batchedUpdate.js 7.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317
  1. // @ts-check
  2. /* eslint-disable no-console */
  3. const { ObjectId, ReadPreference } = require('mongodb')
  4. const READ_PREFERENCE_SECONDARY =
  5. process.env.MONGO_HAS_SECONDARIES === 'true'
  6. ? ReadPreference.secondary.mode
  7. : ReadPreference.secondaryPreferred.mode
  8. const ONE_MONTH_IN_MS = 1000 * 60 * 60 * 24 * 31
  9. let ID_EDGE_PAST
  10. const ID_EDGE_FUTURE = objectIdFromMs(Date.now() + 1000)
  11. let BATCH_DESCENDING
  12. let BATCH_SIZE
  13. let VERBOSE_LOGGING
  14. let BATCH_RANGE_START
  15. let BATCH_RANGE_END
  16. let BATCH_MAX_TIME_SPAN_IN_MS
  17. let BATCHED_UPDATE_RUNNING = false
  18. /**
  19. * @typedef {import("mongodb").Collection} Collection
  20. * @typedef {import("mongodb-legacy").Collection} LegacyCollection
  21. * @typedef {import("mongodb").Document} Document
  22. * @typedef {import("mongodb").FindOptions} FindOptions
  23. * @typedef {import("mongodb").UpdateFilter<Document>} UpdateDocument
  24. */
  25. /**
  26. * @typedef {Object} BatchedUpdateOptions
  27. * @property {string} [BATCH_DESCENDING]
  28. * @property {string} [BATCH_LAST_ID]
  29. * @property {string} [BATCH_MAX_TIME_SPAN_IN_MS]
  30. * @property {string} [BATCH_RANGE_END]
  31. * @property {string} [BATCH_RANGE_START]
  32. * @property {string} [BATCH_SIZE]
  33. * @property {string} [VERBOSE_LOGGING]
  34. */
  35. /**
  36. * @param {BatchedUpdateOptions} options
  37. */
  38. function refreshGlobalOptionsForBatchedUpdate(options = {}) {
  39. options = Object.assign({}, options, process.env)
  40. BATCH_DESCENDING = options.BATCH_DESCENDING === 'true'
  41. BATCH_SIZE = parseInt(options.BATCH_SIZE || '1000', 10) || 1000
  42. VERBOSE_LOGGING = options.VERBOSE_LOGGING === 'true'
  43. if (options.BATCH_LAST_ID) {
  44. BATCH_RANGE_START = objectIdFromInput(options.BATCH_LAST_ID)
  45. } else if (options.BATCH_RANGE_START) {
  46. BATCH_RANGE_START = objectIdFromInput(options.BATCH_RANGE_START)
  47. } else {
  48. if (BATCH_DESCENDING) {
  49. BATCH_RANGE_START = ID_EDGE_FUTURE
  50. } else {
  51. BATCH_RANGE_START = ID_EDGE_PAST
  52. }
  53. }
  54. BATCH_MAX_TIME_SPAN_IN_MS = parseInt(
  55. options.BATCH_MAX_TIME_SPAN_IN_MS || ONE_MONTH_IN_MS.toString(),
  56. 10
  57. )
  58. if (options.BATCH_RANGE_END) {
  59. BATCH_RANGE_END = objectIdFromInput(options.BATCH_RANGE_END)
  60. } else {
  61. if (BATCH_DESCENDING) {
  62. BATCH_RANGE_END = ID_EDGE_PAST
  63. } else {
  64. BATCH_RANGE_END = ID_EDGE_FUTURE
  65. }
  66. }
  67. }
  68. /**
  69. * @param {Collection | LegacyCollection} collection
  70. * @param {Document} query
  71. * @param {ObjectId} start
  72. * @param {ObjectId} end
  73. * @param {Document} projection
  74. * @param {FindOptions} findOptions
  75. * @return {Promise<Array<Document>>}
  76. */
  77. async function getNextBatch(
  78. collection,
  79. query,
  80. start,
  81. end,
  82. projection,
  83. findOptions
  84. ) {
  85. if (BATCH_DESCENDING) {
  86. query._id = {
  87. $gt: end,
  88. $lte: start,
  89. }
  90. } else {
  91. query._id = {
  92. $gt: start,
  93. $lte: end,
  94. }
  95. }
  96. return await collection
  97. .find(query, findOptions)
  98. .project(projection)
  99. .sort({ _id: BATCH_DESCENDING ? -1 : 1 })
  100. .limit(BATCH_SIZE)
  101. .toArray()
  102. }
  103. /**
  104. * @param {Collection | LegacyCollection} collection
  105. * @param {Array<Document>} nextBatch
  106. * @param {UpdateDocument} update
  107. * @return {Promise<void>}
  108. */
  109. async function performUpdate(collection, nextBatch, update) {
  110. await collection.updateMany(
  111. { _id: { $in: nextBatch.map(entry => entry._id) } },
  112. update
  113. )
  114. }
  115. /**
  116. * @param {string} input
  117. * @return {ObjectId}
  118. */
  119. function objectIdFromInput(input) {
  120. if (input.includes('T')) {
  121. const t = new Date(input).getTime()
  122. if (Number.isNaN(t)) throw new Error(`${input} is not a valid date`)
  123. return objectIdFromMs(t)
  124. } else {
  125. return new ObjectId(input)
  126. }
  127. }
  128. /**
  129. * @param {ObjectId} objectId
  130. * @return {string}
  131. */
  132. function renderObjectId(objectId) {
  133. return `${objectId} (${objectId.getTimestamp().toISOString()})`
  134. }
  135. /**
  136. * @param {number} ms
  137. * @return {ObjectId}
  138. */
  139. function objectIdFromMs(ms) {
  140. return ObjectId.createFromTime(ms / 1000)
  141. }
  142. /**
  143. * @param {ObjectId} id
  144. * @return {number}
  145. */
  146. function getMsFromObjectId(id) {
  147. return id.getTimestamp().getTime()
  148. }
  149. /**
  150. * @param {ObjectId} start
  151. * @return {ObjectId}
  152. */
  153. function getNextEnd(start) {
  154. let end
  155. if (BATCH_DESCENDING) {
  156. end = objectIdFromMs(getMsFromObjectId(start) - BATCH_MAX_TIME_SPAN_IN_MS)
  157. if (getMsFromObjectId(end) <= getMsFromObjectId(BATCH_RANGE_END)) {
  158. end = BATCH_RANGE_END
  159. }
  160. } else {
  161. end = objectIdFromMs(getMsFromObjectId(start) + BATCH_MAX_TIME_SPAN_IN_MS)
  162. if (getMsFromObjectId(end) >= getMsFromObjectId(BATCH_RANGE_END)) {
  163. end = BATCH_RANGE_END
  164. }
  165. }
  166. return end
  167. }
  168. /**
  169. * @param {Collection | LegacyCollection} collection
  170. * @return {Promise<ObjectId|null>}
  171. */
  172. async function getIdEdgePast(collection) {
  173. const [first] = await collection
  174. .find({})
  175. .project({ _id: 1 })
  176. .sort({ _id: 1 })
  177. .limit(1)
  178. .toArray()
  179. if (!first) return null
  180. // Go one second further into the past in order to include the first entry via
  181. // first._id > ID_EDGE_PAST
  182. return objectIdFromMs(Math.max(0, getMsFromObjectId(first._id) - 1000))
  183. }
  184. /**
  185. * @param {Collection | LegacyCollection} collection
  186. * @param {Document} query
  187. * @param {UpdateDocument | ((batch: Array<Document>) => Promise<void>)} update
  188. * @param {Document} [projection]
  189. * @param {FindOptions} [findOptions]
  190. * @param {BatchedUpdateOptions} [batchedUpdateOptions]
  191. */
  192. async function batchedUpdate(
  193. collection,
  194. query,
  195. update,
  196. projection,
  197. findOptions,
  198. batchedUpdateOptions
  199. ) {
  200. // only a single batchedUpdate can run at a time due to global variables
  201. if (BATCHED_UPDATE_RUNNING) {
  202. throw new Error('batchedUpdate is already running')
  203. }
  204. try {
  205. BATCHED_UPDATE_RUNNING = true
  206. ID_EDGE_PAST = await getIdEdgePast(collection)
  207. if (!ID_EDGE_PAST) {
  208. console.warn(
  209. `The collection ${collection.collectionName} appears to be empty.`
  210. )
  211. return 0
  212. }
  213. refreshGlobalOptionsForBatchedUpdate(batchedUpdateOptions)
  214. findOptions = findOptions || {}
  215. findOptions.readPreference = READ_PREFERENCE_SECONDARY
  216. projection = projection || { _id: 1 }
  217. let nextBatch
  218. let updated = 0
  219. let start = BATCH_RANGE_START
  220. while (start !== BATCH_RANGE_END) {
  221. let end = getNextEnd(start)
  222. nextBatch = await getNextBatch(
  223. collection,
  224. query,
  225. start,
  226. end,
  227. projection,
  228. findOptions
  229. )
  230. if (nextBatch.length > 0) {
  231. end = nextBatch[nextBatch.length - 1]._id
  232. updated += nextBatch.length
  233. if (VERBOSE_LOGGING) {
  234. console.log(
  235. `Running update on batch with ids ${JSON.stringify(
  236. nextBatch.map(entry => entry._id)
  237. )}`
  238. )
  239. } else {
  240. console.error(`Running update on batch ending ${renderObjectId(end)}`)
  241. }
  242. if (typeof update === 'function') {
  243. await update(nextBatch)
  244. } else {
  245. await performUpdate(collection, nextBatch, update)
  246. }
  247. }
  248. console.error(`Completed batch ending ${renderObjectId(end)}`)
  249. start = end
  250. }
  251. return updated
  252. } finally {
  253. BATCHED_UPDATE_RUNNING = false
  254. }
  255. }
  256. /**
  257. * @param {Collection | LegacyCollection} collection
  258. * @param {Document} query
  259. * @param {UpdateDocument | ((batch: Array<Object>) => Promise<void>)} update
  260. * @param {Document} [projection]
  261. * @param {FindOptions} [findOptions]
  262. * @param {BatchedUpdateOptions} [batchedUpdateOptions]
  263. */
  264. function batchedUpdateWithResultHandling(
  265. collection,
  266. query,
  267. update,
  268. projection,
  269. findOptions,
  270. batchedUpdateOptions
  271. ) {
  272. batchedUpdate(
  273. collection,
  274. query,
  275. update,
  276. projection,
  277. findOptions,
  278. batchedUpdateOptions
  279. )
  280. .then(processed => {
  281. console.error({ processed })
  282. process.exit(0)
  283. })
  284. .catch(error => {
  285. console.error({ error })
  286. process.exit(1)
  287. })
  288. }
  289. module.exports = {
  290. READ_PREFERENCE_SECONDARY,
  291. objectIdFromInput,
  292. renderObjectId,
  293. batchedUpdate,
  294. batchedUpdateWithResultHandling,
  295. }