projects.js 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347
  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. const readOnly = req.swagger.params.readOnly.value
  82. try {
  83. const { startVersion, endVersion, endTimestamp } =
  84. await chunkStore.loadLatestRaw(projectId, { readOnly })
  85. res.json({
  86. startVersion,
  87. endVersion,
  88. endTimestamp,
  89. })
  90. } catch (err) {
  91. if (err instanceof Chunk.NotFoundError) {
  92. render.notFound(res)
  93. } else {
  94. throw err
  95. }
  96. }
  97. }
  98. async function getHistory(req, res, next) {
  99. const projectId = req.swagger.params.project_id.value
  100. const version = req.swagger.params.version.value
  101. try {
  102. const chunk = await chunkStore.loadAtVersion(projectId, version)
  103. const chunkResponse = new ChunkResponse(chunk)
  104. res.json(chunkResponse.toRaw())
  105. } catch (err) {
  106. if (err instanceof Chunk.NotFoundError) {
  107. render.notFound(res)
  108. } else {
  109. throw err
  110. }
  111. }
  112. }
  113. async function getHistoryBefore(req, res, next) {
  114. const projectId = req.swagger.params.project_id.value
  115. const timestamp = req.swagger.params.timestamp.value
  116. try {
  117. const chunk = await chunkStore.loadAtTimestamp(projectId, timestamp)
  118. const chunkResponse = new ChunkResponse(chunk)
  119. res.json(chunkResponse.toRaw())
  120. } catch (err) {
  121. if (err instanceof Chunk.NotFoundError) {
  122. render.notFound(res)
  123. } else {
  124. throw err
  125. }
  126. }
  127. }
  128. async function getZip(req, res, next) {
  129. const projectId = req.swagger.params.project_id.value
  130. const version = req.swagger.params.version.value
  131. const blobStore = new BlobStore(projectId)
  132. let snapshot
  133. try {
  134. snapshot = await getSnapshotAtVersion(projectId, version)
  135. } catch (err) {
  136. if (err instanceof Chunk.NotFoundError) {
  137. return render.notFound(res)
  138. } else {
  139. throw err
  140. }
  141. }
  142. await withTmpDir('get-zip-', async tmpDir => {
  143. const tmpFilename = Path.join(tmpDir, 'project.zip')
  144. const archive = new ProjectArchive(snapshot)
  145. await archive.writeZip(blobStore, tmpFilename)
  146. res.set('Content-Type', 'application/octet-stream')
  147. res.set('Content-Disposition', 'attachment; filename=project.zip')
  148. const stream = fs.createReadStream(tmpFilename)
  149. await pipeline(stream, res)
  150. })
  151. }
  152. async function createZip(req, res, next) {
  153. const projectId = req.swagger.params.project_id.value
  154. const version = req.swagger.params.version.value
  155. try {
  156. const snapshot = await getSnapshotAtVersion(projectId, version)
  157. const zipUrl = await zipStore.getSignedUrl(projectId, version)
  158. // Do not await this; run it in the background.
  159. zipStore.storeZip(projectId, version, snapshot).catch(err => {
  160. logger.error({ err, projectId, version }, 'createZip: storeZip failed')
  161. })
  162. res.status(HTTPStatus.OK).json({ zipUrl })
  163. } catch (error) {
  164. if (error instanceof Chunk.NotFoundError) {
  165. render.notFound(res)
  166. } else {
  167. next(error)
  168. }
  169. }
  170. }
  171. async function deleteProject(req, res, next) {
  172. const projectId = req.swagger.params.project_id.value
  173. const blobStore = new BlobStore(projectId)
  174. await Promise.all([
  175. chunkStore.deleteProjectChunks(projectId),
  176. blobStore.deleteBlobs(),
  177. ])
  178. res.status(HTTPStatus.NO_CONTENT).send()
  179. }
  180. async function createProjectBlob(req, res, next) {
  181. const projectId = req.swagger.params.project_id.value
  182. const expectedHash = req.swagger.params.hash.value
  183. const maxUploadSize = parseInt(config.get('maxFileUploadSize'), 10)
  184. await withTmpDir('blob-', async tmpDir => {
  185. const tmpPath = Path.join(tmpDir, 'content')
  186. const sizeLimit = new StreamSizeLimit(maxUploadSize)
  187. await pipeline(req, sizeLimit, fs.createWriteStream(tmpPath))
  188. if (sizeLimit.sizeLimitExceeded) {
  189. return render.requestEntityTooLarge(res)
  190. }
  191. const hash = await blobHash.fromFile(tmpPath)
  192. if (hash !== expectedHash) {
  193. logger.debug({ hash, expectedHash }, 'Hash mismatch')
  194. return render.conflict(res, 'File hash mismatch')
  195. }
  196. const blobStore = new BlobStore(projectId)
  197. const newBlob = await blobStore.putFile(tmpPath)
  198. try {
  199. const { backupBlob } = await import('../../storage/lib/backupBlob.mjs')
  200. await backupBlob(projectId, newBlob, tmpPath)
  201. } catch (error) {
  202. logger.warn({ error, projectId, hash }, 'Failed to backup blob')
  203. }
  204. res.status(HTTPStatus.CREATED).end()
  205. })
  206. }
  207. async function headProjectBlob(req, res) {
  208. const projectId = req.swagger.params.project_id.value
  209. const hash = req.swagger.params.hash.value
  210. const blobStore = new BlobStore(projectId)
  211. const blob = await blobStore.getBlob(hash)
  212. if (blob) {
  213. res.set('Content-Length', blob.getByteLength())
  214. res.status(200).end()
  215. } else {
  216. res.status(404).end()
  217. }
  218. }
  219. // Support simple, singular ranges starting from zero only, up-to 2MB = 2_000_000, 7 digits
  220. const RANGE_HEADER = /^bytes=0-(\d{1,7})$/
  221. /**
  222. * @param {string} header
  223. * @return {{}|{start: number, end: number}}
  224. * @private
  225. */
  226. function _getRangeOpts(header) {
  227. if (!header) return {}
  228. const match = header.match(RANGE_HEADER)
  229. if (match) {
  230. const end = parseInt(match[1], 10)
  231. return { start: 0, end }
  232. }
  233. return {}
  234. }
  235. async function getProjectBlob(req, res, next) {
  236. const projectId = req.swagger.params.project_id.value
  237. const hash = req.swagger.params.hash.value
  238. const opts = _getRangeOpts(req.swagger.params.range.value || '')
  239. const blobStore = new BlobStore(projectId)
  240. logger.debug({ projectId, hash }, 'getProjectBlob started')
  241. try {
  242. let stream
  243. try {
  244. stream = await blobStore.getStream(hash, opts)
  245. } catch (err) {
  246. if (err instanceof Blob.NotFoundError) {
  247. logger.warn({ projectId, hash }, 'Blob not found')
  248. return res.status(404).end()
  249. } else {
  250. throw err
  251. }
  252. }
  253. res.set('Content-Type', 'application/octet-stream')
  254. try {
  255. await pipeline(stream, res)
  256. } catch (err) {
  257. if (err?.code === 'ERR_STREAM_PREMATURE_CLOSE') {
  258. res.end()
  259. } else {
  260. throw OError.tag(err, 'error transferring stream', { projectId, hash })
  261. }
  262. }
  263. } finally {
  264. logger.debug({ projectId, hash }, 'getProjectBlob finished')
  265. }
  266. }
  267. async function copyProjectBlob(req, res, next) {
  268. const sourceProjectId = req.swagger.params.copyFrom.value
  269. const targetProjectId = req.swagger.params.project_id.value
  270. const blobHash = req.swagger.params.hash.value
  271. // Check that blob exists in source project
  272. const sourceBlobStore = new BlobStore(sourceProjectId)
  273. const targetBlobStore = new BlobStore(targetProjectId)
  274. const [sourceBlob, targetBlob] = await Promise.all([
  275. sourceBlobStore.getBlob(blobHash),
  276. targetBlobStore.getBlob(blobHash),
  277. ])
  278. if (!sourceBlob) {
  279. return render.notFound(res)
  280. }
  281. // Exit early if the blob exists in the target project.
  282. // This will also catch global blobs, which always exist.
  283. if (targetBlob) {
  284. return res.status(HTTPStatus.NO_CONTENT).end()
  285. }
  286. // Otherwise, copy blob from source project to target project
  287. await sourceBlobStore.copyBlob(sourceBlob, targetProjectId)
  288. res.status(HTTPStatus.CREATED).end()
  289. }
  290. async function getSnapshotAtVersion(projectId, version) {
  291. const chunk = await chunkStore.loadAtVersion(projectId, version)
  292. const snapshot = chunk.getSnapshot()
  293. const changes = _.dropRight(
  294. chunk.getChanges(),
  295. chunk.getEndVersion() - version
  296. )
  297. snapshot.applyAll(changes)
  298. return snapshot
  299. }
  300. module.exports = {
  301. initializeProject: expressify(initializeProject),
  302. getLatestContent: expressify(getLatestContent),
  303. getContentAtVersion: expressify(getContentAtVersion),
  304. getLatestHashedContent: expressify(getLatestHashedContent),
  305. getLatestPersistedHistory: expressify(getLatestHistory),
  306. getLatestHistory: expressify(getLatestHistory),
  307. getLatestHistoryRaw: expressify(getLatestHistoryRaw),
  308. getHistory: expressify(getHistory),
  309. getHistoryBefore: expressify(getHistoryBefore),
  310. getZip: expressify(getZip),
  311. createZip: expressify(createZip),
  312. deleteProject: expressify(deleteProject),
  313. createProjectBlob: expressify(createProjectBlob),
  314. getProjectBlob: expressify(getProjectBlob),
  315. headProjectBlob: expressify(headProjectBlob),
  316. copyProjectBlob: expressify(copyProjectBlob),
  317. }