projects.js 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411
  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. persistBuffer,
  20. redisBuffer,
  21. } = require('../../storage')
  22. const render = require('./render')
  23. const expressify = require('./expressify')
  24. const withTmpDir = require('./with_tmp_dir')
  25. const StreamSizeLimit = require('./stream_size_limit')
  26. const pipeline = promisify(Stream.pipeline)
  27. async function initializeProject(req, res, next) {
  28. let projectId = req.swagger.params.body.value.projectId
  29. try {
  30. projectId = await chunkStore.initializeProject(projectId)
  31. res.status(HTTPStatus.OK).json({ projectId })
  32. } catch (err) {
  33. if (err instanceof chunkStore.AlreadyInitialized) {
  34. logger.warn({ err, projectId }, 'failed to initialize')
  35. render.conflict(res)
  36. } else {
  37. throw err
  38. }
  39. }
  40. }
  41. async function getLatestContent(req, res, next) {
  42. const projectId = req.swagger.params.project_id.value
  43. const blobStore = new BlobStore(projectId)
  44. const chunk = await chunkStore.loadLatest(projectId)
  45. const snapshot = chunk.getSnapshot()
  46. snapshot.applyAll(chunk.getChanges())
  47. await snapshot.loadFiles('eager', blobStore)
  48. res.json(snapshot.toRaw())
  49. }
  50. async function getContentAtVersion(req, res, next) {
  51. const projectId = req.swagger.params.project_id.value
  52. const version = req.swagger.params.version.value
  53. const blobStore = new BlobStore(projectId)
  54. const snapshot = await getSnapshotAtVersion(projectId, version)
  55. await snapshot.loadFiles('eager', blobStore)
  56. res.json(snapshot.toRaw())
  57. }
  58. async function getLatestHashedContent(req, res, next) {
  59. const projectId = req.swagger.params.project_id.value
  60. const blobStore = new HashCheckBlobStore(new BlobStore(projectId))
  61. const chunk = await chunkStore.loadLatest(projectId)
  62. const snapshot = chunk.getSnapshot()
  63. snapshot.applyAll(chunk.getChanges())
  64. await snapshot.loadFiles('eager', blobStore)
  65. const rawSnapshot = await snapshot.store(blobStore)
  66. res.json(rawSnapshot)
  67. }
  68. async function getLatestHistory(req, res, next) {
  69. const projectId = req.swagger.params.project_id.value
  70. try {
  71. const chunk = await chunkStore.loadLatest(projectId)
  72. const chunkResponse = new ChunkResponse(chunk)
  73. res.json(chunkResponse.toRaw())
  74. } catch (err) {
  75. if (err instanceof Chunk.NotFoundError) {
  76. render.notFound(res)
  77. } else {
  78. throw err
  79. }
  80. }
  81. }
  82. async function getLatestHistoryRaw(req, res, next) {
  83. const projectId = req.swagger.params.project_id.value
  84. const readOnly = req.swagger.params.readOnly.value
  85. try {
  86. const { startVersion, endVersion, endTimestamp } =
  87. await chunkStore.getLatestChunkMetadata(projectId, { readOnly })
  88. res.json({
  89. startVersion,
  90. endVersion,
  91. endTimestamp,
  92. })
  93. } catch (err) {
  94. if (err instanceof Chunk.NotFoundError) {
  95. render.notFound(res)
  96. } else {
  97. throw err
  98. }
  99. }
  100. }
  101. async function getHistory(req, res, next) {
  102. const projectId = req.swagger.params.project_id.value
  103. const version = req.swagger.params.version.value
  104. try {
  105. const chunk = await chunkStore.loadAtVersion(projectId, version)
  106. const chunkResponse = new ChunkResponse(chunk)
  107. res.json(chunkResponse.toRaw())
  108. } catch (err) {
  109. if (err instanceof Chunk.NotFoundError) {
  110. render.notFound(res)
  111. } else {
  112. throw err
  113. }
  114. }
  115. }
  116. async function getHistoryBefore(req, res, next) {
  117. const projectId = req.swagger.params.project_id.value
  118. const timestamp = req.swagger.params.timestamp.value
  119. try {
  120. const chunk = await chunkStore.loadAtTimestamp(projectId, timestamp)
  121. const chunkResponse = new ChunkResponse(chunk)
  122. res.json(chunkResponse.toRaw())
  123. } catch (err) {
  124. if (err instanceof Chunk.NotFoundError) {
  125. render.notFound(res)
  126. } else {
  127. throw err
  128. }
  129. }
  130. }
  131. /**
  132. * Get all changes since the beginning of history or since a given version
  133. */
  134. async function getChanges(req, res, next) {
  135. const projectId = req.swagger.params.project_id.value
  136. const since = req.swagger.params.since.value ?? 0
  137. if (since < 0) {
  138. // Negative values would cause an infinite loop
  139. return res.status(400).json({
  140. error: `Version out of bounds: ${since}`,
  141. })
  142. }
  143. let chunk
  144. try {
  145. chunk = await chunkStore.loadAtVersion(projectId, since, {
  146. preferNewer: true,
  147. })
  148. } catch (err) {
  149. if (err instanceof Chunk.VersionNotFoundError) {
  150. return res.status(400).json({
  151. error: `Version out of bounds: ${since}`,
  152. })
  153. }
  154. throw err
  155. }
  156. const latestChunkMetadata = await chunkStore.getLatestChunkMetadata(projectId)
  157. // Extract the relevant changes from the chunk that contains the start version
  158. const changes = chunk.getChanges().slice(since - chunk.getStartVersion())
  159. const hasMore = latestChunkMetadata.endVersion > chunk.getEndVersion()
  160. res.json({ changes: changes.map(change => change.toRaw()), hasMore })
  161. }
  162. async function getZip(req, res, next) {
  163. const projectId = req.swagger.params.project_id.value
  164. const version = req.swagger.params.version.value
  165. const blobStore = new BlobStore(projectId)
  166. let snapshot
  167. try {
  168. snapshot = await getSnapshotAtVersion(projectId, version)
  169. } catch (err) {
  170. if (err instanceof Chunk.NotFoundError) {
  171. return render.notFound(res)
  172. } else {
  173. throw err
  174. }
  175. }
  176. await withTmpDir('get-zip-', async tmpDir => {
  177. const tmpFilename = Path.join(tmpDir, 'project.zip')
  178. const archive = new ProjectArchive(snapshot)
  179. await archive.writeZip(blobStore, tmpFilename)
  180. res.set('Content-Type', 'application/octet-stream')
  181. res.set('Content-Disposition', 'attachment; filename=project.zip')
  182. const stream = fs.createReadStream(tmpFilename)
  183. await pipeline(stream, res)
  184. })
  185. }
  186. async function createZip(req, res, next) {
  187. const projectId = req.swagger.params.project_id.value
  188. const version = req.swagger.params.version.value
  189. try {
  190. const snapshot = await getSnapshotAtVersion(projectId, version)
  191. const zipUrl = await zipStore.getSignedUrl(projectId, version)
  192. // Do not await this; run it in the background.
  193. zipStore.storeZip(projectId, version, snapshot).catch(err => {
  194. logger.error({ err, projectId, version }, 'createZip: storeZip failed')
  195. })
  196. res.status(HTTPStatus.OK).json({ zipUrl })
  197. } catch (error) {
  198. if (error instanceof Chunk.NotFoundError) {
  199. render.notFound(res)
  200. } else {
  201. next(error)
  202. }
  203. }
  204. }
  205. async function deleteProject(req, res, next) {
  206. const projectId = req.swagger.params.project_id.value
  207. const blobStore = new BlobStore(projectId)
  208. const farFuture = new Date()
  209. farFuture.setTime(farFuture.getTime() + 7 * 24 * 3600 * 1000)
  210. const limits = {
  211. maxChanges: 0,
  212. minChangeTimestamp: farFuture,
  213. maxChangeTimestamp: farFuture,
  214. autoResync: false,
  215. }
  216. await persistBuffer(projectId, limits)
  217. await redisBuffer.expireProject(projectId)
  218. await Promise.all([
  219. chunkStore.deleteProjectChunks(projectId),
  220. blobStore.deleteBlobs(),
  221. ])
  222. res.status(HTTPStatus.NO_CONTENT).send()
  223. }
  224. async function createProjectBlob(req, res, next) {
  225. const projectId = req.swagger.params.project_id.value
  226. const expectedHash = req.swagger.params.hash.value
  227. const maxUploadSize = parseInt(config.get('maxFileUploadSize'), 10)
  228. await withTmpDir('blob-', async tmpDir => {
  229. const tmpPath = Path.join(tmpDir, 'content')
  230. const sizeLimit = new StreamSizeLimit(maxUploadSize)
  231. await pipeline(req, sizeLimit, fs.createWriteStream(tmpPath))
  232. if (sizeLimit.sizeLimitExceeded) {
  233. logger.warn(
  234. { projectId, expectedHash, maxUploadSize },
  235. 'blob exceeds size threshold'
  236. )
  237. return render.requestEntityTooLarge(res)
  238. }
  239. const hash = await blobHash.fromFile(tmpPath)
  240. if (hash !== expectedHash) {
  241. logger.warn({ projectId, hash, expectedHash }, 'Hash mismatch')
  242. return render.conflict(res, 'File hash mismatch')
  243. }
  244. const blobStore = new BlobStore(projectId)
  245. const newBlob = await blobStore.putFile(tmpPath)
  246. if (config.has('backupStore')) {
  247. try {
  248. const { backupBlob } = await import('../../storage/lib/backupBlob.mjs')
  249. await backupBlob(projectId, newBlob, tmpPath)
  250. } catch (error) {
  251. logger.warn({ error, projectId, hash }, 'Failed to backup blob')
  252. }
  253. }
  254. res.status(HTTPStatus.CREATED).end()
  255. })
  256. }
  257. async function headProjectBlob(req, res) {
  258. const projectId = req.swagger.params.project_id.value
  259. const hash = req.swagger.params.hash.value
  260. const blobStore = new BlobStore(projectId)
  261. const blob = await blobStore.getBlob(hash)
  262. if (blob) {
  263. res.set('Content-Length', blob.getByteLength())
  264. res.status(200).end()
  265. } else {
  266. res.status(404).end()
  267. }
  268. }
  269. // Support simple, singular ranges starting from zero only, up-to 2MB = 2_000_000, 7 digits
  270. const RANGE_HEADER = /^bytes=0-(\d{1,7})$/
  271. /**
  272. * @param {string} header
  273. * @return {{}|{start: number, end: number}}
  274. * @private
  275. */
  276. function _getRangeOpts(header) {
  277. if (!header) return {}
  278. const match = header.match(RANGE_HEADER)
  279. if (match) {
  280. const end = parseInt(match[1], 10)
  281. return { start: 0, end }
  282. }
  283. return {}
  284. }
  285. async function getProjectBlob(req, res, next) {
  286. const projectId = req.swagger.params.project_id.value
  287. const hash = req.swagger.params.hash.value
  288. const opts = _getRangeOpts(req.swagger.params.range.value || '')
  289. const blobStore = new BlobStore(projectId)
  290. logger.debug({ projectId, hash }, 'getProjectBlob started')
  291. try {
  292. let stream
  293. try {
  294. stream = await blobStore.getStream(hash, opts)
  295. } catch (err) {
  296. if (err instanceof Blob.NotFoundError) {
  297. logger.warn({ projectId, hash }, 'Blob not found')
  298. return res.status(404).end()
  299. } else {
  300. throw err
  301. }
  302. }
  303. res.set('Content-Type', 'application/octet-stream')
  304. try {
  305. await pipeline(stream, res)
  306. } catch (err) {
  307. if (err?.code === 'ERR_STREAM_PREMATURE_CLOSE') {
  308. res.end()
  309. } else {
  310. throw OError.tag(err, 'error transferring stream', { projectId, hash })
  311. }
  312. }
  313. } finally {
  314. logger.debug({ projectId, hash }, 'getProjectBlob finished')
  315. }
  316. }
  317. async function copyProjectBlob(req, res, next) {
  318. const sourceProjectId = req.swagger.params.copyFrom.value
  319. const targetProjectId = req.swagger.params.project_id.value
  320. const blobHash = req.swagger.params.hash.value
  321. // Check that blob exists in source project
  322. const sourceBlobStore = new BlobStore(sourceProjectId)
  323. const targetBlobStore = new BlobStore(targetProjectId)
  324. const [sourceBlob, targetBlob] = await Promise.all([
  325. sourceBlobStore.getBlob(blobHash),
  326. targetBlobStore.getBlob(blobHash),
  327. ])
  328. if (!sourceBlob) {
  329. logger.warn(
  330. { sourceProjectId, targetProjectId, blobHash },
  331. 'missing source blob when copying across projects'
  332. )
  333. return render.notFound(res)
  334. }
  335. // Exit early if the blob exists in the target project.
  336. // This will also catch global blobs, which always exist.
  337. if (targetBlob) {
  338. return res.status(HTTPStatus.NO_CONTENT).end()
  339. }
  340. // Otherwise, copy blob from source project to target project
  341. await sourceBlobStore.copyBlob(sourceBlob, targetProjectId)
  342. res.status(HTTPStatus.CREATED).end()
  343. }
  344. async function getSnapshotAtVersion(projectId, version) {
  345. const chunk = await chunkStore.loadAtVersion(projectId, version)
  346. const snapshot = chunk.getSnapshot()
  347. const changes = _.dropRight(
  348. chunk.getChanges(),
  349. chunk.getEndVersion() - version
  350. )
  351. snapshot.applyAll(changes)
  352. return snapshot
  353. }
  354. module.exports = {
  355. initializeProject: expressify(initializeProject),
  356. getLatestContent: expressify(getLatestContent),
  357. getContentAtVersion: expressify(getContentAtVersion),
  358. getLatestHashedContent: expressify(getLatestHashedContent),
  359. getLatestPersistedHistory: expressify(getLatestHistory),
  360. getLatestHistory: expressify(getLatestHistory),
  361. getLatestHistoryRaw: expressify(getLatestHistoryRaw),
  362. getHistory: expressify(getHistory),
  363. getHistoryBefore: expressify(getHistoryBefore),
  364. getChanges: expressify(getChanges),
  365. getZip: expressify(getZip),
  366. createZip: expressify(createZip),
  367. deleteProject: expressify(deleteProject),
  368. createProjectBlob: expressify(createProjectBlob),
  369. getProjectBlob: expressify(getProjectBlob),
  370. headProjectBlob: expressify(headProjectBlob),
  371. copyProjectBlob: expressify(copyProjectBlob),
  372. }