Defaults to "precision-2" **kwargs : optional Extra arguments to send in the body of the request. Returns ------- job_id: str Raises ------ HTTPError If something else went wrong cSsg|] \}}||dqS))labelrr).0speakerrrrr sz#Client.identify..)r threshold) rHrrrrr}rrrmatchingz /identifyrIr)itemsrUrAr?)rrGrrrrzr{r|r}r~rrr3r?r4r'rrridentifys&; zClient.identify job_id every_secondscCsd} |d|}|d}|dvrt|q |dkr0|dtdd }t|||d krD|dtdd }t||t d |S) aRetrieve output of a job (once completed) from pyannoteAI web API Parameters ---------- job_id : str Job ID. Returns ------- job_output : dict Job output Raises ------ PyannoteAIFailedJob If the job failed PyannoteAICanceledJob If the job was canceled HTTPError If something else went wrong NTz/jobs/status) succeededcanceledfailedroutputerrorzPlease contact support.ra You are using periodic polling to retrieve results. If you are running in production, we highly recommend to setup a webhook server to obtain results faster, as soon as they are available. Please check our documentation at https://docs.pyannote.ai/ for more information.) r>r?timesleepr1dictr rrtru)rrr job_statusjobrrrrretrieve-s     zClient.retrieverB)NN) NNNFFFryFN)ry) TrNNNFFFry)r)r rrrr<rrorrr7r>rrApropertyr2setterrJrrrYrrxr(boolrrfloatrrrrrrr+Ws#   q    N (      Sr+)rQimportlib.metadata importlibrgr/rerrtpathlibrtypingrrrrr;rrequests.exceptionsrr metadataversionrF Exceptionr rBytesIOrr+rrrrs"   #