tracker

33.2 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 { syncShow, syncShowDelta, persistSync } from './sync.hl'
13. import { emptyTotals, addTotals, totalsLine } from './sync-helpers.hl'
14. import { planStart, planStep, planIds, planLine, markSynced, isFollowedInRun } from './dailysync.hl'
15. import { adultBackfillQueue, backfillTitle, kindBackfillQueue, kindTitle } 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)
107. r = syncMode == 'delta' && !isFollowedInRun(id) ? syncShowDelta(id) : syncShow(id)
108. // tracker#18 + #28: the record's own fields (title, overview, tagline …) and the cast + crew from the same details answer
109. if (r.details != null) {
110. c = applySynced(id, r.details)
111. if (c.error != '') { console.log('tmdb sync: ' + c.error) }
112. if (c.renamed) { searchTitleRenamed(id) }
113. }
114. // the public lists (catalog.hl, tracker.worldapi.org#13): this show's place by its newest released episode; the sync
115. // also brings TMDB's adult flag (mission 054) → the search's view of it too
116. refreshCatalogShow(id)
117. searchTitleChanged(id)
118. took = now() - t
119. syncTotals = addTotals(syncTotals, r)
120. if (r.error != '') { console.log('tmdb sync: ' + r.error) }
121. // a step blocks every request meanwhile (Saturday Night Live, 53 seasons, 4 requests: ~1.7 s) — say so in the log
122. if (took > 1500) { console.log('tmdb sync: slow step, show ' + id + ', ' + r.requests + ' requests, ' + took + ' ms') }
123. syncNextAt = now() + pauseMsAfter(r)
124. if (syncQueue.length > 0 && syncTotals.shows % 10 == 0) { persistDetails() }
125. if (syncQueue.length == 0) { syncFinish() }
126. return null
127. }
129. // ---- THE ADULT-FLAG BACKFILL (mission 054, backfill.hl backfillTitle) -----------------------------------
130. // The public lists and the search show only titles TMDB says are not adult (shows.hl isPublicTitle; unknown = hidden). At
131. // start the app collects every title whose flag is still unknown and asks TMDB for ONE title per step — like the daily
132. // sync, on the same clock, so requests queued meanwhile are served between two steps: one details request (adult,
133. // poster_path, external ids) + the poster file when missing, then a pause of 260 ms per TMDB API request (≤ 40 / 10 s;
134. // the poster comes from TMDB's image host, not the API, and is not counted). Each answered title goes into the lists
135. // (catalog.hl refreshCatalogShow) and the search at once. RESUMES after a restart: the queue is built from what is
136. // still unknown. Stops when the queue is empty; titles TMDB answers 404 for are marked (`adultCheck`) and stay hidden; a
137. // passing failure (no answer, 429, 5xx) goes to the back of the queue, at most 2 more tries per start.
138. // While the daily sync runs, the backfill waits. Log: start, every TRACKER_BACKFILL_LOG_EVERY (500) titles, done.
139. // TRACKER_BACKFILL=0 switches it off; without TMDB_READ_TOKEN it is off too.
140. backfillOn = env('TRACKER_BACKFILL') != '0' && syncEnabled
141. backfillLogEvery = env('TRACKER_BACKFILL_LOG_EVERY') != null ? toNumber(env('TRACKER_BACKFILL_LOG_EVERY')) : 500
142. backfillQueue = null
143. backfillPos = 0
144. backfillDone = false
145. backfillAt = 0
146. backfillNextAt = 0
147. backfillTries = {}
148. backfillTotals = { titles = 0 adult = 0 notAdult = 0 unknown = 0 posters = 0 ids = 0 requests = 0 errors = 0 retries = 0 }
149. backfillLine = () => {
150. b = backfillTotals
151. return b.titles + '/' + backfillQueue.length + ' titles, adult=' + b.adult + ' notAdult=' + b.notAdult + ' unknown=' + b.unknown + ' posters=' + b.posters
152. + ' ids=' + b.ids + ' requests=' + b.requests + ' errors=' + b.errors + ' retries=' + b.retries + ' seconds=' + ((now() - backfillAt) / 1000)
153. }
154. backfillTick = () => {
155. if (backfillDone) { return null }
156. t = now()
157. if (backfillQueue == null) {
158. backfillQueue = adultBackfillQueue()
159. backfillAt = t
160. console.log('adult backfill: start, ' + backfillQueue.length + ' titles without an adult flag (' + (now() - t) + ' ms to find them)')
161. if (backfillQueue.length == 0) { backfillDone = true console.log('adult backfill: nothing to do') }
162. return null
163. }
164. if (t < backfillNextAt) { return null }
165. id = backfillQueue[backfillPos]
166. backfillPos = backfillPos + 1
167. r = backfillTitle(id)
168. if (r.retry && (backfillTries[id] == null || backfillTries[id] < 2)) {
169. backfillTries[id] = backfillTries[id] == null ? 1 : backfillTries[id] + 1
170. q = backfillQueue
171. q.push(id)
172. backfillQueue = q
173. backfillTotals.retries = backfillTotals.retries + 1
174. } else {
175. backfillTotals.titles = backfillTotals.titles + 1
176. 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 }
177. }
178. backfillTotals.posters = backfillTotals.posters + r.posters
179. backfillTotals.ids = backfillTotals.ids + r.ids
180. backfillTotals.requests = backfillTotals.requests + r.requests
181. backfillTotals.errors = backfillTotals.errors + r.errors
182. if (r.error != '') { console.log('adult backfill: ' + r.error) }
183. if (r.adult != null) {
184. refreshCatalogShow(id)
185. searchTitleChanged(id)
186. }
187. took = now() - t
188. if (took > 1500) { console.log('adult backfill: slow step, show ' + id + ', ' + took + ' ms') }
189. backfillNextAt = now() + r.requests * msPerRequest
190. finished = backfillPos >= backfillQueue.length
191. if (finished || backfillPos % 25 == 0) { persistSync() }
192. if (!finished && backfillTotals.titles > 0 && backfillTotals.titles % backfillLogEvery == 0 && !r.retry) { console.log('adult backfill: ' + backfillLine()) }
193. if (finished) {
194. backfillDone = true
195. console.log('adult backfill done: ' + backfillLine())
196. }
197. return null
198. }
200. // ---- THE DETAILS REPAIR (tracker.worldapi.org#26, details.hl completeStep) ----------------------------------------
201. // Titles the person pages added from a filmography (#16) are MINIMAL (no poster, seasons, ids, cast …), and (tracker#33, mission
202. // 029) the old tracker's migrated rows were never synced (5,697 on the live copy, 691 series without seasons) — shows.hl
203. // isIncomplete. Opening one completes it (components/show.hl); this job completes the rest in the background: at start it
204. // collects every incomplete, non-adult title (the visibly reduced ones first, details.hl detailsRepairQueue) and does ONE per
205. // step on the app's clock, like the adult backfill — a step is the daily sync's work for one title
206. // (1–3 TMDB requests + poster + TVmaze), requests queued meanwhile are served between two steps; then a pause of 260 ms per
207. // TMDB request + 550 ms per TVmaze request (pauseMsAfter: ≤ 40 / 10 s, ≤ 20 / 10 s). RESUMES after a restart (the queue is
208. // what is still incomplete); TMDB 404 marks the title (`detailsCheck`), a passing failure goes to the back of the queue (at
209. // most 2 more tries per start). Waits while the daily sync or the adult backfill runs. Persisted every 10 titles + at the end.
210. // Log: start, every TRACKER_REPAIR_LOG_EVERY (100) titles, each failure, done. TRACKER_REPAIR=0 switches it off (off too
211. // without TMDB_READ_TOKEN). TRACKER_REPAIR_FAST=1 (the gate, fake TMDB only): 20 ms pause instead.
212. repairOn = env('TRACKER_REPAIR') != '0' && syncEnabled
213. repairFast = env('TRACKER_REPAIR_FAST') == '1'
214. repairLogEvery = env('TRACKER_REPAIR_LOG_EVERY') != null ? toNumber(env('TRACKER_REPAIR_LOG_EVERY')) : 100
215. repairQueue = null
216. repairPos = 0
217. repairDone = false
218. repairAt = 0
219. repairNextAt = 0
220. repairTries = {}
221. repairTotals = { titles = 0 complete = 0 failed = 0 adult = 0 newSeasons = 0 newEpisodes = 0 posters = 0 cast = 0 requests = 0 tvmazeRequests = 0 retries = 0 }
222. repairLine = () => {
223. b = repairTotals
224. return b.titles + '/' + repairQueue.length + ' titles, complete=' + b.complete + ' failed=' + b.failed + ' adult=' + b.adult + ' newSeasons=' + b.newSeasons + ' newEpisodes=' + b.newEpisodes
225. + ' posters=' + b.posters + ' cast=' + b.cast + ' requests=' + b.requests + ' tvmazeRequests=' + b.tvmazeRequests + ' retries=' + b.retries + ' seconds=' + ((now() - repairAt) / 1000)
226. }
227. repairTick = () => {
228. if (repairDone) { return null }
229. t = now()
230. if (repairQueue == null) {
231. repairQueue = detailsRepairQueue()
232. repairAt = t
233. console.log('details repair: start, ' + repairQueue.length + ' incomplete titles (' + (now() - t) + ' ms to find them)')
234. if (repairQueue.length == 0) { repairDone = true console.log('details repair: nothing to do') }
235. return null
236. }
237. if (t < repairNextAt) { return null }
238. id = repairQueue[repairPos]
239. r = completeStep(id)
240. took = now() - t
241. if (took > 1500) { console.log('details repair: slow step, show ' + id + ', ' + took + ' ms') }
242. if (r.error != '') { console.log('details repair: ' + r.error) }
243. repairNextAt = now() + (repairFast ? 20 : pauseMsAfter(r))
244. // a series takes several steps: the same title again next time
245. if (!r.done && !r.retry) { return null }
246. repairPos = repairPos + 1
247. if (r.retry && (repairTries[id] == null || repairTries[id] < 2)) {
248. repairTries[id] = repairTries[id] == null ? 1 : repairTries[id] + 1
249. q = repairQueue
250. q.push(id)
251. repairQueue = q
252. repairTotals.retries = repairTotals.retries + 1
253. } else if (!r.skipped) {
254. // done: its counts (a failed first step: the request alone)
255. sum = r.sum
256. repairTotals.titles = repairTotals.titles + 1
257. 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 }
258. repairTotals.newSeasons = repairTotals.newSeasons + sum.newSeasons
259. repairTotals.newEpisodes = repairTotals.newEpisodes + sum.newEpisodes
260. repairTotals.posters = repairTotals.posters + sum.posters
261. repairTotals.cast = repairTotals.cast + sum.cast
262. repairTotals.requests = repairTotals.requests + (sum.steps == 0 ? r.requests : sum.requests)
263. repairTotals.tvmazeRequests = repairTotals.tvmazeRequests + sum.tvmazeRequests
264. }
265. finished = repairPos >= repairQueue.length
266. if (finished || repairPos % 10 == 0) { persistDetails() }
267. if (!finished && !r.skipped && !r.retry && repairTotals.titles > 0 && repairTotals.titles % repairLogEvery == 0) { console.log('details repair: ' + repairLine()) }
268. if (finished) {
269. repairDone = true
270. console.log('details repair done: ' + repairLine())
271. }
272. return null
273. }
275. // ---- THE KIND BACKFILL (tracker.worldapi.org#21, mission 068, backfill.hl kindTitle) ----------------------------------
276. // Every TV title without a stored kind gets one: a title with a show genre (Reality / Talk / News) from the stored data (no
277. // request), any other from TMDB's TV type (ONE request: GET /tv/<id>, then 260 ms pause). Starts once the adult backfill is
278. // 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
279. // /series to /shows at once (catalog.hl refreshCatalogShow). RESUMES after a restart (the queue = every TV title still without
280. // a kind); a 404 stores the genres' answer + `kindCheck`; a passing failure goes to the back of the queue, at most 2 more tries
281. // per start. Log: start, every TRACKER_KINDS_LOG_EVERY (500) titles, done. TRACKER_KINDS=0 switches it off; without
282. // TMDB_READ_TOKEN it is off too (the lists still split by genre: shows.hl titleKind).
283. kindsOn = env('TRACKER_KINDS') != '0' && syncEnabled
284. kindsLogEvery = env('TRACKER_KINDS_LOG_EVERY') != null ? toNumber(env('TRACKER_KINDS_LOG_EVERY')) : 500
285. kindQueue = null
286. kindPos = 0
287. kindDone = false
288. kindAt = 0
289. kindNextAt = 0
290. kindTries = {}
291. kindTotals = { titles = 0 series = 0 shows = 0 fromGenre = 0 moved = 0 requests = 0 errors = 0 retries = 0 }
292. kindLine = () => {
293. k = kindTotals
294. return k.titles + '/' + kindQueue.length + ' titles, series=' + k.series + ' shows=' + k.shows + ' fromGenre=' + k.fromGenre + ' moved=' + k.moved
295. + ' requests=' + k.requests + ' errors=' + k.errors + ' retries=' + k.retries + ' seconds=' + ((now() - kindAt) / 1000)
296. }
297. kindTick = () => {
298. if (kindDone) { return null }
299. t = now()
300. if (kindQueue == null) {
301. kindQueue = kindBackfillQueue()
302. kindAt = t
303. console.log('kind backfill: start, ' + kindQueue.length + ' TV titles without a kind (' + (now() - t) + ' ms to find them)')
304. if (kindQueue.length == 0) { kindDone = true console.log('kind backfill: nothing to do') }
305. return null
306. }
307. if (t < kindNextAt) { return null }
308. id = kindQueue[kindPos]
309. kindPos = kindPos + 1
310. r = kindTitle(id)
311. if (r.retry && (kindTries[id] == null || kindTries[id] < 2)) {
312. kindTries[id] = kindTries[id] == null ? 1 : kindTries[id] + 1
313. q = kindQueue
314. q.push(id)
315. kindQueue = q
316. kindTotals.retries = kindTotals.retries + 1
317. } else {
318. kindTotals.titles = kindTotals.titles + 1
319. if (r.kind == 'show') { kindTotals.shows = kindTotals.shows + 1 } else if (r.kind == 'series') { kindTotals.series = kindTotals.series + 1 }
320. if (r.fromGenre) { kindTotals.fromGenre = kindTotals.fromGenre + 1 }
321. }
322. kindTotals.requests = kindTotals.requests + r.requests
323. kindTotals.errors = kindTotals.errors + r.errors
324. if (r.error != '') { console.log('kind backfill: ' + r.error) }
325. // until now the genres decided (no show genre = series): only a title TMDB's type makes a SHOW changes its list
326. if (r.kind == 'show' && !r.fromGenre) {
327. refreshCatalogShow(id)
328. kindTotals.moved = kindTotals.moved + 1
329. }
330. took = now() - t
331. if (took > 1500) { console.log('kind backfill: slow step, show ' + id + ', ' + took + ' ms') }
332. kindNextAt = now() + r.requests * msPerRequest
333. finished = kindPos >= kindQueue.length
334. if (finished || kindPos % 25 == 0) { persistSync() }
335. if (!finished && kindTotals.titles > 0 && kindTotals.titles % kindsLogEvery == 0 && !r.retry) { console.log('kind backfill: ' + kindLine()) }
336. if (finished) {
337. kindDone = true
338. console.log('kind backfill done: ' + kindLine())
339. }
340. return null
341. }
343. // ---- THE CREDITS JOB (tracker.worldapi.org#28, credits.hl creditsStep) ------------------------------------------------------
344. // Every PUBLIC title (adult == false; adult/unknown are skipped) with a TMDB id gets its whole cast + crew once: at start (after
345. // the adult backfill and the details repair are done) the job collects every such title without `creditsAt` and asks ONE per
346. // step on the app's clock — one request (details + credits / aggregate_credits), then 260 ms (≤ 40 / 10 s); requests queued
347. // meanwhile are served between two steps. Titles the details repair, the daily sync or an import already did carry `creditsAt`
348. // and are not asked. RESUMES after a restart (the queue = what still has no credits); TMDB 404 / adult → `creditsCheck`, never
349. // asked again; a passing failure goes to the back of the queue (at most 2 more tries per start). Waits while the daily sync
350. // runs. Persisted (shows + people only) every 25 titles + at the end. Log: start, every TRACKER_CREDITS_LOG_EVERY (500) titles, failures, slow steps
351. // (> 1.5 s), done. TRACKER_CREDITS=0 switches it off (off too without TMDB_READ_TOKEN). TRACKER_CREDITS_FAST=1 (the gate, fake
352. // TMDB only): 20 ms pause instead.
353. // MEMORY LIMIT: the runtime never frees what a fetch + JSON parse took (mission 061 measured ~6 MB more RSS per title on the real
354. // 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
355. // step the job reads its own RSS (/proc/self/status); above TRACKER_CREDITS_MAX_MB (16000) it PAUSES for this process (logged) and
356. // goes on with what is left after the next restart (deploy, `docker restart`).
357. creditsOn = env('TRACKER_CREDITS') != '0' && syncEnabled
358. creditsFast = env('TRACKER_CREDITS_FAST') == '1'
359. creditsLogEvery = env('TRACKER_CREDITS_LOG_EVERY') != null ? toNumber(env('TRACKER_CREDITS_LOG_EVERY')) : 500
360. creditsQueueIds = null
361. creditsPos = 0
362. creditsDone = false
363. creditsAt = 0
364. creditsNextAt = 0
365. creditsTries = {}
366. creditsMaxMb = env('TRACKER_CREDITS_MAX_MB') != null ? toNumber(env('TRACKER_CREDITS_MAX_MB')) : 16000
367. creditsTotals = { titles = 0 done = 0 failed = 0 cast = 0 crew = 0 people = 0 requests = 0 retries = 0 slowest = 0 }
368. // this process' resident memory in MB (VmRSS), 0 when unknown
369. rssMb = () => {
370. if (!exists('/proc/self/status')) { return 0 }
371. t = readFile('/proc/self/status')
372. i = t != null ? t.indexOf('VmRSS:') : -1
373. if (i < 0) { return 0 }
374. kb = toNumber(t.slice(i + 6, t.indexOf('kB', i)).trim())
375. return kb != null ? (kb - kb % 1024) / 1024 : 0
376. }
377. creditsLine = () => {
378. b = creditsTotals
379. return b.titles + '/' + creditsQueueIds.length + ' titles, done=' + b.done + ' failed=' + b.failed + ' cast=' + b.cast + ' crew=' + b.crew + ' people=' + b.people
380. + ' requests=' + b.requests + ' retries=' + b.retries + ' slowestStepMs=' + b.slowest + ' seconds=' + ((now() - creditsAt) / 1000)
381. }
382. creditsTick = () => {
383. if (creditsDone) { return null }
384. t = now()
385. if (creditsQueueIds == null) {
386. creditsQueueIds = creditsQueue()
387. creditsAt = t
388. console.log('credits job: start, ' + creditsQueueIds.length + ' titles without credits (' + (now() - t) + ' ms to find them)')
389. if (creditsQueueIds.length == 0) { creditsDone = true console.log('credits job: nothing to do') }
390. return null
391. }
392. if (t < creditsNextAt) { return null }
393. mb = rssMb()
394. if (mb > creditsMaxMb) {
395. creditsDone = true
396. persistCredits()
397. console.log('credits job: paused, RSS ' + mb + ' MB > TRACKER_CREDITS_MAX_MB ' + creditsMaxMb + ' — ' + (creditsQueueIds.length - creditsPos) + ' titles left for the next start; ' + creditsLine())
398. return null
399. }
400. id = creditsQueueIds[creditsPos]
401. creditsPos = creditsPos + 1
402. r = creditsStep(id)
403. took = now() - t
404. if (took > creditsTotals.slowest) { creditsTotals.slowest = took }
405. if (took > 1500) { console.log('credits job: slow step, show ' + id + ', ' + took + ' ms, cast=' + r.cast + ' crew=' + r.crew + ' people=' + r.people) }
406. if (r.error != '') { console.log('credits job: ' + r.error) }
407. creditsNextAt = now() + (creditsFast ? 20 : r.requests * msPerRequest)
408. if (r.retry && (creditsTries[id] == null || creditsTries[id] < 2)) {
409. creditsTries[id] = creditsTries[id] == null ? 1 : creditsTries[id] + 1
410. q = creditsQueueIds
411. q.push(id)
412. creditsQueueIds = q
413. creditsTotals.retries = creditsTotals.retries + 1
414. } else if (!r.skipped) {
415. creditsTotals.titles = creditsTotals.titles + 1
416. if (r.error != '') { creditsTotals.failed = creditsTotals.failed + 1 } else { creditsTotals.done = creditsTotals.done + 1 }
417. creditsTotals.cast = creditsTotals.cast + r.cast
418. creditsTotals.crew = creditsTotals.crew + r.crew
419. creditsTotals.people = creditsTotals.people + r.people
420. }
421. creditsTotals.requests = creditsTotals.requests + r.requests
422. finished = creditsPos >= creditsQueueIds.length
423. if (finished || creditsPos % 25 == 0) { persistCredits() }
424. if (!finished && !r.skipped && !r.retry && creditsTotals.titles > 0 && creditsTotals.titles % creditsLogEvery == 0) { console.log('credits job: ' + creditsLine()) }
425. if (finished) {
426. creditsDone = true
427. console.log('credits job done: ' + creditsLine())
428. }
429. return null
430. }
432. // ---- THE DUPLICATE MERGE (tracker.worldapi.org#31, mission 026, merge.hl) ----------------------------------------------------
433. // The old tracker's duplicate titles (same type + TMDB id) become ONE record each — follows, watches, seasons, cast, credits and
434. // timelines moved to the keeper, the others kept as tombstones whose slug and short id lead to it — and seasons that name a title
435. // without being in its list are merged into their twin or linked (merge.hl has the rules). On its own clock (every 0.1 s) ONE
436. // item per tick (a duplicate group, or one title's stray seasons; no TMDB), so pages are served between two items. The TMDB jobs
437. // below (daily sync, backfills, repair, credits, collections) wait until it is done: none of them works on a record that is about
438. // to become a tombstone. RESUMES after a restart (the queue = what is still duplicated / stray). Persisted every 10 items and at
439. // the end. Log: start, one line per tombstone / repaired title, done; with nothing to do one line. TRACKER_MERGE=0 switches it off.
440. mergeOn = env('TRACKER_MERGE') != '0'
441. mergeRun = { started = false done = !mergeOn at = 0 items = 0 slowest = 0 }
442. mergeTick = () => {
443. if (mergeRun.done) { return null }
444. t = now()
445. if (!mergeRun.started) {
446. mergeRun.started = true
447. mergeRun.at = t
448. n = mergePrepare()
449. st = mergeStats()
450. console.log('merge: start, ' + st.groups + ' duplicate groups (' + st.records + ' records), ' + st.strayTitles + ' titles with stray seasons (' + st.ms + ' ms to find them)')
451. if (n == 0) {
452. mergeRun.done = true
453. console.log('merge: nothing to do')
454. }
455. return null
456. }
457. r = mergeStep()
458. mergeRun.items = mergeRun.items + 1
459. took = now() - t
460. if (took > mergeRun.slowest) { mergeRun.slowest = took }
461. if (took > 1500) { console.log('merge: slow step, ' + took + ' ms') }
462. if (r.done || mergeRun.items % 10 == 0) { persistMerge() }
463. if (r.done) {
464. mergeRun.done = true
465. console.log('merge done: ' + mergeLine() + ' items=' + mergeRun.items + ' slowestStepMs=' + mergeRun.slowest + ' seconds=' + ((now() - mergeRun.at) / 1000))
466. }
467. return null
468. }
469. if (!mergeOn) { console.log('merge: off (TRACKER_MERGE=0)') }
470. else {
471. mergeClock = every(0.1)
472. on mergeClock.tick(x) {
473. mergeTick()
474. return null
475. }
476. }
478. // 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
479. // between, only while no daily run is going on; the details repair only when neither has work; then the kind backfill (#21);
480. // the credits job (#28) last
481. clockOn = syncOn || backfillOn || repairOn || kindsOn || creditsOn
482. clockTicks = 0
483. 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') + ')') }
484. if (!backfillOn) { console.log('adult backfill: off (' + (syncEnabled ? 'TRACKER_BACKFILL=0' : 'no TMDB_READ_TOKEN') + ')') }
485. if (!repairOn) { console.log('details repair: off (' + (syncEnabled ? 'TRACKER_REPAIR=0' : 'no TMDB_READ_TOKEN') + ')') }
486. if (!creditsOn) { console.log('credits job: off (' + (syncEnabled ? 'TRACKER_CREDITS=0' : 'no TMDB_READ_TOKEN') + ')') }
487. if (!kindsOn) { console.log('kind backfill: off (' + (syncEnabled ? 'TRACKER_KINDS=0' : 'no TMDB_READ_TOKEN') + ')') }
488. if (clockOn) {
489. syncClock = every(0.1)
490. on syncClock.tick(x) {
491. clockTicks = clockTicks + 1
492. // tracker#31: not before the duplicate merge is done
493. if (!mergeRun.done) { return null }
494. busy = syncQueue.length > 0 || syncPlanning
495. if (syncOn && clockTicks % 10 == 0) { syncTick() }
496. free = !busy && syncQueue.length == 0 && !syncPlanning
497. if (backfillOn && free && !backfillDone) { backfillTick() }
498. else if (repairOn && free && (!backfillOn || backfillDone) && !repairDone) { repairTick() }
499. else if (kindsOn && free && (!backfillOn || backfillDone) && (!repairOn || repairDone) && !kindDone) { kindTick() }
500. else if (creditsOn && free && (!backfillOn || backfillDone) && (!repairOn || repairDone) && (!kindsOn || kindDone)) { creditsTick() }
501. return null
502. }
503. }
505. // ---- THE SHORT-ID BACKFILL (mission 060, tracker.worldapi.org#27, shortids.hl shortIdStep) ----------------------------
506. // 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,
507. // as many records as fit into TRACKER_SHORTID_BUDGET_MS (default 40 ms) — no TMDB, only the tables — so a page waits at
508. // most one batch. The queue is built at start (`ensureShortIds`) from what is still missing = it resumes after a restart.
509. // Persisted every 20 batches and at the end. Log: start, every TRACKER_SHORTID_LOG_EVERY (5000) records, done; with nothing
510. // to do one line. TRACKER_SHORTIDS=0 switches it off (the gate's resume check).
511. shortIdOn = env('TRACKER_SHORTIDS') != '0'
512. shortIdBudget = env('TRACKER_SHORTID_BUDGET_MS') != null ? toNumber(env('TRACKER_SHORTID_BUDGET_MS')) : 40
513. shortIdLogEvery = env('TRACKER_SHORTID_LOG_EVERY') != null ? toNumber(env('TRACKER_SHORTID_LOG_EVERY')) : 5000
514. shortIdRun = { at = 0 batches = 0 assigned = 0 nextLog = 0 slowest = 0 slowestPersist = 0 done = false }
515. shortIdTick = () => {
516. if (shortIdRun.done) { return null }
517. t = now()
518. r = shortIdStep(shortIdBudget)
519. took = now() - t
520. shortIdRun.batches = shortIdRun.batches + 1
521. shortIdRun.assigned = shortIdRun.assigned + r.assigned
522. if (took > shortIdRun.slowest) { shortIdRun.slowest = took }
523. if (r.done || shortIdRun.batches % 20 == 0) {
524. tp = now()
525. persistShortIds()
526. if (now() - tp > shortIdRun.slowestPersist) { shortIdRun.slowestPersist = now() - tp }
527. }
528. if (!r.done && shortIdRun.assigned >= shortIdRun.nextLog) {
529. console.log('short ids: ' + shortIdRun.assigned + ' assigned, ' + r.left + ' left, ' + ((now() - shortIdRun.at) / 1000) + ' s')
530. shortIdRun.nextLog = shortIdRun.nextLog + shortIdLogEvery
531. }
532. if (r.done) {
533. shortIdRun.done = true
534. s = shortIdStats()
535. 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))
536. }
537. return null
538. }
539. if (!shortIdOn) { console.log('short ids: backfill off (TRACKER_SHORTIDS=0), ' + shortIdsLeft() + ' records without one') }
540. else if (shortIdsLeft() == 0) { console.log('short ids: nothing to do') }
541. else {
542. shortIdRun.at = now()
543. shortIdRun.nextLog = shortIdLogEvery
544. console.log('short ids: backfill start, ' + shortIdsLeft() + ' records, ' + shortIdBudget + ' ms per batch')
545. shortIdClock = every(0.1)
546. on shortIdClock.tick(x) {
547. shortIdTick()
548. return null
549. }
550. }
552. // ---- THE COLLECTION SEED (mission 066, tracker.worldapi.org#19, collections.hl seedStep) ------------------------------------
553. // TMDB's movie collections become timelines: ONE movie per step (its details; once per collection the collection's parts), on
554. // 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
555. // sync or the adult backfill runs (they share TMDB's limit and the backfill brings the adult flags it skips by); since the merge
556. // (mission 070) also until the details repair, the kind backfill and the credits job are done (or paused) — the LAST job. The queue =
557. // every non-adult movie without a check row, so it RESUMES after a restart and does nothing once all are checked. Persisted
558. // every TRACKER_COLLECTIONS_PERSIST_EVERY (25) steps and at the end. Log: start, every TRACKER_COLLECTIONS_LOG_EVERY (500) movies, done.
559. // TRACKER_COLLECTIONS=0 switches it off; without TMDB_READ_TOKEN it is off too.
560. collectionsOn = env('TRACKER_COLLECTIONS') != '0' && syncEnabled
561. collectionsLogEvery = env('TRACKER_COLLECTIONS_LOG_EVERY') != null ? toNumber(env('TRACKER_COLLECTIONS_LOG_EVERY')) : 500
562. collectionsPersistEvery = env('TRACKER_COLLECTIONS_PERSIST_EVERY') != null ? toNumber(env('TRACKER_COLLECTIONS_PERSIST_EVERY')) : 25
563. collectionRun = { at = 0 steps = 0 nextAt = 0 done = false started = false slowest = 0 }
564. collectionTick = () => {
565. if (collectionRun.done) { return null }
566. // mission 070 (merge): also behind the details repair, the kind backfill and the credits job — one TMDB pacing at a time
567. if (!mergeRun.done || syncQueue.length > 0 || syncPlanning || (backfillOn && !backfillDone) || (repairOn && !repairDone) || (kindsOn && !kindDone) || (creditsOn && !creditsDone)) { return null }
568. t = now()
569. if (!collectionRun.started) {
570. collectionRun.started = true
571. collectionRun.at = t
572. n = seedPrepare()
573. console.log('collections: start, ' + n + ' movies to check (' + (now() - t) + ' ms to find them)')
574. if (n == 0) { collectionRun.done = true console.log('collections: nothing to do') }
575. return null
576. }
577. if (t < collectionRun.nextAt) { return null }
578. r = seedStep()
579. if (r.done) {
580. collectionRun.done = true
581. seedPersist()
582. console.log('collections done: ' + seedLine() + ' seconds=' + ((now() - collectionRun.at) / 1000) + ' slowestStepMs=' + collectionRun.slowest)
583. return null
584. }
585. collectionRun.steps = collectionRun.steps + 1
586. took = now() - t
587. if (took > collectionRun.slowest) { collectionRun.slowest = took }
588. if (r.error != '') { console.log('collections: ' + r.error) }
589. if (took > 1500) { console.log('collections: slow step, ' + took + ' ms') }
590. collectionRun.nextAt = now() + r.requests * msPerRequest
591. if (collectionRun.steps % collectionsPersistEvery == 0) { seedPersist() }
592. if (collectionRun.steps % collectionsLogEvery == 0) { console.log('collections: ' + seedLine() + ' seconds=' + ((now() - collectionRun.at) / 1000)) }
593. return null
594. }
595. if (!collectionsOn) { console.log('collections: off (' + (syncEnabled ? 'TRACKER_COLLECTIONS=0' : 'no TMDB_READ_TOKEN') + ')') }
596. else {
597. collectionClock = every(0.1)
598. on collectionClock.tick(x) {
599. collectionTick()
600. return null
601. }
602. }