server/lib/activitypub/crawl.ts

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