back_fill_dummy_doc_meta.mjs 4.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162
  1. import { promisify } from 'node:util'
  2. import mongodb from 'mongodb-legacy'
  3. import {
  4. db,
  5. READ_PREFERENCE_SECONDARY,
  6. } from '../app/src/infrastructure/mongodb.js'
  7. import _ from 'lodash'
  8. import LRUCache from 'lru-cache'
  9. import { fileURLToPath } from 'node:url'
  10. import { scriptRunner } from './lib/ScriptRunner.mjs'
  11. const { ObjectId } = mongodb
  12. const sleep = promisify(setTimeout)
  13. const NOW_IN_S = Date.now() / 1000
  14. const ONE_WEEK_IN_S = 60 * 60 * 24 * 7
  15. const TEN_SECONDS = 10 * 1000
  16. const DUMMY_NAME = 'unknown.tex'
  17. const DUMMY_TIME = new Date('2021-04-12T00:00:00.000Z')
  18. let deletedProjectsCache = null
  19. function getSecondsFromObjectId(id) {
  20. return id.getTimestamp().getTime() / 1000
  21. }
  22. async function main(options) {
  23. if (!options) {
  24. options = {}
  25. }
  26. _.defaults(options, {
  27. dryRun: process.env.DRY_RUN === 'true',
  28. cacheSize: parseInt(process.env.CACHE_SIZE, 10) || 100,
  29. firstProjectId: new ObjectId(process.env.FIRST_PROJECT_ID),
  30. incrementByS: parseInt(process.env.INCREMENT_BY_S, 10) || ONE_WEEK_IN_S,
  31. batchSize: parseInt(process.env.BATCH_SIZE, 10) || 1000,
  32. stopAtS: parseInt(process.env.STOP_AT_S, 10) || NOW_IN_S,
  33. letUserDoubleCheckInputsFor:
  34. parseInt(process.env.LET_USER_DOUBLE_CHECK_INPUTS_FOR, 10) || TEN_SECONDS,
  35. })
  36. if (!options.firstProjectId) {
  37. console.error('Set FIRST_PROJECT_ID and re-run.')
  38. process.exit(1)
  39. }
  40. deletedProjectsCache = new LRUCache({
  41. max: options.cacheSize,
  42. })
  43. await letUserDoubleCheckInputs(options)
  44. let startId = options.firstProjectId
  45. let nProcessed = 0
  46. while (getSecondsFromObjectId(startId) <= options.stopAtS) {
  47. const end = getSecondsFromObjectId(startId) + options.incrementByS
  48. let endId = ObjectId.createFromTime(end)
  49. const query = {
  50. project_id: {
  51. // include edge
  52. $gte: startId,
  53. // exclude edge
  54. $lt: endId,
  55. },
  56. deleted: true,
  57. name: {
  58. $exists: false,
  59. },
  60. }
  61. const docs = await db.docs
  62. .find(query, { readPreference: READ_PREFERENCE_SECONDARY })
  63. .project({ _id: 1, project_id: 1 })
  64. .limit(options.batchSize)
  65. .toArray()
  66. if (docs.length) {
  67. const docIds = docs.map(doc => doc._id)
  68. console.log('Back filling dummy meta data for', JSON.stringify(docIds))
  69. await processBatch(docs, options)
  70. nProcessed += docIds.length
  71. if (docs.length === options.batchSize) {
  72. endId = docs[docs.length - 1].project_id
  73. }
  74. }
  75. console.error('Processed %d until %s', nProcessed, endId)
  76. startId = endId
  77. }
  78. }
  79. async function getDeletedProject(projectId) {
  80. const cacheKey = projectId.toString()
  81. if (deletedProjectsCache.has(cacheKey)) {
  82. return deletedProjectsCache.get(cacheKey)
  83. }
  84. const deletedProject = await db.deletedProjects.findOne(
  85. { 'deleterData.deletedProjectId': projectId },
  86. {
  87. projection: {
  88. _id: 1,
  89. 'project.deletedDocs': 1,
  90. },
  91. }
  92. )
  93. deletedProjectsCache.set(cacheKey, deletedProject)
  94. return deletedProject
  95. }
  96. async function processBatch(docs, options) {
  97. for (const doc of docs) {
  98. const { _id: docId, project_id: projectId } = doc
  99. const deletedProject = await getDeletedProject(projectId)
  100. let name = DUMMY_NAME
  101. let deletedAt = DUMMY_TIME
  102. if (deletedProject) {
  103. const project = deletedProject.project
  104. if (project) {
  105. const deletedDoc =
  106. project.deletedDocs &&
  107. project.deletedDocs.find(deletedDoc => docId.equals(deletedDoc._id))
  108. if (deletedDoc) {
  109. console.log('Found deletedDoc for %s', docId)
  110. name = deletedDoc.name
  111. deletedAt = deletedDoc.deletedAt
  112. } else {
  113. console.log('Missing deletedDoc for %s', docId)
  114. }
  115. } else {
  116. console.log('Orphaned deleted doc %s (failed hard deletion)', docId)
  117. }
  118. } else {
  119. console.log('Orphaned deleted doc %s (no deletedProjects entry)', docId)
  120. }
  121. if (options.dryRun) continue
  122. await db.docs.updateOne({ _id: docId }, { $set: { name, deletedAt } })
  123. }
  124. }
  125. async function letUserDoubleCheckInputs(options) {
  126. console.error('Options:', JSON.stringify(options, null, 2))
  127. console.error(
  128. 'Waiting for you to double check inputs for',
  129. options.letUserDoubleCheckInputsFor,
  130. 'ms'
  131. )
  132. await sleep(options.letUserDoubleCheckInputsFor)
  133. }
  134. export default main
  135. if (fileURLToPath(import.meta.url) === process.argv[1]) {
  136. try {
  137. await scriptRunner(async () => await main())
  138. console.error('Done.')
  139. process.exit(0)
  140. } catch (error) {
  141. console.error({ error })
  142. process.exit(1)
  143. }
  144. }