PeerTube/server/lib/activitypub/crawl.ts

61 lines
1.8 KiB
TypeScript
Raw Normal View History

import { retryTransactionWrapper } from '@server/helpers/database-utils'
import * as Bluebird from 'bluebird'
2020-01-31 16:56:52 +01:00
import { URL } from 'url'
2021-03-08 14:24:11 +01:00
import { ActivityPubOrderedCollection } from '../../../shared/models/activitypub'
import { logger } from '../../helpers/logger'
import { doJSONRequest } from '../../helpers/requests'
2021-06-14 18:06:58 +02:00
import { ACTIVITY_PUB, WEBSERVER } from '../../initializers/constants'
2019-03-19 16:23:02 +01:00
type HandlerFunction<T> = (items: T[]) => (Promise<any> | Bluebird<any>)
type CleanerFunction = (startedDate: Date) => (Promise<any> | Bluebird<any>)
2021-03-08 14:24:11 +01:00
async function crawlCollectionPage <T> (argUrl: string, handler: HandlerFunction<T>, cleaner?: CleanerFunction) {
let url = argUrl
logger.info('Crawling ActivityPub data on %s.', url)
2021-06-14 18:06:58 +02:00
const options = { activityPub: true }
2019-03-19 16:23:02 +01:00
const startDate = new Date()
2021-03-08 14:24:11 +01:00
const response = await doJSONRequest<ActivityPubOrderedCollection<T>>(url, options)
const firstBody = response.body
2020-01-31 16:56:52 +01:00
const limit = ACTIVITY_PUB.FETCH_PAGE_LIMIT
let i = 0
let nextLink = firstBody.first
while (nextLink && i < limit) {
let body: any
2019-04-25 14:23:15 +02:00
if (typeof nextLink === 'string') {
// Don't crawl ourselves
2020-01-31 16:56:52 +01:00
const remoteHost = new URL(nextLink).host
if (remoteHost === WEBSERVER.HOST) continue
2021-03-08 14:24:11 +01:00
url = nextLink
2021-03-08 14:24:11 +01:00
const res = await doJSONRequest<ActivityPubOrderedCollection<T>>(url, options)
body = res.body
} else {
// nextLink is already the object we want
body = nextLink
}
nextLink = body.next
i++
if (Array.isArray(body.orderedItems)) {
const items = body.orderedItems
2021-03-08 14:24:11 +01:00
logger.info('Processing %i ActivityPub items for %s.', items.length, url)
await handler(items)
}
}
2019-03-19 16:23:02 +01:00
if (cleaner) await retryTransactionWrapper(cleaner, startDate)
}
export {
crawlCollectionPage
}