back_fill_dummy_doc_meta.mjs 4.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164
  1. import { promisify } from 'util'
  2. import mongodb from 'mongodb-legacy'
  3. import {
  4. db,
  5. waitForDb,
  6. READ_PREFERENCE_SECONDARY,
  7. } from '../app/src/infrastructure/mongodb.js'
  8. import _ from 'lodash'
  9. import LRUCache from 'lru-cache'
  10. import { fileURLToPath } from 'url'
  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. await waitForDb()
  45. let startId = options.firstProjectId
  46. let nProcessed = 0
  47. while (getSecondsFromObjectId(startId) <= options.stopAtS) {
  48. const end = getSecondsFromObjectId(startId) + options.incrementByS
  49. let endId = ObjectId.createFromTime(end)
  50. const query = {
  51. project_id: {
  52. // include edge
  53. $gte: startId,
  54. // exclude edge
  55. $lt: endId,
  56. },
  57. deleted: true,
  58. name: {
  59. $exists: false,
  60. },
  61. }
  62. const docs = await db.docs
  63. .find(query, { readPreference: READ_PREFERENCE_SECONDARY })
  64. .project({ _id: 1, project_id: 1 })
  65. .limit(options.batchSize)
  66. .toArray()
  67. if (docs.length) {
  68. const docIds = docs.map(doc => doc._id)
  69. console.log('Back filling dummy meta data for', JSON.stringify(docIds))
  70. await processBatch(docs, options)
  71. nProcessed += docIds.length
  72. if (docs.length === options.batchSize) {
  73. endId = docs[docs.length - 1].project_id
  74. }
  75. }
  76. console.error('Processed %d until %s', nProcessed, endId)
  77. startId = endId
  78. }
  79. }
  80. async function getDeletedProject(projectId) {
  81. const cacheKey = projectId.toString()
  82. if (deletedProjectsCache.has(cacheKey)) {
  83. return deletedProjectsCache.get(cacheKey)
  84. }
  85. const deletedProject = await db.deletedProjects.findOne(
  86. { 'deleterData.deletedProjectId': projectId },
  87. {
  88. projection: {
  89. _id: 1,
  90. 'project.deletedDocs': 1,
  91. },
  92. }
  93. )
  94. deletedProjectsCache.set(cacheKey, deletedProject)
  95. return deletedProject
  96. }
  97. async function processBatch(docs, options) {
  98. for (const doc of docs) {
  99. const { _id: docId, project_id: projectId } = doc
  100. const deletedProject = await getDeletedProject(projectId)
  101. let name = DUMMY_NAME
  102. let deletedAt = DUMMY_TIME
  103. if (deletedProject) {
  104. const project = deletedProject.project
  105. if (project) {
  106. const deletedDoc =
  107. project.deletedDocs &&
  108. project.deletedDocs.find(deletedDoc => docId.equals(deletedDoc._id))
  109. if (deletedDoc) {
  110. console.log('Found deletedDoc for %s', docId)
  111. name = deletedDoc.name
  112. deletedAt = deletedDoc.deletedAt
  113. } else {
  114. console.log('Missing deletedDoc for %s', docId)
  115. }
  116. } else {
  117. console.log('Orphaned deleted doc %s (failed hard deletion)', docId)
  118. }
  119. } else {
  120. console.log('Orphaned deleted doc %s (no deletedProjects entry)', docId)
  121. }
  122. if (options.dryRun) continue
  123. await db.docs.updateOne({ _id: docId }, { $set: { name, deletedAt } })
  124. }
  125. }
  126. async function letUserDoubleCheckInputs(options) {
  127. console.error('Options:', JSON.stringify(options, null, 2))
  128. console.error(
  129. 'Waiting for you to double check inputs for',
  130. options.letUserDoubleCheckInputsFor,
  131. 'ms'
  132. )
  133. await sleep(options.letUserDoubleCheckInputsFor)
  134. }
  135. export default main
  136. if (fileURLToPath(import.meta.url) === process.argv[1]) {
  137. try {
  138. await main()
  139. console.error('Done.')
  140. process.exit(0)
  141. } catch (error) {
  142. console.error({ error })
  143. process.exit(1)
  144. }
  145. }