Merge pull request #17023 from overleaf/ii-token-access-null-values

[web] Fix projects token access values GitOrigin-RevId: f0c6a4993e42320c06753cb65198138afe55b71a
2024-02-26 09:03:13 +00:00
parent 4a1af0f057
commit b04247dd5a
6 changed files with 425 additions and 5 deletions
@@ -0,0 +1,174 @@
+const { db, waitForDb } = require('../app/src/infrastructure/mongodb')
+const { batchedUpdate } = require('./helpers/batchedUpdate')
+const { ObjectId } = require('mongodb')
+const minimist = require('minimist')
+const CollaboratorsHandler = require('../app/src/Features/Collaborators/CollaboratorsHandler')
+
+const argv = minimist(process.argv.slice(2), {
+  string: ['projects'],
+  boolean: ['dry-run', 'help'],
+  alias: {
+    projects: 'p',
+  },
+  default: {
+    'dry-run': true,
+  },
+})
+
+if (argv.help || argv._.length > 1) {
+  console.error(`Usage: node scripts/remove_deleted_users_from_token_access_refs.js [OPTS]
+    Finds or removes deleted user ids from token access fields
+    "tokenAccessReadOnly_refs" and "tokenAccessReadAndWrite_refs" in the "projects" collection.
+
+    If no projects are specified, all projects will be processed.
+
+    Options:
+
+        --dry-run         finds projects and deleted users but does not do any updates
+        --projects        list of projects ids to be fixed (comma separated)
+    `)
+  process.exit(1)
+}
+
+const DRY_RUN = argv['dry-run']
+const PROJECTS_LIST = argv.projects
+
+async function findUserIds() {
+  const userIds = new Set()
+  const cursor = db.users.find({}, { _id: 1 })
+  for await (const user of cursor) {
+    userIds.add(user._id.toString())
+  }
+  console.log(`=> User ids count: ${userIds.size}`)
+  return userIds
+}
+
+async function fixProjectsWithInvalidTokenAccessRefsIds(
+  DRY_RUN,
+  PROJECTS_LIST
+) {
+  if (DRY_RUN) {
+    console.log('=> Doing dry run')
+  }
+
+  const DELETED_USER_COLLABORATOR_IDS = new Set()
+  const PROJECTS_WITH_DELETED_USER = new Set()
+
+  // get a set of all users ids as an in-memory cache
+  const userIds = await findUserIds()
+
+  // default query for finding all projects with non-empty token access fields
+  let query = {
+    $or: [
+      { 'tokenAccessReadOnly_refs.0': { $exists: true } },
+      { 'tokenAccessReadAndWrite_refs.0': { $exists: true } },
+    ],
+  }
+
+  const projectIds = PROJECTS_LIST?.split(',').map(
+    projectId => new ObjectId(projectId)
+  )
+
+  // query for finding projects passed in as args
+  if (projectIds) {
+    query = { $and: [{ _id: { $in: projectIds } }] }
+  }
+
+  await batchedUpdate(
+    'projects',
+    query,
+    async projects => {
+      for (const project of projects) {
+        // find the set of user ids that are in the token access fields
+        // i.e. the set of collaborators
+        const collaboratorIds = new Set()
+        for (const roUserId of project.tokenAccessReadOnly_refs) {
+          collaboratorIds.add(roUserId.toString())
+        }
+        for (const rwUserId of project.tokenAccessReadAndWrite_refs) {
+          collaboratorIds.add(rwUserId.toString())
+        }
+        // determine which collaborator ids are not in the `users` collection
+        // i.e. the user has been deleted
+        const deletedUserIds = new Set()
+        for (const collaboratorId of collaboratorIds) {
+          if (!userIds.has(collaboratorId)) {
+            deletedUserIds.add(collaboratorId)
+          }
+        }
+
+        // double-check that users doesn't exist in the users collection
+        // we don't want to remove users that were added after the initial query
+        const existingUsersCursor = db.users.find(
+          { _id: { $in: [...deletedUserIds].map(id => new ObjectId(id)) } },
+          { _id: 1 }
+        )
+        for await (const user of existingUsersCursor) {
+          const id = user._id.toString()
+          deletedUserIds.delete(id)
+          // add the user id to the cache
+          userIds.add(id)
+        }
+
+        // remove the actual deleted users
+        for (const deletedUserId of deletedUserIds) {
+          DELETED_USER_COLLABORATOR_IDS.add(deletedUserId)
+          PROJECTS_WITH_DELETED_USER.add(project._id.toString())
+          console.log(
+            '=> Found deleted user id:',
+            deletedUserId,
+            'in project:',
+            project._id.toString()
+          )
+          if (DRY_RUN) {
+            console.log(
+              `=> DRY RUN - would remove deleted ${deletedUserId} from all projects (found in project ${project._id.toString()})`
+            )
+            continue
+          }
+          console.log(
+            `=> Removing deleted ${deletedUserId} from all projects (found in project ${project._id.toString()})`
+          )
+          await CollaboratorsHandler.promises.removeUserFromAllProjects(
+            new ObjectId(deletedUserId)
+          )
+        }
+      }
+    },
+    { tokenAccessReadOnly_refs: 1, tokenAccessReadAndWrite_refs: 1 }
+  )
+
+  console.log(
+    `=> ${DRY_RUN ? 'DRY RUN - would delete' : 'Deleted'} user ids (${
+      DELETED_USER_COLLABORATOR_IDS.size
+    })`
+  )
+  if (DELETED_USER_COLLABORATOR_IDS.size) {
+    console.log(Array.from(DELETED_USER_COLLABORATOR_IDS).join('\n'))
+  }
+  console.log(
+    `=> Projects with deleted user ids (${PROJECTS_WITH_DELETED_USER.size})`
+  )
+  if (PROJECTS_WITH_DELETED_USER.size) {
+    console.log(Array.from(PROJECTS_WITH_DELETED_USER).join('\n'))
+  }
+}
+
+async function main(DRY_RUN, PROJECTS_LIST) {
+  await waitForDb()
+  await fixProjectsWithInvalidTokenAccessRefsIds(DRY_RUN, PROJECTS_LIST)
+}
+
+module.exports = main
+
+if (require.main === module) {
+  main(DRY_RUN, PROJECTS_LIST)
+    .then(() => {
+      console.error('Done')
+      process.exit(0)
+    })
+    .catch(err => {
+      console.error(err)
+      process.exit(1)
+    })
+}