Fix user notifications tests

[github/Chocobozzz/PeerTube.git] / server / lib / activitypub / crawl.ts

diff --git a/server/lib/activitypub/crawl.ts b/server/lib/activitypub/crawl.ts
index 7305b3969b4c5955253bf732b0de49b818ca39e0..0ba59b47dde4c0c90e7d133ab2333f756bdaf27a 100644 (file)
--- a/server/lib/activitypub/crawl.ts
+++ b/server/lib/activitypub/crawl.ts
@@ -1,8 +1,15 @@
-import { ACTIVITY_PUB, JOB_REQUEST_TIMEOUT } from '../../initializers'
+import { ACTIVITY_PUB, JOB_REQUEST_TIMEOUT, WEBSERVER } from '../../initializers/constants'
 import { doRequest } from '../../helpers/requests'
 import { logger } from '../../helpers/logger'
+import * as Bluebird from 'bluebird'
+import { ActivityPubOrderedCollection } from '../../../shared/models/activitypub'
+import { checkUrlsSameHost } from '../../helpers/activitypub'
+import { parse } from "url"
 
-async function crawlCollectionPage <T> (uri: string, handler: (items: T[]) => Promise<any>) {
+type HandlerFunction<T> = (items: T[]) => (Promise<any> | Bluebird<any>)
+type CleanerFunction = (startedDate: Date) => (Promise<any> | Bluebird<any>)
+
+async function crawlCollectionPage <T> (uri: string, handler: HandlerFunction<T>, cleaner?: CleanerFunction) {
   logger.info('Crawling ActivityPub data on %s.', uri)
 
   const options = {
@@ -13,26 +20,34 @@ async function crawlCollectionPage <T> (uri: string, handler: (items: T[]) => Pr
     timeout: JOB_REQUEST_TIMEOUT
   }
 
-  const response = await doRequest(options)
+  const startDate = new Date()
+
+  const response = await doRequest<ActivityPubOrderedCollection<T>>(options)
   const firstBody = response.body
 
   let limit = ACTIVITY_PUB.FETCH_PAGE_LIMIT
   let i = 0
   let nextLink = firstBody.first
   while (nextLink && i < limit) {
+    // Don't crawl ourselves
+    const remoteHost = parse(nextLink).host
+    if (remoteHost === WEBSERVER.HOST) continue
+
     options.uri = nextLink
 
-    const { body } = await doRequest(options)
+    const { body } = await doRequest<ActivityPubOrderedCollection<T>>(options)
     nextLink = body.next
     i++
 
     if (Array.isArray(body.orderedItems)) {
       const items = body.orderedItems
-      logger.info('Processing %i ActivityPub items for %s.', items.length, nextLink)
+      logger.info('Processing %i ActivityPub items for %s.', items.length, options.uri)
 
       await handler(items)
     }
   }
+
+  if (cleaner) await cleaner(startDate)
 }
 
 export {