BackFillDummyDocMetaTests.js 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336
  1. const { exec } = require('child_process')
  2. const { promisify } = require('util')
  3. const { expect } = require('chai')
  4. const logger = require('logger-sharelatex')
  5. const { db, ObjectId } = require('../../../app/src/infrastructure/mongodb')
  6. const DUMMY_NAME = 'unknown.tex'
  7. const DUMMY_TIME = new Date('2021-04-12T00:00:00.000Z')
  8. const ONE_DAY_IN_S = 60 * 60 * 24
  9. const BATCH_SIZE = 3
  10. function getSecondsFromObjectId(id) {
  11. return id.getTimestamp().getTime() / 1000
  12. }
  13. function getObjectIdFromDate(date) {
  14. const seconds = new Date(date).getTime() / 1000
  15. return ObjectId.createFromTime(seconds)
  16. }
  17. describe('BackFillDummyDocMeta', function () {
  18. let docIds
  19. let projectIds
  20. let stopAtSeconds
  21. beforeEach('create docs', async function () {
  22. docIds = []
  23. docIds[0] = getObjectIdFromDate('2021-04-01T00:00:00.000Z')
  24. docIds[1] = getObjectIdFromDate('2021-04-02T00:00:00.000Z')
  25. docIds[2] = getObjectIdFromDate('2021-04-11T00:00:00.000Z')
  26. docIds[3] = getObjectIdFromDate('2021-04-12T00:00:00.000Z')
  27. docIds[4] = getObjectIdFromDate('2021-04-13T00:00:00.000Z')
  28. docIds[5] = getObjectIdFromDate('2021-04-14T00:00:00.000Z')
  29. docIds[6] = getObjectIdFromDate('2021-04-15T00:00:00.000Z')
  30. docIds[7] = getObjectIdFromDate('2021-04-16T00:01:00.000Z')
  31. docIds[8] = getObjectIdFromDate('2021-04-16T00:02:00.000Z')
  32. docIds[9] = getObjectIdFromDate('2021-04-16T00:03:00.000Z')
  33. docIds[10] = getObjectIdFromDate('2021-04-16T00:04:00.000Z')
  34. docIds[11] = getObjectIdFromDate('2021-04-16T00:05:00.000Z')
  35. projectIds = []
  36. projectIds[0] = getObjectIdFromDate('2021-04-01T00:00:00.000Z')
  37. projectIds[1] = getObjectIdFromDate('2021-04-02T00:00:00.000Z')
  38. projectIds[2] = getObjectIdFromDate('2021-04-11T00:00:00.000Z')
  39. projectIds[3] = getObjectIdFromDate('2021-04-12T00:00:00.000Z')
  40. projectIds[4] = getObjectIdFromDate('2021-04-13T00:00:00.000Z')
  41. projectIds[5] = getObjectIdFromDate('2021-04-14T00:00:00.000Z')
  42. projectIds[6] = getObjectIdFromDate('2021-04-15T00:00:00.000Z')
  43. projectIds[7] = getObjectIdFromDate('2021-04-16T00:01:00.000Z')
  44. projectIds[8] = getObjectIdFromDate('2021-04-16T00:02:00.000Z')
  45. projectIds[9] = getObjectIdFromDate('2021-04-16T00:03:00.000Z')
  46. // two docs in the same project
  47. projectIds[10] = projectIds[9]
  48. projectIds[11] = projectIds[4]
  49. stopAtSeconds = new Date('2021-04-17T00:00:00.000Z').getTime() / 1000
  50. })
  51. const now = new Date()
  52. beforeEach('insert doc stubs into docs collection', async function () {
  53. await db.docs.insertMany([
  54. // incomplete, without deletedDocs context
  55. { _id: docIds[0], project_id: projectIds[0], deleted: true },
  56. { _id: docIds[1], project_id: projectIds[1], deleted: true },
  57. { _id: docIds[2], project_id: projectIds[2], deleted: true },
  58. { _id: docIds[3], project_id: projectIds[3], deleted: true },
  59. // incomplete, with deletedDocs context
  60. { _id: docIds[4], project_id: projectIds[4], deleted: true },
  61. // complete
  62. {
  63. _id: docIds[5],
  64. project_id: projectIds[5],
  65. deleted: true,
  66. name: 'foo.tex',
  67. deletedAt: now,
  68. },
  69. // not deleted
  70. { _id: docIds[6], project_id: projectIds[6] },
  71. // multiple in a single batch
  72. { _id: docIds[7], project_id: projectIds[7], deleted: true },
  73. { _id: docIds[8], project_id: projectIds[8], deleted: true },
  74. { _id: docIds[9], project_id: projectIds[9], deleted: true },
  75. // two docs in one project
  76. { _id: docIds[10], project_id: projectIds[10], deleted: true },
  77. { _id: docIds[11], project_id: projectIds[11], deleted: true },
  78. ])
  79. })
  80. beforeEach('insert deleted project context', async function () {
  81. await db.deletedProjects.insertMany([
  82. // projectIds[0] and projectIds[1] have no entry
  83. // hard-deleted
  84. { deleterData: { deletedProjectId: projectIds[2] } },
  85. // soft-deleted, no entry for doc
  86. {
  87. deleterData: { deletedProjectId: projectIds[3] },
  88. project: { deletedDocs: [] },
  89. },
  90. // soft-deleted, has entry for doc
  91. {
  92. deleterData: { deletedProjectId: projectIds[4] },
  93. project: {
  94. deletedDocs: [
  95. { _id: docIds[4], name: 'main.tex', deletedAt: now },
  96. { _id: docIds[11], name: 'main.tex', deletedAt: now },
  97. ],
  98. },
  99. },
  100. ])
  101. })
  102. let options
  103. async function runScript(dryRun) {
  104. options = {
  105. BATCH_SIZE,
  106. CACHE_SIZE: 100,
  107. DRY_RUN: dryRun,
  108. FIRST_PROJECT_ID: projectIds[0].toString(),
  109. INCREMENT_BY_S: ONE_DAY_IN_S,
  110. STOP_AT_S: stopAtSeconds,
  111. // start right away
  112. LET_USER_DOUBLE_CHECK_INPUTS_FOR: 1,
  113. }
  114. let result
  115. try {
  116. result = await promisify(exec)(
  117. Object.entries(options)
  118. .map(([key, value]) => `${key}=${value}`)
  119. .concat(['node', 'scripts/back_fill_dummy_doc_meta.js'])
  120. .join(' ')
  121. )
  122. } catch (error) {
  123. // dump details like exit code, stdErr and stdOut
  124. logger.error({ error }, 'script failed')
  125. throw error
  126. }
  127. let { stderr: stdErr, stdout: stdOut } = result
  128. stdErr = stdErr.split('\n')
  129. stdOut = stdOut
  130. .split('\n')
  131. .filter(line => !line.includes('Using settings from'))
  132. const oneDayFromProjectId9InSeconds =
  133. getSecondsFromObjectId(projectIds[9]) + ONE_DAY_IN_S
  134. const oneDayFromProjectId9AsObjectId = getObjectIdFromDate(
  135. 1000 * oneDayFromProjectId9InSeconds
  136. )
  137. let overlappingPartStdOut
  138. let overlappingPartStdErr
  139. if (dryRun) {
  140. // In dry-run, the previous id will get processed again as the name has not been updated.
  141. overlappingPartStdOut = [
  142. `Back filling dummy meta data for ["${docIds[9]}","${docIds[10]}"]`,
  143. `Orphaned deleted doc ${docIds[9]} (no deletedProjects entry)`,
  144. `Orphaned deleted doc ${docIds[10]} (no deletedProjects entry)`,
  145. ]
  146. overlappingPartStdErr = [
  147. `Processed 11 until ${oneDayFromProjectId9AsObjectId}`,
  148. ]
  149. } else {
  150. // Outside dry-run, the previous id will not match again as the `name` has been back-filled.
  151. overlappingPartStdOut = [
  152. `Back filling dummy meta data for ["${docIds[10]}"]`,
  153. `Orphaned deleted doc ${docIds[10]} (no deletedProjects entry)`,
  154. ]
  155. overlappingPartStdErr = [
  156. `Processed 10 until ${oneDayFromProjectId9AsObjectId}`,
  157. ]
  158. }
  159. expect(stdOut).to.deep.equal([
  160. `Back filling dummy meta data for ["${docIds[0]}"]`,
  161. `Orphaned deleted doc ${docIds[0]} (no deletedProjects entry)`,
  162. `Back filling dummy meta data for ["${docIds[1]}"]`,
  163. `Orphaned deleted doc ${docIds[1]} (no deletedProjects entry)`,
  164. `Back filling dummy meta data for ["${docIds[2]}"]`,
  165. `Orphaned deleted doc ${docIds[2]} (failed hard deletion)`,
  166. `Back filling dummy meta data for ["${docIds[3]}"]`,
  167. `Missing deletedDoc for ${docIds[3]}`,
  168. // two docs in the same project
  169. `Back filling dummy meta data for ["${docIds[4]}","${docIds[11]}"]`,
  170. `Found deletedDoc for ${docIds[4]}`,
  171. `Found deletedDoc for ${docIds[11]}`,
  172. // 7,8,9 are on the same day, but exceed the batch size of 2
  173. `Back filling dummy meta data for ["${docIds[7]}","${docIds[8]}","${docIds[9]}"]`,
  174. `Orphaned deleted doc ${docIds[7]} (no deletedProjects entry)`,
  175. `Orphaned deleted doc ${docIds[8]} (no deletedProjects entry)`,
  176. `Orphaned deleted doc ${docIds[9]} (no deletedProjects entry)`,
  177. // Potential double processing
  178. ...overlappingPartStdOut,
  179. '',
  180. ])
  181. expect(stdErr).to.deep.equal([
  182. ...`Options: ${JSON.stringify(options, null, 2)}`.split('\n'),
  183. 'Waiting for you to double check inputs for 1 ms',
  184. `Processed 1 until ${getObjectIdFromDate('2021-04-02T00:00:00.000Z')}`,
  185. `Processed 2 until ${getObjectIdFromDate('2021-04-03T00:00:00.000Z')}`,
  186. `Processed 2 until ${getObjectIdFromDate('2021-04-04T00:00:00.000Z')}`,
  187. `Processed 2 until ${getObjectIdFromDate('2021-04-05T00:00:00.000Z')}`,
  188. `Processed 2 until ${getObjectIdFromDate('2021-04-06T00:00:00.000Z')}`,
  189. `Processed 2 until ${getObjectIdFromDate('2021-04-07T00:00:00.000Z')}`,
  190. `Processed 2 until ${getObjectIdFromDate('2021-04-08T00:00:00.000Z')}`,
  191. `Processed 2 until ${getObjectIdFromDate('2021-04-09T00:00:00.000Z')}`,
  192. `Processed 2 until ${getObjectIdFromDate('2021-04-10T00:00:00.000Z')}`,
  193. `Processed 2 until ${getObjectIdFromDate('2021-04-11T00:00:00.000Z')}`,
  194. `Processed 3 until ${getObjectIdFromDate('2021-04-12T00:00:00.000Z')}`,
  195. `Processed 4 until ${getObjectIdFromDate('2021-04-13T00:00:00.000Z')}`,
  196. `Processed 6 until ${getObjectIdFromDate('2021-04-14T00:00:00.000Z')}`,
  197. `Processed 6 until ${getObjectIdFromDate('2021-04-15T00:00:00.000Z')}`,
  198. `Processed 6 until ${getObjectIdFromDate('2021-04-16T00:00:00.000Z')}`,
  199. // 7,8,9,10 are on the same day, but exceed the batch size of 3
  200. `Processed 9 until ${projectIds[9]}`,
  201. ...overlappingPartStdErr,
  202. 'Done.',
  203. '',
  204. ])
  205. }
  206. describe('DRY_RUN=true', function () {
  207. beforeEach('run script', async function () {
  208. await runScript(true)
  209. })
  210. it('should leave docs as is', async function () {
  211. const docs = await db.docs.find({}).toArray()
  212. expect(docs).to.deep.equal([
  213. { _id: docIds[0], project_id: projectIds[0], deleted: true },
  214. { _id: docIds[1], project_id: projectIds[1], deleted: true },
  215. { _id: docIds[2], project_id: projectIds[2], deleted: true },
  216. { _id: docIds[3], project_id: projectIds[3], deleted: true },
  217. { _id: docIds[4], project_id: projectIds[4], deleted: true },
  218. {
  219. _id: docIds[5],
  220. project_id: projectIds[5],
  221. deleted: true,
  222. name: 'foo.tex',
  223. deletedAt: now,
  224. },
  225. { _id: docIds[6], project_id: projectIds[6] },
  226. { _id: docIds[7], project_id: projectIds[7], deleted: true },
  227. { _id: docIds[8], project_id: projectIds[8], deleted: true },
  228. { _id: docIds[9], project_id: projectIds[9], deleted: true },
  229. { _id: docIds[10], project_id: projectIds[10], deleted: true },
  230. { _id: docIds[11], project_id: projectIds[11], deleted: true },
  231. ])
  232. })
  233. })
  234. describe('DRY_RUN=false', function () {
  235. beforeEach('run script', async function () {
  236. await runScript(false)
  237. })
  238. it('should back fill name and deletedAt dates into broken docs', async function () {
  239. const docs = await db.docs.find({}).toArray()
  240. expect(docs).to.deep.equal([
  241. {
  242. _id: docIds[0],
  243. project_id: projectIds[0],
  244. deleted: true,
  245. name: DUMMY_NAME,
  246. deletedAt: DUMMY_TIME,
  247. },
  248. {
  249. _id: docIds[1],
  250. project_id: projectIds[1],
  251. deleted: true,
  252. name: DUMMY_NAME,
  253. deletedAt: DUMMY_TIME,
  254. },
  255. {
  256. _id: docIds[2],
  257. project_id: projectIds[2],
  258. deleted: true,
  259. name: DUMMY_NAME,
  260. deletedAt: DUMMY_TIME,
  261. },
  262. {
  263. _id: docIds[3],
  264. project_id: projectIds[3],
  265. deleted: true,
  266. name: DUMMY_NAME,
  267. deletedAt: DUMMY_TIME,
  268. },
  269. {
  270. _id: docIds[4],
  271. project_id: projectIds[4],
  272. deleted: true,
  273. name: 'main.tex',
  274. deletedAt: now,
  275. },
  276. {
  277. _id: docIds[5],
  278. project_id: projectIds[5],
  279. deleted: true,
  280. name: 'foo.tex',
  281. deletedAt: now,
  282. },
  283. { _id: docIds[6], project_id: projectIds[6] },
  284. {
  285. _id: docIds[7],
  286. project_id: projectIds[7],
  287. deleted: true,
  288. name: DUMMY_NAME,
  289. deletedAt: DUMMY_TIME,
  290. },
  291. {
  292. _id: docIds[8],
  293. project_id: projectIds[8],
  294. deleted: true,
  295. name: DUMMY_NAME,
  296. deletedAt: DUMMY_TIME,
  297. },
  298. {
  299. _id: docIds[9],
  300. project_id: projectIds[9],
  301. deleted: true,
  302. name: DUMMY_NAME,
  303. deletedAt: DUMMY_TIME,
  304. },
  305. {
  306. _id: docIds[10],
  307. project_id: projectIds[10],
  308. deleted: true,
  309. name: DUMMY_NAME,
  310. deletedAt: DUMMY_TIME,
  311. },
  312. {
  313. _id: docIds[11],
  314. project_id: projectIds[11],
  315. deleted: true,
  316. name: 'main.tex',
  317. deletedAt: now,
  318. },
  319. ])
  320. })
  321. })
  322. })