projects.js 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429
  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. redisBuffer,
  17. HashCheckBlobStore,
  18. ProjectArchive,
  19. zipStore,
  20. } = require('../../storage')
  21. const render = require('./render')
  22. const expressify = require('./expressify')
  23. const withTmpDir = require('./with_tmp_dir')
  24. const StreamSizeLimit = require('./stream_size_limit')
  25. const { getProjectBlobsBatch } = require('../../storage/lib/blob_store')
  26. const assert = require('../../storage/lib/assert')
  27. const pipeline = promisify(Stream.pipeline)
  28. async function initializeProject(req, res, next) {
  29. let projectId = req.swagger.params.body.value.projectId
  30. try {
  31. projectId = await chunkStore.initializeProject(projectId)
  32. res.status(HTTPStatus.OK).json({ projectId })
  33. } catch (err) {
  34. if (err instanceof chunkStore.AlreadyInitialized) {
  35. logger.warn({ err, projectId }, 'failed to initialize')
  36. render.conflict(res)
  37. } else {
  38. throw err
  39. }
  40. }
  41. }
  42. async function getLatestContent(req, res, next) {
  43. const projectId = req.swagger.params.project_id.value
  44. const blobStore = new BlobStore(projectId)
  45. const chunk = await chunkStore.loadLatest(projectId)
  46. const snapshot = chunk.getSnapshot()
  47. snapshot.applyAll(chunk.getChanges())
  48. await snapshot.loadFiles('eager', blobStore)
  49. res.json(snapshot.toRaw())
  50. }
  51. async function getContentAtVersion(req, res, next) {
  52. const projectId = req.swagger.params.project_id.value
  53. const version = req.swagger.params.version.value
  54. const blobStore = new BlobStore(projectId)
  55. const snapshot = await getSnapshotAtVersion(projectId, version)
  56. await snapshot.loadFiles('eager', blobStore)
  57. res.json(snapshot.toRaw())
  58. }
  59. async function getLatestHashedContent(req, res, next) {
  60. const projectId = req.swagger.params.project_id.value
  61. const blobStore = new HashCheckBlobStore(new BlobStore(projectId))
  62. const chunk = await chunkStore.loadLatest(projectId)
  63. const snapshot = chunk.getSnapshot()
  64. snapshot.applyAll(chunk.getChanges())
  65. await snapshot.loadFiles('eager', blobStore)
  66. const rawSnapshot = await snapshot.store(blobStore)
  67. res.json(rawSnapshot)
  68. }
  69. async function getLatestHistory(req, res, next) {
  70. const projectId = req.swagger.params.project_id.value
  71. try {
  72. const chunk = await chunkStore.loadLatest(projectId)
  73. const chunkResponse = new ChunkResponse(chunk)
  74. res.json(chunkResponse.toRaw())
  75. } catch (err) {
  76. if (err instanceof Chunk.NotFoundError) {
  77. render.notFound(res)
  78. } else {
  79. throw err
  80. }
  81. }
  82. }
  83. async function getLatestHistoryRaw(req, res, next) {
  84. const projectId = req.swagger.params.project_id.value
  85. const readOnly = req.swagger.params.readOnly.value
  86. try {
  87. const { startVersion, endVersion, endTimestamp } =
  88. await chunkStore.getLatestChunkMetadata(projectId, { readOnly })
  89. res.json({
  90. startVersion,
  91. endVersion,
  92. endTimestamp,
  93. })
  94. } catch (err) {
  95. if (err instanceof Chunk.NotFoundError) {
  96. render.notFound(res)
  97. } else {
  98. throw err
  99. }
  100. }
  101. }
  102. async function getHistory(req, res, next) {
  103. const projectId = req.swagger.params.project_id.value
  104. const version = req.swagger.params.version.value
  105. try {
  106. const chunk = await chunkStore.loadAtVersion(projectId, version)
  107. const chunkResponse = new ChunkResponse(chunk)
  108. res.json(chunkResponse.toRaw())
  109. } catch (err) {
  110. if (err instanceof Chunk.NotFoundError) {
  111. render.notFound(res)
  112. } else {
  113. throw err
  114. }
  115. }
  116. }
  117. async function getHistoryBefore(req, res, next) {
  118. const projectId = req.swagger.params.project_id.value
  119. const timestamp = req.swagger.params.timestamp.value
  120. try {
  121. const chunk = await chunkStore.loadAtTimestamp(projectId, timestamp)
  122. const chunkResponse = new ChunkResponse(chunk)
  123. res.json(chunkResponse.toRaw())
  124. } catch (err) {
  125. if (err instanceof Chunk.NotFoundError) {
  126. render.notFound(res)
  127. } else {
  128. throw err
  129. }
  130. }
  131. }
  132. /**
  133. * Get all changes since the beginning of history or since a given version
  134. */
  135. async function getChanges(req, res, next) {
  136. const projectId = req.swagger.params.project_id.value
  137. const since = req.swagger.params.since.value ?? 0
  138. if (since < 0) {
  139. // Negative values would cause an infinite loop
  140. return res.status(400).json({
  141. error: `Version out of bounds: ${since}`,
  142. })
  143. }
  144. try {
  145. const { changes, hasMore } = await chunkStore.getChangesSinceVersion(
  146. projectId,
  147. since
  148. )
  149. res.json({ changes: changes.map(change => change.toRaw()), hasMore })
  150. } catch (err) {
  151. if (err instanceof Chunk.VersionNotFoundError) {
  152. return res.status(400).json({
  153. error: `Version out of bounds: ${since}`,
  154. })
  155. }
  156. throw err
  157. }
  158. }
  159. async function getZip(req, res, next) {
  160. const projectId = req.swagger.params.project_id.value
  161. const version = req.swagger.params.version.value
  162. const blobStore = new BlobStore(projectId)
  163. let snapshot
  164. try {
  165. snapshot = await getSnapshotAtVersion(projectId, version)
  166. } catch (err) {
  167. if (err instanceof Chunk.NotFoundError) {
  168. return render.notFound(res)
  169. } else {
  170. throw err
  171. }
  172. }
  173. await withTmpDir('get-zip-', async tmpDir => {
  174. const tmpFilename = Path.join(tmpDir, 'project.zip')
  175. const archive = new ProjectArchive(snapshot)
  176. await archive.writeZip(blobStore, tmpFilename)
  177. res.set('Content-Type', 'application/octet-stream')
  178. res.set('Content-Disposition', 'attachment; filename=project.zip')
  179. const stream = fs.createReadStream(tmpFilename)
  180. await pipeline(stream, res)
  181. })
  182. }
  183. async function createZip(req, res, next) {
  184. const projectId = req.swagger.params.project_id.value
  185. const version = req.swagger.params.version.value
  186. try {
  187. const snapshot = await getSnapshotAtVersion(projectId, version)
  188. const zipUrl = await zipStore.getSignedUrl(projectId, version)
  189. // Do not await this; run it in the background.
  190. zipStore.storeZip(projectId, version, snapshot).catch(err => {
  191. logger.error({ err, projectId, version }, 'createZip: storeZip failed')
  192. })
  193. res.status(HTTPStatus.OK).json({ zipUrl })
  194. } catch (error) {
  195. if (error instanceof Chunk.NotFoundError) {
  196. render.notFound(res)
  197. } else {
  198. next(error)
  199. }
  200. }
  201. }
  202. async function deleteProject(req, res, next) {
  203. const projectId = req.swagger.params.project_id.value
  204. const blobStore = new BlobStore(projectId)
  205. await Promise.all([
  206. redisBuffer.hardDeleteProject(projectId),
  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. function sumUpByteLength(blobs) {
  343. return blobs.reduce((sum, blob) => sum + blob.getByteLength(), 0)
  344. }
  345. async function getProjectBlobsStats(req, res) {
  346. const projectIds = req.swagger.params.body.value.projectIds
  347. const { blobs } = await getProjectBlobsBatch(
  348. projectIds.map(id => {
  349. if (assert.POSTGRES_ID_REGEXP.test(id)) {
  350. return parseInt(id, 10)
  351. } else {
  352. return id
  353. }
  354. })
  355. )
  356. const sizes = []
  357. for (const projectId of projectIds) {
  358. const projectBlobs = blobs.get(projectId) || []
  359. const textBlobs = projectBlobs.filter(b => b.getStringLength() !== null)
  360. const binaryBlobs = projectBlobs.filter(b => b.getStringLength() === null)
  361. const textBlobBytes = sumUpByteLength(textBlobs)
  362. const binaryBlobBytes = sumUpByteLength(binaryBlobs)
  363. sizes.push({
  364. projectId,
  365. textBlobBytes,
  366. binaryBlobBytes,
  367. totalBytes: textBlobBytes + binaryBlobBytes,
  368. nTextBlobs: textBlobs.length,
  369. nBinaryBlobs: binaryBlobs.length,
  370. })
  371. }
  372. res.json(sizes)
  373. }
  374. module.exports = {
  375. initializeProject: expressify(initializeProject),
  376. getLatestContent: expressify(getLatestContent),
  377. getContentAtVersion: expressify(getContentAtVersion),
  378. getLatestHashedContent: expressify(getLatestHashedContent),
  379. getLatestPersistedHistory: expressify(getLatestHistory),
  380. getLatestHistory: expressify(getLatestHistory),
  381. getLatestHistoryRaw: expressify(getLatestHistoryRaw),
  382. getHistory: expressify(getHistory),
  383. getHistoryBefore: expressify(getHistoryBefore),
  384. getChanges: expressify(getChanges),
  385. getZip: expressify(getZip),
  386. createZip: expressify(createZip),
  387. deleteProject: expressify(deleteProject),
  388. createProjectBlob: expressify(createProjectBlob),
  389. getProjectBlob: expressify(getProjectBlob),
  390. headProjectBlob: expressify(headProjectBlob),
  391. copyProjectBlob: expressify(copyProjectBlob),
  392. getProjectBlobsStats: expressify(getProjectBlobsStats),
  393. }