DocManager.js 8.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297
  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 getTrackedChangesUserIds(projectId) {
  117. const docs = await DocManager.getAllNonDeletedDocs(projectId, {
  118. ranges: true,
  119. })
  120. const userIds = new Set()
  121. for (const doc of docs) {
  122. for (const change of doc.ranges?.changes || []) {
  123. if (change.metadata.user_id === 'anonymous-user') continue
  124. userIds.add(change.metadata.user_id)
  125. }
  126. }
  127. return Array.from(userIds)
  128. },
  129. async projectHasRanges(projectId) {
  130. const docs = await MongoManager.getProjectsDocs(projectId, {}, { _id: 1 })
  131. const docIds = docs.map(doc => doc._id)
  132. for (const docId of docIds) {
  133. const doc = await DocManager.peekDoc(projectId, docId)
  134. if (
  135. (doc.ranges?.comments != null && doc.ranges.comments.length > 0) ||
  136. (doc.ranges?.changes != null && doc.ranges.changes.length > 0)
  137. ) {
  138. return true
  139. }
  140. }
  141. return false
  142. },
  143. async updateDoc(projectId, docId, lines, version, ranges) {
  144. const MAX_ATTEMPTS = 2
  145. for (let attempt = 1; attempt <= MAX_ATTEMPTS; attempt++) {
  146. try {
  147. const { modified, rev } = await DocManager._tryUpdateDoc(
  148. projectId,
  149. docId,
  150. lines,
  151. version,
  152. ranges
  153. )
  154. return { modified, rev }
  155. } catch (err) {
  156. if (err instanceof Errors.DocRevValueError && attempt < MAX_ATTEMPTS) {
  157. // Another updateDoc call was racing with ours.
  158. // Retry once in a bit.
  159. logger.warn(
  160. { projectId, docId, err },
  161. 'detected concurrent updateDoc call'
  162. )
  163. await setTimeout(100 + Math.random() * 100)
  164. continue
  165. } else {
  166. throw err
  167. }
  168. }
  169. }
  170. },
  171. async _tryUpdateDoc(projectId, docId, lines, version, ranges) {
  172. if (lines == null || version == null || ranges == null) {
  173. throw new Error('no lines, version or ranges provided')
  174. }
  175. let doc
  176. try {
  177. doc = await DocManager._getDoc(projectId, docId, {
  178. version: true,
  179. rev: true,
  180. lines: true,
  181. ranges: true,
  182. inS3: true,
  183. })
  184. } catch (err) {
  185. if (err instanceof Errors.NotFoundError) {
  186. doc = null
  187. } else {
  188. throw err
  189. }
  190. }
  191. ranges = RangeManager.jsonRangesToMongo(ranges)
  192. let updateLines, updateRanges, updateVersion
  193. if (doc == null) {
  194. // If the document doesn't exist, we'll make sure to create/update all parts of it.
  195. updateLines = true
  196. updateVersion = true
  197. updateRanges = true
  198. } else {
  199. if (doc.version > version) {
  200. // Reject update when the version was decremented.
  201. // Potential reasons: racing flush, broken history.
  202. throw new Errors.DocVersionDecrementedError('rejecting stale update', {
  203. updateVersion: version,
  204. flushedVersion: doc.version,
  205. })
  206. }
  207. updateLines = !_.isEqual(doc.lines, lines)
  208. updateVersion = doc.version !== version
  209. updateRanges = RangeManager.shouldUpdateRanges(doc.ranges, ranges)
  210. }
  211. let modified = false
  212. let rev = doc?.rev || 0
  213. if (updateLines || updateRanges || updateVersion) {
  214. const update = {}
  215. if (updateLines) {
  216. update.lines = lines
  217. }
  218. if (updateRanges) {
  219. update.ranges = ranges
  220. }
  221. if (updateVersion) {
  222. update.version = version
  223. }
  224. logger.debug(
  225. { projectId, docId, oldVersion: doc?.version, newVersion: version },
  226. 'updating doc'
  227. )
  228. if (updateLines || updateRanges) {
  229. rev += 1 // rev will be incremented in mongo by MongoManager.upsertIntoDocCollection
  230. }
  231. modified = true
  232. await MongoManager.upsertIntoDocCollection(
  233. projectId,
  234. docId,
  235. doc?.rev,
  236. update
  237. )
  238. } else {
  239. logger.debug({ projectId, docId }, 'doc has not changed - not updating')
  240. }
  241. return { modified, rev }
  242. },
  243. async patchDoc(projectId, docId, meta) {
  244. const projection = { _id: 1, deleted: true }
  245. const doc = await MongoManager.findDoc(projectId, docId, projection)
  246. if (!doc) {
  247. throw new Errors.NotFoundError(
  248. `No such project/doc to delete: ${projectId}/${docId}`
  249. )
  250. }
  251. if (meta.deleted && Settings.docstore.archiveOnSoftDelete) {
  252. // The user will not read this doc anytime soon. Flush it out of mongo.
  253. DocArchive.archiveDoc(projectId, docId).catch(err => {
  254. logger.warn(
  255. { projectId, docId, err },
  256. 'archiving a single doc in the background failed'
  257. )
  258. })
  259. }
  260. await MongoManager.patchDoc(projectId, docId, meta)
  261. },
  262. }
  263. module.exports = DocManager