UrlCache.js 7.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258
  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. module.exports = UrlCache = {
  25. downloadUrlToFile(project_id, url, destPath, lastModified, callback) {
  26. if (callback == null) {
  27. callback = function (error) {}
  28. }
  29. return UrlCache._ensureUrlIsInCache(
  30. project_id,
  31. url,
  32. lastModified,
  33. (error, pathToCachedUrl) => {
  34. if (error != null) {
  35. return callback(error)
  36. }
  37. return fs.copyFile(pathToCachedUrl, destPath, function (error) {
  38. if (error != null) {
  39. logger.error(
  40. { err: error, from: pathToCachedUrl, to: destPath },
  41. 'error copying file from cache'
  42. )
  43. return UrlCache._clearUrlDetails(project_id, url, () =>
  44. callback(error)
  45. )
  46. } else {
  47. return callback(error)
  48. }
  49. })
  50. }
  51. )
  52. },
  53. clearProject(project_id, callback) {
  54. if (callback == null) {
  55. callback = function (error) {}
  56. }
  57. return UrlCache._findAllUrlsInProject(project_id, function (error, urls) {
  58. logger.log(
  59. { project_id, url_count: urls.length },
  60. 'clearing project URLs'
  61. )
  62. if (error != null) {
  63. return callback(error)
  64. }
  65. const jobs = Array.from(urls || []).map((url) =>
  66. ((url) => (callback) =>
  67. UrlCache._clearUrlFromCache(project_id, url, function (error) {
  68. if (error != null) {
  69. logger.error(
  70. { err: error, project_id, url },
  71. 'error clearing project URL'
  72. )
  73. }
  74. return callback()
  75. }))(url)
  76. )
  77. return async.series(jobs, callback)
  78. })
  79. },
  80. _ensureUrlIsInCache(project_id, url, lastModified, callback) {
  81. if (callback == null) {
  82. callback = function (error, pathOnDisk) {}
  83. }
  84. if (lastModified != null) {
  85. // MYSQL only stores dates to an accuracy of a second but the incoming lastModified might have milliseconds.
  86. // So round down to seconds
  87. lastModified = new Date(Math.floor(lastModified.getTime() / 1000) * 1000)
  88. }
  89. return UrlCache._doesUrlNeedDownloading(
  90. project_id,
  91. url,
  92. lastModified,
  93. (error, needsDownloading) => {
  94. if (error != null) {
  95. return callback(error)
  96. }
  97. if (needsDownloading) {
  98. logger.log({ url, lastModified }, 'downloading URL')
  99. return UrlFetcher.pipeUrlToFileWithRetry(
  100. url,
  101. UrlCache._cacheFilePathForUrl(project_id, url),
  102. (error) => {
  103. if (error != null) {
  104. return callback(error)
  105. }
  106. return UrlCache._updateOrCreateUrlDetails(
  107. project_id,
  108. url,
  109. lastModified,
  110. (error) => {
  111. if (error != null) {
  112. return callback(error)
  113. }
  114. return callback(
  115. null,
  116. UrlCache._cacheFilePathForUrl(project_id, url)
  117. )
  118. }
  119. )
  120. }
  121. )
  122. } else {
  123. logger.log({ url, lastModified }, 'URL is up to date in cache')
  124. return callback(null, UrlCache._cacheFilePathForUrl(project_id, url))
  125. }
  126. }
  127. )
  128. },
  129. _doesUrlNeedDownloading(project_id, url, lastModified, callback) {
  130. if (callback == null) {
  131. callback = function (error, needsDownloading) {}
  132. }
  133. if (lastModified == null) {
  134. return callback(null, true)
  135. }
  136. return UrlCache._findUrlDetails(project_id, url, function (
  137. error,
  138. urlDetails
  139. ) {
  140. if (error != null) {
  141. return callback(error)
  142. }
  143. if (
  144. urlDetails == null ||
  145. urlDetails.lastModified == null ||
  146. urlDetails.lastModified.getTime() < lastModified.getTime()
  147. ) {
  148. return callback(null, true)
  149. } else {
  150. return callback(null, false)
  151. }
  152. })
  153. },
  154. _cacheFileNameForUrl(project_id, url) {
  155. return project_id + ':' + crypto.createHash('md5').update(url).digest('hex')
  156. },
  157. _cacheFilePathForUrl(project_id, url) {
  158. return `${Settings.path.clsiCacheDir}/${UrlCache._cacheFileNameForUrl(
  159. project_id,
  160. url
  161. )}`
  162. },
  163. _clearUrlFromCache(project_id, url, callback) {
  164. if (callback == null) {
  165. callback = function (error) {}
  166. }
  167. return UrlCache._clearUrlDetails(project_id, url, function (error) {
  168. if (error != null) {
  169. return callback(error)
  170. }
  171. return UrlCache._deleteUrlCacheFromDisk(project_id, url, function (
  172. error
  173. ) {
  174. if (error != null) {
  175. return callback(error)
  176. }
  177. return callback(null)
  178. })
  179. })
  180. },
  181. _deleteUrlCacheFromDisk(project_id, url, callback) {
  182. if (callback == null) {
  183. callback = function (error) {}
  184. }
  185. return fs.unlink(UrlCache._cacheFilePathForUrl(project_id, url), function (
  186. error
  187. ) {
  188. if (error != null && error.code !== 'ENOENT') {
  189. // no error if the file isn't present
  190. return callback(error)
  191. } else {
  192. return callback()
  193. }
  194. })
  195. },
  196. _findUrlDetails(project_id, url, callback) {
  197. if (callback == null) {
  198. callback = function (error, urlDetails) {}
  199. }
  200. const job = (cb) =>
  201. db.UrlCache.findOne({ where: { url, project_id } })
  202. .then((urlDetails) => cb(null, urlDetails))
  203. .error(cb)
  204. return dbQueue.queue.push(job, callback)
  205. },
  206. _updateOrCreateUrlDetails(project_id, url, lastModified, callback) {
  207. if (callback == null) {
  208. callback = function (error) {}
  209. }
  210. const job = (cb) =>
  211. db.UrlCache.findOrCreate({ where: { url, project_id } })
  212. .spread((urlDetails, created) =>
  213. urlDetails
  214. .update({ lastModified })
  215. .then(() => cb())
  216. .error(cb)
  217. )
  218. .error(cb)
  219. return dbQueue.queue.push(job, callback)
  220. },
  221. _clearUrlDetails(project_id, url, callback) {
  222. if (callback == null) {
  223. callback = function (error) {}
  224. }
  225. const job = (cb) =>
  226. db.UrlCache.destroy({ where: { url, project_id } })
  227. .then(() => cb(null))
  228. .error(cb)
  229. return dbQueue.queue.push(job, callback)
  230. },
  231. _findAllUrlsInProject(project_id, callback) {
  232. if (callback == null) {
  233. callback = function (error, urls) {}
  234. }
  235. const job = (cb) =>
  236. db.UrlCache.findAll({ where: { project_id } })
  237. .then((urlEntries) =>
  238. cb(
  239. null,
  240. urlEntries.map((entry) => entry.url)
  241. )
  242. )
  243. .error(cb)
  244. return dbQueue.queue.push(job, callback)
  245. }
  246. }