From 18ecbf37e62cf24d21205eefcffb79ead50472b0 Mon Sep 17 00:00:00 2001
From: Brian Gough <briangough@users.noreply.github.com>
Date: Thu, 25 Nov 2021 14:08:06 +0000
Subject: [PATCH] Merge pull request #5916 from
 overleaf/bg-fix-missing-project-ids

add script to fix missing project ids

GitOrigin-RevId: 5b98705859fa041816d5fb1ad431e868b0e81457
---
 .../scripts/fix_docs_with_missing_project.js  | 102 ++++++++++++++++++
 1 file changed, 102 insertions(+)
 create mode 100644 services/document-updater/scripts/fix_docs_with_missing_project.js

diff --git a/services/document-updater/scripts/fix_docs_with_missing_project.js b/services/document-updater/scripts/fix_docs_with_missing_project.js
new file mode 100644
index 0000000000..c88e71cb01
--- /dev/null
+++ b/services/document-updater/scripts/fix_docs_with_missing_project.js
@@ -0,0 +1,102 @@
+const Settings = require('@overleaf/settings')
+const logger = require('@overleaf/logger')
+const rclient = require('@overleaf/redis-wrapper').createClient(
+  Settings.redis.documentupdater
+)
+const keys = Settings.redis.documentupdater.key_schema
+const ProjectFlusher = require('app/js/ProjectFlusher')
+const DocumentManager = require('app/js/DocumentManager')
+const util = require('util')
+const flushAndDeleteDocWithLock = util.promisify(
+  DocumentManager.flushAndDeleteDocWithLock
+)
+
+const { MongoClient, ObjectId } = require('mongodb')
+
+const clientPromise = MongoClient.connect(
+  Settings.mongo.url,
+  Settings.mongo.options
+)
+
+const db = {}
+clientPromise.then(client => {
+  db.docs = client.db().collection('docs')
+})
+
+async function fixDocsWithMissingProjectIds(dockeys, options) {
+  const docIds = ProjectFlusher._extractIds(dockeys)
+  for (const docId of docIds) {
+    const projectId = await rclient.get(keys.projectKey({ doc_id: docId }))
+    logger.debug({ docId, projectId }, 'checking doc')
+    if (!projectId) {
+      try {
+        await insertMissingProjectId(docId, options)
+      } catch (err) {
+        logger.error({ docId, err }, 'error fixing doc without project id')
+      }
+    }
+  }
+}
+
+async function insertMissingProjectId(docId, options) {
+  const doc = await db.docs.findOne({ _id: ObjectId(docId) })
+  if (!doc) {
+    logger.warn({ docId }, 'doc not found in mongo')
+    return
+  }
+  if (!doc.project_id) {
+    logger.error({ docId }, 'doc does not have project id in mongo')
+    return
+  }
+  logger.debug({ docId, doc }, 'found doc')
+  const projectIdFromMongo = doc.project_id.toString()
+  if (options.dryRun) {
+    logger.info(
+      { projectIdFromMongo, docId },
+      'dry run mode - would insert project id in redis'
+    )
+    return
+  }
+  // set the project id for this doc
+  await rclient.set(keys.projectKey({ doc_id: docId }), projectIdFromMongo)
+  logger.debug({ docId, projectIdFromMongo }, 'inserted project id in redis')
+  if (projectIdFromMongo) {
+    await flushAndDeleteDocWithLock(projectIdFromMongo, docId, {})
+    logger.info(
+      { docId, projectIdFromMongo },
+      'fixed doc with empty project id'
+    )
+  }
+  return projectIdFromMongo
+}
+
+async function findAndProcessDocs(options) {
+  logger.info({ options }, 'fixing docs with missing projcct id')
+  let cursor = 0
+  do {
+    const [newCursor, doclinesKeys] = await rclient.scan(
+      cursor,
+      'MATCH',
+      keys.docLines({ doc_id: '*' }),
+      'COUNT',
+      options.limit
+    )
+    await fixDocsWithMissingProjectIds(doclinesKeys, options)
+    cursor = newCursor
+  } while (cursor !== '0')
+}
+
+clientPromise.then(client => {
+  findAndProcessDocs({ limit: 1000, dryRun: process.env.DRY_RUN !== 'false' })
+    .then(result => {
+      rclient.quit()
+      client.close()
+      console.log('DONE')
+    })
+    .catch(function (error) {
+      if (error) {
+        throw error
+      }
+      return process.exit()
+    })
+})