1 import { ACTIVITY_PUB, JOB_REQUEST_TIMEOUT } from '../../initializers'
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 async function crawlCollectionPage <T> (uri: string, handler: (items: T[]) => (Promise<any> | Bluebird<any>)) {
8 logger.info('Crawling ActivityPub data on %s.', uri)
15 timeout: JOB_REQUEST_TIMEOUT
18 const response = await doRequest<ActivityPubOrderedCollection<T>>(options)
19 const firstBody = response.body
21 let limit = ACTIVITY_PUB.FETCH_PAGE_LIMIT
23 let nextLink = firstBody.first
24 while (nextLink && i < limit) {
25 options.uri = nextLink
27 const { body } = await doRequest<ActivityPubOrderedCollection<T>>(options)
31 if (Array.isArray(body.orderedItems)) {
32 const items = body.orderedItems
33 logger.info('Processing %i ActivityPub items for %s.', items.length, options.uri)