X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=server%2Flib%2Factivitypub%2Fcrawl.ts;h=278abf7de0b3a9637b380521dd9f78158535686f;hb=fd1b2d695320be5d86456c939b371b2e9b4f457b;hp=2675524c63037140e6aab3a00de023751921bb74;hpb=418d092afa81e2c8fe8ac6838fc4b5eb0af6a782;p=github%2FChocobozzz%2FPeerTube.git

diff --git a/server/lib/activitypub/crawl.ts b/server/lib/activitypub/crawl.ts
index 2675524c6..278abf7de 100644
--- a/server/lib/activitypub/crawl.ts
+++ b/server/lib/activitypub/crawl.ts
@@ -1,40 +1,60 @@
-import { ACTIVITY_PUB, JOB_REQUEST_TIMEOUT } from '../../initializers'
-import { doRequest } from '../../helpers/requests'
-import { logger } from '../../helpers/logger'
 import * as Bluebird from 'bluebird'
+import { URL } from 'url'
 import { ActivityPubOrderedCollection } from '../../../shared/models/activitypub'
+import { logger } from '../../helpers/logger'
+import { doJSONRequest } from '../../helpers/requests'
+import { ACTIVITY_PUB, REQUEST_TIMEOUT, WEBSERVER } from '../../initializers/constants'
+
+type HandlerFunction<T> = (items: T[]) => (Promise<any> | Bluebird<any>)
+type CleanerFunction = (startedDate: Date) => (Promise<any> | Bluebird<any>)
 
-async function crawlCollectionPage <T> (uri: string, handler: (items: T[]) => (Promise<any> | Bluebird<any>)) {
-  logger.info('Crawling ActivityPub data on %s.', uri)
+async function crawlCollectionPage <T> (argUrl: string, handler: HandlerFunction<T>, cleaner?: CleanerFunction) {
+  let url = argUrl
+
+  logger.info('Crawling ActivityPub data on %s.', url)
 
   const options = {
-    method: 'GET',
-    uri,
-    json: true,
     activityPub: true,
-    timeout: JOB_REQUEST_TIMEOUT
+    timeout: REQUEST_TIMEOUT
   }
 
-  const response = await doRequest<ActivityPubOrderedCollection<T>>(options)
+  const startDate = new Date()
+
+  const response = await doJSONRequest<ActivityPubOrderedCollection<T>>(url, 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
+
+      url = nextLink
+
+      const res = await doJSONRequest<ActivityPubOrderedCollection<T>>(url, options)
+      body = res.body
+    } else {
+      // nextLink is already the object we want
+      body = nextLink
+    }
 
-    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, options.uri)
+      logger.info('Processing %i ActivityPub items for %s.', items.length, url)
 
       await handler(items)
     }
   }
+
+  if (cleaner) await cleaner(startDate)
 }
 
 export {