Ng5ddlmZddlZddlmZmZmZmZddlZddl m Z ddl m Z m Z ddlmZddlmZddlmZdd lmZdd lmZmZd Zd Zd ZdZdZ d7d8d)Zd9d-Z d:d;d6Z dS)<) annotationsN)IOAnyOptionalSequence)UnstructuredClient) operationsshared)retries)Element)logger) exactly_one)elements_from_dictselements_from_jsoni i g?i@wT.https://api.unstructured.io/general/v0/generalfilename Optional[str] content_typefileOptional[IO[bytes]] file_filenameapi_urlstrapi_keymetadata_filenameretries_initial_interval[int]retries_max_interval Optional[int]retries_exponentOptional[float]retries_max_elapsed_timeretries_connection_errorsOptional[bool]request_kwargsrreturn list[Element]c  t|||r|rtd||}tjdd|vr |ddn|} t || }|Qt |d5}t j|| }dddn #1swxYwYn)|'|td t j|| }tj t j dd |i|  }t| | || || }|j ||}|jdkrt!|jjStd|jd)aPartitions a document using the Unstructured REST API. This is equivalent to running the document through partition. See https://api.unstructured.io/general/docs for the hosted API documentation or https://github.com/Unstructured-IO/unstructured-api for instructions on how to run the API locally as a container. Parameters ---------- filename A string defining the target filename path. content_type A string defining the file content in MIME type file A file-like object using "rb" mode --> open(filename, "rb"). metadata_filename When file is not None, the filename (string) to store in element metadata. E.g. "foo.txt" api_url The URL for the Unstructured API. Defaults to the hosted Unstructured API. api_key The API key to pass to the Unstructured API. retries_initial_interval Defines the time interval (in seconds) to wait before the first retry in case of a request failure. Defaults to 3000. If set should be > 0. retries_max_interval Defines the maximum time interval (in seconds) to wait between retries (the interval between retries is increased as using exponential increase algorithm - this setting limits it). Defaults to 720000. If set should be > 0. retries_exponent Defines the exponential factor to increase the interval between retries. Defaults to 1.5. If set should be > 0.0. retries_max_elapsed_time Defines the maximum time (in seconds) to wait for retries. If exceeded, the original exception is raised. Defaults to 1800000. If set should be > 0. retries_connection_errors Defines whether to retry on connection errors. Defaults to True. request_kwargs Additional parameters to pass to the data field of the request to the Unstructured API. For example the `strategy` parameter. )rrzOnly one of metadata_filename and file_filename is specified. metadata_filename is preferred. file_filename is marked for deprecation.NzuThe file_filename kwarg will be deprecated in a future version of unstructured. Please use metadata_filename instead.z/general/v0/generali) api_key_auth server_urlrb)content file_namezWIf file is specified in partition_via_api, metadata_filename must be specified as well.files)partition_parameters)r$r!rr#rsdk)requestr )textReceive unexpected status code  from the API.)r ValueErrorr warnropenr Filesreadr PartitionRequestPartitionParametersget_retries_configgeneral partition status_coder raw_responser4)rrrrrrrrrr!r#r$r&base_urlr1fr/reqretries_configresponses V/var/www/html/ai-engine/env/lib/python3.11/site-packages/unstructured/partition/api.pypartition_via_apirJs$n---- ]  W    )  4   !6 @ @wtt}}gH 'h G G GC (D ! ! QL"E                   $?  T5FGGG  %#7VVeV~VV   C(";)!9!91  N{$$%H s""!x'<'ABBBB Rh.B R R R   s0)B%%B),B)r1rOptional[retries.RetryConfig]c  d}|jj td|||||fDrd fd } r j jnt}t j|p|d t|p|d t|p|d t|p|d t } t j d| ||n|}|S)aConstructs a RetryConfig object from the provided parameters. If any of the parameters are None, the default values are taken from the SDK configuration or the default constants. If all parameters are None, returns None (and the SDK-managed defaults are used within the client) The solution is not perfect as the RetryConfig object does not include the defaults by itself so we might need to construct it basing on our defaults. Parameters ---------- retries_connection_errors Defines whether to retry on connection errors. If not set the DEFAULT_RETRIES_CONNECTION_ERRORS constant is used. retries_exponent Defines the exponential factor to increase the interval between retries. If set, should be > 0.0 (otherwise the DEFAULT_RETRIES_EXPONENT constant is used) retries_initial_interval Defines the time interval to wait before the first retry in case of a request failure. If set, should be > 0 (otherwise the DEFAULT_RETRIES_INITIAL_INTERVAL_SEC constant is used) retries_max_elapsed_time Defines the maximum time to wait for retries. If exceeded, the original exception is raised. If set, should be > 0 (otherwise the DEFAULT_RETRIES_MAX_ELAPSED_TIME_SEC constant is used) retries_max_interval Defines the maximum time interval to wait between retries. If set, should be > 0 (otherwise the DEFAULT_RETRIES_MAX_INTERVAL_SEC constant is used) sdk The UnstructuredClient object to take the default values from. Nc3K|]}|duV dSNr7).0settings rI z%get_retries_config..s;    t       setting_namer default_valuerr'c>rtj|x}r|S|SrN)getattrbackoff)rSrT setting_valuesdk_default_retries_configs rIget_backoff_defaultz/get_retries_config..get_backoff_defaults3) )$+,F,NP\$]$]]=)(( rRinitial_interval max_intervalexponentmax_elapsed_time)r[r\r]r^rW)strategyrWretry_connection_errors)rSrrTrr'r) sdk_configuration retry_configanyr`!DEFAULT_RETRIES_CONNECTION_ERRORSr BackoffStrategy$DEFAULT_RETRIES_INITIAL_INTERVAL_SEC DEFAULT_RETRIES_MAX_INTERVAL_SECDEFAULT_RETRIES_EXPONENT$DEFAULT_RETRIES_MAX_ELAPSED_TIME_SEC RetryConfig) r$r!rr#rr1rGrZ"default_retries_connneciton_errorsbackoff_strategyrYs @rIr?r?s_JN!$!6!C   %  $ %       1  ! ! ! ! ! !* 3*BN ' > >3 +#2(a&&'9;_``%Y&&~7WXX!]$7$7 D\$]$])a&&'9;_``   "!,$-8*)7     rR filenamesOptional[list[str]] content_typesr/Optional[Sequence[IO[bytes]]]file_filenamesmetadata_filenameslist[list[Element]]c d|d}|r|rtd||}tjd||r/t|t|krtdt j5fd|D}g} t |D]3\} } || } ||| nd} | d| | | ff4tj |||| }dddn #1swxYwYn||r/t|t|krtd |std t|t|krtd g} t |D]3\} }||| nd} || } | d| || ff4tj |||| }|j d kr]g}| }t|dts|g}|D]$}|t|%|Std|j d)aPartitions multiple documents using the Unstructured REST API by batching the documents into a single HTTP request. See https://api.unstructured.io/general/docs for the hosted API documentation or https://github.com/Unstructured-IO/unstructured-api for instructions on how to run the API locally as a container. Parameters ---------- filenames A list of strings defining the target filename paths. content_types A list of strings defining the file contents in MIME types. files A list of file-like object using "rb" mode --> open(filename, "rb"). metadata_filename When file is not None, the filename (string) to store in element metadata. E.g. "foo.txt" api_url The URL for the Unstructured API. Defaults to the hosted Unstructured API. api_key The API key to pass to the Unstructured API. request_kwargs Additional parameters to pass to the data field of the request to the Unstructured API. For example the `strategy` parameter. zapplication/json)ACCEPTzUNSTRUCTURED-API-KEYzOnly one of metadata_filenames and file_filenames is specified. metadata_filenames is preferred. file_filenames is marked for deprecation.NzwThe file_filenames kwarg will be deprecated in a future version of unstructured. Please use metadata_filenames instead.z6content_types and filenames must have the same length.cVg|]%}t|d&S)r,) enter_contextr:)rOrEstacks rI z.partition_multiple_via_api.."s/KKKAU((a77KKKrRr/)headersdatar/z2content_types and files must have the same length.z8metadata_filenames must be specified if files are passedz7metadata_filenames and files must have the same length.r3rr5r6)r8r r9len contextlib ExitStack enumerateappendrequestspostrBjson isinstancelistr)rmror/rqrrrrr&rz_filesirrrrH_file documents response_listdocumentrxs @rIpartition_multiple_via_apirs4H% 'G  n  Y   !+  5     WS//3y>>AAUVV V  ! # # uKKKKKKKEF$U++ I I4$Q<3@3L}Q//RV  w4(FGHHHH}# H                   SS//3u::==QRR R! XWXX X # $ $E 2 2VWW W!%(( F FHAu/rs""""""............22222299999999------333333&&&&&&<<<<<<MMMMMMMM(,$#) '.$$(!#"& $#'C'+&**.(,.204q q q q q hYYYYz&*)-+/*.C.2m m m m m m m rR