U ™ÀÂhÑ%ã@sždZddlmZddlZddlmZddlmZmZm Z m Z m Z ddl Z ddl mZddlmZddlmZdd lmZmZdd lmZGd d „d eeƒZdS) z7Loader that loads data from Sharepoint Document Libraryé)Ú annotationsN)ÚPath)ÚAnyÚIteratorÚListÚOptionalÚSequence)Ú BaseLoader)ÚDocument)ÚField)ÚO365BaseLoaderÚ _FileType)Ú get_parserc@sÖeZdZUdZedƒZded<dZded<dZded <dZ ded <d Z d ed <e   ¡ddZ ded<d Zd ed<eddœdd„ƒZeddœdd„ƒZddœdd„Zdddœdd „Zd!dœd"d#„Zdd$dœd%d&„ZdS)'ÚSharePointLoaderzLoad from `SharePoint`..ÚstrÚdocument_library_idNz Optional[str]Ú folder_pathzOptional[List[str]]Ú object_idsÚ folder_idFzOptional[bool]Ú load_authz .credentialszo365_token.txtrÚ token_pathÚload_extended_metadatazSequence[_FileType])ÚreturncCstjtjtjfS)zfReturn supported file types. Returns: A sequence of supported file types. )r ZDOCZDOCXZPDF©Úself©rúS/tmp/pip-unpacked-wheel-9gdii04g/langchain_community/document_loaders/sharepoint.pyÚ _file_types'szSharePointLoader._file_typesz List[str]cCsddgS)zcReturn required scopes. Returns: List[str]: A list of required scopes. Z sharepointÚbasicrrrrrÚ_scopes/szSharePointLoader._scopeszIterator[Document]c cs˜zddlm}m}Wntk r0tdƒ‚YnX| ¡ ¡ |j¡}t||ƒsbt d|j›dƒ‚t dƒ}|j r6|  |j ¡}t||ƒsšt d|j ›dƒ‚|  |¡D]}t|j d¡ƒ}|jd krÌ| |¡}|jd krð| |¡} |  d |ji¡| |¡D]8} |jd kr|| jd <|jd kr,| j | ¡| Vqúq¤|jr | |j¡}t||ƒsht d|j ›dƒ‚|  |¡D]˜}t|j d¡ƒ}|jd krœ| |¡}|jd krÂ| |¡} |  d |ji¡| |¡D]:} |jd kræ|| jd <|jd krþ| j | ¡| VqÌqr|jr¬| ||j¡D]ˆ}t|j d¡ƒ}|jd krL| |¡}|jd krb| |¡} | |¡D]:} |jd kr†|| jd <|jd krž| j | ¡| Vqlq"|j s”|js”|js”| ¡}t||ƒsàt d ƒ‚|  |¡D]¨}t|j d¡ƒ}|jd kr| |¡}|jd kr*| |¡} | |¡D]Z} | j |j¡|jd kr\|| jd <|jd kr†| j | ¡| j d |ji¡| Vq4qêd S)z¦ Load documents lazily. Use this when working at a large scale. Yields: Document: A document object representing the parsed blob. r)ÚDriveÚFolderzAO365 package not found, please install it with `pip install o365`zThere isn't a Drive with id Ú.ÚdefaultzThere isn't a folder with path ÚidTZsource_full_urlÚauthorized_identitieszUnable to fetch root folderN)Z O365.driver r!Ú ImportErrorZ_authZstorageZ get_driverÚ isinstanceÚ ValueErrorrrZget_item_by_pathZ_load_from_folderrÚmetadataÚgetrr%rÚget_extended_metadataÚupdateZweb_urlZ lazy_parserZget_itemrZ_load_from_object_idsZget_root_folder) rr r!ÚdriveZ blob_parserZ target_folderZblobÚfile_idZauth_identitiesZextended_metadataZ parsed_blobZ blob_partrrrÚ lazy_load7sšÿ                                        ÿzSharePointLoader.lazy_loadr)r.rc Csº| ¡}| d¡}d|j›d|›d}dd|›i}tjd||d}| ¡}g}| d ¡D]Z} |  d ¡rZ|  d ¡ d ¡p–|  d ¡ d ¡p–|  d ¡ d ¡} | rZ|  d¡} | rZ| | ¡qZ|S)a Retrieve the access identities (user/group emails) for a given file. Args: file_id (str): The ID of the file. Returns: List: A list of group names (email addresses) that have access to the file. Ú access_tokenú(https://graph.microsoft.com/v1.0/drives/ú/items/z /permissionsÚ AuthorizationúBearer ÚGET©ÚheadersÚvalueZ grantedToV2ZsiteUserÚuserÚgroupÚemail)Ú_fetch_access_tokenr*rÚrequestsÚrequestÚjsonÚappend) rr.Údatar0Úurlr7ÚresponseZ access_listZ group_namesZ access_dataZ site_datar;rrrr%Œs(  ÿ ÿý  z&SharePointLoader.authorized_identitiesrc Cs0t|jdd}| ¡}W5QRXt |¡}|S)z| Fetch the access token from the token file. Returns: The access token as a dictionary. zutf-8)Úencoding)ÚopenrÚreadr?Úloads)rÚfÚsrArrrr<®s z$SharePointLoader._fetch_access_tokenÚdictc Cs¦| ¡}| d¡}d|j›d|›d}dd|›i}tjd||d}| ¡}| d d ¡| d i¡ d i¡ d d¡| di¡ dd¡ d¡dd| dd¡dœ}|S)a Retrieve extended metadata for a file in SharePoint. As of today, following fields are supported in the extended metadata: - size: size of the source file. - owner: display name of the owner of the source file. - full_path: pretty human readable path of the source file. Args: file_id (str): The ID of the file. Returns: dict: A dictionary containing the extended metadata of the file, including size, owner, and full path. r0r1r2z,?$select=size,createdBy,parentReference,namer3r4r5r6ÚsizerZ createdByr9Z displayNameÚZparentReferenceÚpathú:éÿÿÿÿú/Úname)rKÚownerÚ full_path)r<r*rr=r>r?Úsplit) rr.rAr0rBr7rCr)Zstaged_metadatarrrr+¹s:  ÿ  ÿþ ÿþþý üû z&SharePointLoader.get_extended_metadata)Ú__name__Ú __module__Ú __qualname__Ú__doc__r rÚ__annotations__rrrrrÚhomerrÚpropertyrrr/r%r<r+rrrrrs      U" r)rXÚ __future__rr?ÚpathlibrÚtypingrrrrrr=Zlangchain_core.document_loadersr Zlangchain_core.documentsr Zlangchain_core.pydantic_v1r Z.langchain_community.document_loaders.base_o365r r Z5langchain_community.document_loaders.parsers.registryrrrrrrÚs