projects.js 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346
  1. 'use strict'
  2. const _ = require('lodash')
  3. const Path = require('node:path')
  4. const Stream = require('node:stream')
  5. const HTTPStatus = require('http-status')
  6. const fs = require('node:fs')
  7. const { promisify } = require('node:util')
  8. const config = require('config')
  9. const OError = require('@overleaf/o-error')
  10. const logger = require('@overleaf/logger')
  11. const { Chunk, ChunkResponse, Blob } = require('overleaf-editor-core')
  12. const {
  13. BlobStore,
  14. blobHash,
  15. chunkStore,
  16. HashCheckBlobStore,
  17. ProjectArchive,
  18. zipStore,
  19. } = require('../../storage')
  20. const render = require('./render')
  21. const expressify = require('./expressify')
  22. const withTmpDir = require('./with_tmp_dir')
  23. const StreamSizeLimit = require('./stream_size_limit')
  24. const pipeline = promisify(Stream.pipeline)
  25. async function initializeProject(req, res, next) {
  26. let projectId = req.swagger.params.body.value.projectId
  27. try {
  28. projectId = await chunkStore.initializeProject(projectId)
  29. res.status(HTTPStatus.OK).json({ projectId })
  30. } catch (err) {
  31. if (err instanceof chunkStore.AlreadyInitialized) {
  32. render.conflict(res)
  33. } else {
  34. throw err
  35. }
  36. }
  37. }
  38. async function getLatestContent(req, res, next) {
  39. const projectId = req.swagger.params.project_id.value
  40. const blobStore = new BlobStore(projectId)
  41. const chunk = await chunkStore.loadLatest(projectId)
  42. const snapshot = chunk.getSnapshot()
  43. snapshot.applyAll(chunk.getChanges())
  44. await snapshot.loadFiles('eager', blobStore)
  45. res.json(snapshot.toRaw())
  46. }
  47. async function getContentAtVersion(req, res, next) {
  48. const projectId = req.swagger.params.project_id.value
  49. const version = req.swagger.params.version.value
  50. const blobStore = new BlobStore(projectId)
  51. const snapshot = await getSnapshotAtVersion(projectId, version)
  52. await snapshot.loadFiles('eager', blobStore)
  53. res.json(snapshot.toRaw())
  54. }
  55. async function getLatestHashedContent(req, res, next) {
  56. const projectId = req.swagger.params.project_id.value
  57. const blobStore = new HashCheckBlobStore(new BlobStore(projectId))
  58. const chunk = await chunkStore.loadLatest(projectId)
  59. const snapshot = chunk.getSnapshot()
  60. snapshot.applyAll(chunk.getChanges())
  61. await snapshot.loadFiles('eager', blobStore)
  62. const rawSnapshot = await snapshot.store(blobStore)
  63. res.json(rawSnapshot)
  64. }
  65. async function getLatestHistory(req, res, next) {
  66. const projectId = req.swagger.params.project_id.value
  67. try {
  68. const chunk = await chunkStore.loadLatest(projectId)
  69. const chunkResponse = new ChunkResponse(chunk)
  70. res.json(chunkResponse.toRaw())
  71. } catch (err) {
  72. if (err instanceof Chunk.NotFoundError) {
  73. render.notFound(res)
  74. } else {
  75. throw err
  76. }
  77. }
  78. }
  79. async function getLatestHistoryRaw(req, res, next) {
  80. const projectId = req.swagger.params.project_id.value
  81. try {
  82. const { startVersion, endVersion, endTimestamp } =
  83. await chunkStore.loadLatestRaw(projectId)
  84. res.json({
  85. startVersion,
  86. endVersion,
  87. endTimestamp,
  88. })
  89. } catch (err) {
  90. if (err instanceof Chunk.NotFoundError) {
  91. render.notFound(res)
  92. } else {
  93. throw err
  94. }
  95. }
  96. }
  97. async function getHistory(req, res, next) {
  98. const projectId = req.swagger.params.project_id.value
  99. const version = req.swagger.params.version.value
  100. try {
  101. const chunk = await chunkStore.loadAtVersion(projectId, version)
  102. const chunkResponse = new ChunkResponse(chunk)
  103. res.json(chunkResponse.toRaw())
  104. } catch (err) {
  105. if (err instanceof Chunk.NotFoundError) {
  106. render.notFound(res)
  107. } else {
  108. throw err
  109. }
  110. }
  111. }
  112. async function getHistoryBefore(req, res, next) {
  113. const projectId = req.swagger.params.project_id.value
  114. const timestamp = req.swagger.params.timestamp.value
  115. try {
  116. const chunk = await chunkStore.loadAtTimestamp(projectId, timestamp)
  117. const chunkResponse = new ChunkResponse(chunk)
  118. res.json(chunkResponse.toRaw())
  119. } catch (err) {
  120. if (err instanceof Chunk.NotFoundError) {
  121. render.notFound(res)
  122. } else {
  123. throw err
  124. }
  125. }
  126. }
  127. async function getZip(req, res, next) {
  128. const projectId = req.swagger.params.project_id.value
  129. const version = req.swagger.params.version.value
  130. const blobStore = new BlobStore(projectId)
  131. let snapshot
  132. try {
  133. snapshot = await getSnapshotAtVersion(projectId, version)
  134. } catch (err) {
  135. if (err instanceof Chunk.NotFoundError) {
  136. return render.notFound(res)
  137. } else {
  138. throw err
  139. }
  140. }
  141. await withTmpDir('get-zip-', async tmpDir => {
  142. const tmpFilename = Path.join(tmpDir, 'project.zip')
  143. const archive = new ProjectArchive(snapshot)
  144. await archive.writeZip(blobStore, tmpFilename)
  145. res.set('Content-Type', 'application/octet-stream')
  146. res.set('Content-Disposition', 'attachment; filename=project.zip')
  147. const stream = fs.createReadStream(tmpFilename)
  148. await pipeline(stream, res)
  149. })
  150. }
  151. async function createZip(req, res, next) {
  152. const projectId = req.swagger.params.project_id.value
  153. const version = req.swagger.params.version.value
  154. try {
  155. const snapshot = await getSnapshotAtVersion(projectId, version)
  156. const zipUrl = await zipStore.getSignedUrl(projectId, version)
  157. // Do not await this; run it in the background.
  158. zipStore.storeZip(projectId, version, snapshot).catch(err => {
  159. logger.error({ err, projectId, version }, 'createZip: storeZip failed')
  160. })
  161. res.status(HTTPStatus.OK).json({ zipUrl })
  162. } catch (error) {
  163. if (error instanceof Chunk.NotFoundError) {
  164. render.notFound(res)
  165. } else {
  166. next(error)
  167. }
  168. }
  169. }
  170. async function deleteProject(req, res, next) {
  171. const projectId = req.swagger.params.project_id.value
  172. const blobStore = new BlobStore(projectId)
  173. await Promise.all([
  174. chunkStore.deleteProjectChunks(projectId),
  175. blobStore.deleteBlobs(),
  176. ])
  177. res.status(HTTPStatus.NO_CONTENT).send()
  178. }
  179. async function createProjectBlob(req, res, next) {
  180. const projectId = req.swagger.params.project_id.value
  181. const expectedHash = req.swagger.params.hash.value
  182. const maxUploadSize = parseInt(config.get('maxFileUploadSize'), 10)
  183. await withTmpDir('blob-', async tmpDir => {
  184. const tmpPath = Path.join(tmpDir, 'content')
  185. const sizeLimit = new StreamSizeLimit(maxUploadSize)
  186. await pipeline(req, sizeLimit, fs.createWriteStream(tmpPath))
  187. if (sizeLimit.sizeLimitExceeded) {
  188. return render.requestEntityTooLarge(res)
  189. }
  190. const hash = await blobHash.fromFile(tmpPath)
  191. if (hash !== expectedHash) {
  192. logger.debug({ hash, expectedHash }, 'Hash mismatch')
  193. return render.conflict(res, 'File hash mismatch')
  194. }
  195. const blobStore = new BlobStore(projectId)
  196. const newBlob = await blobStore.putFile(tmpPath)
  197. try {
  198. const { backupBlob } = await import('../../storage/lib/backupBlob.mjs')
  199. await backupBlob(projectId, newBlob, tmpPath)
  200. } catch (error) {
  201. logger.warn({ error, projectId, hash }, 'Failed to backup blob')
  202. }
  203. res.status(HTTPStatus.CREATED).end()
  204. })
  205. }
  206. async function headProjectBlob(req, res) {
  207. const projectId = req.swagger.params.project_id.value
  208. const hash = req.swagger.params.hash.value
  209. const blobStore = new BlobStore(projectId)
  210. const blob = await blobStore.getBlob(hash)
  211. if (blob) {
  212. res.set('Content-Length', blob.getByteLength())
  213. res.status(200).end()
  214. } else {
  215. res.status(404).end()
  216. }
  217. }
  218. // Support simple, singular ranges starting from zero only, up-to 2MB = 2_000_000, 7 digits
  219. const RANGE_HEADER = /^bytes=0-(\d{1,7})$/
  220. /**
  221. * @param {string} header
  222. * @return {{}|{start: number, end: number}}
  223. * @private
  224. */
  225. function _getRangeOpts(header) {
  226. if (!header) return {}
  227. const match = header.match(RANGE_HEADER)
  228. if (match) {
  229. const end = parseInt(match[1], 10)
  230. return { start: 0, end }
  231. }
  232. return {}
  233. }
  234. async function getProjectBlob(req, res, next) {
  235. const projectId = req.swagger.params.project_id.value
  236. const hash = req.swagger.params.hash.value
  237. const opts = _getRangeOpts(req.swagger.params.range.value || '')
  238. const blobStore = new BlobStore(projectId)
  239. logger.debug({ projectId, hash }, 'getProjectBlob started')
  240. try {
  241. let stream
  242. try {
  243. stream = await blobStore.getStream(hash, opts)
  244. } catch (err) {
  245. if (err instanceof Blob.NotFoundError) {
  246. logger.warn({ projectId, hash }, 'Blob not found')
  247. return res.status(404).end()
  248. } else {
  249. throw err
  250. }
  251. }
  252. res.set('Content-Type', 'application/octet-stream')
  253. try {
  254. await pipeline(stream, res)
  255. } catch (err) {
  256. if (err?.code === 'ERR_STREAM_PREMATURE_CLOSE') {
  257. res.end()
  258. } else {
  259. throw OError.tag(err, 'error transferring stream', { projectId, hash })
  260. }
  261. }
  262. } finally {
  263. logger.debug({ projectId, hash }, 'getProjectBlob finished')
  264. }
  265. }
  266. async function copyProjectBlob(req, res, next) {
  267. const sourceProjectId = req.swagger.params.copyFrom.value
  268. const targetProjectId = req.swagger.params.project_id.value
  269. const blobHash = req.swagger.params.hash.value
  270. // Check that blob exists in source project
  271. const sourceBlobStore = new BlobStore(sourceProjectId)
  272. const targetBlobStore = new BlobStore(targetProjectId)
  273. const [sourceBlob, targetBlob] = await Promise.all([
  274. sourceBlobStore.getBlob(blobHash),
  275. targetBlobStore.getBlob(blobHash),
  276. ])
  277. if (!sourceBlob) {
  278. return render.notFound(res)
  279. }
  280. // Exit early if the blob exists in the target project.
  281. // This will also catch global blobs, which always exist.
  282. if (targetBlob) {
  283. return res.status(HTTPStatus.NO_CONTENT).end()
  284. }
  285. // Otherwise, copy blob from source project to target project
  286. await sourceBlobStore.copyBlob(sourceBlob, targetProjectId)
  287. res.status(HTTPStatus.CREATED).end()
  288. }
  289. async function getSnapshotAtVersion(projectId, version) {
  290. const chunk = await chunkStore.loadAtVersion(projectId, version)
  291. const snapshot = chunk.getSnapshot()
  292. const changes = _.dropRight(
  293. chunk.getChanges(),
  294. chunk.getEndVersion() - version
  295. )
  296. snapshot.applyAll(changes)
  297. return snapshot
  298. }
  299. module.exports = {
  300. initializeProject: expressify(initializeProject),
  301. getLatestContent: expressify(getLatestContent),
  302. getContentAtVersion: expressify(getContentAtVersion),
  303. getLatestHashedContent: expressify(getLatestHashedContent),
  304. getLatestPersistedHistory: expressify(getLatestHistory),
  305. getLatestHistory: expressify(getLatestHistory),
  306. getLatestHistoryRaw: expressify(getLatestHistoryRaw),
  307. getHistory: expressify(getHistory),
  308. getHistoryBefore: expressify(getHistoryBefore),
  309. getZip: expressify(getZip),
  310. createZip: expressify(createZip),
  311. deleteProject: expressify(deleteProject),
  312. createProjectBlob: expressify(createProjectBlob),
  313. getProjectBlob: expressify(getProjectBlob),
  314. headProjectBlob: expressify(headProjectBlob),
  315. copyProjectBlob: expressify(copyProjectBlob),
  316. }