delete-orphaned-docs.mjs 2.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125
  1. import fs from 'fs'
  2. import minimist from 'minimist'
  3. import readline from 'readline'
  4. import {
  5. db,
  6. ObjectId,
  7. waitForDb,
  8. } from '../../app/src/infrastructure/mongodb.js'
  9. import DocstoreManagerModule from '../../app/src/Features/Docstore/DocstoreManager.js'
  10. const { promises: DocstoreManager } = DocstoreManagerModule
  11. const argv = minimist(process.argv.slice(2))
  12. const commit = argv.commit !== undefined
  13. const offset = parseInt(argv.offset) || 0
  14. const limit = parseInt(argv.limit) || 0
  15. if (!commit) {
  16. console.log('DOING DRY RUN. TO SAVE CHANGES PASS --commit')
  17. }
  18. const input = fs.createReadStream(argv._[0])
  19. const rl = readline.createInterface({
  20. crlfDelay: Infinity,
  21. input,
  22. })
  23. const orphanedDocs = {}
  24. console.log('Loading Data')
  25. let idx = 0
  26. let processed = 0
  27. rl.on('line', async line => {
  28. if (offset && idx++ < offset) {
  29. return
  30. }
  31. if (limit && processed++ >= limit) {
  32. return
  33. }
  34. let [docId, projectId] = line.split(',')
  35. docId = docId.replace(/^ObjectId\(/, '').replace(/\)$/, '')
  36. projectId = projectId.replace(/^ObjectId\(/, '').replace(/\)$/, '')
  37. try {
  38. docId = new ObjectId(docId).toString()
  39. projectId = new ObjectId(projectId).toString()
  40. } catch (err) {
  41. console.error(`Invalid id: ${docId}, ${projectId}`)
  42. return
  43. }
  44. if (!orphanedDocs[projectId]) {
  45. orphanedDocs[projectId] = []
  46. }
  47. orphanedDocs[projectId].push(docId)
  48. })
  49. rl.on('close', async () => {
  50. const docCount = Object.values(orphanedDocs).reduce((i, v) => i + v.length, 0)
  51. const projectCount = Object.keys(orphanedDocs).length
  52. console.log(`Loaded Data for ${docCount} docs in ${projectCount} Projects`)
  53. await waitForDb()
  54. for (const projectId of Object.keys(orphanedDocs)) {
  55. await deleteOrphanedDocs(projectId, orphanedDocs[projectId])
  56. }
  57. console.log('DONE')
  58. process.exit()
  59. })
  60. async function deleteOrphanedDocs(projectId, docIds) {
  61. try {
  62. if (await projectIdExists(projectId)) {
  63. console.error(`Project id exists: ${projectId}`)
  64. return
  65. }
  66. } catch (err) {
  67. console.error(`Error checking if project exists: ${projectId}`, err.stack)
  68. return
  69. }
  70. console.log(`Delete docs ${docIds.join(', ')} for project ${projectId}`)
  71. if (!commit) {
  72. return
  73. }
  74. try {
  75. await DocstoreManager.destroyProject(projectId)
  76. } catch (err) {
  77. console.error(`Error deleting project ${projectId}`, err)
  78. }
  79. }
  80. async function projectIdExists(projectId) {
  81. // check both projects and deletedProjects to see if project id exists
  82. const [project, deletedProject] = await Promise.all([
  83. findProject(projectId),
  84. findDeletedProject(projectId),
  85. ])
  86. return project !== null || deletedProject !== null
  87. }
  88. async function findProject(projectId) {
  89. return db.projects.findOne(
  90. { _id: new ObjectId(projectId) },
  91. { projection: { _id: 1 } }
  92. )
  93. }
  94. async function findDeletedProject(projectId) {
  95. return db.deletedProjects.findOne(
  96. { 'project._id': new ObjectId(projectId) },
  97. { projection: { _id: 1 } }
  98. )
  99. }