server/lib/activitypub/crawl.ts

   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')
   5
   6 async function crawlCollectionPage <T> (uri: string, handler: (items: T[]) => Promise<any> | Bluebird<any>) {
   7   logger.info('Crawling ActivityPub data on %s.', uri)
   8
   9   const options = {
  10     method: 'GET',
  11     uri,
  12     json: true,
  13     activityPub: true,
  14     timeout: JOB_REQUEST_TIMEOUT
  15   }
  16
  17   const response = await doRequest(options)
  18   const firstBody = response.body
  19
  20   let limit = ACTIVITY_PUB.FETCH_PAGE_LIMIT
  21   let i = 0
  22   let nextLink = firstBody.first
  23   while (nextLink && i < limit) {
  24     options.uri = nextLink
  25
  26     const { body } = await doRequest(options)
  27     nextLink = body.next
  28     i++
  29
  30     if (Array.isArray(body.orderedItems)) {
  31       const items = body.orderedItems
  32       logger.info('Processing %i ActivityPub items for %s.', items.length, options.uri)
  33
  34       await handler(items)
  35     }
  36   }
  37 }
  38
  39 export {
  40   crawlCollectionPage
  41 }