server/lib/activitypub/crawl.ts

   1 import { ACTIVITY_PUB, JOB_REQUEST_TIMEOUT, WEBSERVER } 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'
   6 import { URL } from 'url'
   7
   8 type HandlerFunction<T> = (items: T[]) => (Promise<any> | Bluebird<any>)
   9 type CleanerFunction = (startedDate: Date) => (Promise<any> | Bluebird<any>)
  10
  11 async function crawlCollectionPage <T> (uri: string, handler: HandlerFunction<T>, cleaner?: CleanerFunction) {
  12   logger.info('Crawling ActivityPub data on %s.', uri)
  13
  14   const options = {
  15     method: 'GET',
  16     uri,
  17     json: true,
  18     activityPub: true,
  19     timeout: JOB_REQUEST_TIMEOUT
  20   }
  21
  22   const startDate = new Date()
  23
  24   const response = await doRequest<ActivityPubOrderedCollection<T>>(options)
  25   const firstBody = response.body
  26
  27   const limit = ACTIVITY_PUB.FETCH_PAGE_LIMIT
  28   let i = 0
  29   let nextLink = firstBody.first
  30   while (nextLink && i < limit) {
  31     let body: any
  32
  33     if (typeof nextLink === 'string') {
  34       // Don't crawl ourselves
  35       const remoteHost = new URL(nextLink).host
  36       if (remoteHost === WEBSERVER.HOST) continue
  37
  38       options.uri = nextLink
  39
  40       const res = await doRequest<ActivityPubOrderedCollection<T>>(options)
  41       body = res.body
  42     } else {
  43       // nextLink is already the object we want
  44       body = nextLink
  45     }
  46
  47     nextLink = body.next
  48     i++
  49
  50     if (Array.isArray(body.orderedItems)) {
  51       const items = body.orderedItems
  52       logger.info('Processing %i ActivityPub items for %s.', items.length, options.uri)
  53
  54       await handler(items)
  55     }
  56   }
  57
  58   if (cleaner) await cleaner(startDate)
  59 }
  60
  61 export {
  62   crawlCollectionPage
  63 }