-import { ACTIVITY_PUB, JOB_REQUEST_TIMEOUT } from '../../initializers'
-import { doRequest } from '../../helpers/requests'
+import Bluebird from 'bluebird'
+import { URL } from 'url'
+import { retryTransactionWrapper } from '@server/helpers/database-utils'
+import { ActivityPubOrderedCollection } from '../../../shared/models/activitypub'
import { logger } from '../../helpers/logger'
-import * as Bluebird from 'bluebird'
+import { doJSONRequest } from '../../helpers/requests'
+import { ACTIVITY_PUB, WEBSERVER } from '../../initializers/constants'
-async function crawlCollectionPage <T> (uri: string, handler: (items: T[]) => Promise<any> | Bluebird<any>) {
- logger.info('Crawling ActivityPub data on %s.', uri)
+type HandlerFunction<T> = (items: T[]) => (Promise<any> | Bluebird<any>)
+type CleanerFunction = (startedDate: Date) => Promise<any>
- const options = {
- method: 'GET',
- uri,
- json: true,
- activityPub: true,
- timeout: JOB_REQUEST_TIMEOUT
- }
+async function crawlCollectionPage <T> (argUrl: string, handler: HandlerFunction<T>, cleaner?: CleanerFunction) {
+ let url = argUrl
+
+ logger.info('Crawling ActivityPub data on %s.', url)
+
+ const options = { activityPub: true }
- const response = await doRequest(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(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 retryTransactionWrapper(cleaner, startDate)
}
export {