DocManager.js 7.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283
  1. const MongoManager = require('./MongoManager')
  2. const Errors = require('./Errors')
  3. const logger = require('@overleaf/logger')
  4. const _ = require('lodash')
  5. const DocArchive = require('./DocArchiveManager')
  6. const RangeManager = require('./RangeManager')
  7. const Settings = require('@overleaf/settings')
  8. const { setTimeout } = require('node:timers/promises')
  9. /**
  10. * @import { Document } from 'mongodb'
  11. * @import { WithId } from 'mongodb'
  12. */
  13. const DocManager = {
  14. /**
  15. * @param {string} projectId
  16. * @param {string} docId
  17. * @param {{inS3: boolean}} filter
  18. * @returns {Promise<WithId<Document>>}
  19. * @private
  20. */
  21. async _getDoc(projectId, docId, filter) {
  22. if (filter == null) {
  23. filter = {}
  24. }
  25. if (filter.inS3 !== true) {
  26. throw new Error('must include inS3 when getting doc')
  27. }
  28. const doc = await MongoManager.findDoc(projectId, docId, filter)
  29. if (doc == null) {
  30. throw new Errors.NotFoundError(
  31. `No such doc: ${docId} in project ${projectId}`
  32. )
  33. }
  34. if (doc.inS3) {
  35. await DocArchive.unarchiveDoc(projectId, docId)
  36. return await DocManager._getDoc(projectId, docId, filter)
  37. }
  38. return doc
  39. },
  40. async isDocDeleted(projectId, docId) {
  41. const doc = await MongoManager.findDoc(projectId, docId, {
  42. deleted: true,
  43. })
  44. if (!doc) {
  45. throw new Errors.NotFoundError(
  46. `No such project/doc: ${projectId}/${docId}`
  47. )
  48. }
  49. // `doc.deleted` is `undefined` for non deleted docs
  50. return Boolean(doc.deleted)
  51. },
  52. async getFullDoc(projectId, docId) {
  53. const doc = await DocManager._getDoc(projectId, docId, {
  54. lines: true,
  55. rev: true,
  56. deleted: true,
  57. version: true,
  58. ranges: true,
  59. inS3: true,
  60. })
  61. return doc
  62. },
  63. // returns the doc without any version information
  64. async _peekRawDoc(projectId, docId) {
  65. const doc = await MongoManager.findDoc(projectId, docId, {
  66. lines: true,
  67. rev: true,
  68. deleted: true,
  69. version: true,
  70. ranges: true,
  71. inS3: true,
  72. })
  73. if (doc == null) {
  74. throw new Errors.NotFoundError(
  75. `No such doc: ${docId} in project ${projectId}`
  76. )
  77. }
  78. if (doc.inS3) {
  79. // skip the unarchiving to mongo when getting a doc
  80. const archivedDoc = await DocArchive.getDoc(projectId, docId)
  81. Object.assign(doc, archivedDoc)
  82. }
  83. return doc
  84. },
  85. // get the doc from mongo if possible, or from the persistent store otherwise,
  86. // without unarchiving it (avoids unnecessary writes to mongo)
  87. async peekDoc(projectId, docId) {
  88. const doc = await DocManager._peekRawDoc(projectId, docId)
  89. await MongoManager.checkRevUnchanged(doc)
  90. return doc
  91. },
  92. async getDocLines(projectId, docId) {
  93. const doc = await DocManager._getDoc(projectId, docId, {
  94. lines: true,
  95. inS3: true,
  96. })
  97. if (!doc) throw new Errors.NotFoundError()
  98. if (!Array.isArray(doc.lines)) throw new Errors.DocWithoutLinesError()
  99. return doc.lines.join('\n')
  100. },
  101. async getAllDeletedDocs(projectId, filter) {
  102. return await MongoManager.getProjectsDeletedDocs(projectId, filter)
  103. },
  104. async getAllNonDeletedDocs(projectId, filter) {
  105. await DocArchive.unArchiveAllDocs(projectId)
  106. const docs = await MongoManager.getProjectsDocs(
  107. projectId,
  108. { include_deleted: false },
  109. filter
  110. )
  111. if (docs == null) {
  112. throw new Errors.NotFoundError(`No docs for project ${projectId}`)
  113. }
  114. return docs
  115. },
  116. async projectHasRanges(projectId) {
  117. const docs = await MongoManager.getProjectsDocs(projectId, {}, { _id: 1 })
  118. const docIds = docs.map(doc => doc._id)
  119. for (const docId of docIds) {
  120. const doc = await DocManager.peekDoc(projectId, docId)
  121. if (
  122. (doc.ranges?.comments != null && doc.ranges.comments.length > 0) ||
  123. (doc.ranges?.changes != null && doc.ranges.changes.length > 0)
  124. ) {
  125. return true
  126. }
  127. }
  128. return false
  129. },
  130. async updateDoc(projectId, docId, lines, version, ranges) {
  131. const MAX_ATTEMPTS = 2
  132. for (let attempt = 1; attempt <= MAX_ATTEMPTS; attempt++) {
  133. try {
  134. const { modified, rev } = await DocManager._tryUpdateDoc(
  135. projectId,
  136. docId,
  137. lines,
  138. version,
  139. ranges
  140. )
  141. return { modified, rev }
  142. } catch (err) {
  143. if (err instanceof Errors.DocRevValueError && attempt < MAX_ATTEMPTS) {
  144. // Another updateDoc call was racing with ours.
  145. // Retry once in a bit.
  146. logger.warn(
  147. { projectId, docId, err },
  148. 'detected concurrent updateDoc call'
  149. )
  150. await setTimeout(100 + Math.random() * 100)
  151. continue
  152. } else {
  153. throw err
  154. }
  155. }
  156. }
  157. },
  158. async _tryUpdateDoc(projectId, docId, lines, version, ranges) {
  159. if (lines == null || version == null || ranges == null) {
  160. throw new Error('no lines, version or ranges provided')
  161. }
  162. let doc
  163. try {
  164. doc = await DocManager._getDoc(projectId, docId, {
  165. version: true,
  166. rev: true,
  167. lines: true,
  168. ranges: true,
  169. inS3: true,
  170. })
  171. } catch (err) {
  172. if (err instanceof Errors.NotFoundError) {
  173. doc = null
  174. } else {
  175. throw err
  176. }
  177. }
  178. ranges = RangeManager.jsonRangesToMongo(ranges)
  179. let updateLines, updateRanges, updateVersion
  180. if (doc == null) {
  181. // If the document doesn't exist, we'll make sure to create/update all parts of it.
  182. updateLines = true
  183. updateVersion = true
  184. updateRanges = true
  185. } else {
  186. if (doc.version > version) {
  187. // Reject update when the version was decremented.
  188. // Potential reasons: racing flush, broken history.
  189. throw new Errors.DocVersionDecrementedError('rejecting stale update', {
  190. updateVersion: version,
  191. flushedVersion: doc.version,
  192. })
  193. }
  194. updateLines = !_.isEqual(doc.lines, lines)
  195. updateVersion = doc.version !== version
  196. updateRanges = RangeManager.shouldUpdateRanges(doc.ranges, ranges)
  197. }
  198. let modified = false
  199. let rev = doc?.rev || 0
  200. if (updateLines || updateRanges || updateVersion) {
  201. const update = {}
  202. if (updateLines) {
  203. update.lines = lines
  204. }
  205. if (updateRanges) {
  206. update.ranges = ranges
  207. }
  208. if (updateVersion) {
  209. update.version = version
  210. }
  211. logger.debug(
  212. { projectId, docId, oldVersion: doc?.version, newVersion: version },
  213. 'updating doc'
  214. )
  215. if (updateLines || updateRanges) {
  216. rev += 1 // rev will be incremented in mongo by MongoManager.upsertIntoDocCollection
  217. }
  218. modified = true
  219. await MongoManager.upsertIntoDocCollection(
  220. projectId,
  221. docId,
  222. doc?.rev,
  223. update
  224. )
  225. } else {
  226. logger.debug({ projectId, docId }, 'doc has not changed - not updating')
  227. }
  228. return { modified, rev }
  229. },
  230. async patchDoc(projectId, docId, meta) {
  231. const projection = { _id: 1, deleted: true }
  232. const doc = await MongoManager.findDoc(projectId, docId, projection)
  233. if (!doc) {
  234. throw new Errors.NotFoundError(
  235. `No such project/doc to delete: ${projectId}/${docId}`
  236. )
  237. }
  238. if (meta.deleted && Settings.docstore.archiveOnSoftDelete) {
  239. // The user will not read this doc anytime soon. Flush it out of mongo.
  240. DocArchive.archiveDoc(projectId, docId).catch(err => {
  241. logger.warn(
  242. { projectId, docId, err },
  243. 'archiving a single doc in the background failed'
  244. )
  245. })
  246. }
  247. await MongoManager.patchDoc(projectId, docId, meta)
  248. },
  249. }
  250. module.exports = DocManager