projects.js 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436
  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. let chunk
  145. try {
  146. chunk = await chunkStore.loadAtVersion(projectId, since, {
  147. preferNewer: true,
  148. })
  149. } catch (err) {
  150. if (err instanceof Chunk.VersionNotFoundError) {
  151. return res.status(400).json({
  152. error: `Version out of bounds: ${since}`,
  153. })
  154. }
  155. throw err
  156. }
  157. const latestChunkMetadata = await chunkStore.getLatestChunkMetadata(projectId)
  158. // Extract the relevant changes from the chunk that contains the start version
  159. const changes = chunk.getChanges().slice(since - chunk.getStartVersion())
  160. const hasMore = latestChunkMetadata.endVersion > chunk.getEndVersion()
  161. res.json({ changes: changes.map(change => change.toRaw()), hasMore })
  162. }
  163. async function getZip(req, res, next) {
  164. const projectId = req.swagger.params.project_id.value
  165. const version = req.swagger.params.version.value
  166. const blobStore = new BlobStore(projectId)
  167. let snapshot
  168. try {
  169. snapshot = await getSnapshotAtVersion(projectId, version)
  170. } catch (err) {
  171. if (err instanceof Chunk.NotFoundError) {
  172. return render.notFound(res)
  173. } else {
  174. throw err
  175. }
  176. }
  177. await withTmpDir('get-zip-', async tmpDir => {
  178. const tmpFilename = Path.join(tmpDir, 'project.zip')
  179. const archive = new ProjectArchive(snapshot)
  180. await archive.writeZip(blobStore, tmpFilename)
  181. res.set('Content-Type', 'application/octet-stream')
  182. res.set('Content-Disposition', 'attachment; filename=project.zip')
  183. const stream = fs.createReadStream(tmpFilename)
  184. await pipeline(stream, res)
  185. })
  186. }
  187. async function createZip(req, res, next) {
  188. const projectId = req.swagger.params.project_id.value
  189. const version = req.swagger.params.version.value
  190. try {
  191. const snapshot = await getSnapshotAtVersion(projectId, version)
  192. const zipUrl = await zipStore.getSignedUrl(projectId, version)
  193. // Do not await this; run it in the background.
  194. zipStore.storeZip(projectId, version, snapshot).catch(err => {
  195. logger.error({ err, projectId, version }, 'createZip: storeZip failed')
  196. })
  197. res.status(HTTPStatus.OK).json({ zipUrl })
  198. } catch (error) {
  199. if (error instanceof Chunk.NotFoundError) {
  200. render.notFound(res)
  201. } else {
  202. next(error)
  203. }
  204. }
  205. }
  206. async function deleteProject(req, res, next) {
  207. const projectId = req.swagger.params.project_id.value
  208. const blobStore = new BlobStore(projectId)
  209. await Promise.all([
  210. redisBuffer.hardDeleteProject(projectId),
  211. chunkStore.deleteProjectChunks(projectId),
  212. blobStore.deleteBlobs(),
  213. ])
  214. res.status(HTTPStatus.NO_CONTENT).send()
  215. }
  216. async function createProjectBlob(req, res, next) {
  217. const projectId = req.swagger.params.project_id.value
  218. const expectedHash = req.swagger.params.hash.value
  219. const maxUploadSize = parseInt(config.get('maxFileUploadSize'), 10)
  220. await withTmpDir('blob-', async tmpDir => {
  221. const tmpPath = Path.join(tmpDir, 'content')
  222. const sizeLimit = new StreamSizeLimit(maxUploadSize)
  223. await pipeline(req, sizeLimit, fs.createWriteStream(tmpPath))
  224. if (sizeLimit.sizeLimitExceeded) {
  225. logger.warn(
  226. { projectId, expectedHash, maxUploadSize },
  227. 'blob exceeds size threshold'
  228. )
  229. return render.requestEntityTooLarge(res)
  230. }
  231. const hash = await blobHash.fromFile(tmpPath)
  232. if (hash !== expectedHash) {
  233. logger.warn({ projectId, hash, expectedHash }, 'Hash mismatch')
  234. return render.conflict(res, 'File hash mismatch')
  235. }
  236. const blobStore = new BlobStore(projectId)
  237. const newBlob = await blobStore.putFile(tmpPath)
  238. if (config.has('backupStore')) {
  239. try {
  240. const { backupBlob } = await import('../../storage/lib/backupBlob.mjs')
  241. await backupBlob(projectId, newBlob, tmpPath)
  242. } catch (error) {
  243. logger.warn({ error, projectId, hash }, 'Failed to backup blob')
  244. }
  245. }
  246. res.status(HTTPStatus.CREATED).end()
  247. })
  248. }
  249. async function headProjectBlob(req, res) {
  250. const projectId = req.swagger.params.project_id.value
  251. const hash = req.swagger.params.hash.value
  252. const blobStore = new BlobStore(projectId)
  253. const blob = await blobStore.getBlob(hash)
  254. if (blob) {
  255. res.set('Content-Length', blob.getByteLength())
  256. res.status(200).end()
  257. } else {
  258. res.status(404).end()
  259. }
  260. }
  261. // Support simple, singular ranges starting from zero only, up-to 2MB = 2_000_000, 7 digits
  262. const RANGE_HEADER = /^bytes=0-(\d{1,7})$/
  263. /**
  264. * @param {string} header
  265. * @return {{}|{start: number, end: number}}
  266. * @private
  267. */
  268. function _getRangeOpts(header) {
  269. if (!header) return {}
  270. const match = header.match(RANGE_HEADER)
  271. if (match) {
  272. const end = parseInt(match[1], 10)
  273. return { start: 0, end }
  274. }
  275. return {}
  276. }
  277. async function getProjectBlob(req, res, next) {
  278. const projectId = req.swagger.params.project_id.value
  279. const hash = req.swagger.params.hash.value
  280. const opts = _getRangeOpts(req.swagger.params.range.value || '')
  281. const blobStore = new BlobStore(projectId)
  282. logger.debug({ projectId, hash }, 'getProjectBlob started')
  283. try {
  284. let stream
  285. try {
  286. stream = await blobStore.getStream(hash, opts)
  287. } catch (err) {
  288. if (err instanceof Blob.NotFoundError) {
  289. logger.warn({ projectId, hash }, 'Blob not found')
  290. return res.status(404).end()
  291. } else {
  292. throw err
  293. }
  294. }
  295. res.set('Content-Type', 'application/octet-stream')
  296. try {
  297. await pipeline(stream, res)
  298. } catch (err) {
  299. if (err?.code === 'ERR_STREAM_PREMATURE_CLOSE') {
  300. res.end()
  301. } else {
  302. throw OError.tag(err, 'error transferring stream', { projectId, hash })
  303. }
  304. }
  305. } finally {
  306. logger.debug({ projectId, hash }, 'getProjectBlob finished')
  307. }
  308. }
  309. async function copyProjectBlob(req, res, next) {
  310. const sourceProjectId = req.swagger.params.copyFrom.value
  311. const targetProjectId = req.swagger.params.project_id.value
  312. const blobHash = req.swagger.params.hash.value
  313. // Check that blob exists in source project
  314. const sourceBlobStore = new BlobStore(sourceProjectId)
  315. const targetBlobStore = new BlobStore(targetProjectId)
  316. const [sourceBlob, targetBlob] = await Promise.all([
  317. sourceBlobStore.getBlob(blobHash),
  318. targetBlobStore.getBlob(blobHash),
  319. ])
  320. if (!sourceBlob) {
  321. logger.warn(
  322. { sourceProjectId, targetProjectId, blobHash },
  323. 'missing source blob when copying across projects'
  324. )
  325. return render.notFound(res)
  326. }
  327. // Exit early if the blob exists in the target project.
  328. // This will also catch global blobs, which always exist.
  329. if (targetBlob) {
  330. return res.status(HTTPStatus.NO_CONTENT).end()
  331. }
  332. // Otherwise, copy blob from source project to target project
  333. await sourceBlobStore.copyBlob(sourceBlob, targetProjectId)
  334. res.status(HTTPStatus.CREATED).end()
  335. }
  336. async function getSnapshotAtVersion(projectId, version) {
  337. const chunk = await chunkStore.loadAtVersion(projectId, version)
  338. const snapshot = chunk.getSnapshot()
  339. const changes = _.dropRight(
  340. chunk.getChanges(),
  341. chunk.getEndVersion() - version
  342. )
  343. snapshot.applyAll(changes)
  344. return snapshot
  345. }
  346. function sumUpByteLength(blobs) {
  347. return blobs.reduce((sum, blob) => sum + blob.getByteLength(), 0)
  348. }
  349. async function getProjectBlobsStats(req, res) {
  350. const projectIds = req.swagger.params.body.value.projectIds
  351. const { blobs } = await getProjectBlobsBatch(
  352. projectIds.map(id => {
  353. if (assert.POSTGRES_ID_REGEXP.test(id)) {
  354. return parseInt(id, 10)
  355. } else {
  356. return id
  357. }
  358. })
  359. )
  360. const sizes = []
  361. for (const projectId of projectIds) {
  362. const projectBlobs = blobs.get(projectId) || []
  363. const textBlobs = projectBlobs.filter(b => b.getStringLength() !== null)
  364. const binaryBlobs = projectBlobs.filter(b => b.getStringLength() === null)
  365. const textBlobBytes = sumUpByteLength(textBlobs)
  366. const binaryBlobBytes = sumUpByteLength(binaryBlobs)
  367. sizes.push({
  368. projectId,
  369. textBlobBytes,
  370. binaryBlobBytes,
  371. totalBytes: textBlobBytes + binaryBlobBytes,
  372. nTextBlobs: textBlobs.length,
  373. nBinaryBlobs: binaryBlobs.length,
  374. })
  375. }
  376. res.json(sizes)
  377. }
  378. module.exports = {
  379. initializeProject: expressify(initializeProject),
  380. getLatestContent: expressify(getLatestContent),
  381. getContentAtVersion: expressify(getContentAtVersion),
  382. getLatestHashedContent: expressify(getLatestHashedContent),
  383. getLatestPersistedHistory: expressify(getLatestHistory),
  384. getLatestHistory: expressify(getLatestHistory),
  385. getLatestHistoryRaw: expressify(getLatestHistoryRaw),
  386. getHistory: expressify(getHistory),
  387. getHistoryBefore: expressify(getHistoryBefore),
  388. getChanges: expressify(getChanges),
  389. getZip: expressify(getZip),
  390. createZip: expressify(createZip),
  391. deleteProject: expressify(deleteProject),
  392. createProjectBlob: expressify(createProjectBlob),
  393. getProjectBlob: expressify(getProjectBlob),
  394. headProjectBlob: expressify(headProjectBlob),
  395. copyProjectBlob: expressify(copyProjectBlob),
  396. getProjectBlobsStats: expressify(getProjectBlobsStats),
  397. }