tracker

36.4 KB

1. // lib/jobs.hl — THE BACKGROUND JOBS (mission 028: out of project.hl, unchanged): the daily TMDB sync run, the adult and kind
2. // backfills, the details repair, the credits job, the duplicate merge, the short-id backfill and the collection seed — each with
3. // its queue, its pause, its log lines, and the clocks that drive them (which job runs when, which one waits for which). The work
4. // of ONE step is the topic's own (sync.hl, backfill.hl, details.hl, credits.hl, merge.hl, shortids.hl, collections.hl).
5. // A CLASS, not statics: its state (queues, positions, totals) is rewritten on every step, and a static cannot be reassigned —
6. // project.hl makes the one instance (`jobs = new Jobs()`) after the server, as these lines ran there before; constructing it
7. // starts the clocks and logs which jobs are off.
8. import { env } from 'hl:proc'
9. import { readFile, exists } from 'hl:fs'
10. import { now, timestamp, every } from 'hl:time'
11. import { syncEnabled, pauseMsAfter, msPerRequest } from './tmdb.hl'
12. import { persistSync } from './sync.hl'
13. import { emptyTotals, addTotals, totalsLine } from './sync-helpers.hl'
14. import { planStart, planStep, planIds, planLine, markSynced, syncRunStep } from './dailysync.hl'
15. import { adultBackfillQueue, backfillTitle, kindBackfillQueue, kindTitle, tvmazeStatusQueue, tvmazeStatusTitle } from './backfill.hl'
16. import { detailsRepairQueue, completeStep, persistDetails, applySynced } from './details.hl'
17. import { persistCredits, creditsQueue, creditsStep } from './credits.hl'
18. import { allFollowedShowIds } from './follows.hl'
19. import { refreshCatalogShow } from './catalog.hl'
20. import { searchTitleChanged, searchTitleRenamed } from './search.hl'
21. import { mergePrepare, mergeStep, mergeStats, mergeLine, persistMerge } from './merge.hl'
22. import { shortIdStep, shortIdStats, persistShortIds, shortIdsLeft } from './shortids.hl'
23. import { seedPrepare, seedStep, seedLine, seedPersist } from './collections.hl'
25. // ---- THE DAILY TMDB SYNC (tracker.worldapi.org#9, sync.hl; tracker#18, dailysync.hl) -------------------------
26. // Once a day at TRACKER_SYNC_HOUR (UTC, default 4 — hl:time has no time zones; the container runs UTC). hl:web serves ONE
27. // request at a time, so the run never takes the server for itself: a 1-second clock does ONE step per tick and returns — the
28. // requests that queued meanwhile are served between two ticks, and a tick never starts inside a page render.
29. // tracker#18: the run first asks TMDB's and TVmaze's CHANGE LISTS since the last run (dailysync.hl — one list page per step)
30. // and then syncs only the titles of ours that changed, followed ones first — every title stays current, not only the followed
31. // ones. FULL WALK (every followed show, as before #18) on the first run, after a gap > 14 days, or when a change list fails.
32. // A title's step: sync.hl syncShow (1–3 TMDB requests + maybe its poster, TVmaze) + details.hl applySynced (the record's
33. // title, overview, tagline … and cast/crew from the same details answer). After a step that made n requests the next waits
34. // n × 260 ms (TMDB: ≤ 40 requests / 10 s) + 550 ms per TVmaze request (≤ 20 / 10 s, tracker.worldapi.org#12).
35. // The finished run stores its day (dailysync.hl markSynced). Log: "tmdb sync: start, …", the change lists ("tmdb sync:
36. // changes since …"), errors, slow steps, "tmdb sync done: mode=… changeRequests=… <totals>".
37. // TRACKER_SYNC=0 switches it off; without TMDB_READ_TOKEN it is off too (logged once).
38. syncHour = env('TRACKER_SYNC_HOUR') != null ? toNumber(env('TRACKER_SYNC_HOUR')) : 4
39. syncOn = env('TRACKER_SYNC') != '0' && syncEnabled
40. syncQueue = []
41. syncPlanning = false
42. syncMode = ''
43. syncMark = false
44. syncChangeRequests = 0
45. syncAt = 0
46. syncDay = ''
47. syncNextAt = 0
48. syncTotals = emptyTotals()
49. syncFinish = () => {
50. persistDetails()
51. if (syncMark) { markSynced(syncDay) }
52. console.log('tmdb sync done: mode=' + syncMode + ' changeRequests=' + syncChangeRequests + ' ' + totalsLine(syncTotals) + ' seconds=' + ((now() - syncAt) / 1000))
53. return null
54. }
55. syncTick = () => {
56. t = now()
57. if (syncQueue.length == 0 && !syncPlanning) {
58. stamp = timestamp(t)
59. day = stamp.slice(0, 10)
60. if (toNumber(stamp.slice(11, 13)) != syncHour || day == syncDay) { return null }
61. syncDay = day
62. syncAt = t
63. syncNextAt = 0
64. syncTotals = emptyTotals()
65. syncChangeRequests = 0
66. p = planStart(day, null)
67. syncMode = p.mode
68. syncMark = true
69. if (p.mode == 'delta') {
70. syncPlanning = true
71. console.log('tmdb sync: start, changes ' + p.why)
72. return null
73. }
74. syncQueue = allFollowedShowIds()
75. console.log('tmdb sync: start, ' + syncQueue.length + ' followed shows, full walk (' + p.why + ')')
76. if (syncQueue.length == 0) { syncFinish() }
77. return null
78. }
79. if (t < syncNextAt) { return null }
80. if (syncPlanning) {
81. ps = planStep()
82. syncChangeRequests = syncChangeRequests + ps.requests + ps.tvmazeRequests
83. syncTotals.requests = syncTotals.requests + ps.requests
84. syncTotals.tvmazeRequests = syncTotals.tvmazeRequests + ps.tvmazeRequests
85. syncNextAt = now() + pauseMsAfter(ps)
86. if (!ps.done) { return null }
87. syncPlanning = false
88. console.log('tmdb sync: ' + planLine())
89. if (ps.failed != '') {
90. // a change list could not be read: the old full walk, and the day is not advanced (the next run asks again)
91. syncMode = 'full'
92. syncMark = false
93. syncQueue = allFollowedShowIds()
94. console.log('tmdb sync: change list failed, full walk of ' + syncQueue.length + ' followed shows instead')
95. } else {
96. syncQueue = planIds()
97. }
98. if (syncQueue.length == 0) { syncFinish() }
99. return null
100. }
101. id = syncQueue[0]
102. rest = []
103. let i = 1
104. while (i < syncQueue.length) { rest.push(syncQueue[i]) i = i + 1 }
105. syncQueue = rest
106. // tracker#18: in a delta run a title nobody follows gets the light step (changed seasons, no TVmaze); tracker#35: one only on
107. // TVmaze's change list its TVmaze status (dailysync.hl syncRunStep)
108. r = syncRunStep(id, syncMode == 'delta')
109. // tracker#18 + #28: the record's own fields (title, overview, tagline …) and the cast + crew from the same details answer
110. if (r.details != null) {
111. c = applySynced(id, r.details)
112. if (c.error != '') { console.log('tmdb sync: ' + c.error) }
113. if (c.renamed) { searchTitleRenamed(id) }
114. }
115. // the public lists (catalog.hl, tracker.worldapi.org#13): this show's place by its newest released episode; the sync
116. // also brings TMDB's adult flag (mission 054) → the search's view of it too
117. refreshCatalogShow(id)
118. searchTitleChanged(id)
119. took = now() - t
120. syncTotals = addTotals(syncTotals, r)
121. if (r.error != '') { console.log('tmdb sync: ' + r.error) }
122. // a step blocks every request meanwhile (Saturday Night Live, 53 seasons, 4 requests: ~1.7 s) — say so in the log
123. if (took > 1500) { console.log('tmdb sync: slow step, show ' + id + ', ' + r.requests + ' requests, ' + took + ' ms') }
124. syncNextAt = now() + pauseMsAfter(r)
125. if (syncQueue.length > 0 && syncTotals.shows % 10 == 0) { persistDetails() }
126. if (syncQueue.length == 0) { syncFinish() }
127. return null
128. }
130. // ---- THE ADULT-FLAG BACKFILL (mission 054, backfill.hl backfillTitle) -----------------------------------
131. // The public lists and the search show only titles TMDB says are not adult (shows.hl isPublicTitle; unknown = hidden). At
132. // start the app collects every title whose flag is still unknown and asks TMDB for ONE title per step — like the daily
133. // sync, on the same clock, so requests queued meanwhile are served between two steps: one details request (adult,
134. // poster_path, external ids) + the poster file when missing, then a pause of 260 ms per TMDB API request (≤ 40 / 10 s;
135. // the poster comes from TMDB's image host, not the API, and is not counted). Each answered title goes into the lists
136. // (catalog.hl refreshCatalogShow) and the search at once. RESUMES after a restart: the queue is built from what is
137. // still unknown. Stops when the queue is empty; titles TMDB answers 404 for are marked (`adultCheck`) and stay hidden; a
138. // passing failure (no answer, 429, 5xx) goes to the back of the queue, at most 2 more tries per start.
139. // While the daily sync runs, the backfill waits. Log: start, every TRACKER_BACKFILL_LOG_EVERY (500) titles, done.
140. // TRACKER_BACKFILL=0 switches it off; without TMDB_READ_TOKEN it is off too.
141. backfillOn = env('TRACKER_BACKFILL') != '0' && syncEnabled
142. backfillLogEvery = env('TRACKER_BACKFILL_LOG_EVERY') != null ? toNumber(env('TRACKER_BACKFILL_LOG_EVERY')) : 500
143. backfillQueue = null
144. backfillPos = 0
145. backfillDone = false
146. backfillAt = 0
147. backfillNextAt = 0
148. backfillTries = {}
149. backfillTotals = { titles = 0 adult = 0 notAdult = 0 unknown = 0 posters = 0 ids = 0 requests = 0 errors = 0 retries = 0 }
150. backfillLine = () => {
151. b = backfillTotals
152. return b.titles + '/' + backfillQueue.length + ' titles, adult=' + b.adult + ' notAdult=' + b.notAdult + ' unknown=' + b.unknown + ' posters=' + b.posters
153. + ' ids=' + b.ids + ' requests=' + b.requests + ' errors=' + b.errors + ' retries=' + b.retries + ' seconds=' + ((now() - backfillAt) / 1000)
154. }
155. backfillTick = () => {
156. if (backfillDone) { return null }
157. t = now()
158. if (backfillQueue == null) {
159. backfillQueue = adultBackfillQueue()
160. backfillAt = t
161. console.log('adult backfill: start, ' + backfillQueue.length + ' titles without an adult flag (' + (now() - t) + ' ms to find them)')
162. if (backfillQueue.length == 0) { backfillDone = true console.log('adult backfill: nothing to do') }
163. return null
164. }
165. if (t < backfillNextAt) { return null }
166. id = backfillQueue[backfillPos]
167. backfillPos = backfillPos + 1
168. r = backfillTitle(id)
169. if (r.retry && (backfillTries[id] == null || backfillTries[id] < 2)) {
170. backfillTries[id] = backfillTries[id] == null ? 1 : backfillTries[id] + 1
171. q = backfillQueue
172. q.push(id)
173. backfillQueue = q
174. backfillTotals.retries = backfillTotals.retries + 1
175. } else {
176. backfillTotals.titles = backfillTotals.titles + 1
177. if (r.adult == true) { backfillTotals.adult = backfillTotals.adult + 1 } else if (r.adult == false) { backfillTotals.notAdult = backfillTotals.notAdult + 1 } else { backfillTotals.unknown = backfillTotals.unknown + 1 }
178. }
179. backfillTotals.posters = backfillTotals.posters + r.posters
180. backfillTotals.ids = backfillTotals.ids + r.ids
181. backfillTotals.requests = backfillTotals.requests + r.requests
182. backfillTotals.errors = backfillTotals.errors + r.errors
183. if (r.error != '') { console.log('adult backfill: ' + r.error) }
184. if (r.adult != null) {
185. refreshCatalogShow(id)
186. searchTitleChanged(id)
187. }
188. took = now() - t
189. if (took > 1500) { console.log('adult backfill: slow step, show ' + id + ', ' + took + ' ms') }
190. backfillNextAt = now() + r.requests * msPerRequest
191. finished = backfillPos >= backfillQueue.length
192. if (finished || backfillPos % 25 == 0) { persistSync() }
193. if (!finished && backfillTotals.titles > 0 && backfillTotals.titles % backfillLogEvery == 0 && !r.retry) { console.log('adult backfill: ' + backfillLine()) }
194. if (finished) {
195. backfillDone = true
196. console.log('adult backfill done: ' + backfillLine())
197. }
198. return null
199. }
201. // ---- THE DETAILS REPAIR (tracker.worldapi.org#26, details.hl completeStep) ----------------------------------------
202. // Titles the person pages added from a filmography (#16) are MINIMAL (no poster, seasons, ids, cast …), and (tracker#33, mission
203. // 029) the old tracker's migrated rows were never synced (5,697 on the live copy, 691 series without seasons) — shows.hl
204. // isIncomplete. Opening one completes it (components/show.hl); this job completes the rest in the background: at start it
205. // collects every incomplete, non-adult title (the visibly reduced ones first, details.hl detailsRepairQueue) and does ONE per
206. // step on the app's clock, like the adult backfill — a step is the daily sync's work for one title
207. // (1–3 TMDB requests + poster + TVmaze), requests queued meanwhile are served between two steps; then a pause of 260 ms per
208. // TMDB request + 550 ms per TVmaze request (pauseMsAfter: ≤ 40 / 10 s, ≤ 20 / 10 s). RESUMES after a restart (the queue is
209. // what is still incomplete); TMDB 404 marks the title (`detailsCheck`), a passing failure goes to the back of the queue (at
210. // most 2 more tries per start). Waits while the daily sync or the adult backfill runs. Persisted every 10 titles + at the end.
211. // Log: start, every TRACKER_REPAIR_LOG_EVERY (100) titles, each failure, done. TRACKER_REPAIR=0 switches it off (off too
212. // without TMDB_READ_TOKEN). TRACKER_REPAIR_FAST=1 (the gate, fake TMDB only): 20 ms pause instead.
213. repairOn = env('TRACKER_REPAIR') != '0' && syncEnabled
214. repairFast = env('TRACKER_REPAIR_FAST') == '1'
215. repairLogEvery = env('TRACKER_REPAIR_LOG_EVERY') != null ? toNumber(env('TRACKER_REPAIR_LOG_EVERY')) : 100
216. repairQueue = null
217. repairPos = 0
218. repairDone = false
219. repairAt = 0
220. repairNextAt = 0
221. repairTries = {}
222. repairTotals = { titles = 0 complete = 0 failed = 0 adult = 0 newSeasons = 0 newEpisodes = 0 posters = 0 cast = 0 requests = 0 tvmazeRequests = 0 retries = 0 }
223. repairLine = () => {
224. b = repairTotals
225. return b.titles + '/' + repairQueue.length + ' titles, complete=' + b.complete + ' failed=' + b.failed + ' adult=' + b.adult + ' newSeasons=' + b.newSeasons + ' newEpisodes=' + b.newEpisodes
226. + ' posters=' + b.posters + ' cast=' + b.cast + ' requests=' + b.requests + ' tvmazeRequests=' + b.tvmazeRequests + ' retries=' + b.retries + ' seconds=' + ((now() - repairAt) / 1000)
227. }
228. repairTick = () => {
229. if (repairDone) { return null }
230. t = now()
231. if (repairQueue == null) {
232. repairQueue = detailsRepairQueue()
233. repairAt = t
234. console.log('details repair: start, ' + repairQueue.length + ' incomplete titles (' + (now() - t) + ' ms to find them)')
235. if (repairQueue.length == 0) { repairDone = true console.log('details repair: nothing to do') }
236. return null
237. }
238. if (t < repairNextAt) { return null }
239. id = repairQueue[repairPos]
240. r = completeStep(id)
241. took = now() - t
242. if (took > 1500) { console.log('details repair: slow step, show ' + id + ', ' + took + ' ms') }
243. if (r.error != '') { console.log('details repair: ' + r.error) }
244. repairNextAt = now() + (repairFast ? 20 : pauseMsAfter(r))
245. // a series takes several steps: the same title again next time
246. if (!r.done && !r.retry) { return null }
247. repairPos = repairPos + 1
248. if (r.retry && (repairTries[id] == null || repairTries[id] < 2)) {
249. repairTries[id] = repairTries[id] == null ? 1 : repairTries[id] + 1
250. q = repairQueue
251. q.push(id)
252. repairQueue = q
253. repairTotals.retries = repairTotals.retries + 1
254. } else if (!r.skipped) {
255. // done: its counts (a failed first step: the request alone)
256. sum = r.sum
257. repairTotals.titles = repairTotals.titles + 1
258. if (r.adult == true) { repairTotals.adult = repairTotals.adult + 1 } else if (sum.steps == 0) { repairTotals.failed = repairTotals.failed + 1 } else { repairTotals.complete = repairTotals.complete + 1 }
259. repairTotals.newSeasons = repairTotals.newSeasons + sum.newSeasons
260. repairTotals.newEpisodes = repairTotals.newEpisodes + sum.newEpisodes
261. repairTotals.posters = repairTotals.posters + sum.posters
262. repairTotals.cast = repairTotals.cast + sum.cast
263. repairTotals.requests = repairTotals.requests + (sum.steps == 0 ? r.requests : sum.requests)
264. repairTotals.tvmazeRequests = repairTotals.tvmazeRequests + sum.tvmazeRequests
265. }
266. finished = repairPos >= repairQueue.length
267. if (finished || repairPos % 10 == 0) { persistDetails() }
268. if (!finished && !r.skipped && !r.retry && repairTotals.titles > 0 && repairTotals.titles % repairLogEvery == 0) { console.log('details repair: ' + repairLine()) }
269. if (finished) {
270. repairDone = true
271. console.log('details repair done: ' + repairLine())
272. }
273. return null
274. }
276. // ---- THE TVMAZE STATUS BACKFILL (tracker#35, mission 030, backfill.hl tvmazeStatusTitle) — the details repair's second part ----
277. // The page's status pill shows TVmaze's status when a series has one (fresher than TMDB's). The sync stores it whenever it merges
278. // TVmaze's episodes (followed series, every completion); the series it never reached get it here: once the details repair is
279. // done (same switch, TRACKER_REPAIR) every series with a TVmaze id and no tvmazeStatus, ONE TVmaze request per step, then 550 ms
280. // (TVmaze: ≤ 20 / 10 s). RESUMES after a restart (the queue = what still has none); a 404 marks the series
281. // (`tvmazeStatusCheck`), a passing failure goes to the back of the queue (at most 2 more tries per start). Waits while the daily
282. // sync runs. Persisted every 25 series + at the end. Log: start, every TRACKER_REPAIR_LOG_EVERY series, done.
283. statusQueue = null
284. statusPos = 0
285. statusDone = false
286. statusAt = 0
287. statusNextAt = 0
288. statusTries = {}
289. statusTotals = { series = 0 stored = 0 notFound = 0 requests = 0 errors = 0 retries = 0 }
290. statusLine = () => {
291. b = statusTotals
292. return b.series + '/' + statusQueue.length + ' series, stored=' + b.stored + ' notFound=' + b.notFound + ' tvmazeRequests=' + b.requests + ' errors=' + b.errors + ' retries=' + b.retries + ' seconds=' + ((now() - statusAt) / 1000)
293. }
294. statusTick = () => {
295. if (statusDone) { return null }
296. t = now()
297. if (statusQueue == null) {
298. statusQueue = tvmazeStatusQueue()
299. statusAt = t
300. console.log('tvmaze status: start, ' + statusQueue.length + ' series without a TVmaze status (' + (now() - t) + ' ms to find them)')
301. if (statusQueue.length == 0) { statusDone = true console.log('tvmaze status: nothing to do') }
302. return null
303. }
304. if (t < statusNextAt) { return null }
305. id = statusQueue[statusPos]
306. statusPos = statusPos + 1
307. r = tvmazeStatusTitle(id)
308. if (r.retry && (statusTries[id] == null || statusTries[id] < 2)) {
309. statusTries[id] = statusTries[id] == null ? 1 : statusTries[id] + 1
310. q = statusQueue
311. q.push(id)
312. statusQueue = q
313. statusTotals.retries = statusTotals.retries + 1
314. } else {
315. statusTotals.series = statusTotals.series + 1
316. if (r.status != null) { statusTotals.stored = statusTotals.stored + 1 } else if (r.errors > 0) { statusTotals.notFound = statusTotals.notFound + 1 }
317. }
318. statusTotals.requests = statusTotals.requests + r.tvmazeRequests
319. statusTotals.errors = statusTotals.errors + r.errors
320. if (r.error != '') { console.log('tvmaze status: ' + r.error) }
321. statusNextAt = now() + (repairFast ? 20 : pauseMsAfter(r))
322. finished = statusPos >= statusQueue.length
323. if (finished || statusPos % 25 == 0) { persistSync() }
324. if (!finished && !r.retry && statusTotals.series > 0 && statusTotals.series % repairLogEvery == 0) { console.log('tvmaze status: ' + statusLine()) }
325. if (finished) {
326. statusDone = true
327. console.log('tvmaze status done: ' + statusLine())
328. }
329. return null
330. }
332. // ---- THE KIND BACKFILL (tracker.worldapi.org#21, mission 068, backfill.hl kindTitle) ----------------------------------
333. // Every TV title without a stored kind gets one: a title with a show genre (Reality / Talk / News) from the stored data (no
334. // request), any other from TMDB's TV type (ONE request: GET /tv/<id>, then 260 ms pause). Starts once the adult backfill is
335. // done (or off), waits while the daily sync runs, ONE title per step on the same clock. A title that becomes a SHOW moves from
336. // /series to /shows at once (catalog.hl refreshCatalogShow). RESUMES after a restart (the queue = every TV title still without
337. // a kind); a 404 stores the genres' answer + `kindCheck`; a passing failure goes to the back of the queue, at most 2 more tries
338. // per start. Log: start, every TRACKER_KINDS_LOG_EVERY (500) titles, done. TRACKER_KINDS=0 switches it off; without
339. // TMDB_READ_TOKEN it is off too (the lists still split by genre: shows.hl titleKind).
340. kindsOn = env('TRACKER_KINDS') != '0' && syncEnabled
341. kindsLogEvery = env('TRACKER_KINDS_LOG_EVERY') != null ? toNumber(env('TRACKER_KINDS_LOG_EVERY')) : 500
342. kindQueue = null
343. kindPos = 0
344. kindDone = false
345. kindAt = 0
346. kindNextAt = 0
347. kindTries = {}
348. kindTotals = { titles = 0 series = 0 shows = 0 fromGenre = 0 moved = 0 requests = 0 errors = 0 retries = 0 }
349. kindLine = () => {
350. k = kindTotals
351. return k.titles + '/' + kindQueue.length + ' titles, series=' + k.series + ' shows=' + k.shows + ' fromGenre=' + k.fromGenre + ' moved=' + k.moved
352. + ' requests=' + k.requests + ' errors=' + k.errors + ' retries=' + k.retries + ' seconds=' + ((now() - kindAt) / 1000)
353. }
354. kindTick = () => {
355. if (kindDone) { return null }
356. t = now()
357. if (kindQueue == null) {
358. kindQueue = kindBackfillQueue()
359. kindAt = t
360. console.log('kind backfill: start, ' + kindQueue.length + ' TV titles without a kind (' + (now() - t) + ' ms to find them)')
361. if (kindQueue.length == 0) { kindDone = true console.log('kind backfill: nothing to do') }
362. return null
363. }
364. if (t < kindNextAt) { return null }
365. id = kindQueue[kindPos]
366. kindPos = kindPos + 1
367. r = kindTitle(id)
368. if (r.retry && (kindTries[id] == null || kindTries[id] < 2)) {
369. kindTries[id] = kindTries[id] == null ? 1 : kindTries[id] + 1
370. q = kindQueue
371. q.push(id)
372. kindQueue = q
373. kindTotals.retries = kindTotals.retries + 1
374. } else {
375. kindTotals.titles = kindTotals.titles + 1
376. if (r.kind == 'show') { kindTotals.shows = kindTotals.shows + 1 } else if (r.kind == 'series') { kindTotals.series = kindTotals.series + 1 }
377. if (r.fromGenre) { kindTotals.fromGenre = kindTotals.fromGenre + 1 }
378. }
379. kindTotals.requests = kindTotals.requests + r.requests
380. kindTotals.errors = kindTotals.errors + r.errors
381. if (r.error != '') { console.log('kind backfill: ' + r.error) }
382. // until now the genres decided (no show genre = series): only a title TMDB's type makes a SHOW changes its list
383. if (r.kind == 'show' && !r.fromGenre) {
384. refreshCatalogShow(id)
385. kindTotals.moved = kindTotals.moved + 1
386. }
387. took = now() - t
388. if (took > 1500) { console.log('kind backfill: slow step, show ' + id + ', ' + took + ' ms') }
389. kindNextAt = now() + r.requests * msPerRequest
390. finished = kindPos >= kindQueue.length
391. if (finished || kindPos % 25 == 0) { persistSync() }
392. if (!finished && kindTotals.titles > 0 && kindTotals.titles % kindsLogEvery == 0 && !r.retry) { console.log('kind backfill: ' + kindLine()) }
393. if (finished) {
394. kindDone = true
395. console.log('kind backfill done: ' + kindLine())
396. }
397. return null
398. }
400. // ---- THE CREDITS JOB (tracker.worldapi.org#28, credits.hl creditsStep) ------------------------------------------------------
401. // Every PUBLIC title (adult == false; adult/unknown are skipped) with a TMDB id gets its whole cast + crew once: at start (after
402. // the adult backfill and the details repair are done) the job collects every such title without `creditsAt` and asks ONE per
403. // step on the app's clock — one request (details + credits / aggregate_credits), then 260 ms (≤ 40 / 10 s); requests queued
404. // meanwhile are served between two steps. Titles the details repair, the daily sync or an import already did carry `creditsAt`
405. // and are not asked. RESUMES after a restart (the queue = what still has no credits); TMDB 404 / adult → `creditsCheck`, never
406. // asked again; a passing failure goes to the back of the queue (at most 2 more tries per start). Waits while the daily sync
407. // runs. Persisted (shows + people only) every 25 titles + at the end. Log: start, every TRACKER_CREDITS_LOG_EVERY (500) titles, failures, slow steps
408. // (> 1.5 s), done. TRACKER_CREDITS=0 switches it off (off too without TMDB_READ_TOKEN). TRACKER_CREDITS_FAST=1 (the gate, fake
409. // TMDB only): 20 ms pause instead.
410. // MEMORY LIMIT: the runtime never frees what a fetch + JSON parse took (mission 061 measured ~6 MB more RSS per title on the real
411. // data, 300 fetches of a 440 KB JSON = +2.7 GB in a 10-line script) — the whole job in one process would add ~35 GB. Before every
412. // step the job reads its own RSS (/proc/self/status); above TRACKER_CREDITS_MAX_MB (16000) it PAUSES for this process (logged) and
413. // goes on with what is left after the next restart (deploy, `docker restart`).
414. creditsOn = env('TRACKER_CREDITS') != '0' && syncEnabled
415. creditsFast = env('TRACKER_CREDITS_FAST') == '1'
416. creditsLogEvery = env('TRACKER_CREDITS_LOG_EVERY') != null ? toNumber(env('TRACKER_CREDITS_LOG_EVERY')) : 500
417. creditsQueueIds = null
418. creditsPos = 0
419. creditsDone = false
420. creditsAt = 0
421. creditsNextAt = 0
422. creditsTries = {}
423. creditsMaxMb = env('TRACKER_CREDITS_MAX_MB') != null ? toNumber(env('TRACKER_CREDITS_MAX_MB')) : 16000
424. creditsTotals = { titles = 0 done = 0 failed = 0 cast = 0 crew = 0 people = 0 requests = 0 retries = 0 slowest = 0 }
425. // this process' resident memory in MB (VmRSS), 0 when unknown
426. rssMb = () => {
427. if (!exists('/proc/self/status')) { return 0 }
428. t = readFile('/proc/self/status')
429. i = t != null ? t.indexOf('VmRSS:') : -1
430. if (i < 0) { return 0 }
431. kb = toNumber(t.slice(i + 6, t.indexOf('kB', i)).trim())
432. return kb != null ? (kb - kb % 1024) / 1024 : 0
433. }
434. creditsLine = () => {
435. b = creditsTotals
436. return b.titles + '/' + creditsQueueIds.length + ' titles, done=' + b.done + ' failed=' + b.failed + ' cast=' + b.cast + ' crew=' + b.crew + ' people=' + b.people
437. + ' requests=' + b.requests + ' retries=' + b.retries + ' slowestStepMs=' + b.slowest + ' seconds=' + ((now() - creditsAt) / 1000)
438. }
439. creditsTick = () => {
440. if (creditsDone) { return null }
441. t = now()
442. if (creditsQueueIds == null) {
443. creditsQueueIds = creditsQueue()
444. creditsAt = t
445. console.log('credits job: start, ' + creditsQueueIds.length + ' titles without credits (' + (now() - t) + ' ms to find them)')
446. if (creditsQueueIds.length == 0) { creditsDone = true console.log('credits job: nothing to do') }
447. return null
448. }
449. if (t < creditsNextAt) { return null }
450. mb = rssMb()
451. if (mb > creditsMaxMb) {
452. creditsDone = true
453. persistCredits()
454. console.log('credits job: paused, RSS ' + mb + ' MB > TRACKER_CREDITS_MAX_MB ' + creditsMaxMb + ' — ' + (creditsQueueIds.length - creditsPos) + ' titles left for the next start; ' + creditsLine())
455. return null
456. }
457. id = creditsQueueIds[creditsPos]
458. creditsPos = creditsPos + 1
459. r = creditsStep(id)
460. took = now() - t
461. if (took > creditsTotals.slowest) { creditsTotals.slowest = took }
462. if (took > 1500) { console.log('credits job: slow step, show ' + id + ', ' + took + ' ms, cast=' + r.cast + ' crew=' + r.crew + ' people=' + r.people) }
463. if (r.error != '') { console.log('credits job: ' + r.error) }
464. creditsNextAt = now() + (creditsFast ? 20 : r.requests * msPerRequest)
465. if (r.retry && (creditsTries[id] == null || creditsTries[id] < 2)) {
466. creditsTries[id] = creditsTries[id] == null ? 1 : creditsTries[id] + 1
467. q = creditsQueueIds
468. q.push(id)
469. creditsQueueIds = q
470. creditsTotals.retries = creditsTotals.retries + 1
471. } else if (!r.skipped) {
472. creditsTotals.titles = creditsTotals.titles + 1
473. if (r.error != '') { creditsTotals.failed = creditsTotals.failed + 1 } else { creditsTotals.done = creditsTotals.done + 1 }
474. creditsTotals.cast = creditsTotals.cast + r.cast
475. creditsTotals.crew = creditsTotals.crew + r.crew
476. creditsTotals.people = creditsTotals.people + r.people
477. }
478. creditsTotals.requests = creditsTotals.requests + r.requests
479. finished = creditsPos >= creditsQueueIds.length
480. if (finished || creditsPos % 25 == 0) { persistCredits() }
481. if (!finished && !r.skipped && !r.retry && creditsTotals.titles > 0 && creditsTotals.titles % creditsLogEvery == 0) { console.log('credits job: ' + creditsLine()) }
482. if (finished) {
483. creditsDone = true
484. console.log('credits job done: ' + creditsLine())
485. }
486. return null
487. }
489. // ---- THE DUPLICATE MERGE (tracker.worldapi.org#31, mission 026, merge.hl) ----------------------------------------------------
490. // The old tracker's duplicate titles (same type + TMDB id) become ONE record each — follows, watches, seasons, cast, credits and
491. // timelines moved to the keeper, the others kept as tombstones whose slug and short id lead to it — and seasons that name a title
492. // without being in its list are merged into their twin or linked, and (tracker#39) a season's double episodes into one (merge.hl
493. // has the rules). On its own clock (every 0.1 s) ONE
494. // item per tick (a duplicate group, or one title's stray seasons; no TMDB), so pages are served between two items. The TMDB jobs
495. // below (daily sync, backfills, repair, credits, collections) wait until it is done: none of them works on a record that is about
496. // to become a tombstone. RESUMES after a restart (the queue = what is still duplicated / stray). Persisted every 10 items and at
497. // the end. Log: start, one line per tombstone / repaired title, done; with nothing to do one line. TRACKER_MERGE=0 switches it off.
498. mergeOn = env('TRACKER_MERGE') != '0'
499. mergeRun = { started = false done = !mergeOn at = 0 items = 0 slowest = 0 }
500. mergeTick = () => {
501. if (mergeRun.done) { return null }
502. t = now()
503. if (!mergeRun.started) {
504. mergeRun.started = true
505. mergeRun.at = t
506. n = mergePrepare()
507. st = mergeStats()
508. console.log('merge: start, ' + st.groups + ' duplicate groups (' + st.records + ' records), ' + st.strayTitles + ' titles with stray seasons, ' + st.doubleSeasons + ' seasons with double episodes (' + st.ms + ' ms to find them)')
509. if (n == 0) {
510. mergeRun.done = true
511. console.log('merge: nothing to do')
512. }
513. return null
514. }
515. r = mergeStep()
516. mergeRun.items = mergeRun.items + 1
517. took = now() - t
518. if (took > mergeRun.slowest) { mergeRun.slowest = took }
519. if (took > 1500) { console.log('merge: slow step, ' + took + ' ms') }
520. if (r.done || mergeRun.items % 10 == 0) { persistMerge() }
521. if (r.done) {
522. mergeRun.done = true
523. console.log('merge done: ' + mergeLine() + ' items=' + mergeRun.items + ' slowestStepMs=' + mergeRun.slowest + ' seconds=' + ((now() - mergeRun.at) / 1000))
524. }
525. return null
526. }
527. if (!mergeOn) { console.log('merge: off (TRACKER_MERGE=0)') }
528. else {
529. mergeClock = every(0.1)
530. on mergeClock.tick(x) {
531. mergeTick()
532. return null
533. }
534. }
536. // ONE clock for all five (a tick every 0.1 s): the daily sync's step once a second as before; the backfill's steps in
537. // between, only while no daily run is going on; the details repair only when neither has work (then its TVmaze status part,
538. // tracker#35); then the kind backfill (#21);
539. // the credits job (#28) last
540. clockOn = syncOn || backfillOn || repairOn || kindsOn || creditsOn
541. clockTicks = 0
542. if (syncOn) { console.log('tmdb sync: daily at ' + syncHour + ':00 UTC') } else { console.log('tmdb sync: off (' + (syncEnabled ? 'TRACKER_SYNC=0' : 'no TMDB_READ_TOKEN') + ')') }
543. if (!backfillOn) { console.log('adult backfill: off (' + (syncEnabled ? 'TRACKER_BACKFILL=0' : 'no TMDB_READ_TOKEN') + ')') }
544. if (!repairOn) { console.log('details repair: off (' + (syncEnabled ? 'TRACKER_REPAIR=0' : 'no TMDB_READ_TOKEN') + ')') }
545. if (!creditsOn) { console.log('credits job: off (' + (syncEnabled ? 'TRACKER_CREDITS=0' : 'no TMDB_READ_TOKEN') + ')') }
546. if (!kindsOn) { console.log('kind backfill: off (' + (syncEnabled ? 'TRACKER_KINDS=0' : 'no TMDB_READ_TOKEN') + ')') }
547. if (clockOn) {
548. syncClock = every(0.1)
549. on syncClock.tick(x) {
550. clockTicks = clockTicks + 1
551. // tracker#31: not before the duplicate merge is done
552. if (!mergeRun.done) { return null }
553. busy = syncQueue.length > 0 || syncPlanning
554. if (syncOn && clockTicks % 10 == 0) { syncTick() }
555. free = !busy && syncQueue.length == 0 && !syncPlanning
556. if (backfillOn && free && !backfillDone) { backfillTick() }
557. else if (repairOn && free && (!backfillOn || backfillDone) && !repairDone) { repairTick() }
558. else if (repairOn && free && (!backfillOn || backfillDone) && !statusDone) { statusTick() }
559. else if (kindsOn && free && (!backfillOn || backfillDone) && (!repairOn || (repairDone && statusDone)) && !kindDone) { kindTick() }
560. else if (creditsOn && free && (!backfillOn || backfillDone) && (!repairOn || (repairDone && statusDone)) && (!kindsOn || kindDone)) { creditsTick() }
561. return null
562. }
563. }
565. // ---- THE SHORT-ID BACKFILL (mission 060, tracker.worldapi.org#27, shortids.hl shortIdStep) ----------------------------
566. // Every show and person without a short id gets one in the background: on its own clock (every 0.1 s) ONE batch per tick,
567. // as many records as fit into TRACKER_SHORTID_BUDGET_MS (default 40 ms) — no TMDB, only the tables — so a page waits at
568. // most one batch. The queue is built at start (`ensureShortIds`) from what is still missing = it resumes after a restart.
569. // Persisted every 20 batches and at the end. Log: start, every TRACKER_SHORTID_LOG_EVERY (5000) records, done; with nothing
570. // to do one line. TRACKER_SHORTIDS=0 switches it off (the gate's resume check).
571. shortIdOn = env('TRACKER_SHORTIDS') != '0'
572. shortIdBudget = env('TRACKER_SHORTID_BUDGET_MS') != null ? toNumber(env('TRACKER_SHORTID_BUDGET_MS')) : 40
573. shortIdLogEvery = env('TRACKER_SHORTID_LOG_EVERY') != null ? toNumber(env('TRACKER_SHORTID_LOG_EVERY')) : 5000
574. shortIdRun = { at = 0 batches = 0 assigned = 0 nextLog = 0 slowest = 0 slowestPersist = 0 done = false }
575. shortIdTick = () => {
576. if (shortIdRun.done) { return null }
577. t = now()
578. r = shortIdStep(shortIdBudget)
579. took = now() - t
580. shortIdRun.batches = shortIdRun.batches + 1
581. shortIdRun.assigned = shortIdRun.assigned + r.assigned
582. if (took > shortIdRun.slowest) { shortIdRun.slowest = took }
583. if (r.done || shortIdRun.batches % 20 == 0) {
584. tp = now()
585. persistShortIds()
586. if (now() - tp > shortIdRun.slowestPersist) { shortIdRun.slowestPersist = now() - tp }
587. }
588. if (!r.done && shortIdRun.assigned >= shortIdRun.nextLog) {
589. console.log('short ids: ' + shortIdRun.assigned + ' assigned, ' + r.left + ' left, ' + ((now() - shortIdRun.at) / 1000) + ' s')
590. shortIdRun.nextLog = shortIdRun.nextLog + shortIdLogEvery
591. }
592. if (r.done) {
593. shortIdRun.done = true
594. s = shortIdStats()
595. console.log('short ids done: assigned=' + shortIdRun.assigned + ' oldKept=' + s.oldKept + ' batches=' + shortIdRun.batches + ' slowestBatchMs=' + shortIdRun.slowest + ' slowestPersistMs=' + shortIdRun.slowestPersist + ' seconds=' + ((now() - shortIdRun.at) / 1000))
596. }
597. return null
598. }
599. if (!shortIdOn) { console.log('short ids: backfill off (TRACKER_SHORTIDS=0), ' + shortIdsLeft() + ' records without one') }
600. else if (shortIdsLeft() == 0) { console.log('short ids: nothing to do') }
601. else {
602. shortIdRun.at = now()
603. shortIdRun.nextLog = shortIdLogEvery
604. console.log('short ids: backfill start, ' + shortIdsLeft() + ' records, ' + shortIdBudget + ' ms per batch')
605. shortIdClock = every(0.1)
606. on shortIdClock.tick(x) {
607. shortIdTick()
608. return null
609. }
610. }
612. // ---- THE COLLECTION SEED (mission 066, tracker.worldapi.org#19, collections.hl seedStep) ------------------------------------
613. // TMDB's movie collections become timelines: ONE movie per step (its details; once per collection the collection's parts), on
614. // its own 0.1 s clock like the adult backfill — a pause of 260 ms per TMDB request made (≤ 40 / 10 s). It WAITS while the daily
615. // sync or the adult backfill runs (they share TMDB's limit and the backfill brings the adult flags it skips by); since the merge
616. // (mission 070) also until the details repair, the kind backfill and the credits job are done (or paused) — the LAST job. The queue =
617. // every non-adult movie without a check row, so it RESUMES after a restart and does nothing once all are checked. Persisted
618. // every TRACKER_COLLECTIONS_PERSIST_EVERY (25) steps and at the end. Log: start, every TRACKER_COLLECTIONS_LOG_EVERY (500) movies, done.
619. // TRACKER_COLLECTIONS=0 switches it off; without TMDB_READ_TOKEN it is off too.
620. collectionsOn = env('TRACKER_COLLECTIONS') != '0' && syncEnabled
621. collectionsLogEvery = env('TRACKER_COLLECTIONS_LOG_EVERY') != null ? toNumber(env('TRACKER_COLLECTIONS_LOG_EVERY')) : 500
622. collectionsPersistEvery = env('TRACKER_COLLECTIONS_PERSIST_EVERY') != null ? toNumber(env('TRACKER_COLLECTIONS_PERSIST_EVERY')) : 25
623. collectionRun = { at = 0 steps = 0 nextAt = 0 done = false started = false slowest = 0 }
624. collectionTick = () => {
625. if (collectionRun.done) { return null }
626. // mission 070 (merge): also behind the details repair, the kind backfill and the credits job — one TMDB pacing at a time
627. if (!mergeRun.done || syncQueue.length > 0 || syncPlanning || (backfillOn && !backfillDone) || (repairOn && (!repairDone || !statusDone)) || (kindsOn && !kindDone) || (creditsOn && !creditsDone)) { return null }
628. t = now()
629. if (!collectionRun.started) {
630. collectionRun.started = true
631. collectionRun.at = t
632. n = seedPrepare()
633. console.log('collections: start, ' + n + ' movies to check (' + (now() - t) + ' ms to find them)')
634. if (n == 0) { collectionRun.done = true console.log('collections: nothing to do') }
635. return null
636. }
637. if (t < collectionRun.nextAt) { return null }
638. r = seedStep()
639. if (r.done) {
640. collectionRun.done = true
641. seedPersist()
642. console.log('collections done: ' + seedLine() + ' seconds=' + ((now() - collectionRun.at) / 1000) + ' slowestStepMs=' + collectionRun.slowest)
643. return null
644. }
645. collectionRun.steps = collectionRun.steps + 1
646. took = now() - t
647. if (took > collectionRun.slowest) { collectionRun.slowest = took }
648. if (r.error != '') { console.log('collections: ' + r.error) }
649. if (took > 1500) { console.log('collections: slow step, ' + took + ' ms') }
650. collectionRun.nextAt = now() + r.requests * msPerRequest
651. if (collectionRun.steps % collectionsPersistEvery == 0) { seedPersist() }
652. if (collectionRun.steps % collectionsLogEvery == 0) { console.log('collections: ' + seedLine() + ' seconds=' + ((now() - collectionRun.at) / 1000)) }
653. return null
654. }
655. if (!collectionsOn) { console.log('collections: off (' + (syncEnabled ? 'TRACKER_COLLECTIONS=0' : 'no TMDB_READ_TOKEN') + ')') }
656. else {
657. collectionClock = every(0.1)
658. on collectionClock.tick(x) {
659. collectionTick()
660. return null
661. }
662. }