Add script to remove emails with commas, taken from a CSV file (#25107)

* Add script to remove emails with commas and replace them emails to support with encoded former emails

* Enhance RemoveEmailsWithCommasScriptTest to verify unchanged user data

GitOrigin-RevId: 6961995f2a143ac1c53bc2eeb183808a4be7dd02

services/web/scripts/remove_emails_with_commas.mjs

@@ -0,0 +1,124 @@
+// @ts-check
+
+import minimist from 'minimist'
+import fs from 'node:fs/promises'
+import * as csv from 'csv'
+import { promisify } from 'node:util'
+import UserAuditLogHandler from '../app/src/Features/User/UserAuditLogHandler.js'
+import { db } from '../app/src/infrastructure/mongodb.js'
+
+const CSV_FILENAME = '/tmp/emails-with-commas.csv'
+
+/**
+ * @type {(csvString: string) => Promise<string[][]>}
+ */
+const parseAsync = promisify(csv.parse)
+
+function usage() {
+  console.log('Usage: node remove_emails_with_commas.mjs')
+  console.log(`Read emails from ${CSV_FILENAME} and remove them from users.`)
+  console.log('Add support+<encoded_email>@overleaf.com instead.')
+  console.log('Options:')
+  console.log('  --commit       apply the changes\n')
+  process.exit(0)
+}
+
+const { commit, help } = minimist(process.argv.slice(2), {
+  boolean: ['commit', 'help'],
+  alias: { help: 'h' },
+  default: { commit: false },
+})
+
+async function consumeCsvFileAndUpdate() {
+  console.time('remove_emails_with_commas')
+
+  const csvContent = await fs.readFile(CSV_FILENAME, 'utf8')
+  const rows = await parseAsync(csvContent)
+  const emailsWithComma = rows.map(row => row[0])
+
+  console.log('Total emails in the CSV:', emailsWithComma.length)
+
+  const unexpectedValidEmails = emailsWithComma.filter(
+    str => !str.includes(',')
+  )
+  if (unexpectedValidEmails.length > 0) {
+    throw new Error(
+      'CSV file contains unexpected valid emails: ' +
+        JSON.stringify(emailsWithComma)
+    )
+  }
+
+  let updatedUsersCount = 0
+  for (const oldEmail of emailsWithComma) {
+    const encodedEmail = oldEmail
+      .replaceAll('_', '_5f')
+      .replaceAll('@', '_40')
+      .replaceAll(',', '_2c')
+      .replaceAll('<', '_60')
+      .replaceAll('>', '_62')
+
+    const newEmail = `support+${encodedEmail}@overleaf.com`
+
+    console.log(oldEmail, '->', newEmail)
+
+    const user = await db.users.findOne({ email: oldEmail })
+
+    if (!user) {
+      console.log('User not found for email:', oldEmail)
+      continue
+    }
+
+    if (commit) {
+      await db.users.updateOne(
+        { _id: user._id },
+        {
+          $set: { email: newEmail },
+          $pull: { emails: { email: oldEmail } },
+        }
+      )
+      await db.users.updateOne(
+        { _id: user._id },
+        {
+          $addToSet: {
+            emails: {
+              email: newEmail,
+              createdAt: Date.now(),
+              reversedHostname: 'moc.faelrevo',
+            },
+          },
+        }
+      )
+
+      await UserAuditLogHandler.promises.addEntry(
+        user._id,
+        'remove-email',
+        undefined,
+        undefined,
+        {
+          removedEmail: oldEmail,
+          script: true,
+          note: 'remove primary email containing commas',
+        }
+      )
+      updatedUsersCount++
+    }
+  }
+
+  console.log('Updated users:', updatedUsersCount)
+
+  if (!commit) {
+    console.log('Note: this was a dry-run. No changes were made.')
+  }
+  console.log()
+  console.timeEnd('remove_emails_with_commas')
+  console.log()
+}
+
+try {
+  if (help) usage()
+  else await consumeCsvFileAndUpdate()
+  process.exit(0)
+} catch (error) {
+  console.error(error)
+  process.exit(1)
+}

services/web/test/acceptance/src/RemoveEmailsWithCommasScriptTest.mjs

@@ -0,0 +1,226 @@
+import { promisify } from 'node:util'
+import { exec } from 'node:child_process'
+import { expect } from 'chai'
+import { filterOutput } from './helpers/settings.mjs'
+import { db, ObjectId } from '../../../app/src/infrastructure/mongodb.js'
+import fs from 'node:fs/promises'
+
+const CSV_FILENAME = '/tmp/emails-with-commas.csv'
+
+async function runScript(commit) {
+  const result = await promisify(exec)(
+    ['node', 'scripts/remove_emails_with_commas.mjs', commit && '--commit']
+      .filter(Boolean)
+      .join(' ')
+  )
+  return {
+    ...result,
+    stdout: result.stdout.split('\n').filter(filterOutput),
+  }
+}
+
+function createUser(email, emails) {
+  return {
+    _id: new ObjectId(),
+    email,
+    emails,
+  }
+}
+
+describe('scripts/remove_emails_with_commas', function () {
+  let user, unchangedUser
+
+  beforeEach(async function () {
+    await fs.writeFile(
+      CSV_FILENAME,
+      '"user,email@test.com"\n"user,another@test.com"\n'
+    )
+  })
+
+  afterEach(async function () {
+    try {
+      await fs.unlink(CSV_FILENAME)
+    } catch (err) {
+      // Ignore errors if file doesn't exist
+    }
+  })
+
+  describe('when removing email addresses with commas', function () {
+    beforeEach(async function () {
+      user = createUser('user,email@test.com', [
+        {
+          email: 'user,email@test.com',
+          createdAt: new Date(),
+          reversedHostname: 'moc.tset',
+        },
+      ])
+      await db.users.insertOne(user)
+
+      unchangedUser = createUser('john.doe@example.com', [
+        {
+          email: 'john.doe@example.com',
+          createdAt: new Date(),
+          reversedHostname: 'moc.elpmaxe',
+        },
+      ])
+      await db.users.insertOne(unchangedUser)
+    })
+
+    afterEach(async function () {
+      await db.users.deleteOne({ _id: user._id })
+    })
+
+    it('should replace emails with commas with encoded support emails', async function () {
+      const r = await runScript(true)
+
+      expect(r.stdout).to.include(
+        'user,email@test.com -> support+user_2cemail_40test.com@overleaf.com'
+      )
+      expect(r.stdout).to.include('Updated users: 1')
+
+      const updatedUser = await db.users.findOne({ _id: user._id })
+      expect(updatedUser.email).to.equal(
+        'support+user_2cemail_40test.com@overleaf.com'
+      )
+      expect(updatedUser.emails).to.have.length(1)
+      expect(updatedUser.emails[0].email).to.equal(
+        'support+user_2cemail_40test.com@overleaf.com'
+      )
+      expect(updatedUser.emails[0].reversedHostname).to.equal('moc.faelrevo')
+
+      const unchanged = await db.users.findOne({ _id: unchangedUser._id })
+
+      expect(unchanged.emails).to.have.length(1)
+      expect(unchanged.email).to.equal('john.doe@example.com')
+      expect(unchanged.emails[0].email).to.equal('john.doe@example.com')
+    })
+
+    it('should not modify anything in dry run mode', async function () {
+      const r = await runScript(false)
+
+      expect(r.stdout).to.include(
+        'user,email@test.com -> support+user_2cemail_40test.com@overleaf.com'
+      )
+      expect(r.stdout).to.include(
+        'Note: this was a dry-run. No changes were made.'
+      )
+
+      const updatedUser = await db.users.findOne({ _id: user._id })
+      expect(updatedUser.email).to.equal('user,email@test.com')
+      expect(updatedUser.emails).to.have.length(1)
+      expect(updatedUser.emails[0].email).to.equal('user,email@test.com')
+    })
+  })
+
+  describe('when handling multiple email replacements', function () {
+    beforeEach(async function () {
+      user = createUser('user,email@test.com', [
+        {
+          email: 'user,email@test.com',
+          createdAt: new Date(),
+          reversedHostname: 'moc.tset',
+        },
+        {
+          email: 'normal@test.com',
+          createdAt: new Date(),
+          reversedHostname: 'moc.tset',
+        },
+      ])
+      await db.users.insertOne(user)
+    })
+
+    afterEach(async function () {
+      await db.users.deleteOne({ _id: user._id })
+    })
+
+    it('should only replace primary email with comma and keep other emails', async function () {
+      const r = await runScript(true)
+
+      expect(r.stdout).to.include(
+        'user,email@test.com -> support+user_2cemail_40test.com@overleaf.com'
+      )
+      expect(r.stdout).to.include('Updated users: 1')
+
+      const updatedUser = await db.users.findOne({ _id: user._id })
+      expect(updatedUser.email).to.equal(
+        'support+user_2cemail_40test.com@overleaf.com'
+      )
+      expect(updatedUser.emails).to.have.length(2)
+      expect(updatedUser.emails[0].email).to.equal('normal@test.com')
+      expect(updatedUser.emails[1].email).to.equal(
+        'support+user_2cemail_40test.com@overleaf.com'
+      )
+    })
+  })
+
+  describe('when handling special characters in emails', function () {
+    beforeEach(async function () {
+      await fs.writeFile(
+        CSV_FILENAME,
+        '"user,email@test.com"\n",<user@test.com>"\n"user_special@test.co,"\n'
+      )
+
+      user = createUser('user,email@test.com', [
+        {
+          email: 'user,email@test.com',
+          createdAt: new Date(),
+          reversedHostname: 'moc.tset',
+        },
+      ])
+
+      await db.users.insertOne(user)
+
+      const user2 = createUser('user<>@test.com', [
+        {
+          email: 'user<>@test.com',
+          createdAt: new Date(),
+          reversedHostname: 'moc.tset',
+        },
+      ])
+
+      await db.users.insertOne(user2)
+    })
+
+    afterEach(async function () {
+      await db.users.deleteMany({
+        email: {
+          $in: [
+            'support+user_2cemail_40test.com@overleaf.com',
+            'support+user_60_62_40test.com@overleaf.com',
+          ],
+        },
+      })
+    })
+
+    it('should correctly encode various special characters', async function () {
+      const r = await runScript(true)
+
+      expect(r.stdout).to.include(
+        'user,email@test.com -> support+user_2cemail_40test.com@overleaf.com'
+      )
+      expect(r.stdout).to.include(
+        ',<user@test.com> -> support+_2c_60user_40test.com_62@overleaf.com'
+      )
+
+      const updatedUser1 = await db.users.findOne({ _id: user._id })
+      expect(updatedUser1.email).to.equal(
+        'support+user_2cemail_40test.com@overleaf.com'
+      )
+    })
+  })
+
+  describe('when user does not exist', function () {
+    beforeEach(async function () {
+      await fs.writeFile(CSV_FILENAME, '"nonexistent,email@test.com"\n')
+    })
+
+    it('should handle missing users gracefully', async function () {
+      const r = await runScript(true)
+
+      expect(r.stdout).to.include(
+        'User not found for email: nonexistent,email@test.com'
+      )
+      expect(r.stdout).to.include('Updated users: 0')
+    })
+  })
+})