ProjectDeleter.mjs 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380
  1. import _ from 'lodash'
  2. import {
  3. db,
  4. ObjectId,
  5. READ_PREFERENCE_SECONDARY,
  6. } from '../../infrastructure/mongodb.mjs'
  7. import Modules from '../../infrastructure/Modules.mjs'
  8. import { callbackify } from 'node:util'
  9. import { Project } from '../../models/Project.mjs'
  10. import { DeletedProject } from '../../models/DeletedProject.mjs'
  11. import { ProjectAuditLogEntry } from '../../models/ProjectAuditLogEntry.mjs'
  12. import Errors from '../Errors/Errors.js'
  13. import logger from '@overleaf/logger'
  14. import Settings from '@overleaf/settings'
  15. import DocumentUpdaterHandler from '../DocumentUpdater/DocumentUpdaterHandler.mjs'
  16. import TagsHandler from '../Tags/TagsHandler.mjs'
  17. import ProjectDetailsHandler from './ProjectDetailsHandler.mjs'
  18. import CollaboratorsHandler from '../Collaborators/CollaboratorsHandler.mjs'
  19. import CollaboratorsGetter from '../Collaborators/CollaboratorsGetter.mjs'
  20. import DocstoreManager from '../Docstore/DocstoreManager.mjs'
  21. import EditorRealTimeController from '../Editor/EditorRealTimeController.mjs'
  22. import HistoryManager from '../History/HistoryManager.mjs'
  23. import ChatApiHandler from '../Chat/ChatApiHandler.mjs'
  24. import { promiseMapWithLimit } from '@overleaf/promise-utils'
  25. const PROJECT_EXPIRATION_BATCH_SIZE = 10000
  26. export default {
  27. markAsDeletedByExternalSource: callbackify(markAsDeletedByExternalSource),
  28. unmarkAsDeletedByExternalSource: callbackify(unmarkAsDeletedByExternalSource),
  29. deleteUsersProjects: callbackify(deleteUsersProjects),
  30. expireDeletedProjectsAfterDuration: callbackify(
  31. expireDeletedProjectsAfterDuration
  32. ),
  33. restoreProject: callbackify(restoreProject),
  34. archiveProject: callbackify(archiveProject),
  35. unarchiveProject: callbackify(unarchiveProject),
  36. trashProject: callbackify(trashProject),
  37. untrashProject: callbackify(untrashProject),
  38. deleteProject: callbackify(deleteProject),
  39. undeleteProject: callbackify(undeleteProject),
  40. expireDeletedProject: callbackify(expireDeletedProject),
  41. promises: {
  42. archiveProject,
  43. unarchiveProject,
  44. trashProject,
  45. untrashProject,
  46. deleteProject,
  47. undeleteProject,
  48. expireDeletedProject,
  49. markAsDeletedByExternalSource,
  50. unmarkAsDeletedByExternalSource,
  51. deleteUsersProjects,
  52. expireDeletedProjectsAfterDuration,
  53. restoreProject,
  54. },
  55. }
  56. async function markAsDeletedByExternalSource(projectId) {
  57. logger.debug(
  58. { projectId },
  59. 'marking project as deleted by external data source'
  60. )
  61. await Project.updateOne(
  62. { _id: projectId },
  63. { deletedByExternalDataSource: true }
  64. ).exec()
  65. EditorRealTimeController.emitToRoom(
  66. projectId,
  67. 'projectRenamedOrDeletedByExternalSource'
  68. )
  69. }
  70. async function unmarkAsDeletedByExternalSource(projectId) {
  71. await Project.updateOne(
  72. { _id: projectId },
  73. { deletedByExternalDataSource: false }
  74. ).exec()
  75. }
  76. async function deleteUsersProjects(userId) {
  77. const projects = await Project.find({ owner_ref: userId }).exec()
  78. logger.info(
  79. { userId, projectCount: projects.length },
  80. 'found user projects to delete'
  81. )
  82. await promiseMapWithLimit(5, projects, project => deleteProject(project._id))
  83. logger.info({ userId }, 'deleted all user projects')
  84. await CollaboratorsHandler.promises.removeUserFromAllProjects(userId)
  85. }
  86. async function expireDeletedProjectsAfterDuration() {
  87. const deletedProjects = await DeletedProject.find(
  88. {
  89. 'deleterData.deletedAt': {
  90. $lt: new Date(Date.now() - Settings.projectHardDeletionDelay),
  91. },
  92. project: { $type: 'object' },
  93. },
  94. { 'deleterData.deletedProjectId': 1 }
  95. )
  96. .limit(PROJECT_EXPIRATION_BATCH_SIZE)
  97. .read(READ_PREFERENCE_SECONDARY)
  98. const projectIds = _.shuffle(
  99. deletedProjects.map(
  100. deletedProject => deletedProject.deleterData.deletedProjectId
  101. )
  102. )
  103. logger.info(
  104. {
  105. projectCount: projectIds.length,
  106. retentionPeriodInDays:
  107. Settings.projectHardDeletionDelay / (1000 * 60 * 60 * 24),
  108. },
  109. 'expiring batch of deleted projects'
  110. )
  111. try {
  112. for (const projectId of projectIds) {
  113. await expireDeletedProject(projectId)
  114. }
  115. logger.info(
  116. { projectCount: projectIds.length },
  117. 'batch of deleted projects expired successfully'
  118. )
  119. } catch (error) {
  120. logger.warn(
  121. { error },
  122. 'something went wrong expiring batch of deleted projects'
  123. )
  124. throw error
  125. }
  126. }
  127. async function restoreProject(projectId) {
  128. await Project.updateOne(
  129. { _id: projectId },
  130. { $unset: { archived: true } }
  131. ).exec()
  132. }
  133. async function archiveProject(projectId, userId) {
  134. await Project.updateOne(
  135. { _id: projectId },
  136. {
  137. $addToSet: { archived: new ObjectId(userId) },
  138. $pull: { trashed: new ObjectId(userId) },
  139. }
  140. )
  141. }
  142. async function unarchiveProject(projectId, userId) {
  143. await Project.updateOne(
  144. { _id: projectId },
  145. { $pull: { archived: new ObjectId(userId) } }
  146. )
  147. }
  148. async function trashProject(projectId, userId) {
  149. await Project.updateOne(
  150. { _id: projectId },
  151. {
  152. $addToSet: { trashed: new ObjectId(userId) },
  153. $pull: { archived: new ObjectId(userId) },
  154. }
  155. )
  156. }
  157. async function untrashProject(projectId, userId) {
  158. await Project.updateOne(
  159. { _id: projectId },
  160. { $pull: { trashed: new ObjectId(userId) } }
  161. )
  162. }
  163. async function deleteProject(projectId, options = {}) {
  164. try {
  165. const project = await Project.findOne({ _id: projectId }).exec()
  166. if (!project) {
  167. throw new Errors.NotFoundError('project not found')
  168. }
  169. await DocumentUpdaterHandler.promises.flushProjectToMongoAndDelete(
  170. projectId
  171. )
  172. try {
  173. // OPTIMIZATION: flush docs out of mongo
  174. await DocstoreManager.promises.archiveProject(projectId)
  175. } catch (err) {
  176. // It is OK to fail here, the docs will get hard-deleted eventually after
  177. // the grace-period for soft-deleted projects has passed.
  178. logger.warn(
  179. { projectId, err },
  180. 'failed archiving doc via docstore as part of project soft-deletion'
  181. )
  182. }
  183. const memberIds = await CollaboratorsGetter.promises.getMemberIds(projectId)
  184. // fire these jobs in the background
  185. for (const memberId of memberIds) {
  186. TagsHandler.promises
  187. .removeProjectFromAllTags(memberId, projectId)
  188. .catch(err => {
  189. logger.err(
  190. { err, memberId, projectId },
  191. 'failed to remove project from tags'
  192. )
  193. })
  194. }
  195. const deleterData = {
  196. deletedAt: new Date(),
  197. deleterId:
  198. options.deleterUser != null ? options.deleterUser._id : undefined,
  199. deleterIpAddress: options.ipAddress,
  200. deletedProjectId: project._id,
  201. deletedProjectOwnerId: project.owner_ref,
  202. deletedProjectCollaboratorIds: project.collaberator_refs,
  203. deletedProjectReadOnlyIds: project.readOnly_refs,
  204. deletedProjectReviewerIds: project.reviewer_refs,
  205. deletedProjectReadWriteTokenAccessIds:
  206. project.tokenAccessReadAndWrite_refs,
  207. deletedProjectOverleafId: project.overleaf
  208. ? project.overleaf.id
  209. : undefined,
  210. deletedProjectOverleafHistoryId:
  211. project.overleaf && project.overleaf.history
  212. ? project.overleaf.history.id
  213. : undefined,
  214. deletedProjectReadOnlyTokenAccessIds: project.tokenAccessReadOnly_refs,
  215. deletedProjectReadWriteToken: project.tokens.readAndWrite,
  216. deletedProjectReadOnlyToken: project.tokens.readOnly,
  217. deletedProjectLastUpdatedAt: project.lastUpdated,
  218. }
  219. Object.keys(deleterData).forEach(key =>
  220. deleterData[key] === undefined ? delete deleterData[key] : ''
  221. )
  222. await DeletedProject.updateOne(
  223. { 'deleterData.deletedProjectId': projectId },
  224. { project, deleterData },
  225. { upsert: true }
  226. )
  227. await Project.deleteOne({ _id: projectId }).exec()
  228. logger.info(
  229. { projectId, userId: project.owner_ref },
  230. 'successfully deleted project'
  231. )
  232. } catch (err) {
  233. logger.warn({ err }, 'problem deleting project')
  234. throw err
  235. }
  236. }
  237. async function undeleteProject(projectId, options = {}) {
  238. projectId = new ObjectId(projectId)
  239. const deletedProject = await DeletedProject.findOne({
  240. 'deleterData.deletedProjectId': projectId,
  241. }).exec()
  242. if (!deletedProject) {
  243. throw new Errors.NotFoundError('project_not_found')
  244. }
  245. if (!deletedProject.project) {
  246. throw new Errors.NotFoundError('project_too_old_to_restore')
  247. }
  248. const restored = new Project(deletedProject.project)
  249. if (options.userId) {
  250. restored.owner_ref = options.userId
  251. }
  252. // if we're undeleting, we want the document to show up
  253. restored.name = await ProjectDetailsHandler.promises.generateUniqueName(
  254. deletedProject.deleterData.deletedProjectOwnerId,
  255. restored.name + ' (Restored)'
  256. )
  257. restored.archived = undefined
  258. if (restored.deletedDocs && restored.deletedDocs.length > 0) {
  259. await promiseMapWithLimit(10, restored.deletedDocs, async deletedDoc => {
  260. // back fill context of deleted docs
  261. const { _id: docId, name, deletedAt } = deletedDoc
  262. await DocstoreManager.promises.deleteDoc(
  263. projectId,
  264. docId,
  265. name,
  266. deletedAt
  267. )
  268. })
  269. restored.deletedDocs = []
  270. }
  271. // we can't use Mongoose to re-insert the project, as it won't
  272. // create a new document with an _id already specified. We need to
  273. // insert it directly into the collection
  274. await db.projects.insertOne(restored)
  275. await DeletedProject.deleteOne({ _id: deletedProject._id }).exec()
  276. }
  277. async function expireDeletedProject(projectId) {
  278. try {
  279. logger.info({ projectId }, 'expiring deleted project')
  280. const activeProject = await Project.findById(projectId).exec()
  281. if (activeProject) {
  282. // That project is active. The deleted project record might be there
  283. // because of an incomplete delete or undelete operation. Clean it up and
  284. // return.
  285. logger.info(
  286. { projectId },
  287. 'deleted project record found but project is active'
  288. )
  289. await DeletedProject.deleteOne({
  290. 'deleterData.deletedProjectId': projectId,
  291. })
  292. return
  293. }
  294. const deletedProject = await DeletedProject.findOne({
  295. 'deleterData.deletedProjectId': projectId,
  296. }).exec()
  297. if (!deletedProject) {
  298. throw new Errors.NotFoundError(
  299. `No deleted project found for project id ${projectId}`
  300. )
  301. }
  302. if (!deletedProject.project) {
  303. logger.warn(
  304. { projectId },
  305. `Attempted to expire already-expired deletedProject`
  306. )
  307. return
  308. }
  309. const userId = deletedProject.deleterData?.deletedProjectOwnerId?.toString()
  310. const historyId =
  311. deletedProject.project.overleaf &&
  312. deletedProject.project.overleaf.history &&
  313. deletedProject.project.overleaf.history.id
  314. logger.info({ projectId, userId }, 'destroying expired project data')
  315. await Promise.all([
  316. DocstoreManager.promises.destroyProject(deletedProject.project._id),
  317. HistoryManager.promises.deleteProject(
  318. deletedProject.project._id,
  319. historyId
  320. ),
  321. ChatApiHandler.promises.destroyProject(deletedProject.project._id),
  322. ProjectAuditLogEntry.deleteMany({ projectId }),
  323. Modules.promises.hooks.fire('projectExpired', deletedProject.project._id),
  324. ])
  325. logger.info(
  326. { projectId, userId },
  327. 'redacting PII from the deleted project record'
  328. )
  329. await DeletedProject.updateOne(
  330. {
  331. _id: deletedProject._id,
  332. },
  333. {
  334. $set: {
  335. 'deleterData.deleterIpAddress': null,
  336. project: null,
  337. },
  338. }
  339. ).exec()
  340. logger.info({ projectId, userId }, 'expired deleted project successfully')
  341. } catch (error) {
  342. logger.warn({ projectId, error }, 'error expiring deleted project')
  343. throw error
  344. }
  345. }