from http import HTTPStatus from sanic import response, Request, Blueprint from sanic_ext import validate from sanic_ext.extensions.openapi import openapi from wrolpi.api_utils import json_response from wrolpi.common import logger, wrol_mode_check, api_param_limiter from wrolpi.errors import ValidationError from wrolpi.events import Events from wrolpi.schema import JSONErrorResponse from wrolpi.switches import register_switch_handler, ActivateSwitchMethod from . import lib, schema NAME = 'archive' archive_bp = Blueprint('Archive', '/api/archive') logger = logger.getChild(__name__) @archive_bp.get('/') @openapi.description('Get an archive') @openapi.response(HTTPStatus.NOT_FOUND, JSONErrorResponse) async def get_archive(_: Request, archive_id: int): archive = lib.get_archive(archive_id=archive_id) archive_file_group = archive.file_group.__json__() history = [i.file_group.__json__() for i in archive.history] return json_response({'file_group': archive_file_group, 'history': history}) @archive_bp.delete('/', name='archive_delete_one') @archive_bp.delete('/', name='archive_delete_many') @openapi.description('Delete Archives') @openapi.response(HTTPStatus.NOT_FOUND, JSONErrorResponse) @wrol_mode_check async def delete_archive(_: Request, archive_ids: str): try: archive_ids = [int(i) for i in str(archive_ids).split(',')] except ValueError: raise ValidationError('Could not parse archive ids') lib.delete_archives(*archive_ids) return response.empty() @archive_bp.get('/domains') @openapi.summary('Get a list of all Domains and their Archive statistics') @openapi.response(200, schema.GetDomainsResponse, "The list of domains") async def get_domains(_: Request): domains = lib.get_domains() return json_response({'domains': domains, 'totals': {'domains': len(domains)}}) archive_limit_limiter = api_param_limiter(100) @archive_bp.post('/search') @openapi.definition( summary='A File search with more filtering related to Archives', body=schema.ArchiveSearchRequest, ) @openapi.response(HTTPStatus.OK, schema.ArchiveSearchResponse) @openapi.response(HTTPStatus.NOT_FOUND, JSONErrorResponse) @validate(schema.ArchiveSearchRequest) async def search_archives(_: Request, body: schema.ArchiveSearchRequest): search_str = body.search_str domain = body.domain limit = archive_limit_limiter(body.limit) offset = body.offset or 0 file_groups, total = lib.search_archives(search_str, domain, limit, offset, body.order_by, body.tag_names, body.headline) ret = dict(file_groups=file_groups, totals=dict(file_groups=total)) return json_response(ret) @archive_bp.get('/upload') @openapi.definition( summary='A message to confirm to the user that they have the correct upload URL.' ) async def get_upload_singlefile(request: Request): return response.text('This is the URL to upload files using the SingleFile browser extension.' ' This requires a POST request.') @register_switch_handler('singlefile_upload_switch_handler') async def singlefile_upload_switch_handler(singlefile): """Used by `post_upload_singlefile` to upload a single file""" from wrolpi.downloader import download_manager, Download from . import ArchiveDownloader archive = await lib.singlefile_to_archive(singlefile) logger.info(f'Created Archive from upload: {archive}') name = archive.file_group.title or archive.file_group.url Events.send_archive_uploaded(f'Created Archive from upload: {name}', url=archive.location) url = archive.file_group.url if url and (download := Download.get_by_url(url)): if (download.is_failed or download.is_deferred) and download.downloader == ArchiveDownloader.name: # Download was attempted and failed, user manually archived the URL. download_manager.delete_download(download.id) singlefile_upload_switch_handler: ActivateSwitchMethod @archive_bp.post('/upload') @openapi.definition( summary='Upload SingleFile from SingleFile browser extension and convert it to an Archive.' ) async def post_upload_singlefile(request: Request): url = request.form['url'][0] singlefile = request.files['singlefile_contents'][0].body logger.info(f'Got Archive upload of {len(singlefile)} bytes for URL: {url}') # Get the URL to ensure that the singlefile is valid. lib.get_url_from_singlefile(singlefile) # Send processing to background task so the extension can continue. singlefile_upload_switch_handler.activate_switch(dict(singlefile=singlefile)) # Return empty json response because SingleFile extension expects a JSON response. return json_response(dict(), status=HTTPStatus.OK)