UrlCache.js 7.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275
  1. /* eslint-disable
  2. camelcase,
  3. handle-callback-err,
  4. no-return-assign,
  5. */
  6. // TODO: This file was created by bulk-decaffeinate.
  7. // Fix any style issues and re-enable lint.
  8. /*
  9. * decaffeinate suggestions:
  10. * DS101: Remove unnecessary use of Array.from
  11. * DS102: Remove unnecessary code created because of implicit returns
  12. * DS207: Consider shorter variations of null checks
  13. * Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
  14. */
  15. let UrlCache
  16. const db = require('./db')
  17. const dbQueue = require('./DbQueue')
  18. const UrlFetcher = require('./UrlFetcher')
  19. const Settings = require('settings-sharelatex')
  20. const crypto = require('crypto')
  21. const fs = require('fs')
  22. const logger = require('logger-sharelatex')
  23. const async = require('async')
  24. const Metrics = require('./Metrics')
  25. module.exports = UrlCache = {
  26. downloadUrlToFile(project_id, url, destPath, lastModified, callback) {
  27. if (callback == null) {
  28. callback = function (error) {}
  29. }
  30. return UrlCache._ensureUrlIsInCache(
  31. project_id,
  32. url,
  33. lastModified,
  34. (error, pathToCachedUrl) => {
  35. if (error != null) {
  36. return callback(error)
  37. }
  38. return fs.copyFile(pathToCachedUrl, destPath, function (error) {
  39. if (error != null) {
  40. logger.error(
  41. { err: error, from: pathToCachedUrl, to: destPath },
  42. 'error copying file from cache'
  43. )
  44. return UrlCache._clearUrlDetails(project_id, url, () =>
  45. callback(error)
  46. )
  47. } else {
  48. return callback(error)
  49. }
  50. })
  51. }
  52. )
  53. },
  54. clearProject(project_id, callback) {
  55. if (callback == null) {
  56. callback = function (error) {}
  57. }
  58. return UrlCache._findAllUrlsInProject(project_id, function (error, urls) {
  59. logger.log(
  60. { project_id, url_count: urls.length },
  61. 'clearing project URLs'
  62. )
  63. if (error != null) {
  64. return callback(error)
  65. }
  66. const jobs = Array.from(urls || []).map((url) =>
  67. ((url) => (callback) =>
  68. UrlCache._clearUrlFromCache(project_id, url, function (error) {
  69. if (error != null) {
  70. logger.error(
  71. { err: error, project_id, url },
  72. 'error clearing project URL'
  73. )
  74. }
  75. return callback()
  76. }))(url)
  77. )
  78. return async.series(jobs, callback)
  79. })
  80. },
  81. _ensureUrlIsInCache(project_id, url, lastModified, callback) {
  82. if (callback == null) {
  83. callback = function (error, pathOnDisk) {}
  84. }
  85. if (lastModified != null) {
  86. // MYSQL only stores dates to an accuracy of a second but the incoming lastModified might have milliseconds.
  87. // So round down to seconds
  88. lastModified = new Date(Math.floor(lastModified.getTime() / 1000) * 1000)
  89. }
  90. return UrlCache._doesUrlNeedDownloading(
  91. project_id,
  92. url,
  93. lastModified,
  94. (error, needsDownloading) => {
  95. if (error != null) {
  96. return callback(error)
  97. }
  98. if (needsDownloading) {
  99. logger.log({ url, lastModified }, 'downloading URL')
  100. return UrlFetcher.pipeUrlToFileWithRetry(
  101. url,
  102. UrlCache._cacheFilePathForUrl(project_id, url),
  103. (error) => {
  104. if (error != null) {
  105. return callback(error)
  106. }
  107. return UrlCache._updateOrCreateUrlDetails(
  108. project_id,
  109. url,
  110. lastModified,
  111. (error) => {
  112. if (error != null) {
  113. return callback(error)
  114. }
  115. return callback(
  116. null,
  117. UrlCache._cacheFilePathForUrl(project_id, url)
  118. )
  119. }
  120. )
  121. }
  122. )
  123. } else {
  124. logger.log({ url, lastModified }, 'URL is up to date in cache')
  125. return callback(null, UrlCache._cacheFilePathForUrl(project_id, url))
  126. }
  127. }
  128. )
  129. },
  130. _doesUrlNeedDownloading(project_id, url, lastModified, callback) {
  131. if (callback == null) {
  132. callback = function (error, needsDownloading) {}
  133. }
  134. if (lastModified == null) {
  135. return callback(null, true)
  136. }
  137. return UrlCache._findUrlDetails(project_id, url, function (
  138. error,
  139. urlDetails
  140. ) {
  141. if (error != null) {
  142. return callback(error)
  143. }
  144. if (
  145. urlDetails == null ||
  146. urlDetails.lastModified == null ||
  147. urlDetails.lastModified.getTime() < lastModified.getTime()
  148. ) {
  149. return callback(null, true)
  150. } else {
  151. return callback(null, false)
  152. }
  153. })
  154. },
  155. _cacheFileNameForUrl(project_id, url) {
  156. return project_id + ':' + crypto.createHash('md5').update(url).digest('hex')
  157. },
  158. _cacheFilePathForUrl(project_id, url) {
  159. return `${Settings.path.clsiCacheDir}/${UrlCache._cacheFileNameForUrl(
  160. project_id,
  161. url
  162. )}`
  163. },
  164. _clearUrlFromCache(project_id, url, callback) {
  165. if (callback == null) {
  166. callback = function (error) {}
  167. }
  168. return UrlCache._clearUrlDetails(project_id, url, function (error) {
  169. if (error != null) {
  170. return callback(error)
  171. }
  172. return UrlCache._deleteUrlCacheFromDisk(project_id, url, function (
  173. error
  174. ) {
  175. if (error != null) {
  176. return callback(error)
  177. }
  178. return callback(null)
  179. })
  180. })
  181. },
  182. _deleteUrlCacheFromDisk(project_id, url, callback) {
  183. if (callback == null) {
  184. callback = function (error) {}
  185. }
  186. return fs.unlink(UrlCache._cacheFilePathForUrl(project_id, url), function (
  187. error
  188. ) {
  189. if (error != null && error.code !== 'ENOENT') {
  190. // no error if the file isn't present
  191. return callback(error)
  192. } else {
  193. return callback()
  194. }
  195. })
  196. },
  197. _findUrlDetails(project_id, url, callback) {
  198. if (callback == null) {
  199. callback = function (error, urlDetails) {}
  200. }
  201. const timer = new Metrics.Timer('db-find-url-details')
  202. const job = (cb) =>
  203. db.UrlCache.findOne({ where: { url, project_id } })
  204. .then((urlDetails) => cb(null, urlDetails))
  205. .error(cb)
  206. dbQueue.queue.push(job, (error, urlDetails) => {
  207. timer.done()
  208. callback(error, urlDetails)
  209. })
  210. },
  211. _updateOrCreateUrlDetails(project_id, url, lastModified, callback) {
  212. if (callback == null) {
  213. callback = function (error) {}
  214. }
  215. const timer = new Metrics.Timer('db-update-or-create-url-details')
  216. const job = (cb) =>
  217. db.UrlCache.findOrCreate({ where: { url, project_id } })
  218. .spread((urlDetails, created) =>
  219. urlDetails
  220. .update({ lastModified })
  221. .then(() => cb())
  222. .error(cb)
  223. )
  224. .error(cb)
  225. dbQueue.queue.push(job, (error) => {
  226. timer.done()
  227. callback(error)
  228. })
  229. },
  230. _clearUrlDetails(project_id, url, callback) {
  231. if (callback == null) {
  232. callback = function (error) {}
  233. }
  234. const timer = new Metrics.Timer('db-clear-url-details')
  235. const job = (cb) =>
  236. db.UrlCache.destroy({ where: { url, project_id } })
  237. .then(() => cb(null))
  238. .error(cb)
  239. dbQueue.queue.push(job, (error) => {
  240. timer.done()
  241. callback(error)
  242. })
  243. },
  244. _findAllUrlsInProject(project_id, callback) {
  245. if (callback == null) {
  246. callback = function (error, urls) {}
  247. }
  248. const timer = new Metrics.Timer('db-find-urls-in-project')
  249. const job = (cb) =>
  250. db.UrlCache.findAll({ where: { project_id } })
  251. .then((urlEntries) =>
  252. cb(
  253. null,
  254. urlEntries.map((entry) => entry.url)
  255. )
  256. )
  257. .error(cb)
  258. dbQueue.queue.push(job, (err, urls) => {
  259. timer.done()
  260. callback(err, urls)
  261. })
  262. }
  263. }