DocArchiveManager.js 4.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180
  1. const { callbackify } = require('util')
  2. const MongoManager = require('./MongoManager').promises
  3. const Errors = require('./Errors')
  4. const logger = require('logger-sharelatex')
  5. const settings = require('settings-sharelatex')
  6. const crypto = require('crypto')
  7. const Streamifier = require('streamifier')
  8. const RangeManager = require('./RangeManager')
  9. const PersistorManager = require('./PersistorManager')
  10. const asyncPool = require('tiny-async-pool')
  11. const PARALLEL_JOBS = 5
  12. module.exports = {
  13. archiveAllDocs: callbackify(archiveAllDocs),
  14. archiveDoc: callbackify(archiveDoc),
  15. unArchiveAllDocs: callbackify(unArchiveAllDocs),
  16. unarchiveDoc: callbackify(unarchiveDoc),
  17. destroyAllDocs: callbackify(destroyAllDocs),
  18. destroyDoc: callbackify(destroyDoc),
  19. promises: {
  20. archiveAllDocs,
  21. archiveDoc,
  22. unArchiveAllDocs,
  23. unarchiveDoc,
  24. destroyAllDocs,
  25. destroyDoc,
  26. },
  27. }
  28. async function archiveAllDocs(projectId) {
  29. const docs = await MongoManager.getProjectsDocs(
  30. projectId,
  31. { include_deleted: true },
  32. { lines: true, ranges: true, rev: true, inS3: true }
  33. )
  34. if (!docs) {
  35. throw new Errors.NotFoundError(`No docs for project ${projectId}`)
  36. }
  37. const docsToArchive = docs.filter((doc) => !doc.inS3)
  38. if (docsToArchive.length) {
  39. await asyncPool(PARALLEL_JOBS, docsToArchive, (doc) =>
  40. archiveDoc(projectId, doc)
  41. )
  42. }
  43. }
  44. async function archiveDoc(projectId, doc) {
  45. logger.log(
  46. { project_id: projectId, doc_id: doc._id },
  47. 'sending doc to persistor'
  48. )
  49. const key = `${projectId}/${doc._id}`
  50. if (doc.lines == null) {
  51. throw new Error('doc has no lines')
  52. }
  53. const json = JSON.stringify({
  54. lines: doc.lines,
  55. ranges: doc.ranges,
  56. schema_v: 1,
  57. })
  58. // this should never happen, but protects against memory-corruption errors that
  59. // have happened in the past
  60. if (json.indexOf('\u0000') > -1) {
  61. const error = new Error('null bytes detected')
  62. logger.err({ err: error, doc }, error.message)
  63. throw error
  64. }
  65. const md5 = crypto.createHash('md5').update(json).digest('hex')
  66. const stream = Streamifier.createReadStream(json)
  67. await PersistorManager.sendStream(settings.docstore.bucket, key, stream, {
  68. sourceMd5: md5,
  69. })
  70. await MongoManager.markDocAsArchived(doc._id, doc.rev)
  71. }
  72. async function unArchiveAllDocs(projectId) {
  73. const docs = await MongoManager.getArchivedProjectDocs(projectId)
  74. if (!docs) {
  75. throw new Errors.NotFoundError(`No docs for project ${projectId}`)
  76. }
  77. if (!docs.length) {
  78. // asyncPool will throw an error with an empty array
  79. return
  80. }
  81. await asyncPool(PARALLEL_JOBS, docs, (doc) =>
  82. unarchiveDoc(projectId, doc._id)
  83. )
  84. }
  85. async function unarchiveDoc(projectId, docId) {
  86. logger.log(
  87. { project_id: projectId, doc_id: docId },
  88. 'getting doc from persistor'
  89. )
  90. const key = `${projectId}/${docId}`
  91. const sourceMd5 = await PersistorManager.getObjectMd5Hash(
  92. settings.docstore.bucket,
  93. key
  94. )
  95. const stream = await PersistorManager.getObjectStream(
  96. settings.docstore.bucket,
  97. key
  98. )
  99. stream.resume()
  100. const json = await _streamToString(stream)
  101. const md5 = crypto.createHash('md5').update(json).digest('hex')
  102. if (sourceMd5 !== md5) {
  103. throw new Errors.Md5MismatchError('md5 mismatch when downloading doc', {
  104. key,
  105. sourceMd5,
  106. md5,
  107. })
  108. }
  109. const doc = JSON.parse(json)
  110. const mongoDoc = {}
  111. if (doc.schema_v === 1 && doc.lines != null) {
  112. mongoDoc.lines = doc.lines
  113. if (doc.ranges != null) {
  114. mongoDoc.ranges = RangeManager.jsonRangesToMongo(doc.ranges)
  115. }
  116. } else if (Array.isArray(doc)) {
  117. mongoDoc.lines = doc
  118. } else {
  119. throw new Error("I don't understand the doc format in s3")
  120. }
  121. await MongoManager.upsertIntoDocCollection(projectId, docId, mongoDoc)
  122. await PersistorManager.deleteObject(settings.docstore.bucket, key)
  123. }
  124. async function destroyAllDocs(projectId) {
  125. const docs = await MongoManager.getProjectsDocs(
  126. projectId,
  127. { include_deleted: true },
  128. { _id: 1 }
  129. )
  130. if (docs && docs.length) {
  131. await asyncPool(PARALLEL_JOBS, docs, (doc) =>
  132. destroyDoc(projectId, doc._id)
  133. )
  134. }
  135. }
  136. async function destroyDoc(projectId, docId) {
  137. logger.log(
  138. { project_id: projectId, doc_id: docId },
  139. 'removing doc from mongo and persistor'
  140. )
  141. const doc = await MongoManager.findDoc(projectId, docId, {
  142. inS3: 1,
  143. })
  144. if (!doc) {
  145. throw new Errors.NotFoundError('Doc not found in Mongo')
  146. }
  147. if (doc.inS3) {
  148. await PersistorManager.deleteObject(
  149. settings.docstore.bucket,
  150. `${projectId}/${docId}`
  151. )
  152. }
  153. await MongoManager.destroyDoc(docId)
  154. }
  155. async function _streamToString(stream) {
  156. const chunks = []
  157. return new Promise((resolve, reject) => {
  158. stream.on('data', (chunk) => chunks.push(chunk))
  159. stream.on('error', reject)
  160. stream.on('end', () => resolve(Buffer.concat(chunks).toString('utf8')))
  161. })
  162. }