tracker

35.5 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 Genre from './components/genre.hl'
38. import { ensureIndex, searchTitleChanged } from './search.hl'
39. import { syncEnabled, syncShow, persistSync, emptyTotals, addTotals, totalsLine, pauseMsAfter, msPerRequest, adultBackfillQueue, backfillTitle } from './tmdbsync.hl'
40. import { detailsRepairQueue, completeStep, completeLine, persistDetails, persistCredits, applyCredits, creditsQueue, creditsStep } from './details.hl'
41. import { allFollowedShowIds } from './follows.hl'
42. import { refreshCatalogShow } from './catalog.hl'
43. import { ensureShortIds, shortIdStep, shortIdPath, shortIdsLeft, shortIdStats, persistShortIds } from './shortids.hl'
44. import { guestPersonId } from './people.hl'
45. import { personById } from './shows.hl'
46. import { dark, darker } from './shared/tokens.hl'
48. static appTitle = "tracker"
50. // ---- THE INSTALLABLE APP (tracker.worldapi.org#10), the same way calendar.worldapi.org does it: hl:web's own web app
51. // manifest (/__hl/manifest.webmanifest, linked from every head with the apple-touch-icon and theme-color) and service
52. // worker (/__hl/sw.js) from these settings — no JavaScript of ours. Icons in icons/ (icon.svg is the source, the PNGs
53. // are rendered from it with rsvg-convert, README "Icons"). The theme colour is the header's background (darker), the
54. // splash background the page's (dark) — both off the tokens, not written out again.
55. appThemeColor = darker.value
56. appBackgroundColor = dark.value
57. appIcons = [
58. { src = '/icons/icon-192.png' sizes = '192x192' purpose = 'any' }
59. { src = '/icons/icon-512.png' sizes = '512x512' purpose = 'any' }
60. { src = '/icons/icon-192.png' sizes = '192x192' purpose = 'maskable' }
61. { src = '/icons/icon-512.png' sizes = '512x512' purpose = 'maskable' }
62. ]
63. appTouchIcon = '/icons/apple-touch-icon.png'
64. appFavicon = '/icons/favicon.svg'
65. // OFFLINE: only the shell — the header and `/` (components/main.hl says "You are offline" while the browser has no
66. // network). The data pages are not kept: without a network they get hl:web's "Unavailable offline" page.
67. offline = [ Home ]
69. styles = Styles
71. port = env('TRACKER_PORT') != null ? toNumber(env('TRACKER_PORT')) : 45008
72. watching = env('TRACKER_WATCH') != '0'
73. // read by hl:web's own manifest config (WebFramework.hl `cfg.sessionDir`), not the constructor call below
74. sessionDir = env('TRACKER_SESSIONS') != null ? env('TRACKER_SESSIONS') : null
76. // ---- THE LOGIN BUTTON'S RETURN (ident README "How apps use ident"), copied from calendar.worldapi.org's project.hl
77. // BACK TO THE PAGE: /login.js puts `?next=` into the button's return URL at the click. Only a same-origin PATH
78. // goes (one `/`, URL-safe characters, ≤ 500). Anything else → `/`.
79. nextChars = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789-._~/?&=%+,;@!$()*:'
80. safePath = (want) => {
81. if (want == null || hlTypeName(want) != 'String' || want == '' || want.length > 500) { return '/' }
82. if (want.slice(0, 1) != '/' || want.slice(0, 2) == '//' || want.slice(0, 7) == '/login/') { return '/' }
83. let i = 0
84. while (i < want.length) {
85. if (!nextChars.includes(want[i])) { return '/' }
86. i = i + 1
87. }
88. return want
89. }
91. // A FAILED LOGIN is a page (components/loginfailed.hl): the reason is parked in the session, then → /login/failed
92. failed = (req, why) => {
93. let s = req.session
94. let fresh = s == null
95. if (fresh) { s = server.sessions.mint() }
96. s.data.loginError = why
97. server.sessions.save(s)
98. let res = new Response('login failed: ' + why, { status = 302 headers = { 'Location' = '/login/failed' 'Cache-Control' = 'no-store' 'Content-Type' = 'text/plain; charset=utf-8' } })
99. if (fresh) { res.headers['Set-Cookie'] = server.sessions.cookieHeader(s.id) }
100. return res
101. }
103. // the function route gets the cookie's session as req.session (hybriel #11); none yet → minted here
104. loginCallback = (route, req) => {
105. if (req.method != 'GET') { return failed(req, 'GET only') }
106. let q = req.query != null ? req.query : {}
107. let code = q.ident_code
108. if (code == null || code == '') { return failed(req, 'ident sent no login code') }
109. let x = exchangeCode(code)
110. if (x.error != null) { return failed(req, x.error) }
111. let u = ensureUser(x.identity)
112. if (u == null) { return failed(req, 'could not store the user') }
113. let s = req.session
114. let fresh = s == null
115. if (fresh) { s = server.sessions.mint() }
116. s.user = { id = u.id }
117. s.data.tag = randomBytes(16)
118. s.data.loginError = null
119. server.sessions.save(s)
120. let res = new Response('logged in', { status = 302 headers = { 'Location' = safePath(q.next) 'Cache-Control' = 'no-store' 'Content-Type' = 'text/plain; charset=utf-8' } })
121. if (fresh) { res.headers['Set-Cookie'] = server.sessions.cookieHeader(s.id) }
122. return res
123. }
125. // ---- /posters/<name> (tracker.worldapi.org#4): a show's poster image, served dynamically from
126. // storage/mpackdb/posters/ (the poster's file name is the show's oldId + its migrated extension,
127. // shows.hl posterName) — falls back to a placeholder svg when the real file isn't there yet (the
128. // mongo export never included the actual poster bytes, see the report's `open`).
129. postersDir = (env('TRACKER_STORAGE') != null ? env('TRACKER_STORAGE') : './storage/mpackdb') + '/posters'
130. placeholderPoster = './assets/poster-placeholder.svg'
131. posterNameChars = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789.-_'
132. isSafePosterName = (name) => {
133. if (name == null || hlTypeName(name) != 'String' || name == '' || name.length > 100) { return false }
134. let i = 0
135. while (i < name.length) {
136. if (!posterNameChars.includes(name[i])) { return false }
137. i = i + 1
138. }
139. return true
140. }
141. endsWithStr = (s, suffix) => { return s.length >= suffix.length && s.slice(s.length - suffix.length) == suffix }
142. mimeOfPoster = (name) => {
143. if (endsWithStr(name, '.png')) { return 'image/png' }
144. if (endsWithStr(name, '.webp')) { return 'image/webp' }
145. if (endsWithStr(name, '.svg')) { return 'image/svg+xml' }
146. return 'image/jpeg'
147. }
148. posterRoute = (route, req) => {
149. let name = route.params != null ? route.params.name : null
150. if (!isSafePosterName(name)) { return new Response('bad poster name', { status = 400 headers = { 'Content-Type' = 'text/plain; charset=utf-8' } }) }
151. let path = postersDir + '/' + name
152. if (exists(path)) { return new Response(readBytes(path), { headers = { 'Content-Type' = mimeOfPoster(name) 'Cache-Control' = 'public, max-age=86400' } }) }
153. return new Response(readBytes(placeholderPoster), { headers = { 'Content-Type' = 'image/svg+xml' 'Cache-Control' = 'public, max-age=3600' } })
154. }
156. // ---- /profiles/<name> (tracker.worldapi.org#16): a person's photo, from storage/mpackdb/profiles/ (people.hl syncProfile
157. // downloads it on the first visit of the person page) — the same checks and placeholder as /posters/<name>
158. profilesDir = (env('TRACKER_STORAGE') != null ? env('TRACKER_STORAGE') : './storage/mpackdb') + '/profiles'
159. profileRoute = (route, req) => {
160. let name = route.params != null ? route.params.name : null
161. if (!isSafePosterName(name)) { return new Response('bad photo name', { status = 400 headers = { 'Content-Type' = 'text/plain; charset=utf-8' } }) }
162. let path = profilesDir + '/' + name
163. if (exists(path)) { return new Response(readBytes(path), { headers = { 'Content-Type' = mimeOfPoster(name) 'Cache-Control' = 'public, max-age=86400' } }) }
164. return new Response(readBytes(placeholderPoster), { headers = { 'Content-Type' = 'image/svg+xml' 'Cache-Control' = 'public, max-age=3600' } })
165. }
167. // ---- the old addresses of the personal lists (tracker.worldapi.org#8): they all live under /my/ now —
168. // a permanent redirect so old links keep working (architect's decision)
169. movedTo = (path) => { return new Response('moved to ' + path, { status = 301 headers = { 'Location' = path 'Content-Type' = 'text/plain; charset=utf-8' } }) }
170. unwatchedMoved = (route, req) => { return movedTo('/my/unwatched') }
171. scheduleMoved = (route, req) => { return movedTo('/my/schedule') }
172. // mission 058: the show page's genre pills linked `/genre/<x>` (no page there) — the genre pages are `/genres/<x>`
173. genreMoved = (route, req) => {
174. let g = route.params != null ? route.params.genre : null
175. return movedTo(isSafePosterName(g) ? '/genres/' + g : '/')
176. }
177. // mission 060 (tracker.worldapi.org#27): `/<shortId>` → 301 to the title's or person's page (shortids.hl); upper case is
178. // read as lower case; anything else is a 404
179. shortIdRoute = (route, req) => {
180. let sid = route.params != null ? route.params.sid : null
181. let path = sid != null && hlTypeName(sid) == 'String' ? shortIdPath(sid.toLowerCase()) : null
182. if (path == null) { return new Response('not found', { status = 404 headers = { 'Content-Type' = 'text/plain; charset=utf-8' } }) }
183. return movedTo(path)
184. }
186. // mission 062: a GUEST STAR's link (components/show.hl castRowsOf) — guests are no person records until someone opens one:
187. // `/person/tmdb/<tmdbId>?show=<show id>` creates the person (people.hl guestPersonId: name + role from that title, a short id) or
188. // finds the one we have, then 302 → `/person/<slug>` (the page fills from TMDB as for any new person). A function route: the
189. // browser loads it itself (hl:web's client router hands every non-component link to the browser)
190. guestRoute = (route, req) => {
191. let t = route.params != null ? route.params.tmdb : null
192. let tmdbId = t != null && hlTypeName(t) == 'String' && t.length <= 10 ? toNumber(t) : null
193. if (tmdbId == null || tmdbId < 1) { return new Response('not found', { status = 404 headers = { 'Content-Type' = 'text/plain; charset=utf-8' } }) }
194. let q = req.query != null ? req.query : {}
195. let r = guestPersonId(tmdbId, q.show)
196. let p = r.id != null ? personById(r.id) : null
197. if (p == null || p.urlSegment == null) { return new Response('not found: ' + (r.error != null ? r.error : 'no such person'), { status = 404 headers = { 'Content-Type' = 'text/plain; charset=utf-8' } }) }
198. return new Response('see /person/' + p.urlSegment, { status = 302 headers = { 'Location' = '/person/' + p.urlSegment 'Cache-Control' = 'no-store' 'Content-Type' = 'text/plain; charset=utf-8' } })
199. }
201. routes = [
202. { pattern = "/favicon.ico" file = "./icons/favicon.ico" headers = { 'Cache-Control' = 'no-cache' } }
203. // the installable app (tracker.worldapi.org#10): the icons (manifest and service worker are hl:web's own, from appIcons / offline)
204. { pattern = "/icons/icon-192.png" file = "./icons/icon-192.png" headers = { 'Cache-Control' = 'no-cache' } }
205. { pattern = "/icons/icon-512.png" file = "./icons/icon-512.png" headers = { 'Cache-Control' = 'no-cache' } }
206. { pattern = "/icons/apple-touch-icon.png" file = "./icons/apple-touch-icon.png" headers = { 'Cache-Control' = 'no-cache' } }
207. { pattern = "/icons/favicon.svg" file = "./icons/favicon.svg" headers = { 'Cache-Control' = 'no-cache' } }
208. { pattern = "/login/callback" function = loginCallback }
209. { pattern = "/login/failed" component = LoginFailed }
210. { pattern = "/login.js" file = "./login.js" headers = { 'Cache-Control' = 'no-cache' } }
211. { pattern = "/posters/:name" function = posterRoute }
212. // tracker.worldapi.org#13: every movie / every series, 24 per page — the page is a PATH segment (`/movies/page/2`), a
213. // page component cannot read `?page=` (hybriel#11) — and the movies the user follows
214. { pattern = "/movies" component = Movies }
215. { pattern = "/movies/page/:page" component = Movies }
216. { pattern = "/shows" component = AllShows }
217. { pattern = "/shows/page/:page" component = AllShows }
218. { pattern = "/shows/:slug" component = Show }
219. { pattern = "/my/shows" component = MyShows }
220. { pattern = "/my/movies" component = MyMovies }
221. { pattern = "/my/unwatched" component = Unwatched }
222. { pattern = "/my/schedule" component = Schedule }
223. { pattern = "/unwatched" function = unwatchedMoved }
224. { pattern = "/schedule" function = scheduleMoved }
225. // tracker.worldapi.org#14: the search — `/search/<text>` (hl:web gives a page no query string, so not `?q=`)
226. { pattern = "/search/:q?" component = Search }
227. // tracker.worldapi.org#16: a person's page (the show page's cast and the search link here) and their photo
228. { pattern = "/person/tmdb/:tmdb" function = guestRoute }
229. { pattern = "/person/:slug" component = Person }
230. { pattern = "/profiles/:name" function = profileRoute }
231. // mission 058 (tracker.worldapi.org#25): a genre's movies and series, like /movies (the page in the path)
232. { pattern = "/genres/:genre" component = Genre }
233. { pattern = "/genres/:genre/page/:page" component = Genre }
234. { pattern = "/genre/:genre" function = genreMoved }
235. { pattern = "/" component = Home }
236. // mission 060 (tracker.worldapi.org#27): a short id → its page (LAST: every named one-segment route above wins)
237. { pattern = "/:sid" function = shortIdRoute }
238. ]
240. // WHO GETS THE PUSH: the login state reaches the tabs of one session (components/main.hl trackerSignedIn/Out).
241. tagOf = (session) => { return session != null && session.data != null ? session.data.tag : null }
242. audience = {
243. trackerSignedIn = (tag, session) => { return tag != null && tagOf(session) == tag }
244. trackerSignedOut = (tag, session) => { return tag != null && tagOf(session) == tag }
245. }
247. // cookies are per host, not per port: an own name keeps this app's session apart from
248. // ident's own (and from any other worldapi app sharing a dev host), see ident README "Design
249. // tokens" / hybriel#10 hybriel#17.
250. sessionCookie = 'trackersid'
251. // 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
252. // first search
253. searchIndexReady = ensureIndex()
254. // mission 060: every short id in use (one map over shows AND persons) and the records still without one, before the first request
255. shortIdsReady = ensureShortIds()
256. server =new WebFramework(routes = routes, styles = styles, port = port, sessionCookie = sessionCookie, watchMode = watching)
258. // ---- THE DAILY TMDB SYNC (tracker.worldapi.org#9, tmdbsync.hl) --------------------------------------
259. // Once a day at TRACKER_SYNC_HOUR (UTC, default 4 — hl:time has no time zones; the container runs UTC) every
260. // followed show is synced from TMDB. hl:web serves ONE request at a time, so the run never takes the server
261. // for itself: a 1-second clock does ONE show per tick (1–3 TMDB requests + maybe its poster) and returns — the
262. // requests that queued meanwhile are served between two ticks, and a tick never starts inside a page render.
263. // After a show that made n requests the next waits n × 260 ms (TMDB: ≤ 40 requests / 10 s) + 550 ms per TVmaze lookup
264. // (≤ 20 / 10 s, tracker.worldapi.org#12).
265. // TRACKER_SYNC=0 switches it off; without TMDB_READ_TOKEN it is off too (logged once).
266. syncHour = env('TRACKER_SYNC_HOUR') != null ? toNumber(env('TRACKER_SYNC_HOUR')) : 4
267. syncOn = env('TRACKER_SYNC') != '0' && syncEnabled
268. syncQueue = []
269. syncAt = 0
270. syncDay = ''
271. syncNextAt = 0
272. syncTotals = emptyTotals()
273. syncTick = () => {
274. let t = now()
275. if (syncQueue.length == 0) {
276. let stamp = timestamp(t)
277. let day = stamp.slice(0, 10)
278. if (toNumber(stamp.slice(11, 13)) != syncHour || day == syncDay) { return null }
279. syncDay = day
280. syncQueue = allFollowedShowIds()
281. syncAt = t
282. syncNextAt = 0
283. syncTotals = emptyTotals()
284. console.log('tmdb sync: start, ' + syncQueue.length + ' followed shows')
285. return null
286. }
287. if (t < syncNextAt) { return null }
288. let id = syncQueue[0]
289. let rest = []
290. let i = 1
291. while (i < syncQueue.length) { rest.push(syncQueue[i]) i = i + 1 }
292. syncQueue = rest
293. let r = syncShow(id)
294. // tracker#28: the cast + crew came with the details request (details.hl applyCredits)
295. if (r.details != null) {
296. let c = applyCredits(id, r.details)
297. if (c.error != '') { console.log('tmdb sync: ' + c.error) }
298. }
299. // the public lists (catalog.hl, tracker.worldapi.org#13): this show's place by its newest released episode; the sync
300. // also brings TMDB's adult flag (mission 054) → the search's view of it too
301. refreshCatalogShow(id)
302. searchTitleChanged(id)
303. let took = now() - t
304. syncTotals = addTotals(syncTotals, r)
305. if (r.error != '') { console.log('tmdb sync: ' + r.error) }
306. // a step blocks every request meanwhile (Saturday Night Live, 53 seasons, 4 requests: ~1.7 s) — say so in the log
307. if (took > 1500) { console.log('tmdb sync: slow step, show ' + id + ', ' + r.requests + ' requests, ' + took + ' ms') }
308. syncNextAt = now() + pauseMsAfter(r)
309. if (syncQueue.length == 0 || syncTotals.shows % 10 == 0) { persistDetails() }
310. if (syncQueue.length == 0) { console.log('tmdb sync done: ' + totalsLine(syncTotals) + ' seconds=' + ((now() - syncAt) / 1000)) }
311. return null
312. }
314. // ---- THE ADULT-FLAG BACKFILL (mission 054, tmdbsync.hl backfillTitle) -----------------------------------
315. // The public lists and the search show only titles TMDB says are not adult (shows.hl isPublicTitle; unknown = hidden). At
316. // start the app collects every title whose flag is still unknown and asks TMDB for ONE title per step — like the daily
317. // sync, on the same clock, so requests queued meanwhile are served between two steps: one details request (adult,
318. // poster_path, external ids) + the poster file when missing, then a pause of 260 ms per TMDB API request (≤ 40 / 10 s;
319. // the poster comes from TMDB's image host, not the API, and is not counted). Each answered title goes into the lists
320. // (catalog.hl refreshCatalogShow) and the search at once. RESUMES after a restart: the queue is built from what is
321. // still unknown. Stops when the queue is empty; titles TMDB answers 404 for are marked (`adultCheck`) and stay hidden; a
322. // passing failure (no answer, 429, 5xx) goes to the back of the queue, at most 2 more tries per start.
323. // While the daily sync runs, the backfill waits. Log: start, every TRACKER_BACKFILL_LOG_EVERY (500) titles, done.
324. // TRACKER_BACKFILL=0 switches it off; without TMDB_READ_TOKEN it is off too.
325. backfillOn = env('TRACKER_BACKFILL') != '0' && syncEnabled
326. backfillLogEvery = env('TRACKER_BACKFILL_LOG_EVERY') != null ? toNumber(env('TRACKER_BACKFILL_LOG_EVERY')) : 500
327. backfillQueue = null
328. backfillPos = 0
329. backfillDone = false
330. backfillAt = 0
331. backfillNextAt = 0
332. backfillTries = {}
333. backfillTotals = { titles = 0 adult = 0 notAdult = 0 unknown = 0 posters = 0 ids = 0 requests = 0 errors = 0 retries = 0 }
334. backfillLine = () => {
335. let b = backfillTotals
336. return b.titles + '/' + backfillQueue.length + ' titles, adult=' + b.adult + ' notAdult=' + b.notAdult + ' unknown=' + b.unknown + ' posters=' + b.posters
337. + ' ids=' + b.ids + ' requests=' + b.requests + ' errors=' + b.errors + ' retries=' + b.retries + ' seconds=' + ((now() - backfillAt) / 1000)
338. }
339. backfillTick = () => {
340. if (backfillDone) { return null }
341. let t = now()
342. if (backfillQueue == null) {
343. backfillQueue = adultBackfillQueue()
344. backfillAt = t
345. console.log('adult backfill: start, ' + backfillQueue.length + ' titles without an adult flag (' + (now() - t) + ' ms to find them)')
346. if (backfillQueue.length == 0) { backfillDone = true console.log('adult backfill: nothing to do') }
347. return null
348. }
349. if (t < backfillNextAt) { return null }
350. let id = backfillQueue[backfillPos]
351. backfillPos = backfillPos + 1
352. let r = backfillTitle(id)
353. if (r.retry && (backfillTries[id] == null || backfillTries[id] < 2)) {
354. backfillTries[id] = backfillTries[id] == null ? 1 : backfillTries[id] + 1
355. let q = backfillQueue
356. q.push(id)
357. backfillQueue = q
358. backfillTotals.retries = backfillTotals.retries + 1
359. } else {
360. backfillTotals.titles = backfillTotals.titles + 1
361. 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 }
362. }
363. backfillTotals.posters = backfillTotals.posters + r.posters
364. backfillTotals.ids = backfillTotals.ids + r.ids
365. backfillTotals.requests = backfillTotals.requests + r.requests
366. backfillTotals.errors = backfillTotals.errors + r.errors
367. if (r.error != '') { console.log('adult backfill: ' + r.error) }
368. if (r.adult != null) {
369. refreshCatalogShow(id)
370. searchTitleChanged(id)
371. }
372. let took = now() - t
373. if (took > 1500) { console.log('adult backfill: slow step, show ' + id + ', ' + took + ' ms') }
374. backfillNextAt = now() + r.requests * msPerRequest
375. let finished = backfillPos >= backfillQueue.length
376. if (finished || backfillPos % 25 == 0) { persistSync() }
377. if (!finished && backfillTotals.titles > 0 && backfillTotals.titles % backfillLogEvery == 0 && !r.retry) { console.log('adult backfill: ' + backfillLine()) }
378. if (finished) {
379. backfillDone = true
380. console.log('adult backfill done: ' + backfillLine())
381. }
382. return null
383. }
385. // ---- THE DETAILS REPAIR (tracker.worldapi.org#26, details.hl completeStep) ----------------------------------------
386. // Titles the person pages added from a filmography (#16) are MINIMAL (no poster, seasons, ids, cast …). Opening one completes
387. // it (components/show.hl); this job completes the rest in the background: at start it collects every incomplete, non-adult
388. // 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
389. // (1–3 TMDB requests + poster + TVmaze), requests queued meanwhile are served between two steps; then a pause of 260 ms per
390. // TMDB request + 550 ms per TVmaze request (pauseMsAfter: ≤ 40 / 10 s, ≤ 20 / 10 s). RESUMES after a restart (the queue is
391. // what is still incomplete); TMDB 404 marks the title (`detailsCheck`), a passing failure goes to the back of the queue (at
392. // most 2 more tries per start). Waits while the daily sync or the adult backfill runs. Persisted every 10 titles + at the end.
393. // Log: start, every TRACKER_REPAIR_LOG_EVERY (100) titles, each failure, done. TRACKER_REPAIR=0 switches it off (off too
394. // without TMDB_READ_TOKEN). TRACKER_REPAIR_FAST=1 (the gate, fake TMDB only): 20 ms pause instead.
395. repairOn = env('TRACKER_REPAIR') != '0' && syncEnabled
396. repairFast = env('TRACKER_REPAIR_FAST') == '1'
397. repairLogEvery = env('TRACKER_REPAIR_LOG_EVERY') != null ? toNumber(env('TRACKER_REPAIR_LOG_EVERY')) : 100
398. repairQueue = null
399. repairPos = 0
400. repairDone = false
401. repairAt = 0
402. repairNextAt = 0
403. repairTries = {}
404. repairTotals = { titles = 0 complete = 0 failed = 0 adult = 0 newSeasons = 0 newEpisodes = 0 posters = 0 cast = 0 requests = 0 tvmazeRequests = 0 retries = 0 }
405. repairLine = () => {
406. let b = repairTotals
407. return b.titles + '/' + repairQueue.length + ' titles, complete=' + b.complete + ' failed=' + b.failed + ' adult=' + b.adult + ' newSeasons=' + b.newSeasons + ' newEpisodes=' + b.newEpisodes
408. + ' posters=' + b.posters + ' cast=' + b.cast + ' requests=' + b.requests + ' tvmazeRequests=' + b.tvmazeRequests + ' retries=' + b.retries + ' seconds=' + ((now() - repairAt) / 1000)
409. }
410. repairTick = () => {
411. if (repairDone) { return null }
412. let t = now()
413. if (repairQueue == null) {
414. repairQueue = detailsRepairQueue()
415. repairAt = t
416. console.log('details repair: start, ' + repairQueue.length + ' incomplete titles (' + (now() - t) + ' ms to find them)')
417. if (repairQueue.length == 0) { repairDone = true console.log('details repair: nothing to do') }
418. return null
419. }
420. if (t < repairNextAt) { return null }
421. let id = repairQueue[repairPos]
422. let r = completeStep(id)
423. let took = now() - t
424. if (took > 1500) { console.log('details repair: slow step, show ' + id + ', ' + took + ' ms') }
425. if (r.error != '') { console.log('details repair: ' + r.error) }
426. repairNextAt = now() + (repairFast ? 20 : pauseMsAfter(r))
427. // a series takes several steps: the same title again next time
428. if (!r.done && !r.retry) { return null }
429. repairPos = repairPos + 1
430. if (r.retry && (repairTries[id] == null || repairTries[id] < 2)) {
431. repairTries[id] = repairTries[id] == null ? 1 : repairTries[id] + 1
432. let q = repairQueue
433. q.push(id)
434. repairQueue = q
435. repairTotals.retries = repairTotals.retries + 1
436. } else if (!r.skipped) {
437. // done: its counts (a failed first step: the request alone)
438. let sum = r.sum
439. repairTotals.titles = repairTotals.titles + 1
440. 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 }
441. repairTotals.newSeasons = repairTotals.newSeasons + sum.newSeasons
442. repairTotals.newEpisodes = repairTotals.newEpisodes + sum.newEpisodes
443. repairTotals.posters = repairTotals.posters + sum.posters
444. repairTotals.cast = repairTotals.cast + sum.cast
445. repairTotals.requests = repairTotals.requests + (sum.steps == 0 ? r.requests : sum.requests)
446. repairTotals.tvmazeRequests = repairTotals.tvmazeRequests + sum.tvmazeRequests
447. }
448. let finished = repairPos >= repairQueue.length
449. if (finished || repairPos % 10 == 0) { persistDetails() }
450. if (!finished && !r.skipped && !r.retry && repairTotals.titles > 0 && repairTotals.titles % repairLogEvery == 0) { console.log('details repair: ' + repairLine()) }
451. if (finished) {
452. repairDone = true
453. console.log('details repair done: ' + repairLine())
454. }
455. return null
456. }
458. // ---- THE CREDITS JOB (tracker.worldapi.org#28, details.hl creditsStep) ------------------------------------------------------
459. // Every PUBLIC title (adult == false; adult/unknown are skipped) with a TMDB id gets its whole cast + crew once: at start (after
460. // the adult backfill and the details repair are done) the job collects every such title without `creditsAt` and asks ONE per
461. // step on the app's clock — one request (details + credits / aggregate_credits), then 260 ms (≤ 40 / 10 s); requests queued
462. // meanwhile are served between two steps. Titles the details repair, the daily sync or an import already did carry `creditsAt`
463. // and are not asked. RESUMES after a restart (the queue = what still has no credits); TMDB 404 / adult → `creditsCheck`, never
464. // asked again; a passing failure goes to the back of the queue (at most 2 more tries per start). Waits while the daily sync
465. // runs. Persisted (shows + people only) every 25 titles + at the end. Log: start, every TRACKER_CREDITS_LOG_EVERY (500) titles, failures, slow steps
466. // (> 1.5 s), done. TRACKER_CREDITS=0 switches it off (off too without TMDB_READ_TOKEN). TRACKER_CREDITS_FAST=1 (the gate, fake
467. // TMDB only): 20 ms pause instead.
468. // MEMORY LIMIT: the runtime never frees what a fetch + JSON parse took (mission 061 measured ~6 MB more RSS per title on the real
469. // 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
470. // step the job reads its own RSS (/proc/self/status); above TRACKER_CREDITS_MAX_MB (16000) it PAUSES for this process (logged) and
471. // goes on with what is left after the next restart (deploy, `docker restart`).
472. creditsOn = env('TRACKER_CREDITS') != '0' && syncEnabled
473. creditsFast = env('TRACKER_CREDITS_FAST') == '1'
474. creditsLogEvery = env('TRACKER_CREDITS_LOG_EVERY') != null ? toNumber(env('TRACKER_CREDITS_LOG_EVERY')) : 500
475. creditsQueueIds = null
476. creditsPos = 0
477. creditsDone = false
478. creditsAt = 0
479. creditsNextAt = 0
480. creditsTries = {}
481. creditsMaxMb = env('TRACKER_CREDITS_MAX_MB') != null ? toNumber(env('TRACKER_CREDITS_MAX_MB')) : 16000
482. creditsTotals = { titles = 0 done = 0 failed = 0 cast = 0 crew = 0 people = 0 requests = 0 retries = 0 slowest = 0 }
483. // this process' resident memory in MB (VmRSS), 0 when unknown
484. rssMb = () => {
485. if (!exists('/proc/self/status')) { return 0 }
486. let t = readFile('/proc/self/status')
487. let i = t != null ? t.indexOf('VmRSS:') : -1
488. if (i < 0) { return 0 }
489. let kb = toNumber(t.slice(i + 6, t.indexOf('kB', i)).trim())
490. return kb != null ? (kb - kb % 1024) / 1024 : 0
491. }
492. creditsLine = () => {
493. let b = creditsTotals
494. return b.titles + '/' + creditsQueueIds.length + ' titles, done=' + b.done + ' failed=' + b.failed + ' cast=' + b.cast + ' crew=' + b.crew + ' people=' + b.people
495. + ' requests=' + b.requests + ' retries=' + b.retries + ' slowestStepMs=' + b.slowest + ' seconds=' + ((now() - creditsAt) / 1000)
496. }
497. creditsTick = () => {
498. if (creditsDone) { return null }
499. let t = now()
500. if (creditsQueueIds == null) {
501. creditsQueueIds = creditsQueue()
502. creditsAt = t
503. console.log('credits job: start, ' + creditsQueueIds.length + ' titles without credits (' + (now() - t) + ' ms to find them)')
504. if (creditsQueueIds.length == 0) { creditsDone = true console.log('credits job: nothing to do') }
505. return null
506. }
507. if (t < creditsNextAt) { return null }
508. let mb = rssMb()
509. if (mb > creditsMaxMb) {
510. creditsDone = true
511. persistCredits()
512. console.log('credits job: paused, RSS ' + mb + ' MB > TRACKER_CREDITS_MAX_MB ' + creditsMaxMb + ' — ' + (creditsQueueIds.length - creditsPos) + ' titles left for the next start; ' + creditsLine())
513. return null
514. }
515. let id = creditsQueueIds[creditsPos]
516. creditsPos = creditsPos + 1
517. let r = creditsStep(id)
518. let took = now() - t
519. if (took > creditsTotals.slowest) { creditsTotals.slowest = took }
520. if (took > 1500) { console.log('credits job: slow step, show ' + id + ', ' + took + ' ms, cast=' + r.cast + ' crew=' + r.crew + ' people=' + r.people) }
521. if (r.error != '') { console.log('credits job: ' + r.error) }
522. creditsNextAt = now() + (creditsFast ? 20 : r.requests * msPerRequest)
523. if (r.retry && (creditsTries[id] == null || creditsTries[id] < 2)) {
524. creditsTries[id] = creditsTries[id] == null ? 1 : creditsTries[id] + 1
525. let q = creditsQueueIds
526. q.push(id)
527. creditsQueueIds = q
528. creditsTotals.retries = creditsTotals.retries + 1
529. } else if (!r.skipped) {
530. creditsTotals.titles = creditsTotals.titles + 1
531. if (r.error != '') { creditsTotals.failed = creditsTotals.failed + 1 } else { creditsTotals.done = creditsTotals.done + 1 }
532. creditsTotals.cast = creditsTotals.cast + r.cast
533. creditsTotals.crew = creditsTotals.crew + r.crew
534. creditsTotals.people = creditsTotals.people + r.people
535. }
536. creditsTotals.requests = creditsTotals.requests + r.requests
537. let finished = creditsPos >= creditsQueueIds.length
538. if (finished || creditsPos % 25 == 0) { persistCredits() }
539. if (!finished && !r.skipped && !r.retry && creditsTotals.titles > 0 && creditsTotals.titles % creditsLogEvery == 0) { console.log('credits job: ' + creditsLine()) }
540. if (finished) {
541. creditsDone = true
542. console.log('credits job done: ' + creditsLine())
543. }
544. return null
545. }
547. // 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
548. // between, only while no daily run is going on; the details repair only when neither has work; the credits job (#28) last
549. clockOn = syncOn || backfillOn || repairOn || creditsOn
550. clockTicks = 0
551. 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') + ')') }
552. if (!backfillOn) { console.log('adult backfill: off (' + (syncEnabled ? 'TRACKER_BACKFILL=0' : 'no TMDB_READ_TOKEN') + ')') }
553. if (!repairOn) { console.log('details repair: off (' + (syncEnabled ? 'TRACKER_REPAIR=0' : 'no TMDB_READ_TOKEN') + ')') }
554. if (!creditsOn) { console.log('credits job: off (' + (syncEnabled ? 'TRACKER_CREDITS=0' : 'no TMDB_READ_TOKEN') + ')') }
555. if (clockOn) {
556. syncClock = every(0.1)
557. on syncClock.tick(x) {
558. clockTicks = clockTicks + 1
559. let busy = syncQueue.length > 0
560. if (syncOn && clockTicks % 10 == 0) { syncTick() }
561. let free = !busy && syncQueue.length == 0
562. if (backfillOn && free && !backfillDone) { backfillTick() }
563. else if (repairOn && free && (!backfillOn || backfillDone) && !repairDone) { repairTick() }
564. else if (creditsOn && free && (!backfillOn || backfillDone) && (!repairOn || repairDone)) { creditsTick() }
565. return null
566. }
567. }
569. // ---- THE SHORT-ID BACKFILL (mission 060, tracker.worldapi.org#27, shortids.hl shortIdStep) ----------------------------
570. // 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,
571. // as many records as fit into TRACKER_SHORTID_BUDGET_MS (default 40 ms) — no TMDB, only the tables — so a page waits at
572. // most one batch. The queue is built at start (`ensureShortIds`) from what is still missing = it resumes after a restart.
573. // Persisted every 20 batches and at the end. Log: start, every TRACKER_SHORTID_LOG_EVERY (5000) records, done; with nothing
574. // to do one line. TRACKER_SHORTIDS=0 switches it off (the gate's resume check).
575. shortIdOn = env('TRACKER_SHORTIDS') != '0'
576. shortIdBudget = env('TRACKER_SHORTID_BUDGET_MS') != null ? toNumber(env('TRACKER_SHORTID_BUDGET_MS')) : 40
577. shortIdLogEvery = env('TRACKER_SHORTID_LOG_EVERY') != null ? toNumber(env('TRACKER_SHORTID_LOG_EVERY')) : 5000
578. shortIdRun = { at = 0 batches = 0 assigned = 0 nextLog = 0 slowest = 0 slowestPersist = 0 done = false }
579. shortIdTick = () => {
580. if (shortIdRun.done) { return null }
581. let t = now()
582. let r = shortIdStep(shortIdBudget)
583. let took = now() - t
584. shortIdRun.batches = shortIdRun.batches + 1
585. shortIdRun.assigned = shortIdRun.assigned + r.assigned
586. if (took > shortIdRun.slowest) { shortIdRun.slowest = took }
587. if (r.done || shortIdRun.batches % 20 == 0) {
588. let tp = now()
589. persistShortIds()
590. if (now() - tp > shortIdRun.slowestPersist) { shortIdRun.slowestPersist = now() - tp }
591. }
592. if (!r.done && shortIdRun.assigned >= shortIdRun.nextLog) {
593. console.log('short ids: ' + shortIdRun.assigned + ' assigned, ' + r.left + ' left, ' + ((now() - shortIdRun.at) / 1000) + ' s')
594. shortIdRun.nextLog = shortIdRun.nextLog + shortIdLogEvery
595. }
596. if (r.done) {
597. shortIdRun.done = true
598. let s = shortIdStats()
599. 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))
600. }
601. return null
602. }
603. if (!shortIdOn) { console.log('short ids: backfill off (TRACKER_SHORTIDS=0), ' + shortIdsLeft() + ' records without one') }
604. else if (shortIdsLeft() == 0) { console.log('short ids: nothing to do') }
605. else {
606. shortIdRun.at = now()
607. shortIdRun.nextLog = shortIdLogEvery
608. console.log('short ids: backfill start, ' + shortIdsLeft() + ' records, ' + shortIdBudget + ' ms per batch')
609. shortIdClock = every(0.1)
610. on shortIdClock.tick(x) {
611. shortIdTick()
612. return null
613. }
614. }
616. on Error(e) { console.log('error absorbed: ' + e.message) }