123 lines
4.9 KiB
Python
123 lines
4.9 KiB
Python
from typing import Dict, List, Optional
|
|
from common.config.app_settings import app_settings
|
|
from datetime import datetime, timedelta, timezone
|
|
from backend.sharepoint.sharepoint_graph_client import SharePointGraphClient
|
|
from common.constants.region import UserRegion
|
|
from backend.document.document_manager import DocumentManager
|
|
from backend.content.constants import (
|
|
ContentSource,
|
|
ContentMediaType,
|
|
ContentDataFormat,
|
|
ContentFileConstants,
|
|
)
|
|
from backend.content.models import ContentDirectory, ContentFolderDoc
|
|
|
|
|
|
class ContentSharePointManager:
|
|
def __init__(self) -> None:
|
|
self.sharepoint_client = SharePointGraphClient(
|
|
tenant_id="cf151ee8-5c2c-4fe7-a1c4-809ba43c9f24",
|
|
client_id="a91e3c1d-d55d-423d-83cb-a736df229817",
|
|
client_secret=app_settings.FREELEAPS_WWW_AS_AZURE_CLIENT_SECRET,
|
|
host_name="netorgft10898514.sharepoint.com",
|
|
site_name="MediaRoom",
|
|
)
|
|
self.sharepoint_client.set_document_scope(
|
|
document_library="freeleaps", root_folder="production"
|
|
)
|
|
self.share_point_file_expiry = timedelta(hours=24)
|
|
|
|
def __generate_created_by(self, folder_name: str) -> str:
|
|
return "content-service-" + folder_name.replace("/", "-").lower()
|
|
|
|
async def __get_or_initialize_folder(self, folder_name: str, region: UserRegion):
|
|
current_time = datetime.now(timezone.utc)
|
|
folder = await ContentFolderDoc.find_one(
|
|
ContentFolderDoc.folder_name == folder_name,
|
|
ContentFolderDoc.region == region.value,
|
|
)
|
|
|
|
if folder is None:
|
|
folder = ContentFolderDoc(
|
|
folder_name=folder_name,
|
|
content_directories=[],
|
|
udpate_time=current_time,
|
|
update_source=ContentSource.SHAREPOINT,
|
|
valid_thru=current_time + self.share_point_file_expiry,
|
|
region=region,
|
|
)
|
|
else:
|
|
folder.content_directories.clear()
|
|
|
|
return folder
|
|
|
|
async def __process_file(
|
|
self, sp_file: Dict, folder_name: str, content_directory: ContentDirectory
|
|
):
|
|
file_name = sp_file["name"]
|
|
created_by = self.__generate_created_by(folder_name)
|
|
document_manager = DocumentManager()
|
|
|
|
content_type_mapping = {
|
|
ContentFileConstants.COVER_FILE_NAME: "cover_document_id",
|
|
ContentFileConstants.SUMMARY_FILE_NAME: "summary_text_new",
|
|
ContentFileConstants.TITLE_FILE_NAME: "title_text_new",
|
|
ContentFileConstants.CONTENT_LINK_FILE_NAME: "content_link",
|
|
ContentFileConstants.CONTENT_HTML_FILE_NAME: "content_html_new",
|
|
ContentFileConstants.CONTENT_TEXT_FILE_NAME: "content_text_new",
|
|
ContentFileConstants.CONTENT_PDF_FILE_NAME: "content_document_id",
|
|
}
|
|
|
|
content_property = content_type_mapping.get(file_name.lower())
|
|
if content_property:
|
|
content_file_content = self.sharepoint_client.get_file_content(
|
|
sp_file["id"]
|
|
)
|
|
saved_document_id = await document_manager.save_document_file(
|
|
created_by, file_name, content_file_content
|
|
)
|
|
setattr(content_directory, content_property, saved_document_id)
|
|
|
|
async def retrieve_directories_for_folder(
|
|
self, folder_name: str, region: UserRegion
|
|
):
|
|
content_folder_name = folder_name + "/" + region.name
|
|
sp_folders = self.sharepoint_client.list_sub_folders(content_folder_name)
|
|
current_time = datetime.now(timezone.utc)
|
|
folder = await self.__get_or_initialize_folder(folder_name, region)
|
|
|
|
for sp_folder in sp_folders:
|
|
content_directory = ContentDirectory(
|
|
content_name=sp_folder["name"],
|
|
cover_document_id=None,
|
|
summary_text=None,
|
|
title_text=None,
|
|
content_link=None,
|
|
content_document_id=None,
|
|
)
|
|
|
|
sp_files = self.sharepoint_client.list_files(
|
|
content_folder_name + "/" + sp_folder["name"]
|
|
)
|
|
for sp_file in sp_files:
|
|
await self.__process_file(sp_file, folder_name, content_directory)
|
|
|
|
folder.content_directories.append(content_directory)
|
|
|
|
folder.udpate_time = current_time
|
|
folder.update_source = ContentSource.SHAREPOINT
|
|
folder.valid_thru = current_time + self.share_point_file_expiry
|
|
await folder.save()
|
|
|
|
async def retrieve_directories_for_all_folders(self):
|
|
current_time = datetime.now(timezone.utc)
|
|
expired_folders = await ContentFolderDoc.find(
|
|
ContentFolderDoc.update_source == ContentSource.SHAREPOINT,
|
|
ContentFolderDoc.valid_thru < current_time,
|
|
).to_list()
|
|
|
|
for folder in expired_folders:
|
|
await self.retrieve_directories_for_folder(
|
|
folder.folder_name, folder.region
|
|
)
|