Merge pull request #2963 from overleaf/cmg-archive-state-script

Script to convert old archived state into new array state

GitOrigin-RevId: 64cd3a236dc627f922488e1fa9d8aa2e7df1eb15

services/web/scripts/convert_archived_state.js

@@ -0,0 +1,71 @@
+const _ = require('lodash')
+
+const WRITE_CONCURRENCY = parseInt(process.env.WRITE_CONCURRENCY, 10) || 10
+
+const { batchedUpdate } = require('./helpers/batchedUpdate')
+const { promiseMapWithLimit } = require('../app/src/util/promises')
+
+async function main() {
+  await batchedUpdate(
+    'projects',
+    { archived: false },
+    {
+      $set: { archived: [] },
+      $unset: { trashed: '' } // lest any v1 trashed bits be left behind
+    }
+  )
+
+  console.log('Done, moving to archived projects')
+
+  await batchedUpdate('projects', { archived: true }, performUpdate, {
+    _id: 1,
+    owner_ref: 1,
+    collaberator_refs: 1,
+    readOnly_refs: 1,
+    tokenAccessReadAndWrite_refs: 1,
+    tokenAccessReadOnly_refs: 1
+  })
+}
+
+main()
+  .then(() => {
+    process.exit(0)
+  })
+  .catch(error => {
+    console.error({ error })
+    process.exit(1)
+  })
+
+async function performUpdate(collection, nextBatch) {
+  await promiseMapWithLimit(WRITE_CONCURRENCY, nextBatch, project =>
+    setArchived(collection, project)
+  )
+}
+
+async function setArchived(collection, project) {
+  const archived = calculateArchivedArray(project)
+
+  return collection.updateOne(
+    { _id: project._id },
+    {
+      $set: { archived: archived },
+      $unset: { trashed: '' } // lest any v1 trashed bits be left behind
+    }
+  )
+}
+
+function calculateArchivedArray(project) {
+  return _.unionWith(
+    [project.owner_ref],
+    project.collaberator_refs,
+    project.readOnly_refs,
+    project.tokenAccessReadAndWrite_refs,
+    project.tokenAccessReadOnly_refs,
+    _objectIdEquals
+  )
+}
+
+function _objectIdEquals(firstVal, secondVal) {
+  // For use as a comparator for unionWith
+  return firstVal.toString() === secondVal.toString()
+}

services/web/scripts/helpers/batchedUpdate.js

@@ -8,24 +8,27 @@ if (process.env.BATCH_LAST_ID) {
   BATCH_LAST_ID = ObjectId(process.env.BATCH_LAST_ID)
 }
 
-async function getNextBatch(collection, query, maxId) {
+async function getNextBatch(collection, query, maxId, projection) {
   if (maxId) {
     query['_id'] = { $gt: maxId }
   }
   const entries = await collection
-    .find(query, { _id: 1 })
+    .find(query, projection)
     .sort({ _id: 1 })
     .limit(BATCH_SIZE)
     .setReadPreference(ReadPreference.SECONDARY)
     .toArray()
-  return entries.map(entry => entry._id)
+  return entries
 }
 
 async function performUpdate(collection, nextBatch, update) {
-  return collection.updateMany({ _id: { $in: nextBatch } }, update)
+  return collection.updateMany(
+    { _id: { $in: nextBatch.map(entry => entry._id) } },
+    update
+  )
 }
 
-async function batchedUpdate(collectionName, query, update) {
+async function batchedUpdate(collectionName, query, update, projection) {
   // Apparently the mongo driver returns the connection too early.
   // Some secondary connections are not ready as it returns, leading to
   //  failing cursor actions with a readPreference set to 'secondary'.
@@ -35,14 +38,23 @@ async function batchedUpdate(collectionName, query, update) {
   const db = await getNativeDb()
   const collection = db.collection(collectionName)
 
+  projection = projection || { _id: 1 }
   let nextBatch
   let updated = 0
   let maxId = BATCH_LAST_ID
-  while ((nextBatch = await getNextBatch(collection, query, maxId)).length) {
-    maxId = nextBatch[nextBatch.length - 1]
+  while (
+    (nextBatch = await getNextBatch(collection, query, maxId, projection))
+      .length
+  ) {
+    maxId = nextBatch[nextBatch.length - 1]._id
     updated += nextBatch.length
     console.log(JSON.stringify(nextBatch))
-    await performUpdate(collection, nextBatch, update)
+
+    if (typeof update === 'function') {
+      await update(collection, nextBatch)
+    } else {
+      await performUpdate(collection, nextBatch, update)
+    }
   }
   return updated
 }