remove_unconfirmed_emails.mjs 7.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265
  1. // @ts-check
  2. import minimist from 'minimist'
  3. import { batchedUpdate } from '@overleaf/mongo-utils/batchedUpdate.js'
  4. import { db, ObjectId } from '../app/src/infrastructure/mongodb.js'
  5. import UserAuditLogHandler from '../app/src/Features/User/UserAuditLogHandler.js'
  6. import fs from 'node:fs/promises'
  7. import * as csv from 'csv'
  8. import { promisify } from 'node:util'
  9. import _ from 'lodash'
  10. import { scriptRunner } from './lib/ScriptRunner.mjs'
  11. const CSV_FILENAME = '/tmp/remove_unconfirmed_emails.csv'
  12. /**
  13. * @type {(records: string[][]) => Promise<string>}
  14. */
  15. const stringifyAsync = promisify(csv.stringify)
  16. /**
  17. * @type {(csvString: string) => Promise<string[][]>}
  18. */
  19. const parseAsync = promisify(csv.parse)
  20. function usage() {
  21. console.log('Usage: node remove_unconfirmed_emails.mjs')
  22. console.log('Removes unconfirmed emails from users')
  23. console.log('Options:')
  24. console.log(
  25. '' +
  26. ' --generate generate the CSV file (remove_unconfirmed_emails.csv) containing the emails to remove\n' +
  27. ' --consume consume the CSV file (remove_unconfirmed_emails.csv) and remove the emails (by default it is a dry-run)\n' +
  28. ' --commit apply the changes (to be used with --consume)\n'
  29. )
  30. process.exit(0)
  31. }
  32. const { generate, consume, commit, help } = minimist(process.argv.slice(2), {
  33. boolean: ['generate', 'consume', 'commit', 'help'],
  34. alias: { help: 'h' },
  35. default: { generate: false, consume: false, commit: false },
  36. })
  37. async function generateCsvFile(trackProgress) {
  38. console.time('generate_csv')
  39. let processedUsersCount = 0
  40. let skippedUnconfirmedPrimaries = 0
  41. let totalEmailsToRemove = 0
  42. let totalUsersInCsv = 0
  43. const records = [['User ID', 'Email', 'Sign Up Date']]
  44. await batchedUpdate(
  45. db.users,
  46. {
  47. $and: [
  48. { emails: { $exists: true } },
  49. { emails: { $not: { $size: 0 } } },
  50. // Warning: this also matches unconfirmed primary emails
  51. {
  52. emails: {
  53. $elemMatch: {
  54. $or: [{ confirmedAt: { $exists: false } }, { confirmedAt: null }],
  55. },
  56. },
  57. },
  58. ],
  59. },
  60. async users => {
  61. console.log('Process', users.length, 'users')
  62. processedUsersCount += users.length
  63. for (const user of users) {
  64. const unconfirmedSecondaries = user.emails.filter(
  65. email => !email.confirmedAt && email.email !== user.email
  66. )
  67. if (unconfirmedSecondaries.length === 0) {
  68. // Users can have been selected because of their unconfirmed primary email
  69. // we don't want to remove those
  70. skippedUnconfirmedPrimaries++
  71. continue
  72. }
  73. for (const email of unconfirmedSecondaries) {
  74. records.push([
  75. user._id.toString(),
  76. email.email,
  77. user.signUpDate.toISOString(),
  78. ])
  79. }
  80. totalUsersInCsv++
  81. totalEmailsToRemove += unconfirmedSecondaries.length
  82. }
  83. },
  84. { _id: 1, signUpDate: 1, emails: 1, email: 1 },
  85. undefined,
  86. { trackProgress }
  87. )
  88. const csvContent = await stringifyAsync(records)
  89. await fs.writeFile(CSV_FILENAME, csvContent)
  90. console.log()
  91. console.log('Processed users:', processedUsersCount)
  92. console.log()
  93. console.log('Generated CSV file:', CSV_FILENAME)
  94. console.log('Total emails in the CSV:', totalEmailsToRemove)
  95. console.log('Total users in the CSV:', totalUsersInCsv)
  96. console.log(
  97. 'Unconfirmed primary emails (skipped):',
  98. skippedUnconfirmedPrimaries
  99. )
  100. console.log()
  101. console.timeEnd('generate_csv')
  102. console.log()
  103. }
  104. async function consumeCsvFile() {
  105. console.time('consume_csv')
  106. const csvContent = await fs.readFile(CSV_FILENAME, 'utf8')
  107. const rows = await parseAsync(csvContent)
  108. rows.shift() // Remove header row
  109. const emailsByUserId = {}
  110. for (const [userId, email] of rows) {
  111. if (!emailsByUserId[userId]) {
  112. emailsByUserId[userId] = []
  113. }
  114. emailsByUserId[userId].push(email)
  115. }
  116. const userIds = Object.keys(emailsByUserId)
  117. let processedUsersCount = 0
  118. let removedEmailsCount = 0
  119. let totalModifiedUsersCount = 0
  120. const skippedEmail = {
  121. userNotFound: 0,
  122. nowConfirmed: 0,
  123. nowPrimary: 0,
  124. nowRemoved: 0,
  125. }
  126. console.log('Total emails in the CSV:', rows.length)
  127. console.log('Total users in the CSV:', userIds.length)
  128. for (const userId of userIds) {
  129. const emailsToRemove = emailsByUserId[userId]
  130. const user = await db.users.findOne({ _id: new ObjectId(userId) })
  131. if (!user) {
  132. skippedEmail.userNotFound += emailsToRemove.length
  133. continue
  134. }
  135. const emailsToRemoveNow = emailsToRemove.filter(email => {
  136. const currentEmail = user.emails.find(e => e.email === email)
  137. if (!currentEmail) {
  138. skippedEmail.nowRemoved++
  139. return false
  140. }
  141. if (currentEmail.confirmedAt) {
  142. skippedEmail.nowConfirmed++
  143. return false
  144. }
  145. if (currentEmail.email === user.email) {
  146. skippedEmail.nowPrimary++
  147. return false
  148. }
  149. return true
  150. })
  151. removedEmailsCount += emailsToRemoveNow.length
  152. if (commit && emailsToRemoveNow.length > 0) {
  153. for (const email of emailsToRemove) {
  154. await UserAuditLogHandler.promises.addEntry(
  155. userId,
  156. 'remove-email',
  157. undefined,
  158. undefined,
  159. {
  160. removedEmail: email,
  161. script: true,
  162. note: 'remove unconfirmed secondary emails',
  163. }
  164. )
  165. }
  166. const updated = await db.users.updateOne(
  167. { _id: new ObjectId(userId) },
  168. { $pull: { emails: { email: { $in: emailsToRemove } } } }
  169. )
  170. totalModifiedUsersCount += updated.modifiedCount
  171. }
  172. processedUsersCount++
  173. if (processedUsersCount % 100 === 0) {
  174. console.log('Processed', processedUsersCount, 'users')
  175. }
  176. }
  177. console.log()
  178. if (!commit) {
  179. console.log('Dry-run, use --commit to apply changes')
  180. console.log('This would be the result:')
  181. console.log()
  182. }
  183. console.log('Total emails in the CSV:', rows.length)
  184. console.log('Total users in the CSV:', userIds.length)
  185. console.log('Total users processed:', processedUsersCount)
  186. console.log('Total emails removed:', removedEmailsCount)
  187. console.log('Skipped emails:', _.sum(Object.values(skippedEmail)))
  188. console.log(' - User not found:', skippedEmail.userNotFound)
  189. console.log(' - Email now confirmed:', skippedEmail.nowConfirmed)
  190. console.log(' - Email now primary:', skippedEmail.nowPrimary)
  191. console.log(' - Email now removed:', skippedEmail.nowRemoved)
  192. console.log()
  193. if (commit) {
  194. console.log('Total users modified:', totalModifiedUsersCount)
  195. } else {
  196. console.log('Note: this was a dry-run. No changes were made.')
  197. }
  198. console.log()
  199. console.timeEnd('consume_csv')
  200. console.log()
  201. }
  202. async function main(trackProgress) {
  203. if (help) {
  204. return usage()
  205. }
  206. if (!generate && !consume) {
  207. console.error('Error: Either --generate or --consume must be specified')
  208. return usage()
  209. }
  210. if (generate && consume) {
  211. console.error('Error: Cannot use both --generate and --consume together')
  212. return usage()
  213. }
  214. if (commit && !consume) {
  215. console.error('Error: --commit can only be used with --consume')
  216. return usage()
  217. }
  218. if (generate) {
  219. await generateCsvFile(trackProgress)
  220. } else if (consume) {
  221. await consumeCsvFile()
  222. }
  223. }
  224. try {
  225. await scriptRunner(main)
  226. process.exit(0)
  227. } catch (error) {
  228. console.error(error)
  229. process.exit(1)
  230. }