]>
Commit | Line | Data |
---|---|---|
1 | import { ACTIVITY_PUB, JOB_REQUEST_TIMEOUT } from '../../initializers' | |
2 | import { doRequest } from '../../helpers/requests' | |
3 | import { logger } from '../../helpers/logger' | |
4 | import * as Bluebird from 'bluebird' | |
5 | import { ActivityPubOrderedCollection } from '../../../shared/models/activitypub' | |
6 | ||
7 | type HandlerFunction<T> = (items: T[]) => (Promise<any> | Bluebird<any>) | |
8 | type CleanerFunction = (startedDate: Date) => (Promise<any> | Bluebird<any>) | |
9 | ||
10 | async function crawlCollectionPage <T> (uri: string, handler: HandlerFunction<T>, cleaner?: CleanerFunction) { | |
11 | logger.info('Crawling ActivityPub data on %s.', uri) | |
12 | ||
13 | const options = { | |
14 | method: 'GET', | |
15 | uri, | |
16 | json: true, | |
17 | activityPub: true, | |
18 | timeout: JOB_REQUEST_TIMEOUT | |
19 | } | |
20 | ||
21 | const startDate = new Date() | |
22 | ||
23 | const response = await doRequest<ActivityPubOrderedCollection<T>>(options) | |
24 | const firstBody = response.body | |
25 | ||
26 | let limit = ACTIVITY_PUB.FETCH_PAGE_LIMIT | |
27 | let i = 0 | |
28 | let nextLink = firstBody.first | |
29 | while (nextLink && i < limit) { | |
30 | options.uri = nextLink | |
31 | ||
32 | const { body } = await doRequest<ActivityPubOrderedCollection<T>>(options) | |
33 | nextLink = body.next | |
34 | i++ | |
35 | ||
36 | if (Array.isArray(body.orderedItems)) { | |
37 | const items = body.orderedItems | |
38 | logger.info('Processing %i ActivityPub items for %s.', items.length, options.uri) | |
39 | ||
40 | await handler(items) | |
41 | } | |
42 | } | |
43 | ||
44 | if (cleaner) await cleaner(startDate) | |
45 | } | |
46 | ||
47 | export { | |
48 | crawlCollectionPage | |
49 | } |