1 import { ACTIVITY_PUB, JOB_REQUEST_TIMEOUT } from '../../initializers'
2 import { doRequest } from '../../helpers/requests'
3 import { logger } from '../../helpers/logger'
4 import Bluebird = require('bluebird')
6 async function crawlCollectionPage <T> (uri: string, handler: (items: T[]) => Promise<any> | Bluebird<any>) {
7 logger.info('Crawling ActivityPub data on %s.', uri)
14 timeout: JOB_REQUEST_TIMEOUT
17 const response = await doRequest(options)
18 const firstBody = response.body
20 let limit = ACTIVITY_PUB.FETCH_PAGE_LIMIT
22 let nextLink = firstBody.first
23 while (nextLink && i < limit) {
24 options.uri = nextLink
26 const { body } = await doRequest(options)
30 if (Array.isArray(body.orderedItems)) {
31 const items = body.orderedItems
32 logger.info('Processing %i ActivityPub items for %s.', items.length, options.uri)