Files
nmemmert 1d43875e5a Fix 11 bugs: restore crash, draft leak, email failures, memory leaks
Critical fixes:
- sanitizeLoadedHitStats/VisitorStats: restore full state shape so a
  snapshot restore no longer crashes hit-counting middleware (missing
  byPathReal, byPathBot, byDayReal, byDayBot, botReasons, ipHashIndex)
- /questions/share/🆔 read state.questions only, not draft questions
- inbound-email: validate date with Number.isFinite before toISOString
- study-reminders: wrap each send in try/catch so one failure doesn't
  block remaining users; persist sent-markers after each success

Security:
- getClientIp: use req.ip (trust-proxy-resolved) instead of raw
  x-forwarded-for header to prevent IP spoofing
- env-snapshot.env: delete immediately after backup tar stream ends
  so secrets don't linger on disk between exports

Correctness / UX:
- contact form: email failures no longer 500 the user after the
  submission is already saved; log and fall through instead
- study-account profile: cap data URI avatar at 6 MB
- admin enrollment PATCH: validate slug against study catalog
- signup: return 503 at MAX_STUDY_USERS instead of silently dropping
  oldest accounts

Memory leaks:
- contactHits, downloadHits Maps: prune stale entries at 5000 entries
- resendEmailSubmissionIndex: trim to 2000 entries (oldest first)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-07-16 07:48:54 -04:00

252 lines
9.1 KiB
JavaScript

import { mkdir, mkdtemp, readdir, rm, cp, writeFile, unlink } from 'node:fs/promises'
import os from 'node:os'
import path from 'node:path'
import express from 'express'
import * as tar from 'tar'
import { requireAdminAuth } from '../auth.js'
import { DATA_DIR } from '../config.js'
import { state } from '../state.js'
import {
loadHitStatsFromDisk,
loadVisitorStatsFromDisk,
loadContactSubmissionsFromDisk,
loadReplyTemplatesFromDisk,
loadReplyHistoryFromDisk,
loadQuestionsFromDisk,
loadDraftQuestionsFromDisk,
loadStudyUsersFromDisk,
loadStudyCommunityFromDisk,
loadStudyRemindersFromDisk,
loadStudyCommentsFromDisk,
loadStudyCertificatesFromDisk,
loadEpisodeScriptsFromDisk,
migrateStudyNotesIfNeeded,
loadDownloadCountsFromDisk,
loadQrCodesFromDisk,
loadEpisodePlaysFromDisk,
loadPodcastChecklistFromDisk,
loadAnalyticsEventsFromDisk,
createBackupSnapshot,
refreshContentCaches,
} from '../data.js'
// Portable server settings captured into env-snapshot.env at export time so a
// backup carries the runtime configuration too (Docker deployments have no
// .env file — settings live in container env vars). Machine-specific vars
// (PORT, NODE_ENV, SITEFORGE_DATA_DIR, ALLOW_INSECURE_COOKIES, build info)
// are deliberately excluded. Values are NOT applied automatically on restore.
const ENV_SNAPSHOT_KEYS = [
'ADMIN_PASSWORD',
'RESEND_API_KEY',
'RESEND_CONTACTS_API_KEY',
'RESEND_WEBHOOK_TOKEN',
'INBOUND_EMAIL_SECRET',
'RESEND_FROM',
'RESEND_TO',
'RESEND_REPLY_TO',
'RESEND_AUTOMATION_WELCOME',
'RESEND_SEGMENT_ID',
'RESEND_REMINDER_SUBJECT',
'RESEND_WELCOME_SUBJECT',
'RESEND_WELCOME_IMAGE_URL',
'RESEND_WELCOME_EPISODE_URL',
'RESEND_WELCOME_WEBSITE_URL',
'RESEND_WELCOME_SPOTIFY_URL',
'RESEND_WELCOME_APPLE_URL',
'RESEND_WELCOME_AMAZON_URL',
'CACHE_PURGE_WEBHOOK_URL',
'DEPLOY_WEBHOOK_URL',
'CONTACT_EMAIL_COOLDOWN_MS',
'TRUST_PROXY_HOPS',
'TITUS_STUDY_FILE',
'TITUS_STUDY_DOWNLOAD_NAME',
]
// Write env snapshot into DATA_DIR for inclusion in the archive, then delete
// it immediately after the stream finishes so secrets don't linger on disk.
async function writeEnvSnapshot() {
const lines = [
'# Siteforge environment snapshot — regenerated on every full backup export.',
'# Contains secrets (admin password, API keys): keep this backup private.',
'# These values are NOT applied automatically on restore. Set them as',
'# container environment variables (or in .env) on the new server.',
`# Exported at ${new Date().toISOString()}`,
'',
]
for (const key of ENV_SNAPSHOT_KEYS) {
const value = process.env[key]
if (typeof value === 'string' && value.trim() !== '') lines.push(`${key}=${value}`)
}
await writeFile(path.join(DATA_DIR, 'env-snapshot.env'), `${lines.join('\n')}\n`, 'utf8')
}
async function deleteEnvSnapshot() {
await unlink(path.join(DATA_DIR, 'env-snapshot.env')).catch(() => {})
}
// Files that identify an archive as a Siteforge data backup. At least one
// must be present at the top level of an uploaded archive before we restore.
const KNOWN_DATA_FILES = [
'admin-content.json',
'admin-content-draft.json',
'study-users.json',
'hit-stats.json',
'visitor-stats.json',
'questions.json',
]
// Wait for every queued disk write so the archive reflects current state.
async function flushPendingWrites() {
await Promise.allSettled([
state.hitStatsWritePromise,
state.visitorStatsWritePromise,
state.contactSubmissionsWritePromise,
state.questionsWritePromise,
state.draftQuestionsWritePromise,
state.replyTemplatesWritePromise,
state.replyHistoryWritePromise,
state.podcastChecklistWritePromise,
state.studyUsersWritePromise,
state.studyCommunityWritePromise,
state.studyRemindersWritePromise,
state.studyCommentsWritePromise,
state.studyCertificatesWritePromise,
state.episodeScriptsWritePromise,
state.downloadCountsWritePromise,
state.episodePlaysWritePromise,
state.analyticsEventsWritePromise,
state.qrCodesWritePromise,
...state.studyNotesWriteQueues.values(),
...state.studyProgressWriteQueues.values(),
])
}
// Rebuild all in-memory state from whatever is now on disk (mirrors startup).
async function reloadStateFromDisk() {
state.studyNotesCache.clear()
state.studyNotesWriteQueues.clear()
state.studyProgressCache.clear()
state.studyProgressWriteQueues.clear()
// Restored study users may not match current sessions — force re-login.
state.studySessions.clear()
await Promise.all([
loadHitStatsFromDisk(),
loadVisitorStatsFromDisk(),
loadContactSubmissionsFromDisk(),
loadReplyTemplatesFromDisk(),
loadReplyHistoryFromDisk(),
loadQuestionsFromDisk(),
loadDraftQuestionsFromDisk(),
loadStudyUsersFromDisk(),
loadStudyCommunityFromDisk(),
loadStudyRemindersFromDisk(),
loadStudyCommentsFromDisk(),
loadStudyCertificatesFromDisk(),
loadEpisodeScriptsFromDisk(),
migrateStudyNotesIfNeeded(),
loadDownloadCountsFromDisk(),
loadQrCodesFromDisk(),
loadEpisodePlaysFromDisk(),
loadPodcastChecklistFromDisk(),
loadAnalyticsEventsFromDisk(),
refreshContentCaches(),
])
}
export function register(app) {
// Download the entire data directory as a tar.gz (excluding the automatic
// snapshot folder, which is derived from the other files).
app.get('/api/admin-backup/export', requireAdminAuth, async (_req, res) => {
try {
await flushPendingWrites()
await writeEnvSnapshot()
const entries = (await readdir(DATA_DIR)).filter(name => name !== 'backups')
if (entries.length === 0) {
res.status(500).json({ message: 'Data directory is empty — nothing to export.' })
return
}
const stamp = new Date().toISOString().replace(/[:.]/g, '-')
res.setHeader('Content-Type', 'application/gzip')
res.setHeader('Content-Disposition', `attachment; filename="siteforge-data-${stamp}.tar.gz"`)
const archive = tar.create({ gzip: true, cwd: DATA_DIR, portable: true }, entries)
archive.on('error', err => {
console.error('[admin-backup] export stream failed:', err)
res.destroy(err)
})
archive.on('end', () => { deleteEnvSnapshot() })
res.on('close', () => { deleteEnvSnapshot() })
archive.pipe(res)
} catch (err) {
console.error('[admin-backup] export failed:', err)
deleteEnvSnapshot()
if (!res.headersSent) res.status(500).json({ message: 'Full backup export failed.' })
}
})
// Restore the entire data directory from an uploaded tar.gz produced by the
// export endpoint, then reload all in-memory state from the restored files.
app.post(
'/api/admin-backup/import',
requireAdminAuth,
express.raw({ type: () => true, limit: '500mb' }),
async (req, res) => {
let workDir = null
try {
const body = req.body
if (!Buffer.isBuffer(body) || body.length === 0) {
res.status(400).json({ message: 'Upload the .tar.gz file produced by the full backup export.' })
return
}
if (body[0] !== 0x1f || body[1] !== 0x8b) {
res.status(400).json({ message: 'File is not a gzip archive (.tar.gz expected).' })
return
}
workDir = await mkdtemp(path.join(os.tmpdir(), 'siteforge-import-'))
const archivePath = path.join(workDir, 'import.tar.gz')
await writeFile(archivePath, body)
const extractDir = path.join(workDir, 'extracted')
await mkdir(extractDir)
// node-tar strips absolute paths and rejects entries that escape cwd.
await tar.extract({ file: archivePath, cwd: extractDir })
const extractedEntries = await readdir(extractDir)
if (!extractedEntries.some(name => KNOWN_DATA_FILES.includes(name))) {
res.status(400).json({ message: 'Archive does not look like a Siteforge data backup.' })
return
}
// Settle queued writes so nothing overwrites the restored files, and
// keep a snapshot of the pre-import state in the backups folder.
await flushPendingWrites()
await createBackupSnapshot('pre-import')
// Replace current data with the archive contents. The snapshot folder
// is preserved unless the archive itself contains one.
const currentEntries = await readdir(DATA_DIR)
for (const name of currentEntries) {
if (name === 'backups' && !extractedEntries.includes('backups')) continue
await rm(path.join(DATA_DIR, name), { recursive: true, force: true })
}
for (const name of extractedEntries) {
await cp(path.join(extractDir, name), path.join(DATA_DIR, name), { recursive: true })
}
await reloadStateFromDisk()
await createBackupSnapshot('post-import')
res.json({ ok: true, restoredEntries: extractedEntries.length })
} catch (err) {
console.error('[admin-backup] import failed:', err)
res.status(500).json({ message: 'Full data restore failed. Check the server logs — data may need manual attention.' })
} finally {
if (workDir) await rm(workDir, { recursive: true, force: true }).catch(() => {})
}
},
)
}