overleaf/services/web/scripts/delete_dangling_file_refs.js
Brian Gough 5a3318f5b3 Merge pull request #8701 from overleaf/bg-simple-iterable-paths
simple iterable paths

GitOrigin-RevId: f6906016888ccfc95c88858bdac4d2633fc1c5f4
2022-07-05 08:03:43 +00:00

129 lines
3.3 KiB
JavaScript

/**
* This script deletes dangling doc and file refs in projects
*/
const minimist = require('minimist')
const { ObjectId } = require('mongodb')
const { db, waitForDb } = require('../app/src/infrastructure/mongodb')
const Errors = require('../app/src/Features/Errors/Errors')
const FileStoreHandler = require('../app/src/Features/FileStore/FileStoreHandler')
const ProjectEntityMongoUpdateHandler = require('../app/src/Features/Project/ProjectEntityMongoUpdateHandler')
const { iterablePaths } = require('../app/src/Features/Project/IterablePath')
const OPTIONS = parseArgs()
function parseArgs() {
const argv = minimist(process.argv.slice(2), {
boolean: ['dry-run'],
default: { 'dry-run': true },
})
const dryRun = argv['dry-run']
const projectIds = argv._
if (projectIds.length === 0) {
console.log(`Usage: ${process.argv[1]} [--no-dry-run] PROJECT_ID ...`)
process.exit(0)
}
return { projectIds, dryRun }
}
async function main() {
await waitForDb()
const projects = await getProjects()
for (const project of projects) {
await processProject(project)
}
if (OPTIONS.dryRun) {
console.log(
'\nThis was a dry run. Re-run with --no-dry-run to delete broken refs.'
)
}
}
async function getProjects() {
const projectIds = OPTIONS.projectIds.map(id => ObjectId(id))
const projects = await db.projects
.find(
{ _id: { $in: projectIds } },
{ projection: { _id: 1, rootFolder: 1 } }
)
.toArray()
return projects
}
async function processProject(project) {
console.log(`Processing project ${project._id}`)
const { docIds, fileIds } = findRefsInFolder(project.rootFolder[0])
for (const docId of docIds) {
if (!(await docExists(docId))) {
await deleteDoc(project._id, docId)
}
}
for (const fileId of fileIds) {
if (!(await fileExists(project._id, fileId))) {
await deleteFile(project._id, fileId)
}
}
}
function findRefsInFolder(folder) {
let docIds = folder.docs.map(doc => doc._id)
let fileIds = folder.fileRefs.map(file => file._id)
for (const subfolder of iterablePaths(folder, 'folders')) {
const subrefs = findRefsInFolder(subfolder)
docIds = docIds.concat(subrefs.docIds)
fileIds = fileIds.concat(subrefs.fileIds)
}
return { docIds, fileIds }
}
async function docExists(docId) {
const doc = await db.docs.findOne({ _id: docId })
return doc != null
}
async function fileExists(projectId, fileId) {
try {
// Getting the file size to avoid downloading the whole file
await FileStoreHandler.promises.getFileSize(projectId, fileId)
} catch (err) {
if (err instanceof Errors.NotFoundError) {
return false
}
throw err
}
return true
}
async function deleteDoc(projectId, docId) {
console.log(` * Deleting bad doc ref ${docId}`)
if (!OPTIONS.dryRun) {
await ProjectEntityMongoUpdateHandler.promises.deleteEntity(
projectId,
docId,
'doc'
)
}
}
async function deleteFile(projectId, fileId) {
console.log(` * Deleting bad file ref ${fileId}`)
if (!OPTIONS.dryRun) {
await ProjectEntityMongoUpdateHandler.promises.deleteEntity(
projectId,
fileId,
'file'
)
}
}
main()
.then(() => {
process.exit(0)
})
.catch(error => {
console.error({ error })
process.exit(1)
})