projects.js 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396
  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. logger.warn({ err, projectId }, 'failed to initialize')
  33. render.conflict(res)
  34. } else {
  35. throw err
  36. }
  37. }
  38. }
  39. async function getLatestContent(req, res, next) {
  40. const projectId = req.swagger.params.project_id.value
  41. const blobStore = new BlobStore(projectId)
  42. const chunk = await chunkStore.loadLatest(projectId)
  43. const snapshot = chunk.getSnapshot()
  44. snapshot.applyAll(chunk.getChanges())
  45. await snapshot.loadFiles('eager', blobStore)
  46. res.json(snapshot.toRaw())
  47. }
  48. async function getContentAtVersion(req, res, next) {
  49. const projectId = req.swagger.params.project_id.value
  50. const version = req.swagger.params.version.value
  51. const blobStore = new BlobStore(projectId)
  52. const snapshot = await getSnapshotAtVersion(projectId, version)
  53. await snapshot.loadFiles('eager', blobStore)
  54. res.json(snapshot.toRaw())
  55. }
  56. async function getLatestHashedContent(req, res, next) {
  57. const projectId = req.swagger.params.project_id.value
  58. const blobStore = new HashCheckBlobStore(new BlobStore(projectId))
  59. const chunk = await chunkStore.loadLatest(projectId)
  60. const snapshot = chunk.getSnapshot()
  61. snapshot.applyAll(chunk.getChanges())
  62. await snapshot.loadFiles('eager', blobStore)
  63. const rawSnapshot = await snapshot.store(blobStore)
  64. res.json(rawSnapshot)
  65. }
  66. async function getLatestHistory(req, res, next) {
  67. const projectId = req.swagger.params.project_id.value
  68. try {
  69. const chunk = await chunkStore.loadLatest(projectId)
  70. const chunkResponse = new ChunkResponse(chunk)
  71. res.json(chunkResponse.toRaw())
  72. } catch (err) {
  73. if (err instanceof Chunk.NotFoundError) {
  74. render.notFound(res)
  75. } else {
  76. throw err
  77. }
  78. }
  79. }
  80. async function getLatestHistoryRaw(req, res, next) {
  81. const projectId = req.swagger.params.project_id.value
  82. const readOnly = req.swagger.params.readOnly.value
  83. try {
  84. const { startVersion, endVersion, endTimestamp } =
  85. await chunkStore.getLatestChunkMetadata(projectId, { readOnly })
  86. res.json({
  87. startVersion,
  88. endVersion,
  89. endTimestamp,
  90. })
  91. } catch (err) {
  92. if (err instanceof Chunk.NotFoundError) {
  93. render.notFound(res)
  94. } else {
  95. throw err
  96. }
  97. }
  98. }
  99. async function getHistory(req, res, next) {
  100. const projectId = req.swagger.params.project_id.value
  101. const version = req.swagger.params.version.value
  102. try {
  103. const chunk = await chunkStore.loadAtVersion(projectId, version)
  104. const chunkResponse = new ChunkResponse(chunk)
  105. res.json(chunkResponse.toRaw())
  106. } catch (err) {
  107. if (err instanceof Chunk.NotFoundError) {
  108. render.notFound(res)
  109. } else {
  110. throw err
  111. }
  112. }
  113. }
  114. async function getHistoryBefore(req, res, next) {
  115. const projectId = req.swagger.params.project_id.value
  116. const timestamp = req.swagger.params.timestamp.value
  117. try {
  118. const chunk = await chunkStore.loadAtTimestamp(projectId, timestamp)
  119. const chunkResponse = new ChunkResponse(chunk)
  120. res.json(chunkResponse.toRaw())
  121. } catch (err) {
  122. if (err instanceof Chunk.NotFoundError) {
  123. render.notFound(res)
  124. } else {
  125. throw err
  126. }
  127. }
  128. }
  129. /**
  130. * Get all changes since the beginning of history or since a given version
  131. */
  132. async function getChanges(req, res, next) {
  133. const projectId = req.swagger.params.project_id.value
  134. const since = req.swagger.params.since.value ?? 0
  135. if (since < 0) {
  136. // Negative values would cause an infinite loop
  137. return res.status(400).json({
  138. error: `Version out of bounds: ${since}`,
  139. })
  140. }
  141. let chunk
  142. try {
  143. chunk = await chunkStore.loadAtVersion(projectId, since, {
  144. preferNewer: true,
  145. })
  146. } catch (err) {
  147. if (err instanceof Chunk.VersionNotFoundError) {
  148. return res.status(400).json({
  149. error: `Version out of bounds: ${since}`,
  150. })
  151. }
  152. throw err
  153. }
  154. const latestChunkMetadata = await chunkStore.getLatestChunkMetadata(projectId)
  155. // Extract the relevant changes from the chunk that contains the start version
  156. const changes = chunk.getChanges().slice(since - chunk.getStartVersion())
  157. const hasMore = latestChunkMetadata.endVersion > chunk.getEndVersion()
  158. res.json({ changes: changes.map(change => change.toRaw()), hasMore })
  159. }
  160. async function getZip(req, res, next) {
  161. const projectId = req.swagger.params.project_id.value
  162. const version = req.swagger.params.version.value
  163. const blobStore = new BlobStore(projectId)
  164. let snapshot
  165. try {
  166. snapshot = await getSnapshotAtVersion(projectId, version)
  167. } catch (err) {
  168. if (err instanceof Chunk.NotFoundError) {
  169. return render.notFound(res)
  170. } else {
  171. throw err
  172. }
  173. }
  174. await withTmpDir('get-zip-', async tmpDir => {
  175. const tmpFilename = Path.join(tmpDir, 'project.zip')
  176. const archive = new ProjectArchive(snapshot)
  177. await archive.writeZip(blobStore, tmpFilename)
  178. res.set('Content-Type', 'application/octet-stream')
  179. res.set('Content-Disposition', 'attachment; filename=project.zip')
  180. const stream = fs.createReadStream(tmpFilename)
  181. await pipeline(stream, res)
  182. })
  183. }
  184. async function createZip(req, res, next) {
  185. const projectId = req.swagger.params.project_id.value
  186. const version = req.swagger.params.version.value
  187. try {
  188. const snapshot = await getSnapshotAtVersion(projectId, version)
  189. const zipUrl = await zipStore.getSignedUrl(projectId, version)
  190. // Do not await this; run it in the background.
  191. zipStore.storeZip(projectId, version, snapshot).catch(err => {
  192. logger.error({ err, projectId, version }, 'createZip: storeZip failed')
  193. })
  194. res.status(HTTPStatus.OK).json({ zipUrl })
  195. } catch (error) {
  196. if (error instanceof Chunk.NotFoundError) {
  197. render.notFound(res)
  198. } else {
  199. next(error)
  200. }
  201. }
  202. }
  203. async function deleteProject(req, res, next) {
  204. const projectId = req.swagger.params.project_id.value
  205. const blobStore = new BlobStore(projectId)
  206. await Promise.all([
  207. chunkStore.deleteProjectChunks(projectId),
  208. blobStore.deleteBlobs(),
  209. ])
  210. res.status(HTTPStatus.NO_CONTENT).send()
  211. }
  212. async function createProjectBlob(req, res, next) {
  213. const projectId = req.swagger.params.project_id.value
  214. const expectedHash = req.swagger.params.hash.value
  215. const maxUploadSize = parseInt(config.get('maxFileUploadSize'), 10)
  216. await withTmpDir('blob-', async tmpDir => {
  217. const tmpPath = Path.join(tmpDir, 'content')
  218. const sizeLimit = new StreamSizeLimit(maxUploadSize)
  219. await pipeline(req, sizeLimit, fs.createWriteStream(tmpPath))
  220. if (sizeLimit.sizeLimitExceeded) {
  221. logger.warn(
  222. { projectId, expectedHash, maxUploadSize },
  223. 'blob exceeds size threshold'
  224. )
  225. return render.requestEntityTooLarge(res)
  226. }
  227. const hash = await blobHash.fromFile(tmpPath)
  228. if (hash !== expectedHash) {
  229. logger.warn({ projectId, hash, expectedHash }, 'Hash mismatch')
  230. return render.conflict(res, 'File hash mismatch')
  231. }
  232. const blobStore = new BlobStore(projectId)
  233. const newBlob = await blobStore.putFile(tmpPath)
  234. if (config.has('backupStore')) {
  235. try {
  236. const { backupBlob } = await import('../../storage/lib/backupBlob.mjs')
  237. await backupBlob(projectId, newBlob, tmpPath)
  238. } catch (error) {
  239. logger.warn({ error, projectId, hash }, 'Failed to backup blob')
  240. }
  241. }
  242. res.status(HTTPStatus.CREATED).end()
  243. })
  244. }
  245. async function headProjectBlob(req, res) {
  246. const projectId = req.swagger.params.project_id.value
  247. const hash = req.swagger.params.hash.value
  248. const blobStore = new BlobStore(projectId)
  249. const blob = await blobStore.getBlob(hash)
  250. if (blob) {
  251. res.set('Content-Length', blob.getByteLength())
  252. res.status(200).end()
  253. } else {
  254. res.status(404).end()
  255. }
  256. }
  257. // Support simple, singular ranges starting from zero only, up-to 2MB = 2_000_000, 7 digits
  258. const RANGE_HEADER = /^bytes=0-(\d{1,7})$/
  259. /**
  260. * @param {string} header
  261. * @return {{}|{start: number, end: number}}
  262. * @private
  263. */
  264. function _getRangeOpts(header) {
  265. if (!header) return {}
  266. const match = header.match(RANGE_HEADER)
  267. if (match) {
  268. const end = parseInt(match[1], 10)
  269. return { start: 0, end }
  270. }
  271. return {}
  272. }
  273. async function getProjectBlob(req, res, next) {
  274. const projectId = req.swagger.params.project_id.value
  275. const hash = req.swagger.params.hash.value
  276. const opts = _getRangeOpts(req.swagger.params.range.value || '')
  277. const blobStore = new BlobStore(projectId)
  278. logger.debug({ projectId, hash }, 'getProjectBlob started')
  279. try {
  280. let stream
  281. try {
  282. stream = await blobStore.getStream(hash, opts)
  283. } catch (err) {
  284. if (err instanceof Blob.NotFoundError) {
  285. logger.warn({ projectId, hash }, 'Blob not found')
  286. return res.status(404).end()
  287. } else {
  288. throw err
  289. }
  290. }
  291. res.set('Content-Type', 'application/octet-stream')
  292. try {
  293. await pipeline(stream, res)
  294. } catch (err) {
  295. if (err?.code === 'ERR_STREAM_PREMATURE_CLOSE') {
  296. res.end()
  297. } else {
  298. throw OError.tag(err, 'error transferring stream', { projectId, hash })
  299. }
  300. }
  301. } finally {
  302. logger.debug({ projectId, hash }, 'getProjectBlob finished')
  303. }
  304. }
  305. async function copyProjectBlob(req, res, next) {
  306. const sourceProjectId = req.swagger.params.copyFrom.value
  307. const targetProjectId = req.swagger.params.project_id.value
  308. const blobHash = req.swagger.params.hash.value
  309. // Check that blob exists in source project
  310. const sourceBlobStore = new BlobStore(sourceProjectId)
  311. const targetBlobStore = new BlobStore(targetProjectId)
  312. const [sourceBlob, targetBlob] = await Promise.all([
  313. sourceBlobStore.getBlob(blobHash),
  314. targetBlobStore.getBlob(blobHash),
  315. ])
  316. if (!sourceBlob) {
  317. logger.warn(
  318. { sourceProjectId, targetProjectId, blobHash },
  319. 'missing source blob when copying across projects'
  320. )
  321. return render.notFound(res)
  322. }
  323. // Exit early if the blob exists in the target project.
  324. // This will also catch global blobs, which always exist.
  325. if (targetBlob) {
  326. return res.status(HTTPStatus.NO_CONTENT).end()
  327. }
  328. // Otherwise, copy blob from source project to target project
  329. await sourceBlobStore.copyBlob(sourceBlob, targetProjectId)
  330. res.status(HTTPStatus.CREATED).end()
  331. }
  332. async function getSnapshotAtVersion(projectId, version) {
  333. const chunk = await chunkStore.loadAtVersion(projectId, version)
  334. const snapshot = chunk.getSnapshot()
  335. const changes = _.dropRight(
  336. chunk.getChanges(),
  337. chunk.getEndVersion() - version
  338. )
  339. snapshot.applyAll(changes)
  340. return snapshot
  341. }
  342. module.exports = {
  343. initializeProject: expressify(initializeProject),
  344. getLatestContent: expressify(getLatestContent),
  345. getContentAtVersion: expressify(getContentAtVersion),
  346. getLatestHashedContent: expressify(getLatestHashedContent),
  347. getLatestPersistedHistory: expressify(getLatestHistory),
  348. getLatestHistory: expressify(getLatestHistory),
  349. getLatestHistoryRaw: expressify(getLatestHistoryRaw),
  350. getHistory: expressify(getHistory),
  351. getHistoryBefore: expressify(getHistoryBefore),
  352. getChanges: expressify(getChanges),
  353. getZip: expressify(getZip),
  354. createZip: expressify(createZip),
  355. deleteProject: expressify(deleteProject),
  356. createProjectBlob: expressify(createProjectBlob),
  357. getProjectBlob: expressify(getProjectBlob),
  358. headProjectBlob: expressify(headProjectBlob),
  359. copyProjectBlob: expressify(copyProjectBlob),
  360. }