1 import { ACTIVITY_PUB, JOB_REQUEST_TIMEOUT } from '../../initializers/constants'
2 import { doRequest } from '../../helpers/requests'
3 import { logger } from '../../helpers/logger'
4 import * as Bluebird from 'bluebird'
5 import { ActivityPubOrderedCollection } from '../../../shared/models/activitypub'
7 type HandlerFunction<T> = (items: T[]) => (Promise<any> | Bluebird<any>)
8 type CleanerFunction = (startedDate: Date) => (Promise<any> | Bluebird<any>)
10 async function crawlCollectionPage <T> (uri: string, handler: HandlerFunction<T>, cleaner?: CleanerFunction) {
11 logger.info('Crawling ActivityPub data on %s.', uri)
18 timeout: JOB_REQUEST_TIMEOUT
21 const startDate = new Date()
23 const response = await doRequest<ActivityPubOrderedCollection<T>>(options)
24 const firstBody = response.body
26 let limit = ACTIVITY_PUB.FETCH_PAGE_LIMIT
28 let nextLink = firstBody.first
29 while (nextLink && i < limit) {
30 options.uri = nextLink
32 const { body } = await doRequest<ActivityPubOrderedCollection<T>>(options)
36 if (Array.isArray(body.orderedItems)) {
37 const items = body.orderedItems
38 logger.info('Processing %i ActivityPub items for %s.', items.length, options.uri)
44 if (cleaner) await cleaner(startDate)