remove_unconfirmed_emails.mjs 7.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262
  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. const CSV_FILENAME = '/tmp/remove_unconfirmed_emails.csv'
  11. /**
  12. * @type {(records: string[][]) => Promise<string>}
  13. */
  14. const stringifyAsync = promisify(csv.stringify)
  15. /**
  16. * @type {(csvString: string) => Promise<string[][]>}
  17. */
  18. const parseAsync = promisify(csv.parse)
  19. function usage() {
  20. console.log('Usage: node remove_unconfirmed_emails.mjs')
  21. console.log('Removes unconfirmed emails from users')
  22. console.log('Options:')
  23. console.log(
  24. '' +
  25. ' --generate generate the CSV file (remove_unconfirmed_emails.csv) containing the emails to remove\n' +
  26. ' --consume consume the CSV file (remove_unconfirmed_emails.csv) and remove the emails (by default it is a dry-run)\n' +
  27. ' --commit apply the changes (to be used with --consume)\n'
  28. )
  29. process.exit(0)
  30. }
  31. const { generate, consume, commit, help } = minimist(process.argv.slice(2), {
  32. boolean: ['generate', 'consume', 'commit', 'help'],
  33. alias: { help: 'h' },
  34. default: { generate: false, consume: false, commit: false },
  35. })
  36. async function generateCsvFile() {
  37. console.time('generate_csv')
  38. let processedUsersCount = 0
  39. let skippedUnconfirmedPrimaries = 0
  40. let totalEmailsToRemove = 0
  41. let totalUsersInCsv = 0
  42. const records = [['User ID', 'Email', 'Sign Up Date']]
  43. await batchedUpdate(
  44. db.users,
  45. {
  46. $and: [
  47. { emails: { $exists: true } },
  48. { emails: { $not: { $size: 0 } } },
  49. // Warning: this also matches unconfirmed primary emails
  50. {
  51. emails: {
  52. $elemMatch: {
  53. $or: [{ confirmedAt: { $exists: false } }, { confirmedAt: null }],
  54. },
  55. },
  56. },
  57. ],
  58. },
  59. async users => {
  60. console.log('Process', users.length, 'users')
  61. processedUsersCount += users.length
  62. for (const user of users) {
  63. const unconfirmedSecondaries = user.emails.filter(
  64. email => !email.confirmedAt && email.email !== user.email
  65. )
  66. if (unconfirmedSecondaries.length === 0) {
  67. // Users can have been selected because of their unconfirmed primary email
  68. // we don't want to remove those
  69. skippedUnconfirmedPrimaries++
  70. continue
  71. }
  72. for (const email of unconfirmedSecondaries) {
  73. records.push([
  74. user._id.toString(),
  75. email.email,
  76. user.signUpDate.toISOString(),
  77. ])
  78. }
  79. totalUsersInCsv++
  80. totalEmailsToRemove += unconfirmedSecondaries.length
  81. }
  82. },
  83. { _id: 1, signUpDate: 1, emails: 1, email: 1 }
  84. )
  85. const csvContent = await stringifyAsync(records)
  86. await fs.writeFile(CSV_FILENAME, csvContent)
  87. console.log()
  88. console.log('Processed users:', processedUsersCount)
  89. console.log()
  90. console.log('Generated CSV file:', CSV_FILENAME)
  91. console.log('Total emails in the CSV:', totalEmailsToRemove)
  92. console.log('Total users in the CSV:', totalUsersInCsv)
  93. console.log(
  94. 'Unconfirmed primary emails (skipped):',
  95. skippedUnconfirmedPrimaries
  96. )
  97. console.log()
  98. console.timeEnd('generate_csv')
  99. console.log()
  100. }
  101. async function consumeCsvFile() {
  102. console.time('consume_csv')
  103. const csvContent = await fs.readFile(CSV_FILENAME, 'utf8')
  104. const rows = await parseAsync(csvContent)
  105. rows.shift() // Remove header row
  106. const emailsByUserId = {}
  107. for (const [userId, email] of rows) {
  108. if (!emailsByUserId[userId]) {
  109. emailsByUserId[userId] = []
  110. }
  111. emailsByUserId[userId].push(email)
  112. }
  113. const userIds = Object.keys(emailsByUserId)
  114. let processedUsersCount = 0
  115. let removedEmailsCount = 0
  116. let totalModifiedUsersCount = 0
  117. const skippedEmail = {
  118. userNotFound: 0,
  119. nowConfirmed: 0,
  120. nowPrimary: 0,
  121. nowRemoved: 0,
  122. }
  123. console.log('Total emails in the CSV:', rows.length)
  124. console.log('Total users in the CSV:', userIds.length)
  125. for (const userId of userIds) {
  126. const emailsToRemove = emailsByUserId[userId]
  127. const user = await db.users.findOne({ _id: new ObjectId(userId) })
  128. if (!user) {
  129. skippedEmail.userNotFound += emailsToRemove.length
  130. continue
  131. }
  132. const emailsToRemoveNow = emailsToRemove.filter(email => {
  133. const currentEmail = user.emails.find(e => e.email === email)
  134. if (!currentEmail) {
  135. skippedEmail.nowRemoved++
  136. return false
  137. }
  138. if (currentEmail.confirmedAt) {
  139. skippedEmail.nowConfirmed++
  140. return false
  141. }
  142. if (currentEmail.email === user.email) {
  143. skippedEmail.nowPrimary++
  144. return false
  145. }
  146. return true
  147. })
  148. removedEmailsCount += emailsToRemoveNow.length
  149. if (commit && emailsToRemoveNow.length > 0) {
  150. for (const email of emailsToRemove) {
  151. await UserAuditLogHandler.promises.addEntry(
  152. userId,
  153. 'remove-email',
  154. undefined,
  155. undefined,
  156. {
  157. removedEmail: email,
  158. script: true,
  159. note: 'remove unconfirmed secondary emails',
  160. }
  161. )
  162. }
  163. const updated = await db.users.updateOne(
  164. { _id: new ObjectId(userId) },
  165. { $pull: { emails: { email: { $in: emailsToRemove } } } }
  166. )
  167. totalModifiedUsersCount += updated.modifiedCount
  168. }
  169. processedUsersCount++
  170. if (processedUsersCount % 100 === 0) {
  171. console.log('Processed', processedUsersCount, 'users')
  172. }
  173. }
  174. console.log()
  175. if (!commit) {
  176. console.log('Dry-run, use --commit to apply changes')
  177. console.log('This would be the result:')
  178. console.log()
  179. }
  180. console.log('Total emails in the CSV:', rows.length)
  181. console.log('Total users in the CSV:', userIds.length)
  182. console.log('Total users processed:', processedUsersCount)
  183. console.log('Total emails removed:', removedEmailsCount)
  184. console.log('Skipped emails:', _.sum(Object.values(skippedEmail)))
  185. console.log(' - User not found:', skippedEmail.userNotFound)
  186. console.log(' - Email now confirmed:', skippedEmail.nowConfirmed)
  187. console.log(' - Email now primary:', skippedEmail.nowPrimary)
  188. console.log(' - Email now removed:', skippedEmail.nowRemoved)
  189. console.log()
  190. if (commit) {
  191. console.log('Total users modified:', totalModifiedUsersCount)
  192. } else {
  193. console.log('Note: this was a dry-run. No changes were made.')
  194. }
  195. console.log()
  196. console.timeEnd('consume_csv')
  197. console.log()
  198. }
  199. async function main() {
  200. if (help) {
  201. return usage()
  202. }
  203. if (!generate && !consume) {
  204. console.error('Error: Either --generate or --consume must be specified')
  205. return usage()
  206. }
  207. if (generate && consume) {
  208. console.error('Error: Cannot use both --generate and --consume together')
  209. return usage()
  210. }
  211. if (commit && !consume) {
  212. console.error('Error: --commit can only be used with --consume')
  213. return usage()
  214. }
  215. if (generate) {
  216. await generateCsvFile()
  217. } else if (consume) {
  218. await consumeCsvFile()
  219. }
  220. }
  221. try {
  222. await main()
  223. process.exit(0)
  224. } catch (error) {
  225. console.error(error)
  226. process.exit(1)
  227. }