MVP, working sentry scraper

This commit is contained in:
Mike Conrad
2025-05-18 20:44:42 -04:00
parent a34b0899ce
commit 8ea49772df
29 changed files with 724 additions and 481 deletions

View File

@ -0,0 +1,80 @@
import Replay from '#models/replay'
import env from '#start/env'
import type { HttpContext } from '@adonisjs/core/http'
const SENTRY_TOKEN = env.get('SENTRY_TOKEN')
const SENTRY_ORG = env.get('SENTRY_ORG')
let recordsUpdated = 0
export default class ReplaysController {
async index({ request, response }: HttpContext) {
const {statsPeriod, start, end} = request.qs()
recordsUpdated = 0
let queryString: string = '?statsPeriod=24h'// Default in case none is provided
if (statsPeriod) {
queryString = `?statsPeriod=${statsPeriod}`
} else if (start && end) {
queryString = `?start=${start}&end=${end}`
}
const replays = await fetchBatch(`https://sentry.io/api/0/organizations/${SENTRY_ORG}/replays/${queryString}`)
return response.json(replays)
}
}
async function fetchBatch(url: string) {
const options: RequestInit = {
headers: {
Authorization: `Bearer ${SENTRY_TOKEN}`
}
}
const req = await fetch(url, options)
const resp = await req.json() as unknown
const replays = await resp.data as unknown
const headers = await req.headers
const cleanedData = replays.map(record => sanitizeInput(record, Replay.allowedFields))
let updated = await Replay.updateOrCreateMany('id', cleanedData )
recordsUpdated = recordsUpdated + updated.length
const pagination = parseSentryLinkHeader(headers.get('link'))
if (pagination.hasNextResults == true) {
console.log('fetching', pagination.next)
await fetchBatch(pagination.next)
}
console.log('no more results')
return {recordsUpdated}
}
function parseSentryLinkHeader(header:string) {
const links = header.split(',').map(part => part.trim())
const result = {}
for (const link of links) {
const match = link.match(/<([^>]+)>;\s*rel="([^"]+)";\s*results="([^"]+)";\s*cursor="([^"]+)"/)
if (!match) continue
const [, url, rel, results] = match
if (rel === 'previous') {
result.previous = url
result.hasPreviousResults = results === 'true'
} else if (rel === 'next') {
result.next = url
result.hasNextResults = results === 'true'
}
}
return result
}
function sanitizeInput(data: Record<string, any>, allowedFields: string[]) {
return allowedFields.reduce((acc, key) => {
if (key in data) acc[key] = data[key]
return acc
}, {} as Record<string, any>)
}