o ®©ZhY$ã@sŽdZddlmZddlZddlmZddlmZmZm Z m Z m Z ddl Z ddl mZddlmZddlmZdd lmZGd d „d eeƒZdS) z7Loader that loads data from Sharepoint Document Libraryé)Ú annotationsN)ÚPath)ÚAnyÚDictÚIteratorÚListÚOptional)Ú BaseLoader)ÚDocument)ÚField)ÚO365BaseLoaderc@sºeZdZUdZedƒZded< dZded< dZded < dZ ded < d Z d ed < e   ¡ddZ ded< d Zd ed< ed$dd„ƒZd%dd„Zd&dd„Zd'dd „Zd(d"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_metadataÚreturnú List[str]cCsddgS)zcReturn required scopes. Returns: List[str]: A list of required scopes. Z sharepointÚbasic©)Úselfrrúf/var/www/html/lang_env/lib/python3.10/site-packages/langchain_community/document_loaders/sharepoint.pyÚ_scopes%szSharePointLoader._scopesúIterator[Document]c csŠz ddlm}m}Wn tytdƒ‚w| ¡ ¡ |j¡}t||ƒs/t d|j›dƒ‚|j r“|  |j ¡}t||ƒsFt d|j ›dƒ‚|  |¡D]G}t |j d¡ƒ}|jdur_| |¡}|jdurq| |¡}| d |ji¡|j |¡D]} |jdurƒ|| jd <|jdurŽ| j |¡| VqwqK|jr÷| |j¡}t||ƒsªt d|j ›dƒ‚|  |¡D]G}t |j d¡ƒ}|jdurÃ| |¡}|jdurÕ| |¡}| d |ji¡|j |¡D]} |jdurç|| jd <|jdurò| j |¡| VqÛq¯|jrH| ||j¡D]E}t |j d¡ƒ}|jdur| |¡}|jdur"| |¡}|j |¡D]} |jdur5|| jd <|jdurA| j |¡| Vq(q|j s½|js¿|jsÁ| ¡}t||ƒsbt d ƒ‚|  |¡D][}t |j d¡ƒ}|jdur|| |¡}|jdur‡| |¡}|j |¡D]-} | j |j¡|jdur¡|| jd <|jdur¶| j |¡| j d |ji¡| Vqqgd Sd Sd Sd 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.zThere isn't a folder with path ÚidTZsource_full_urlÚauthorized_identitieszUnable to fetch root folderN)Z O365.driverrÚ ImportErrorZ_authZstorageZ get_driverÚ isinstanceÚ ValueErrorrZget_item_by_pathZ_load_from_folderrÚmetadataÚgetrr"rÚget_extended_metadataÚupdateZweb_urlZ _blob_parserZ lazy_parserZget_itemrZ_load_from_object_idsZget_root_folder) rrrÚdriveZ target_folderZblobÚfile_idZauth_identitiesZextended_metadataZ parsed_blobZ blob_partrrrÚ lazy_load-s¨€ ÿÿ           û          û         û         ÿ ÷ özSharePointLoader.lazy_loadr+rc Csº| ¡}| d¡}d|j›d|›d}dd|›i}tjd||d}| ¡}g}| d ¡D]-} |  d ¡rZ|  d ¡ d ¡pK|  d ¡ d ¡pK|  d ¡ d ¡} | rZ|  d¡} | rZ| | ¡q-|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+Údatar-Úurlr4ÚresponseZ access_listZ group_namesZ access_dataZ site_datar8rrrr"s2 ÿÿÿ ÿý  €z&SharePointLoader.authorized_identitiesrcCsDt|jdd }| ¡}Wdƒn1swYt |¡}|S)z| Fetch the access token from the token file. Returns: The access token as a dictionary. zutf-8)ÚencodingN)ÚopenrÚreadr<Úloads)rÚfÚsr>rrrr9£s  ÿ z$SharePointLoader._fetch_access_tokenrc Cs¦| ¡}| d¡}d|j›d|›d}dd|›i}tjd||d}| ¡}| d d ¡| d i¡ d i¡ d d¡| di¡ dd¡ d¡dd| dd¡dœ}|S)a Retrieve extended metadata for a file in SharePoint. As of today, following fields are supported in the extended metadata: - size: size of the source file. - owner: display name of the owner of the source file. - full_path: pretty human readable path of the source file. Args: file_id (str): The ID of the file. Returns: dict: A dictionary containing the extended metadata of the file, including size, owner, and full path. r-r.r/z,?$select=size,createdBy,parentReference,namer0r1r2r3ÚsizerZ createdByr6Z displayNameÚZparentReferenceÚpathú:éÿÿÿÿú/Úname)rGÚownerÚ full_path)r9r'rr:r;r<Úsplit) rr+r>r-r?r4r@r&Zstaged_metadatarrrr(®s2 ÿÿÿ   þý üû z&SharePointLoader.get_extended_metadata)rr)rr)r+rrr)rr)r+rrr)Ú__name__Ú __module__Ú __qualname__Ú__doc__r rÚ__annotations__rrrrrÚhomerrÚpropertyrr,r"r9r(rrrrr s,         T " r )rTÚ __future__rr<ÚpathlibrÚtypingrrrrrr:Zlangchain_core.document_loadersr Zlangchain_core.documentsr Zpydanticr Z.langchain_community.document_loaders.base_o365r r rrrrÚs