From c538091fa810c3824c16fedf3cc80edf6179dfd1 Mon Sep 17 00:00:00 2001
From: Domagoj Kriskovic <dom.kriskovic@overleaf.com>
Date: Wed, 2 Jul 2025 11:32:48 +0200
Subject: [PATCH] Add script to check and fix duplicate collaborators in
 projects (#26572)

* [web] Add script to check and fix duplicate collaborators in projects

* use batchedUpdate

* project-id param and BATCH_RANGE_START,

GitOrigin-RevId: 451549eaff255dfae3e55515786d7a68184d557f
---
 .../scripts/check_duplicate_collaborators.mjs | 157 ++++++++++++++++++
 1 file changed, 157 insertions(+)
 create mode 100644 services/web/scripts/check_duplicate_collaborators.mjs

diff --git a/services/web/scripts/check_duplicate_collaborators.mjs b/services/web/scripts/check_duplicate_collaborators.mjs
new file mode 100644
index 0000000000..cf4ba7eafd
--- /dev/null
+++ b/services/web/scripts/check_duplicate_collaborators.mjs
@@ -0,0 +1,157 @@
+#!/usr/bin/env node
+
+/**
+ * Script to check for and optionally fix duplicate collaborators in projects
+ *
+ * A duplicate collaborator is when the same user id appears in multiple collaborator
+ * arrays for the same project (collaberator_refs, readOnly_refs, reviewer_refs, etc.)
+ *
+ * If "--fix" is used, this script will remove users from higher privilege roles and keeps them in lower privilege roles
+ *
+ * Usage:
+ *   node scripts/check_duplicate_collaborators.mjs [--fix] [--project-id=<id>]
+ */
+
+import {
+  batchedUpdate,
+  READ_PREFERENCE_SECONDARY,
+} from '@overleaf/mongo-utils/batchedUpdate.js'
+import { db, ObjectId } from '../app/src/infrastructure/mongodb.js'
+import minimist from 'minimist'
+import { scriptRunner } from './lib/ScriptRunner.mjs'
+
+const args = minimist(process.argv.slice(2), {
+  boolean: ['fix'],
+  string: ['project-id', 'start-date', 'end-date'],
+  default: {
+    fix: false,
+  },
+})
+
+async function fixDuplicateCollaborators(project, trackProgress) {
+  const dryRun = !args.fix
+  const removeCollaboratorRefs = []
+  const removeReviewerRefs = []
+
+  for (const reviewerRef of project.reviewer_refs || []) {
+    if (includesId(project.readOnly_refs, reviewerRef)) {
+      removeReviewerRefs.push(reviewerRef) // remove from reviewer_refs (keep read-only)
+    }
+    if (includesId(project.collaberator_refs, reviewerRef)) {
+      removeCollaboratorRefs.push(reviewerRef) // remove from collaberator_refs (keep reviewer)
+    }
+  }
+
+  if (
+    !dryRun &&
+    (removeCollaboratorRefs.length > 0 || removeReviewerRefs.length > 0)
+  ) {
+    await db.projects.updateOne(
+      { _id: project._id },
+      {
+        $pull: {
+          collaberator_refs: { $in: removeCollaboratorRefs },
+          reviewer_refs: { $in: removeReviewerRefs },
+        },
+      }
+    )
+  }
+
+  const action = args.fix ? 'Removed' : 'Found duplicates in'
+
+  if (removeCollaboratorRefs.length > 0) {
+    trackProgress(
+      `${action} collaborators from project ${project._id}:`,
+      removeCollaboratorRefs
+    )
+  }
+  if (removeReviewerRefs.length > 0) {
+    trackProgress(
+      `${action} reviewers from project ${project._id}:`,
+      removeReviewerRefs
+    )
+  }
+}
+
+async function main(trackProgress) {
+  if (!args['start-date'] && !args['project-id']) {
+    console.error(
+      'Please provide either --start-date or --project-id argument.'
+    )
+    process.exit(1)
+  }
+
+  if (args['project-id']) {
+    const projectId = new ObjectId(args['project-id'])
+    const project = await db.projects.findOne(
+      { _id: projectId },
+      {
+        readPreference: READ_PREFERENCE_SECONDARY,
+        projection: {
+          _id: 1,
+          collaberator_refs: 1,
+          readOnly_refs: 1,
+          reviewer_refs: 1,
+        },
+      }
+    )
+
+    if (!project) {
+      console.error(`Project with id ${projectId} not found`)
+      process.exit(1)
+    }
+
+    await fixDuplicateCollaborators(project, trackProgress)
+
+    return
+  }
+
+  let projectsProcessed = 0
+  await batchedUpdate(
+    db.projects,
+    {
+      reviewer_refs: { $ne: [] },
+      $or: [{ readOnly_refs: { $ne: [] } }, { collaberator_refs: { $ne: [] } }],
+    },
+    /**
+     * @param {Array<Project>} projects
+     * @return {Promise<void>}
+     */
+    async function projects(projects) {
+      for (const project of projects) {
+        projectsProcessed += 1
+        if (projectsProcessed % 10000 === 0) {
+          console.log(projectsProcessed, 'projects processed')
+        }
+
+        await fixDuplicateCollaborators(project, trackProgress)
+      }
+    },
+    {
+      _id: 1,
+      collaberator_refs: 1,
+      readOnly_refs: 1,
+      reviewer_refs: 1,
+    },
+    undefined,
+    {
+      trackProgress,
+      BATCH_RANGE_START: new Date(args['start-date']).toISOString(),
+      BATCH_RANGE_END: args['end-date']
+        ? new Date(args['end-date']).toISOString()
+        : new Date().toISOString(),
+    }
+  )
+}
+
+function includesId(array, id) {
+  return array?.some(item => item.toString() === id.toString())
+}
+
+try {
+  await scriptRunner(main)
+  process.exit()
+} catch (error) {
+  console.error(error)
+  process.exit(1)
+}