tracker

33.0 KB

1. // project.hl — tracker.worldapi.org: STEP 1 (tracker.worldapi.org#1), an empty shell. Login copied unchanged from
2. // calendar.worldapi.org (rejected once for a centered sign-in and no header selector — architect, 2026-09-27):
3. // ident only, no own passwords (README "How apps use ident" of ident.worldapi.org), the identity selector in the
4. // header (components/main.hl). No shows, no data yet — later steps come from the creator (CONCEPT.md).
5. //
6. // the header's ident-selector / "Log in with ident" -> <ident>/login?key=&return=<this app>/login/callback
7. // /login/callback?ident_code=<code> the app's server exchanges the code for the identity's short id
8. // (POST <ident>/api/exchange) and signs this app's OWN session in
9. // (session.user = { id = <users @id> }, users.hl)
10. // the header's "Log out" button signs this app's session out via the face trackerLogOut (components/main.hl)
11. // — ident's own session, and the browser's ident cookie, are untouched
12. //
13. // Config (env; no committed secret — the first deploy registers this app in ident and sets these, done by the
14. // architect, as with ident's own .env for SMTP):
15. // TRACKER_PORT (45008), TRACKER_URL (this app's own public address, for the return URL),
16. // IDENT_URL (https://ident.worldapi.org), TRACKER_KEY (pk_…), TRACKER_SECRET (sk_…)
17. import WebFramework from 'hl:web'
18. import { Response } from 'hl:http1'
19. import { randomBytes } from 'hl:crypto'
20. import { env } from 'hl:proc'
21. import { readBytes, readFile, exists } from 'hl:fs'
22. import { now, timestamp, every } from 'hl:time'
24. import Styles from './styles.hl'
25. import { exchangeCode, ensureUser } from './users.hl'
26. import Home from './components/home.hl'
27. import LoginFailed from './components/loginfailed.hl'
28. import Show from './components/show.hl'
29. import Unwatched from './components/unwatched.hl'
30. import Schedule from './components/schedule.hl'
31. import MyShows from './components/myshows.hl'
32. import MyMovies from './components/mymovies.hl'
33. import Movies from './components/movies.hl'
34. import AllShows from './components/allshows.hl'
35. import Search from './components/search.hl'
36. import Person from './components/person.hl'
37. import { ensureIndex, searchTitleChanged, searchTitleRenamed } from './search.hl'
38. import { syncEnabled, syncShow, syncShowDelta, persistSync, emptyTotals, addTotals, totalsLine, pauseMsAfter, msPerRequest, adultBackfillQueue, backfillTitle } from './tmdbsync.hl'
39. import { detailsRepairQueue, completeStep, completeLine, persistDetails, persistCredits, applySynced, creditsQueue, creditsStep } from './details.hl'
40. import { planStart, planStep, planIds, planLine, markSynced, clearCopiedSummaries, isFollowedInRun } from './deltasync.hl'
41. import { allFollowedShowIds } from './follows.hl'
42. import { refreshCatalogShow } from './catalog.hl'
43. import { dark, darker } from './shared/tokens.hl'
45. static appTitle = "tracker"
47. // ---- THE INSTALLABLE APP (tracker.worldapi.org#10), the same way calendar.worldapi.org does it: hl:web's own web app
48. // manifest (/__hl/manifest.webmanifest, linked from every head with the apple-touch-icon and theme-color) and service
49. // worker (/__hl/sw.js) from these settings — no JavaScript of ours. Icons in icons/ (icon.svg is the source, the PNGs
50. // are rendered from it with rsvg-convert, README "Icons"). The theme colour is the header's background (darker), the
51. // splash background the page's (dark) — both off the tokens, not written out again.
52. appThemeColor = darker.value
53. appBackgroundColor = dark.value
54. appIcons = [
55. { src = '/icons/icon-192.png' sizes = '192x192' purpose = 'any' }
56. { src = '/icons/icon-512.png' sizes = '512x512' purpose = 'any' }
57. { src = '/icons/icon-192.png' sizes = '192x192' purpose = 'maskable' }
58. { src = '/icons/icon-512.png' sizes = '512x512' purpose = 'maskable' }
59. ]
60. appTouchIcon = '/icons/apple-touch-icon.png'
61. appFavicon = '/icons/favicon.svg'
62. // OFFLINE: only the shell — the header and `/` (components/main.hl says "You are offline" while the browser has no
63. // network). The data pages are not kept: without a network they get hl:web's "Unavailable offline" page.
64. offline = [ Home ]
66. styles = Styles
68. port = env('TRACKER_PORT') != null ? toNumber(env('TRACKER_PORT')) : 45008
69. watching = env('TRACKER_WATCH') != '0'
70. // read by hl:web's own manifest config (WebFramework.hl `cfg.sessionDir`), not the constructor call below
71. sessionDir = env('TRACKER_SESSIONS') != null ? env('TRACKER_SESSIONS') : null
73. // ---- THE LOGIN BUTTON'S RETURN (ident README "How apps use ident"), copied from calendar.worldapi.org's project.hl
74. // BACK TO THE PAGE: /login.js puts `?next=` into the button's return URL at the click. Only a same-origin PATH
75. // goes (one `/`, URL-safe characters, ≤ 500). Anything else → `/`.
76. nextChars = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789-._~/?&=%+,;@!$()*:'
77. safePath = (want) => {
78. if (want == null || hlTypeName(want) != 'String' || want == '' || want.length > 500) { return '/' }
79. if (want.slice(0, 1) != '/' || want.slice(0, 2) == '//' || want.slice(0, 7) == '/login/') { return '/' }
80. let i = 0
81. while (i < want.length) {
82. if (!nextChars.includes(want[i])) { return '/' }
83. i = i + 1
84. }
85. return want
86. }
88. // A FAILED LOGIN is a page (components/loginfailed.hl): the reason is parked in the session, then → /login/failed
89. failed = (req, why) => {
90. let s = req.session
91. let fresh = s == null
92. if (fresh) { s = server.sessions.mint() }
93. s.data.loginError = why
94. server.sessions.save(s)
95. let res = new Response('login failed: ' + why, { status = 302 headers = { 'Location' = '/login/failed' 'Cache-Control' = 'no-store' 'Content-Type' = 'text/plain; charset=utf-8' } })
96. if (fresh) { res.headers['Set-Cookie'] = server.sessions.cookieHeader(s.id) }
97. return res
98. }
100. // the function route gets the cookie's session as req.session (hybriel #11); none yet → minted here
101. loginCallback = (route, req) => {
102. if (req.method != 'GET') { return failed(req, 'GET only') }
103. let q = req.query != null ? req.query : {}
104. let code = q.ident_code
105. if (code == null || code == '') { return failed(req, 'ident sent no login code') }
106. let x = exchangeCode(code)
107. if (x.error != null) { return failed(req, x.error) }
108. let u = ensureUser(x.identity)
109. if (u == null) { return failed(req, 'could not store the user') }
110. let s = req.session
111. let fresh = s == null
112. if (fresh) { s = server.sessions.mint() }
113. s.user = { id = u.id }
114. s.data.tag = randomBytes(16)
115. s.data.loginError = null
116. server.sessions.save(s)
117. let res = new Response('logged in', { status = 302 headers = { 'Location' = safePath(q.next) 'Cache-Control' = 'no-store' 'Content-Type' = 'text/plain; charset=utf-8' } })
118. if (fresh) { res.headers['Set-Cookie'] = server.sessions.cookieHeader(s.id) }
119. return res
120. }
122. // ---- /posters/<name> (tracker.worldapi.org#4): a show's poster image, served dynamically from
123. // storage/mpackdb/posters/ (the poster's file name is the show's oldId + its migrated extension,
124. // shows.hl posterName) — falls back to a placeholder svg when the real file isn't there yet (the
125. // mongo export never included the actual poster bytes, see the report's `open`).
126. postersDir = (env('TRACKER_STORAGE') != null ? env('TRACKER_STORAGE') : './storage/mpackdb') + '/posters'
127. placeholderPoster = './assets/poster-placeholder.svg'
128. posterNameChars = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789.-_'
129. isSafePosterName = (name) => {
130. if (name == null || hlTypeName(name) != 'String' || name == '' || name.length > 100) { return false }
131. let i = 0
132. while (i < name.length) {
133. if (!posterNameChars.includes(name[i])) { return false }
134. i = i + 1
135. }
136. return true
137. }
138. endsWithStr = (s, suffix) => { return s.length >= suffix.length && s.slice(s.length - suffix.length) == suffix }
139. mimeOfPoster = (name) => {
140. if (endsWithStr(name, '.png')) { return 'image/png' }
141. if (endsWithStr(name, '.webp')) { return 'image/webp' }
142. if (endsWithStr(name, '.svg')) { return 'image/svg+xml' }
143. return 'image/jpeg'
144. }
145. posterRoute = (route, req) => {
146. let name = route.params != null ? route.params.name : null
147. if (!isSafePosterName(name)) { return new Response('bad poster name', { status = 400 headers = { 'Content-Type' = 'text/plain; charset=utf-8' } }) }
148. let path = postersDir + '/' + name
149. if (exists(path)) { return new Response(readBytes(path), { headers = { 'Content-Type' = mimeOfPoster(name) 'Cache-Control' = 'public, max-age=86400' } }) }
150. return new Response(readBytes(placeholderPoster), { headers = { 'Content-Type' = 'image/svg+xml' 'Cache-Control' = 'public, max-age=3600' } })
151. }
153. // ---- /profiles/<name> (tracker.worldapi.org#16): a person's photo, from storage/mpackdb/profiles/ (people.hl syncProfile
154. // downloads it on the first visit of the person page) — the same checks and placeholder as /posters/<name>
155. profilesDir = (env('TRACKER_STORAGE') != null ? env('TRACKER_STORAGE') : './storage/mpackdb') + '/profiles'
156. profileRoute = (route, req) => {
157. let name = route.params != null ? route.params.name : null
158. if (!isSafePosterName(name)) { return new Response('bad photo name', { status = 400 headers = { 'Content-Type' = 'text/plain; charset=utf-8' } }) }
159. let path = profilesDir + '/' + name
160. if (exists(path)) { return new Response(readBytes(path), { headers = { 'Content-Type' = mimeOfPoster(name) 'Cache-Control' = 'public, max-age=86400' } }) }
161. return new Response(readBytes(placeholderPoster), { headers = { 'Content-Type' = 'image/svg+xml' 'Cache-Control' = 'public, max-age=3600' } })
162. }
164. // ---- the old addresses of the personal lists (tracker.worldapi.org#8): they all live under /my/ now —
165. // a permanent redirect so old links keep working (architect's decision)
166. movedTo = (path) => { return new Response('moved to ' + path, { status = 301 headers = { 'Location' = path 'Content-Type' = 'text/plain; charset=utf-8' } }) }
167. unwatchedMoved = (route, req) => { return movedTo('/my/unwatched') }
168. scheduleMoved = (route, req) => { return movedTo('/my/schedule') }
170. routes = [
171. { pattern = "/favicon.ico" file = "./icons/favicon.ico" headers = { 'Cache-Control' = 'no-cache' } }
172. // the installable app (tracker.worldapi.org#10): the icons (manifest and service worker are hl:web's own, from appIcons / offline)
173. { pattern = "/icons/icon-192.png" file = "./icons/icon-192.png" headers = { 'Cache-Control' = 'no-cache' } }
174. { pattern = "/icons/icon-512.png" file = "./icons/icon-512.png" headers = { 'Cache-Control' = 'no-cache' } }
175. { pattern = "/icons/apple-touch-icon.png" file = "./icons/apple-touch-icon.png" headers = { 'Cache-Control' = 'no-cache' } }
176. { pattern = "/icons/favicon.svg" file = "./icons/favicon.svg" headers = { 'Cache-Control' = 'no-cache' } }
177. { pattern = "/login/callback" function = loginCallback }
178. { pattern = "/login/failed" component = LoginFailed }
179. { pattern = "/login.js" file = "./login.js" headers = { 'Cache-Control' = 'no-cache' } }
180. { pattern = "/posters/:name" function = posterRoute }
181. // tracker.worldapi.org#13: every movie / every series, 24 per page — the page is a PATH segment (`/movies/page/2`), a
182. // page component cannot read `?page=` (hybriel#11) — and the movies the user follows
183. { pattern = "/movies" component = Movies }
184. { pattern = "/movies/page/:page" component = Movies }
185. { pattern = "/shows" component = AllShows }
186. { pattern = "/shows/page/:page" component = AllShows }
187. { pattern = "/shows/:slug" component = Show }
188. { pattern = "/my/shows" component = MyShows }
189. { pattern = "/my/movies" component = MyMovies }
190. { pattern = "/my/unwatched" component = Unwatched }
191. { pattern = "/my/schedule" component = Schedule }
192. { pattern = "/unwatched" function = unwatchedMoved }
193. { pattern = "/schedule" function = scheduleMoved }
194. // tracker.worldapi.org#14: the search — `/search/<text>` (hl:web gives a page no query string, so not `?q=`)
195. { pattern = "/search/:q?" component = Search }
196. // tracker.worldapi.org#16: a person's page (the show page's cast and the search link here) and their photo
197. { pattern = "/person/:slug" component = Person }
198. { pattern = "/profiles/:name" function = profileRoute }
199. { pattern = "/" component = Home }
200. ]
202. // WHO GETS THE PUSH: the login state reaches the tabs of one session (components/main.hl trackerSignedIn/Out).
203. tagOf = (session) => { return session != null && session.data != null ? session.data.tag : null }
204. audience = {
205. trackerSignedIn = (tag, session) => { return tag != null && tagOf(session) == tag }
206. trackerSignedOut = (tag, session) => { return tag != null && tagOf(session) == tag }
207. }
209. // cookies are per host, not per port: an own name keeps this app's session apart from
210. // ident's own (and from any other worldapi app sharing a dev host), see ident README "Design
211. // tokens" / hybriel#10 hybriel#17.
212. sessionCookie = 'trackersid'
213. // tracker.worldapi.org#14: the search index (search.hl) is built once here at boot (~0.6 s on the real data), not by the
214. // first search
215. searchIndexReady = ensureIndex()
216. // tracker#18, ONCE (deltasync.hl clearCopiedSummaries): a `summary` that is only the migrated copy of tmdbSummary is cleared — the
217. // current TMDB text shows; the creator's own texts stay
218. summariesCleared = clearCopiedSummaries()
219. if (summariesCleared.ran) { console.log('summaries: cleared ' + summariesCleared.cleared + ' copies of tmdbSummary, kept ' + summariesCleared.kept + ' own texts, errors ' + summariesCleared.errors + ', ' + summariesCleared.ms + ' ms') }
220. server = new WebFramework(routes = routes, styles = styles, port = port, sessionCookie = sessionCookie, watchMode = watching)
222. // ---- THE DAILY TMDB SYNC (tracker.worldapi.org#9, tmdbsync.hl; tracker#18, deltasync.hl) -------------------------
223. // Once a day at TRACKER_SYNC_HOUR (UTC, default 4 — hl:time has no time zones; the container runs UTC). hl:web serves ONE
224. // 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
225. // requests that queued meanwhile are served between two ticks, and a tick never starts inside a page render.
226. // tracker#18: the run first asks TMDB's and TVmaze's CHANGE LISTS since the last run (deltasync.hl — one list page per step)
227. // and then syncs only the titles of ours that changed, followed ones first — every title stays current, not only the followed
228. // ones. FULL WALK (every followed show, as before #18) on the first run, after a gap > 14 days, or when a change list fails.
229. // A title's step: tmdbsync.hl syncShow (1–3 TMDB requests + maybe its poster, TVmaze) + details.hl applySynced (the record's
230. // title, overview, tagline … and cast/crew from the same details answer). After a step that made n requests the next waits
231. // n × 260 ms (TMDB: ≤ 40 requests / 10 s) + 550 ms per TVmaze request (≤ 20 / 10 s, tracker.worldapi.org#12).
232. // The finished run stores its day (deltasync.hl markSynced). Log: "tmdb sync: start, …", the change lists ("tmdb sync:
233. // changes since …"), errors, slow steps, "tmdb sync done: mode=… changeRequests=… <totals>".
234. // TRACKER_SYNC=0 switches it off; without TMDB_READ_TOKEN it is off too (logged once).
235. syncHour = env('TRACKER_SYNC_HOUR') != null ? toNumber(env('TRACKER_SYNC_HOUR')) : 4
236. syncOn = env('TRACKER_SYNC') != '0' && syncEnabled
237. syncQueue = []
238. syncPlanning = false
239. syncMode = ''
240. syncMark = false
241. syncChangeRequests = 0
242. syncAt = 0
243. syncDay = ''
244. syncNextAt = 0
245. syncTotals = emptyTotals()
246. syncFinish = () => {
247. persistDetails()
248. if (syncMark) { markSynced(syncDay) }
249. console.log('tmdb sync done: mode=' + syncMode + ' changeRequests=' + syncChangeRequests + ' ' + totalsLine(syncTotals) + ' seconds=' + ((now() - syncAt) / 1000))
250. return null
251. }
252. syncTick = () => {
253. let t = now()
254. if (syncQueue.length == 0 && !syncPlanning) {
255. let stamp = timestamp(t)
256. let day = stamp.slice(0, 10)
257. if (toNumber(stamp.slice(11, 13)) != syncHour || day == syncDay) { return null }
258. syncDay = day
259. syncAt = t
260. syncNextAt = 0
261. syncTotals = emptyTotals()
262. syncChangeRequests = 0
263. let p = planStart(day, null)
264. syncMode = p.mode
265. syncMark = true
266. if (p.mode == 'delta') {
267. syncPlanning = true
268. console.log('tmdb sync: start, changes ' + p.why)
269. return null
270. }
271. syncQueue = allFollowedShowIds()
272. console.log('tmdb sync: start, ' + syncQueue.length + ' followed shows, full walk (' + p.why + ')')
273. if (syncQueue.length == 0) { syncFinish() }
274. return null
275. }
276. if (t < syncNextAt) { return null }
277. if (syncPlanning) {
278. let ps = planStep()
279. syncChangeRequests = syncChangeRequests + ps.requests + ps.tvmazeRequests
280. syncTotals.requests = syncTotals.requests + ps.requests
281. syncTotals.tvmazeRequests = syncTotals.tvmazeRequests + ps.tvmazeRequests
282. syncNextAt = now() + pauseMsAfter(ps)
283. if (!ps.done) { return null }
284. syncPlanning = false
285. console.log('tmdb sync: ' + planLine())
286. if (ps.failed != '') {
287. // a change list could not be read: the old full walk, and the day is not advanced (the next run asks again)
288. syncMode = 'full'
289. syncMark = false
290. syncQueue = allFollowedShowIds()
291. console.log('tmdb sync: change list failed, full walk of ' + syncQueue.length + ' followed shows instead')
292. } else {
293. syncQueue = planIds()
294. }
295. if (syncQueue.length == 0) { syncFinish() }
296. return null
297. }
298. let id = syncQueue[0]
299. let rest = []
300. let i = 1
301. while (i < syncQueue.length) { rest.push(syncQueue[i]) i = i + 1 }
302. syncQueue = rest
303. // tracker#18: in a delta run a title nobody follows gets the light step (changed seasons, no TVmaze)
304. let r = syncMode == 'delta' && !isFollowedInRun(id) ? syncShowDelta(id) : syncShow(id)
305. // tracker#18 + #28: the record's own fields (title, overview, tagline …) and the cast + crew from the same details answer
306. if (r.details != null) {
307. let c = applySynced(id, r.details)
308. if (c.error != '') { console.log('tmdb sync: ' + c.error) }
309. if (c.renamed) { searchTitleRenamed(id) }
310. }
311. // the public lists (catalog.hl, tracker.worldapi.org#13): this show's place by its newest released episode; the sync
312. // also brings TMDB's adult flag (mission 054) → the search's view of it too
313. refreshCatalogShow(id)
314. searchTitleChanged(id)
315. let took = now() - t
316. syncTotals = addTotals(syncTotals, r)
317. if (r.error != '') { console.log('tmdb sync: ' + r.error) }
318. // a step blocks every request meanwhile (Saturday Night Live, 53 seasons, 4 requests: ~1.7 s) — say so in the log
319. if (took > 1500) { console.log('tmdb sync: slow step, show ' + id + ', ' + r.requests + ' requests, ' + took + ' ms') }
320. syncNextAt = now() + pauseMsAfter(r)
321. if (syncQueue.length > 0 && syncTotals.shows % 10 == 0) { persistDetails() }
322. if (syncQueue.length == 0) { syncFinish() }
323. return null
324. }
326. // ---- THE ADULT-FLAG BACKFILL (mission 054, tmdbsync.hl backfillTitle) -----------------------------------
327. // The public lists and the search show only titles TMDB says are not adult (shows.hl isPublicTitle; unknown = hidden). At
328. // start the app collects every title whose flag is still unknown and asks TMDB for ONE title per step — like the daily
329. // sync, on the same clock, so requests queued meanwhile are served between two steps: one details request (adult,
330. // poster_path, external ids) + the poster file when missing, then a pause of 260 ms per TMDB API request (≤ 40 / 10 s;
331. // the poster comes from TMDB's image host, not the API, and is not counted). Each answered title goes into the lists
332. // (catalog.hl refreshCatalogShow) and the search at once. RESUMES after a restart: the queue is built from what is
333. // still unknown. Stops when the queue is empty; titles TMDB answers 404 for are marked (`adultCheck`) and stay hidden; a
334. // passing failure (no answer, 429, 5xx) goes to the back of the queue, at most 2 more tries per start.
335. // While the daily sync runs, the backfill waits. Log: start, every TRACKER_BACKFILL_LOG_EVERY (500) titles, done.
336. // TRACKER_BACKFILL=0 switches it off; without TMDB_READ_TOKEN it is off too.
337. backfillOn = env('TRACKER_BACKFILL') != '0' && syncEnabled
338. backfillLogEvery = env('TRACKER_BACKFILL_LOG_EVERY') != null ? toNumber(env('TRACKER_BACKFILL_LOG_EVERY')) : 500
339. backfillQueue = null
340. backfillPos = 0
341. backfillDone = false
342. backfillAt = 0
343. backfillNextAt = 0
344. backfillTries = {}
345. backfillTotals = { titles = 0 adult = 0 notAdult = 0 unknown = 0 posters = 0 ids = 0 requests = 0 errors = 0 retries = 0 }
346. backfillLine = () => {
347. let b = backfillTotals
348. return b.titles + '/' + backfillQueue.length + ' titles, adult=' + b.adult + ' notAdult=' + b.notAdult + ' unknown=' + b.unknown + ' posters=' + b.posters
349. + ' ids=' + b.ids + ' requests=' + b.requests + ' errors=' + b.errors + ' retries=' + b.retries + ' seconds=' + ((now() - backfillAt) / 1000)
350. }
351. backfillTick = () => {
352. if (backfillDone) { return null }
353. let t = now()
354. if (backfillQueue == null) {
355. backfillQueue = adultBackfillQueue()
356. backfillAt = t
357. console.log('adult backfill: start, ' + backfillQueue.length + ' titles without an adult flag (' + (now() - t) + ' ms to find them)')
358. if (backfillQueue.length == 0) { backfillDone = true console.log('adult backfill: nothing to do') }
359. return null
360. }
361. if (t < backfillNextAt) { return null }
362. let id = backfillQueue[backfillPos]
363. backfillPos = backfillPos + 1
364. let r = backfillTitle(id)
365. if (r.retry && (backfillTries[id] == null || backfillTries[id] < 2)) {
366. backfillTries[id] = backfillTries[id] == null ? 1 : backfillTries[id] + 1
367. let q = backfillQueue
368. q.push(id)
369. backfillQueue = q
370. backfillTotals.retries = backfillTotals.retries + 1
371. } else {
372. backfillTotals.titles = backfillTotals.titles + 1
373. 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 }
374. }
375. backfillTotals.posters = backfillTotals.posters + r.posters
376. backfillTotals.ids = backfillTotals.ids + r.ids
377. backfillTotals.requests = backfillTotals.requests + r.requests
378. backfillTotals.errors = backfillTotals.errors + r.errors
379. if (r.error != '') { console.log('adult backfill: ' + r.error) }
380. if (r.adult != null) {
381. refreshCatalogShow(id)
382. searchTitleChanged(id)
383. }
384. let took = now() - t
385. if (took > 1500) { console.log('adult backfill: slow step, show ' + id + ', ' + took + ' ms') }
386. backfillNextAt = now() + r.requests * msPerRequest
387. let finished = backfillPos >= backfillQueue.length
388. if (finished || backfillPos % 25 == 0) { persistSync() }
389. if (!finished && backfillTotals.titles > 0 && backfillTotals.titles % backfillLogEvery == 0 && !r.retry) { console.log('adult backfill: ' + backfillLine()) }
390. if (finished) {
391. backfillDone = true
392. console.log('adult backfill done: ' + backfillLine())
393. }
394. return null
395. }
397. // ---- THE DETAILS REPAIR (tracker.worldapi.org#26, details.hl completeStep) ----------------------------------------
398. // Titles the person pages added from a filmography (#16) are MINIMAL (no poster, seasons, ids, cast …). Opening one completes
399. // it (components/show.hl); this job completes the rest in the background: at start it collects every incomplete, non-adult
400. // title and does ONE per step on the app's clock, like the adult backfill — a step is the daily sync's work for one title
401. // (1–3 TMDB requests + poster + TVmaze), requests queued meanwhile are served between two steps; then a pause of 260 ms per
402. // TMDB request + 550 ms per TVmaze request (pauseMsAfter: ≤ 40 / 10 s, ≤ 20 / 10 s). RESUMES after a restart (the queue is
403. // what is still incomplete); TMDB 404 marks the title (`detailsCheck`), a passing failure goes to the back of the queue (at
404. // most 2 more tries per start). Waits while the daily sync or the adult backfill runs. Persisted every 10 titles + at the end.
405. // Log: start, every TRACKER_REPAIR_LOG_EVERY (100) titles, each failure, done. TRACKER_REPAIR=0 switches it off (off too
406. // without TMDB_READ_TOKEN). TRACKER_REPAIR_FAST=1 (the gate, fake TMDB only): 20 ms pause instead.
407. repairOn = env('TRACKER_REPAIR') != '0' && syncEnabled
408. repairFast = env('TRACKER_REPAIR_FAST') == '1'
409. repairLogEvery = env('TRACKER_REPAIR_LOG_EVERY') != null ? toNumber(env('TRACKER_REPAIR_LOG_EVERY')) : 100
410. repairQueue = null
411. repairPos = 0
412. repairDone = false
413. repairAt = 0
414. repairNextAt = 0
415. repairTries = {}
416. repairTotals = { titles = 0 complete = 0 failed = 0 adult = 0 newSeasons = 0 newEpisodes = 0 posters = 0 cast = 0 requests = 0 tvmazeRequests = 0 retries = 0 }
417. repairLine = () => {
418. let b = repairTotals
419. return b.titles + '/' + repairQueue.length + ' titles, complete=' + b.complete + ' failed=' + b.failed + ' adult=' + b.adult + ' newSeasons=' + b.newSeasons + ' newEpisodes=' + b.newEpisodes
420. + ' posters=' + b.posters + ' cast=' + b.cast + ' requests=' + b.requests + ' tvmazeRequests=' + b.tvmazeRequests + ' retries=' + b.retries + ' seconds=' + ((now() - repairAt) / 1000)
421. }
422. repairTick = () => {
423. if (repairDone) { return null }
424. let t = now()
425. if (repairQueue == null) {
426. repairQueue = detailsRepairQueue()
427. repairAt = t
428. console.log('details repair: start, ' + repairQueue.length + ' incomplete titles (' + (now() - t) + ' ms to find them)')
429. if (repairQueue.length == 0) { repairDone = true console.log('details repair: nothing to do') }
430. return null
431. }
432. if (t < repairNextAt) { return null }
433. let id = repairQueue[repairPos]
434. let r = completeStep(id)
435. let took = now() - t
436. if (took > 1500) { console.log('details repair: slow step, show ' + id + ', ' + took + ' ms') }
437. if (r.error != '') { console.log('details repair: ' + r.error) }
438. repairNextAt = now() + (repairFast ? 20 : pauseMsAfter(r))
439. // a series takes several steps: the same title again next time
440. if (!r.done && !r.retry) { return null }
441. repairPos = repairPos + 1
442. if (r.retry && (repairTries[id] == null || repairTries[id] < 2)) {
443. repairTries[id] = repairTries[id] == null ? 1 : repairTries[id] + 1
444. let q = repairQueue
445. q.push(id)
446. repairQueue = q
447. repairTotals.retries = repairTotals.retries + 1
448. } else if (!r.skipped) {
449. // done: its counts (a failed first step: the request alone)
450. let sum = r.sum
451. repairTotals.titles = repairTotals.titles + 1
452. 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 }
453. repairTotals.newSeasons = repairTotals.newSeasons + sum.newSeasons
454. repairTotals.newEpisodes = repairTotals.newEpisodes + sum.newEpisodes
455. repairTotals.posters = repairTotals.posters + sum.posters
456. repairTotals.cast = repairTotals.cast + sum.cast
457. repairTotals.requests = repairTotals.requests + (sum.steps == 0 ? r.requests : sum.requests)
458. repairTotals.tvmazeRequests = repairTotals.tvmazeRequests + sum.tvmazeRequests
459. }
460. let finished = repairPos >= repairQueue.length
461. if (finished || repairPos % 10 == 0) { persistDetails() }
462. if (!finished && !r.skipped && !r.retry && repairTotals.titles > 0 && repairTotals.titles % repairLogEvery == 0) { console.log('details repair: ' + repairLine()) }
463. if (finished) {
464. repairDone = true
465. console.log('details repair done: ' + repairLine())
466. }
467. return null
468. }
470. // ---- THE CREDITS JOB (tracker.worldapi.org#28, details.hl creditsStep) ------------------------------------------------------
471. // Every PUBLIC title (adult == false; adult/unknown are skipped) with a TMDB id gets its whole cast + crew once: at start (after
472. // the adult backfill and the details repair are done) the job collects every such title without `creditsAt` and asks ONE per
473. // step on the app's clock — one request (details + credits / aggregate_credits), then 260 ms (≤ 40 / 10 s); requests queued
474. // meanwhile are served between two steps. Titles the details repair, the daily sync or an import already did carry `creditsAt`
475. // and are not asked. RESUMES after a restart (the queue = what still has no credits); TMDB 404 / adult → `creditsCheck`, never
476. // asked again; a passing failure goes to the back of the queue (at most 2 more tries per start). Waits while the daily sync
477. // runs. Persisted (shows + people only) every 25 titles + at the end. Log: start, every TRACKER_CREDITS_LOG_EVERY (500) titles, failures, slow steps
478. // (> 1.5 s), done. TRACKER_CREDITS=0 switches it off (off too without TMDB_READ_TOKEN). TRACKER_CREDITS_FAST=1 (the gate, fake
479. // TMDB only): 20 ms pause instead.
480. // MEMORY LIMIT: the runtime never frees what a fetch + JSON parse took (mission 061 measured ~6 MB more RSS per title on the real
481. // 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
482. // step the job reads its own RSS (/proc/self/status); above TRACKER_CREDITS_MAX_MB (16000) it PAUSES for this process (logged) and
483. // goes on with what is left after the next restart (deploy, `docker restart`).
484. creditsOn = env('TRACKER_CREDITS') != '0' && syncEnabled
485. creditsFast = env('TRACKER_CREDITS_FAST') == '1'
486. creditsLogEvery = env('TRACKER_CREDITS_LOG_EVERY') != null ? toNumber(env('TRACKER_CREDITS_LOG_EVERY')) : 500
487. creditsQueueIds = null
488. creditsPos = 0
489. creditsDone = false
490. creditsAt = 0
491. creditsNextAt = 0
492. creditsTries = {}
493. creditsMaxMb = env('TRACKER_CREDITS_MAX_MB') != null ? toNumber(env('TRACKER_CREDITS_MAX_MB')) : 16000
494. creditsTotals = { titles = 0 done = 0 failed = 0 cast = 0 crew = 0 people = 0 requests = 0 retries = 0 slowest = 0 }
495. // this process' resident memory in MB (VmRSS), 0 when unknown
496. rssMb = () => {
497. if (!exists('/proc/self/status')) { return 0 }
498. let t = readFile('/proc/self/status')
499. let i = t != null ? t.indexOf('VmRSS:') : -1
500. if (i < 0) { return 0 }
501. let kb = toNumber(t.slice(i + 6, t.indexOf('kB', i)).trim())
502. return kb != null ? (kb - kb % 1024) / 1024 : 0
503. }
504. creditsLine = () => {
505. let b = creditsTotals
506. return b.titles + '/' + creditsQueueIds.length + ' titles, done=' + b.done + ' failed=' + b.failed + ' cast=' + b.cast + ' crew=' + b.crew + ' people=' + b.people
507. + ' requests=' + b.requests + ' retries=' + b.retries + ' slowestStepMs=' + b.slowest + ' seconds=' + ((now() - creditsAt) / 1000)
508. }
509. creditsTick = () => {
510. if (creditsDone) { return null }
511. let t = now()
512. if (creditsQueueIds == null) {
513. creditsQueueIds = creditsQueue()
514. creditsAt = t
515. console.log('credits job: start, ' + creditsQueueIds.length + ' titles without credits (' + (now() - t) + ' ms to find them)')
516. if (creditsQueueIds.length == 0) { creditsDone = true console.log('credits job: nothing to do') }
517. return null
518. }
519. if (t < creditsNextAt) { return null }
520. let mb = rssMb()
521. if (mb > creditsMaxMb) {
522. creditsDone = true
523. persistCredits()
524. console.log('credits job: paused, RSS ' + mb + ' MB > TRACKER_CREDITS_MAX_MB ' + creditsMaxMb + ' — ' + (creditsQueueIds.length - creditsPos) + ' titles left for the next start; ' + creditsLine())
525. return null
526. }
527. let id = creditsQueueIds[creditsPos]
528. creditsPos = creditsPos + 1
529. let r = creditsStep(id)
530. let took = now() - t
531. if (took > creditsTotals.slowest) { creditsTotals.slowest = took }
532. if (took > 1500) { console.log('credits job: slow step, show ' + id + ', ' + took + ' ms, cast=' + r.cast + ' crew=' + r.crew + ' people=' + r.people) }
533. if (r.error != '') { console.log('credits job: ' + r.error) }
534. creditsNextAt = now() + (creditsFast ? 20 : r.requests * msPerRequest)
535. if (r.retry && (creditsTries[id] == null || creditsTries[id] < 2)) {
536. creditsTries[id] = creditsTries[id] == null ? 1 : creditsTries[id] + 1
537. let q = creditsQueueIds
538. q.push(id)
539. creditsQueueIds = q
540. creditsTotals.retries = creditsTotals.retries + 1
541. } else if (!r.skipped) {
542. creditsTotals.titles = creditsTotals.titles + 1
543. if (r.error != '') { creditsTotals.failed = creditsTotals.failed + 1 } else { creditsTotals.done = creditsTotals.done + 1 }
544. creditsTotals.cast = creditsTotals.cast + r.cast
545. creditsTotals.crew = creditsTotals.crew + r.crew
546. creditsTotals.people = creditsTotals.people + r.people
547. }
548. creditsTotals.requests = creditsTotals.requests + r.requests
549. let finished = creditsPos >= creditsQueueIds.length
550. if (finished || creditsPos % 25 == 0) { persistCredits() }
551. if (!finished && !r.skipped && !r.retry && creditsTotals.titles > 0 && creditsTotals.titles % creditsLogEvery == 0) { console.log('credits job: ' + creditsLine()) }
552. if (finished) {
553. creditsDone = true
554. console.log('credits job done: ' + creditsLine())
555. }
556. return null
557. }
559. // ONE clock for all four (a tick every 0.1 s): the daily sync's step once a second as before; the backfill's steps in
560. // between, only while no daily run is going on; the details repair only when neither has work; the credits job (#28) last
561. clockOn = syncOn || backfillOn || repairOn || creditsOn
562. clockTicks = 0
563. 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') + ')') }
564. if (!backfillOn) { console.log('adult backfill: off (' + (syncEnabled ? 'TRACKER_BACKFILL=0' : 'no TMDB_READ_TOKEN') + ')') }
565. if (!repairOn) { console.log('details repair: off (' + (syncEnabled ? 'TRACKER_REPAIR=0' : 'no TMDB_READ_TOKEN') + ')') }
566. if (!creditsOn) { console.log('credits job: off (' + (syncEnabled ? 'TRACKER_CREDITS=0' : 'no TMDB_READ_TOKEN') + ')') }
567. if (clockOn) {
568. syncClock = every(0.1)
569. on syncClock.tick(x) {
570. clockTicks = clockTicks + 1
571. let busy = syncQueue.length > 0 || syncPlanning
572. if (syncOn && clockTicks % 10 == 0) { syncTick() }
573. let free = !busy && syncQueue.length == 0 && !syncPlanning
574. if (backfillOn && free && !backfillDone) { backfillTick() }
575. else if (repairOn && free && (!backfillOn || backfillDone) && !repairDone) { repairTick() }
576. else if (creditsOn && free && (!backfillOn || backfillDone) && (!repairOn || repairDone)) { creditsTick() }
577. return null
578. }
579. }
581. on Error(e) { console.log('error absorbed: ' + e.message) }