Merge pull request #5917 from overleaf/bg-clean-up-pathname-script

[document-updater] clean up empty pathname script

GitOrigin-RevId: 6d7e47081de892e96ba93ca273662ec4397df796

services/document-updater/scripts/delete_docs_with_empty_pathnames.js → services/document-updater/scripts/fix_docs_with_empty_pathnames.js

@@ -6,25 +6,15 @@ const rclient = require('@overleaf/redis-wrapper').createClient(
 const keys = Settings.redis.documentupdater.key_schema
 const ProjectFlusher = require('app/js/ProjectFlusher')
 const DocumentManager = require('app/js/DocumentManager')
-// const RedisManager = require('app/js/RedisManager')
 const util = require('util')
 const flushAndDeleteDocWithLock = util.promisify(
   DocumentManager.flushAndDeleteDocWithLock
 )
-// const getDoc = util.promisify((projectId, docId, cb) => {
-//   RedisManager.getDoc(projectId, docId, (err, ...results) => {
-//     cb(err, results)
-//   })
-// })
-
-// const removeDocFromMemory = util.promisify(RedisManager.removeDocFromMemory)
 
 async function flushAndDeleteDocs(dockeys, options) {
-  logger.debug({ dockeys }, 'trying keys')
   const docIds = ProjectFlusher._extractIds(dockeys)
   for (const docId of docIds) {
     const pathname = await rclient.get(keys.pathname({ doc_id: docId }))
-    logger.debug({ docId, pathname }, 'checking doc')
     if (!pathname) {
       const projectId = await rclient.get(keys.projectKey({ doc_id: docId }))
       if (!projectId) {
@@ -40,54 +30,6 @@ async function flushAndDeleteDocs(dockeys, options) {
   }
 }
 
-// async function deleteDanglingDoc(projectId, docId, pathname, options) {
-//   const [
-//     // eslint-disable-next-line no-unused-vars
-//     _docLines,
-//     version,
-//     // eslint-disable-next-line no-unused-vars
-//     _ranges,
-//     // eslint-disable-next-line no-unused-vars
-//     _pathname,
-//     projectHistoryId,
-//     unflushedTime,
-//     lastUpdatedAt,
-//     lastUpdatedBy,
-//   ] = await getDoc(projectId, docId)
-//   logger.warn(
-//     {
-//       projectId,
-//       docId,
-//       pathname,
-//       version,
-//       projectHistoryId,
-//       unflushedTime,
-//       lastUpdatedAt,
-//       lastUpdatedBy,
-//     },
-//     'missing project id'
-//   )
-//   if (options.dryRun) {
-//     logger.info(
-//       { projectId, docId, pathname },
-//       'dry run mode - would delete doc with empty pathname and project id'
-//     )
-//     return
-//   }
-//   logger.info(
-//     { projectId, docId, pathname },
-//     'deleting doc with empty pathname and project id'
-//   )
-//   try {
-//     await removeDocFromMemory(projectId, docId)
-//   } catch (err) {
-//     logger.error(
-//       { projectId, docId, pathname, err },
-//       'error deleting doc with empty pathname and project id'
-//     )
-//   }
-// }
-
 async function flushAndDeleteDoc(projectId, docId, pathname, options) {
   if (options.dryRun) {
     logger.info(
@@ -121,14 +63,14 @@ async function cleanUpDocs(options) {
       'COUNT',
       options.limit
     )
-    logger.debug({ count: doclinesKeys.length }, 'found docs')
     await flushAndDeleteDocs(doclinesKeys, options)
     cursor = newCursor
   } while (cursor !== '0')
 }
 
-cleanUpDocs({ limit: 1000, dryRun: true })
+cleanUpDocs({ limit: 1000, dryRun: process.env.DRY_RUN !== 'false' })
   .then(result => {
+    rclient.quit()
     console.log('DONE')
   })
   .catch(function (error) {