tracker

6.7 KB

1. // tests/realdata-056.mjs — mission 056, NOT the gate: the merged app (#13–#16 + 054 + 056) on a COPY of live storage/ AFTER the
2. // adult backfill ran on it (STATUS.md mission 056 has the procedure). Signed out at 390 + 1280: every public page type (home,
3. // /movies + page 2, /shows + page 2, two searches, a person page, a show page) — full load (navigate → document complete →
4. // socket open → images in view loaded), header one row, screenshot. Then every /shows/<slug> link of ALL /movies + /shows pages,
5. // the searches and the person's filmography → $REAL_OUT/slugs.txt (space separated, for tools/check-public-slugs.hl); an
6. // adult title's page signed out = "Not found". Signed in (cookie from .scratch/real-sid): /my/* + the show page + the adult
7. // page (visible when followed). Env: REAL_PORT (8730), REAL_CHROME (8731), REAL_OUT (/tmp/w056/real), REAL_PERSON (a person
8. // slug), REAL_SHOW (a show path; default the first on /my/shows), REAL_ADULT (an adult title's slug), REAL_SEARCHES.
9. import { writeFileSync, readFileSync, mkdirSync } from 'node:fs';
10. import { launchBrowser } from './cdp.mjs';
11. process.env.HL_CHROME ||= '/opt/google/chrome/chrome';
12. const BASE = `http://127.0.0.1:${process.env.REAL_PORT || 8730}`, OUT = process.env.REAL_OUT || '/tmp/w056/real';
13. const SID = readFileSync('.scratch/real-sid', 'utf8').trim();
14. const PERSON = process.env.REAL_PERSON || 'Jared-Harris', ADULT = process.env.REAL_ADULT || null;
15. const SEARCHES = (process.env.REAL_SEARCHES || 'the,sex,porn,xxx,girl,love').split(',');
16. mkdirSync(OUT, { recursive: true });
17. const browser = await launchBrowser({ debugPortRange: [Number(process.env.REAL_CHROME || 8731), Number(process.env.REAL_CHROME || 8731)] });
18. const page = await browser.newPage();
19. const ev = (e) => page.evaluate(e);
20. await page.send('Network.enable', {});
21. const hydrated = () => page.waitFor('!!window.__hl && window.__hl.socket && window.__hl.socket.readyState === 1 && document.readyState === "complete"', { timeout: 60000, label: 'hydrated' });
22. const imgsDone = () => page.waitFor(`Array.from(document.querySelectorAll('img')).filter(i => { const r = i.getBoundingClientRect(); return r.top < innerHeight && r.bottom > 0 && r.left < innerWidth && r.right > 0 }).every(i => i.complete)`, { timeout: 30000, label: 'imgs' }).catch(() => 'imgs timeout');
23. const shot = async (n) => writeFileSync(`${OUT}/${n}.png`, Buffer.from((await page.send('Page.captureScreenshot', { format: 'png' })).data, 'base64'));
24. const header = () => ev(`JSON.stringify((() => { const h = document.querySelector('application-header').getBoundingClientRect();
25. const b = document.querySelector('application-header .brand').getBoundingClientRect();
26. const parts = Array.from(document.querySelectorAll('application-header .brand, #searchlink, #selector, #loginbutton, #logout')).map(e => e.getBoundingClientRect()).filter(r => r.width > 0);
27. return { h: Math.round(h.height), oneRow: parts.every(r => r.top < b.bottom && r.bottom > b.top), sw: document.documentElement.scrollWidth }; })())`).then(JSON.parse);
28. const out = [];
29. const load = async (w, name, path, extra = '') => {
30. const t = Date.now();
31. await page.goto(BASE + path, { timeout: 60000 });
32. await hydrated();
33. if (extra) await page.waitFor(extra, { timeout: 120000, label: name + ' ready' });
34. await imgsDone();
35. const ms = Date.now() - t;
36. const hd = await header();
37. const h1 = await ev(`(document.querySelector('main h1') || {}).textContent || ''`);
38. out.push(`${w} ${path} ${ms} ms header=${hd.h}px oneRow=${hd.oneRow} scrollW=${hd.sw} h1="${h1.slice(0, 40)}"`);
39. await shot(`${name}-${w}`);
40. return { ms, hd };
41. };
42. const slugsOf = (html) => [...html.matchAll(/href="\/shows\/([^"\/?#]+)"/g)].map(m => decodeURIComponent(m[1]));
44. // ---- signed out ----
45. let show = process.env.REAL_SHOW || null;
46. for (const [w, h, mobile] of [[390, 844, true], [1280, 900, false]]) {
47. await page.send('Emulation.setDeviceMetricsOverride', { width: w, height: h, deviceScaleFactor: mobile ? 2 : 1, mobile });
48. for (const [name, path] of [['home', '/'], ['movies', '/movies'], ['movies-page-2', '/movies/page/2'], ['shows', '/shows'], ['shows-page-2', '/shows/page/2'], ['search-the', '/search/the'], ['search-sex', '/search/sex']])
49. await load(w, 'out-' + name, path);
50. // the person page: the first visit fills the filmography from TMDB (writes to the copy), so wait for it
51. await load(w, 'out-person', '/person/' + PERSON, `!document.querySelector('#personloading')`);
52. if (!show) show = '/shows/' + (await ev(`(document.querySelector('a.film-row') || { getAttribute: () => '' }).getAttribute('href')`)).split('/').pop();
53. await load(w, 'out-show', show);
54. }
55. // every link of the public lists, the searches, the person's filmography → their slugs
56. const slugs = new Set();
57. // the pagination is buttons (no /page/N links in the HTML) and a page past the end shows the last one: ask page after page
58. // until one brings no new title
59. let pagesAsked = 0;
60. for (const base of ['/movies', '/shows']) {
61. for (let p = 1; p < 1000; p++) {
62. const found = slugsOf(await (await fetch(BASE + base + (p > 1 ? '/page/' + p : ''))).text());
63. pagesAsked++;
64. const before = slugs.size;
65. found.forEach(s => slugs.add(s));
66. if (slugs.size === before) break;
67. }
68. }
69. slugsOf(await (await fetch(BASE + '/')).text()).forEach(s => slugs.add(s));
70. for (const q of SEARCHES) slugsOf(await (await fetch(BASE + '/search/' + encodeURIComponent(q))).text()).forEach(s => slugs.add(s));
71. const personSlugs = slugsOf(await (await fetch(BASE + '/person/' + PERSON)).text());
72. personSlugs.forEach(s => slugs.add(s));
73. writeFileSync(`${OUT}/slugs.txt`, [...slugs].join(' '));
74. out.push(`slugs: ${slugs.size} distinct from ${pagesAsked} list pages + / + ${SEARCHES.length} searches + the person (${personSlugs.length}) → ${OUT}/slugs.txt`);
75. if (ADULT) {
76. const html = await (await fetch(BASE + '/shows/' + ADULT)).text();
77. const h1 = (html.match(/<show-page[^>]*>[\s\S]*?<h1[^>]*>([^<]*)<\/h1>/) || [])[1];
78. out.push(`adult page signed out /shows/${ADULT}: h1="${h1}" slugInState=${html.includes('"urlSegment":"' + ADULT + '"')}`);
79. }
81. // ---- signed in (the creator) ----
82. await page.send('Network.setCookie', { name: 'trackersid', value: SID, url: BASE });
83. for (const [w, h, mobile] of [[390, 844, true], [1280, 900, false]]) {
84. await page.send('Emulation.setDeviceMetricsOverride', { width: w, height: h, deviceScaleFactor: mobile ? 2 : 1, mobile });
85. for (const [name, path] of [['my-shows', '/my/shows'], ['my-movies', '/my/movies'], ['my-unwatched', '/my/unwatched'], ['my-schedule', '/my/schedule'], ['home', '/']])
86. await load(w, 'in-' + name, path);
87. await load(w, 'in-show', show);
88. if (ADULT) await load(w, 'in-adult', '/shows/' + ADULT);
89. }
90. console.log(out.join('\n'));
91. console.log('problems', JSON.stringify(page.problems ? page.problems() : []));
92. await browser.close();