delete-orphaned-docs.mjs 2.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119
  1. import fs from 'fs'
  2. import minimist from 'minimist'
  3. import readline from 'readline'
  4. import { db, ObjectId } from '../../app/src/infrastructure/mongodb.js'
  5. import DocstoreManagerModule from '../../app/src/Features/Docstore/DocstoreManager.js'
  6. const { promises: DocstoreManager } = DocstoreManagerModule
  7. const argv = minimist(process.argv.slice(2))
  8. const commit = argv.commit !== undefined
  9. const offset = parseInt(argv.offset) || 0
  10. const limit = parseInt(argv.limit) || 0
  11. if (!commit) {
  12. console.log('DOING DRY RUN. TO SAVE CHANGES PASS --commit')
  13. }
  14. const input = fs.createReadStream(argv._[0])
  15. const rl = readline.createInterface({
  16. crlfDelay: Infinity,
  17. input,
  18. })
  19. const orphanedDocs = {}
  20. console.log('Loading Data')
  21. let idx = 0
  22. let processed = 0
  23. rl.on('line', async line => {
  24. if (offset && idx++ < offset) {
  25. return
  26. }
  27. if (limit && processed++ >= limit) {
  28. return
  29. }
  30. let [docId, projectId] = line.split(',')
  31. docId = docId.replace(/^ObjectId\(/, '').replace(/\)$/, '')
  32. projectId = projectId.replace(/^ObjectId\(/, '').replace(/\)$/, '')
  33. try {
  34. docId = new ObjectId(docId).toString()
  35. projectId = new ObjectId(projectId).toString()
  36. } catch (err) {
  37. console.error(`Invalid id: ${docId}, ${projectId}`)
  38. return
  39. }
  40. if (!orphanedDocs[projectId]) {
  41. orphanedDocs[projectId] = []
  42. }
  43. orphanedDocs[projectId].push(docId)
  44. })
  45. rl.on('close', async () => {
  46. const docCount = Object.values(orphanedDocs).reduce((i, v) => i + v.length, 0)
  47. const projectCount = Object.keys(orphanedDocs).length
  48. console.log(`Loaded Data for ${docCount} docs in ${projectCount} Projects`)
  49. for (const projectId of Object.keys(orphanedDocs)) {
  50. await deleteOrphanedDocs(projectId, orphanedDocs[projectId])
  51. }
  52. console.log('DONE')
  53. process.exit()
  54. })
  55. async function deleteOrphanedDocs(projectId, docIds) {
  56. try {
  57. if (await projectIdExists(projectId)) {
  58. console.error(`Project id exists: ${projectId}`)
  59. return
  60. }
  61. } catch (err) {
  62. console.error(`Error checking if project exists: ${projectId}`, err.stack)
  63. return
  64. }
  65. console.log(`Delete docs ${docIds.join(', ')} for project ${projectId}`)
  66. if (!commit) {
  67. return
  68. }
  69. try {
  70. await DocstoreManager.destroyProject(projectId)
  71. } catch (err) {
  72. console.error(`Error deleting project ${projectId}`, err)
  73. }
  74. }
  75. async function projectIdExists(projectId) {
  76. // check both projects and deletedProjects to see if project id exists
  77. const [project, deletedProject] = await Promise.all([
  78. findProject(projectId),
  79. findDeletedProject(projectId),
  80. ])
  81. return project !== null || deletedProject !== null
  82. }
  83. async function findProject(projectId) {
  84. return db.projects.findOne(
  85. { _id: new ObjectId(projectId) },
  86. { projection: { _id: 1 } }
  87. )
  88. }
  89. async function findDeletedProject(projectId) {
  90. return db.deletedProjects.findOne(
  91. { 'project._id': new ObjectId(projectId) },
  92. { projection: { _id: 1 } }
  93. )
  94. }