projects.js 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455
  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. BatchBlobStore,
  15. blobHash,
  16. chunkStore,
  17. redisBuffer,
  18. HashCheckBlobStore,
  19. ProjectArchive,
  20. zipStore,
  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 { getProjectBlobsBatch } = require('../../storage/lib/blob_store')
  27. const assert = require('../../storage/lib/assert')
  28. const pipeline = promisify(Stream.pipeline)
  29. async function initializeProject(req, res, next) {
  30. let projectId = req.swagger.params.body.value.projectId
  31. try {
  32. projectId = await chunkStore.initializeProject(projectId)
  33. res.status(HTTPStatus.OK).json({ projectId })
  34. } catch (err) {
  35. if (err instanceof chunkStore.AlreadyInitialized) {
  36. logger.warn({ err, projectId }, 'failed to initialize')
  37. render.conflict(res)
  38. } else {
  39. throw err
  40. }
  41. }
  42. }
  43. async function getLatestContent(req, res, next) {
  44. const projectId = req.swagger.params.project_id.value
  45. const blobStore = new BlobStore(projectId)
  46. const chunk = await chunkStore.loadLatest(projectId)
  47. const snapshot = chunk.getSnapshot()
  48. snapshot.applyAll(chunk.getChanges())
  49. await snapshot.loadFiles('eager', blobStore)
  50. res.json(snapshot.toRaw())
  51. }
  52. async function getContentAtVersion(req, res, next) {
  53. const projectId = req.swagger.params.project_id.value
  54. const version = req.swagger.params.version.value
  55. const blobStore = new BlobStore(projectId)
  56. const snapshot = await getSnapshotAtVersion(projectId, version)
  57. await snapshot.loadFiles('eager', blobStore)
  58. res.json(snapshot.toRaw())
  59. }
  60. async function getLatestHashedContent(req, res, next) {
  61. const projectId = req.swagger.params.project_id.value
  62. const blobStore = new HashCheckBlobStore(new BlobStore(projectId))
  63. const chunk = await chunkStore.loadLatest(projectId)
  64. const snapshot = chunk.getSnapshot()
  65. snapshot.applyAll(chunk.getChanges())
  66. await snapshot.loadFiles('eager', blobStore)
  67. const rawSnapshot = await snapshot.store(blobStore)
  68. res.json(rawSnapshot)
  69. }
  70. async function getLatestHistory(req, res, next) {
  71. const projectId = req.swagger.params.project_id.value
  72. try {
  73. const chunk = await chunkStore.loadLatest(projectId)
  74. const chunkResponse = new ChunkResponse(chunk)
  75. res.json(chunkResponse.toRaw())
  76. } catch (err) {
  77. if (err instanceof Chunk.NotFoundError) {
  78. render.notFound(res)
  79. } else {
  80. throw err
  81. }
  82. }
  83. }
  84. async function getLatestHistoryRaw(req, res, next) {
  85. const projectId = req.swagger.params.project_id.value
  86. const readOnly = req.swagger.params.readOnly.value
  87. try {
  88. const { startVersion, endVersion, endTimestamp } =
  89. await chunkStore.getLatestChunkMetadata(projectId, { readOnly })
  90. res.json({
  91. startVersion,
  92. endVersion,
  93. endTimestamp,
  94. })
  95. } catch (err) {
  96. if (err instanceof Chunk.NotFoundError) {
  97. render.notFound(res)
  98. } else {
  99. throw err
  100. }
  101. }
  102. }
  103. async function getHistory(req, res, next) {
  104. const projectId = req.swagger.params.project_id.value
  105. const version = req.swagger.params.version.value
  106. try {
  107. const chunk = await chunkStore.loadAtVersion(projectId, version)
  108. const chunkResponse = new ChunkResponse(chunk)
  109. res.json(chunkResponse.toRaw())
  110. } catch (err) {
  111. if (err instanceof Chunk.NotFoundError) {
  112. render.notFound(res)
  113. } else {
  114. throw err
  115. }
  116. }
  117. }
  118. async function getHistoryBefore(req, res, next) {
  119. const projectId = req.swagger.params.project_id.value
  120. const timestamp = req.swagger.params.timestamp.value
  121. try {
  122. const chunk = await chunkStore.loadAtTimestamp(projectId, timestamp)
  123. const chunkResponse = new ChunkResponse(chunk)
  124. res.json(chunkResponse.toRaw())
  125. } catch (err) {
  126. if (err instanceof Chunk.NotFoundError) {
  127. render.notFound(res)
  128. } else {
  129. throw err
  130. }
  131. }
  132. }
  133. /**
  134. * Get all changes since the beginning of history or since a given version
  135. */
  136. async function getChanges(req, res, next) {
  137. const projectId = req.swagger.params.project_id.value
  138. const since = req.swagger.params.since.value ?? 0
  139. if (since < 0) {
  140. // Negative values would cause an infinite loop
  141. return res.status(400).json({
  142. error: `Version out of bounds: ${since}`,
  143. })
  144. }
  145. try {
  146. const { changes, hasMore } = await chunkStore.getChangesSinceVersion(
  147. projectId,
  148. since
  149. )
  150. res.json({ changes: changes.map(change => change.toRaw()), hasMore })
  151. } catch (err) {
  152. if (err instanceof Chunk.VersionNotFoundError) {
  153. return res.status(400).json({
  154. error: `Version out of bounds: ${since}`,
  155. })
  156. }
  157. throw err
  158. }
  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. redisBuffer.hardDeleteProject(projectId),
  208. chunkStore.deleteProjectChunks(projectId),
  209. blobStore.deleteBlobs(),
  210. ])
  211. res.status(HTTPStatus.NO_CONTENT).send()
  212. }
  213. async function createProjectBlob(req, res, next) {
  214. const projectId = req.swagger.params.project_id.value
  215. const expectedHash = req.swagger.params.hash.value
  216. const maxUploadSize = parseInt(config.get('maxFileUploadSize'), 10)
  217. await withTmpDir('blob-', async tmpDir => {
  218. const tmpPath = Path.join(tmpDir, 'content')
  219. const sizeLimit = new StreamSizeLimit(maxUploadSize)
  220. await pipeline(req, sizeLimit, fs.createWriteStream(tmpPath))
  221. if (sizeLimit.sizeLimitExceeded) {
  222. logger.warn(
  223. { projectId, expectedHash, maxUploadSize },
  224. 'blob exceeds size threshold'
  225. )
  226. return render.requestEntityTooLarge(res)
  227. }
  228. const hash = await blobHash.fromFile(tmpPath)
  229. if (hash !== expectedHash) {
  230. logger.warn({ projectId, hash, expectedHash }, 'Hash mismatch')
  231. return render.conflict(res, 'File hash mismatch')
  232. }
  233. const blobStore = new BlobStore(projectId)
  234. const newBlob = await blobStore.putFile(tmpPath)
  235. if (config.has('backupStore')) {
  236. try {
  237. const { backupBlob } = await import('../../storage/lib/backupBlob.mjs')
  238. await backupBlob(projectId, newBlob, tmpPath)
  239. } catch (error) {
  240. logger.warn({ error, projectId, hash }, 'Failed to backup blob')
  241. }
  242. }
  243. res.status(HTTPStatus.CREATED).end()
  244. })
  245. }
  246. async function headProjectBlob(req, res) {
  247. const projectId = req.swagger.params.project_id.value
  248. const hash = req.swagger.params.hash.value
  249. const blobStore = new BlobStore(projectId)
  250. const blob = await blobStore.getBlob(hash)
  251. if (blob) {
  252. res.set('Content-Length', blob.getByteLength())
  253. res.status(200).end()
  254. } else {
  255. res.status(404).end()
  256. }
  257. }
  258. // Support simple, singular ranges starting from zero only, up-to 2MB = 2_000_000, 7 digits
  259. const RANGE_HEADER = /^bytes=0-(\d{1,7})$/
  260. /**
  261. * @param {string} header
  262. * @return {{}|{start: number, end: number}}
  263. * @private
  264. */
  265. function _getRangeOpts(header) {
  266. if (!header) return {}
  267. const match = header.match(RANGE_HEADER)
  268. if (match) {
  269. const end = parseInt(match[1], 10)
  270. return { start: 0, end }
  271. }
  272. return {}
  273. }
  274. async function getProjectBlob(req, res, next) {
  275. const projectId = req.swagger.params.project_id.value
  276. const hash = req.swagger.params.hash.value
  277. const opts = _getRangeOpts(req.swagger.params.range.value || '')
  278. const blobStore = new BlobStore(projectId)
  279. logger.debug({ projectId, hash }, 'getProjectBlob started')
  280. try {
  281. let stream
  282. try {
  283. stream = await blobStore.getStream(hash, opts)
  284. } catch (err) {
  285. if (err instanceof Blob.NotFoundError) {
  286. logger.warn({ projectId, hash }, 'Blob not found')
  287. return res.status(404).end()
  288. } else {
  289. throw err
  290. }
  291. }
  292. res.set('Content-Type', 'application/octet-stream')
  293. try {
  294. await pipeline(stream, res)
  295. } catch (err) {
  296. if (err?.code === 'ERR_STREAM_PREMATURE_CLOSE') {
  297. res.end()
  298. } else {
  299. throw OError.tag(err, 'error transferring stream', { projectId, hash })
  300. }
  301. }
  302. } finally {
  303. logger.debug({ projectId, hash }, 'getProjectBlob finished')
  304. }
  305. }
  306. async function copyProjectBlob(req, res, next) {
  307. const sourceProjectId = req.swagger.params.copyFrom.value
  308. const targetProjectId = req.swagger.params.project_id.value
  309. const blobHash = req.swagger.params.hash.value
  310. // Check that blob exists in source project
  311. const sourceBlobStore = new BlobStore(sourceProjectId)
  312. const targetBlobStore = new BlobStore(targetProjectId)
  313. const [sourceBlob, targetBlob] = await Promise.all([
  314. sourceBlobStore.getBlob(blobHash),
  315. targetBlobStore.getBlob(blobHash),
  316. ])
  317. if (!sourceBlob) {
  318. logger.warn(
  319. { sourceProjectId, targetProjectId, blobHash },
  320. 'missing source blob when copying across projects'
  321. )
  322. return render.notFound(res)
  323. }
  324. // Exit early if the blob exists in the target project.
  325. // This will also catch global blobs, which always exist.
  326. if (targetBlob) {
  327. return res.status(HTTPStatus.NO_CONTENT).end()
  328. }
  329. // Otherwise, copy blob from source project to target project
  330. await sourceBlobStore.copyBlob(sourceBlob, targetProjectId)
  331. res.status(HTTPStatus.CREATED).end()
  332. }
  333. async function getSnapshotAtVersion(projectId, version) {
  334. const chunk = await chunkStore.loadAtVersion(projectId, version)
  335. const snapshot = chunk.getSnapshot()
  336. const changes = _.dropRight(
  337. chunk.getChanges(),
  338. chunk.getEndVersion() - version
  339. )
  340. snapshot.applyAll(changes)
  341. return snapshot
  342. }
  343. function sumUpByteLength(blobs) {
  344. return blobs.reduce((sum, blob) => sum + blob.getByteLength(), 0)
  345. }
  346. async function getBlobStats(req, res) {
  347. const projectId = req.swagger.params.project_id.value
  348. const blobHashes = req.swagger.params.body.value.blobHashes || []
  349. for (const hash of blobHashes) {
  350. assert.blobHash(hash, 'bad hash')
  351. }
  352. const blobStore = new BlobStore(projectId)
  353. const batchBlobStore = new BatchBlobStore(blobStore)
  354. await batchBlobStore.preload(Array.from(blobHashes))
  355. const blobs = Array.from(batchBlobStore.blobs.values()).filter(Boolean)
  356. const textBlobs = blobs.filter(b => b.getStringLength() !== null)
  357. const binaryBlobs = blobs.filter(b => b.getStringLength() === null)
  358. const textBlobBytes = sumUpByteLength(textBlobs)
  359. const binaryBlobBytes = sumUpByteLength(binaryBlobs)
  360. res.json({
  361. projectId,
  362. textBlobBytes,
  363. binaryBlobBytes,
  364. totalBytes: textBlobBytes + binaryBlobBytes,
  365. nTextBlobs: textBlobs.length,
  366. nBinaryBlobs: binaryBlobs.length,
  367. })
  368. }
  369. async function getProjectBlobsStats(req, res) {
  370. const projectIds = req.swagger.params.body.value.projectIds
  371. const { blobs } = await getProjectBlobsBatch(
  372. projectIds.map(id => {
  373. if (assert.POSTGRES_ID_REGEXP.test(id)) {
  374. return parseInt(id, 10)
  375. } else {
  376. return id
  377. }
  378. })
  379. )
  380. const sizes = []
  381. for (const projectId of projectIds) {
  382. const projectBlobs = blobs.get(projectId) || []
  383. const textBlobs = projectBlobs.filter(b => b.getStringLength() !== null)
  384. const binaryBlobs = projectBlobs.filter(b => b.getStringLength() === null)
  385. const textBlobBytes = sumUpByteLength(textBlobs)
  386. const binaryBlobBytes = sumUpByteLength(binaryBlobs)
  387. sizes.push({
  388. projectId,
  389. textBlobBytes,
  390. binaryBlobBytes,
  391. totalBytes: textBlobBytes + binaryBlobBytes,
  392. nTextBlobs: textBlobs.length,
  393. nBinaryBlobs: binaryBlobs.length,
  394. })
  395. }
  396. res.json(sizes)
  397. }
  398. module.exports = {
  399. initializeProject: expressify(initializeProject),
  400. getLatestContent: expressify(getLatestContent),
  401. getContentAtVersion: expressify(getContentAtVersion),
  402. getLatestHashedContent: expressify(getLatestHashedContent),
  403. getLatestPersistedHistory: expressify(getLatestHistory),
  404. getLatestHistory: expressify(getLatestHistory),
  405. getLatestHistoryRaw: expressify(getLatestHistoryRaw),
  406. getHistory: expressify(getHistory),
  407. getHistoryBefore: expressify(getHistoryBefore),
  408. getChanges: expressify(getChanges),
  409. getZip: expressify(getZip),
  410. createZip: expressify(createZip),
  411. deleteProject: expressify(deleteProject),
  412. createProjectBlob: expressify(createProjectBlob),
  413. getProjectBlob: expressify(getProjectBlob),
  414. headProjectBlob: expressify(headProjectBlob),
  415. copyProjectBlob: expressify(copyProjectBlob),
  416. getBlobStats: expressify(getBlobStats),
  417. getProjectBlobsStats: expressify(getProjectBlobsStats),
  418. }