Fix incorrect IDs in AP federation

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

diff --git a/server/lib/activitypub/crawl.ts b/server/lib/activitypub/crawl.ts
index db9ce3293de0475825b276594dea75bba10ae457..eeafdf4ba8d04848242020af795f8fb4388cda8b 100644 (file)
--- a/server/lib/activitypub/crawl.ts
+++ b/server/lib/activitypub/crawl.ts
@@ -1,9 +1,14 @@
-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 { URL } from 'url'
 
-async function crawlCollectionPage <T> (uri: string, handler: (items: T[]) => Promise<any> | Bluebird<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 = {
@@ -14,16 +19,31 @@ 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
+  const limit = ACTIVITY_PUB.FETCH_PAGE_LIMIT
   let i = 0
   let nextLink = firstBody.first
   while (nextLink && i < limit) {
-    options.uri = nextLink
+    let body: any
+
+    if (typeof nextLink === 'string') {
+      // Don't crawl ourselves
+      const remoteHost = new URL(nextLink).host
+      if (remoteHost === WEBSERVER.HOST) continue
+
+      options.uri = nextLink
+
+      const res = await doRequest<ActivityPubOrderedCollection<T>>(options)
+      body = res.body
+    } else {
+      // nextLink is already the object we want
+      body = nextLink
+    }
 
-    const { body } = await doRequest(options)
     nextLink = body.next
     i++
 
@@ -34,6 +54,8 @@ async function crawlCollectionPage <T> (uri: string, handler: (items: T[]) => Pr
       await handler(items)
     }
   }
+
+  if (cleaner) await cleaner(startDate)
 }
 
 export {