tracker

7.3 KB

1. // lib/sync-helpers.hl — the small pieces of the sync (sync.hl) that need no table and no request: placeholder titles, day
2. // numbers, TVmaze's HTML stripped, whether TVmaze numbers a show like we do, which seasons to ask, batches, changed texts,
3. // migrated rows, and the totals of a run (mission 028: out of tmdbsync.hl).
4. import { textOr } from './util.hl'
6. // ---- TVmaze merge (tracker.worldapi.org#15) ------------------------------------------------------
7. // a title that is none yet: empty, "TBA"/"TBD", or "Episode <n>" (what TMDB and TVmaze name an untitled episode; also
8. // French/German/Spanish: real data has TVmaze "Épisode 1" where TMDB says "Episode 1")
9. static placeholderWords = ['episode', 'épisode', 'Épisode', 'folge', 'episodio']
10. static isPlaceholderTitle = (&v) => {
11. t = textOr(v)
12. if (t == null) { return true }
13. l = t.trim().toLowerCase()
14. if (l == 'tba' || l == 'tbd' || l == 'tba.' || l == 'to be announced') { return true }
15. parts = l.split(' ')
16. if (parts.length != 2 || !placeholderWords.includes(parts[0]) || parts[1].length == 0) { return false }
17. let i = 0
18. while (i < parts[1].length) {
19. if (!'0123456789'.includes(parts[1][i])) { return false }
20. i = i + 1
21. }
22. return true
23. }
25. // 'YYYY-MM-DD' → a day number (only differences matter), null when it is no such date
26. static dayNumber = (&v) => {
27. t = textOr(v)
28. if (t == null || t.length < 10) { return null }
29. y = toNumber(t.slice(0, 4))
30. m = toNumber(t.slice(5, 7))
31. d = toNumber(t.slice(8, 10))
32. if (y == null || m == null || d == null || m < 1 || m > 12) { return null }
33. before = [0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304, 334]
34. py = m <= 2 ? y - 1 : y
35. leaps = (py - py % 4) / 4 - (py - py % 100) / 100 + (py - py % 400) / 400
36. return y * 365 + leaps + before[m - 1] + d
37. }
38. static daysApart = (&a, &b) => {
39. x = dayNumber(a)
40. y = dayNumber(b)
41. if (x == null || y == null) { return null }
42. return x > y ? x - y : y - x
43. }
45. // TVmaze's text without its HTML (tracker#18: `tvmazeSummary`): tags dropped — a block tag (p, br, div, li) becomes a space —
46. // the common entities decoded, white space collapsed; null when nothing is left
47. static blockTags = ['p', 'br', 'div', 'li', 'ul', 'ol', 'h1', 'h2', 'h3', 'h4', 'blockquote']
48. static entities = [['&nbsp;', ' '], ['&quot;', '"'], ['&#39;', "'"], ['&apos;', "'"], ['&lt;', '<'], ['&gt;', '>'], ['&ndash;', '–'], ['&mdash;', '—'], ['&hellip;', '…'], ['&amp;', '&']]
49. static stripHtml = (&v) => {
50. t = textOr(v)
51. if (t == null) { return null }
52. let out = ''
53. let i = 0
54. while (i < t.length) {
55. if (t[i] == '<') {
56. let close = t.indexOf('>', i)
57. if (close < 0) { close = t.length }
58. let name = t.slice(i + 1, close).trim().toLowerCase()
59. if (name.startsWith('/')) { name = name.slice(1) }
60. let sp = name.indexOf(' ')
61. if (sp >= 0) { name = name.slice(0, sp) }
62. if (name.endsWith('/')) { name = name.slice(0, name.length - 1) }
63. if (blockTags.includes(name)) { out = out + ' ' }
64. i = close + 1
65. } else {
66. // a line break / tab (a control character) is a space too (strings have no escapes to name them)
67. out = out + (t.charCodeAt(i) < 32 ? ' ' : t[i])
68. i = i + 1
69. }
70. }
71. for (e of entities) { out = out.replaceAll(e[0], e[1]) }
72. words = []
73. for (w of out.split(' ')) { if (w != '') { words.push(w) } }
74. return words.length > 0 ? words.join(' ') : null
75. }
77. // do we and TVmaze number a show the same way? Of the episodes both know (same S/E), at least 80 % must air within a day
78. // of each other OR carry the same title (a streaming show TVmaze dates all on its release day, TMDB week by week, still
79. // agrees by title). A show we have episodes of but TVmaze numbers none of the same (e.g. seasons by year) never agrees.
80. // Answers { known, compared, agree, ok }.
81. static sameTitle = (&a, &b) => { return !isPlaceholderTitle(a) && !isPlaceholderTitle(b) && a.trim().toLowerCase() == b.trim().toLowerCase() }
82. static numberingAgrees = (&epByNumber, &list) => {
83. out = { known = 0 compared = 0 agree = 0 ok = true }
84. localCount = epByNumber.keys().length
85. for (te of list) {
86. let e = epByNumber[te.season + 'x' + te.number]
87. if (e != null) {
88. out.known = out.known + 1
89. let apart = daysApart(e.release, te.airdate)
90. let titled = !isPlaceholderTitle(e.title) && !isPlaceholderTitle(te.name)
91. if (apart != null || titled) {
92. out.compared = out.compared + 1
93. if ((apart != null && apart <= 1) || sameTitle(e.title, te.name)) { out.agree = out.agree + 1 }
94. }
95. }
96. }
97. out.ok = !((localCount > 0 && out.known == 0) || (out.compared > 0 && out.agree * 5 < out.compared * 4))
98. return out
99. }
101. // which TMDB seasons to sync: every numbered one with episodes; 0 only when the show has a season 0 already
102. static wantedSeasonNumbers = (&details, &localByNumber) => {
103. out = []
104. if (details.seasons == null) { return out }
105. for (s of details.seasons) {
106. let n = s.season_number
107. if (n != null && (n > 0 || localByNumber['' + n] != null) && (s.episode_count == null || s.episode_count > 0 || localByNumber['' + n] != null)) { out.push(n) }
108. }
109. return out
110. }
112. // the seasons in batches of 20 (TMDB's append_to_response limit)
113. static batchesOf = (&list, &size) => {
114. out = []
115. let cur = []
116. for (x of list) {
117. cur.push(x)
118. if (cur.length == size) { out.push(cur) cur = [] }
119. }
120. if (cur.length > 0) { out.push(cur) }
121. return out
122. }
124. // fields of an episode/season from TMDB that differ from the stored record (null/empty never overwrites)
125. static changedText = (stored, &fresh) => { return fresh != null && fresh != stored }
126. // mission 070 (conductor: no step destroys migrated data): a season / episode the old tracker's migration brought (its `oldId`
127. // is neither `tmdb-…` nor `tvmaze-…`) keeps its FIRST title / summary in `migratedTitle` / `migratedSummary` before TMDB's text
128. // replaces it (set once; shown nowhere) — the daily delta run now also refreshes seasons of titles nobody follows
129. static isMigratedRow = (&r) => { return r.oldId != null && ('' + r.oldId).slice(0, 5) != 'tmdb-' && ('' + r.oldId).slice(0, 7) != 'tvmaze-' }
131. // totals of a run: the per-show counts added up
132. static emptyTotals = () => { return { shows = 0 requests = 0 tvmazeRequests = 0 newSeasons = 0 newEpisodes = 0 updatedEpisodes = 0 updatedSeasons = 0 posters = 0 ids = 0 tvmazeEpisodes = 0 tvmazeSeasons = 0 tvmazeFilled = 0 tvmazeSkipped = 0 errors = 0 } }
133. static addTotals = (t, &r) => {
134. o = t
135. o.shows = o.shows + 1
136. o.requests = o.requests + r.requests
137. o.tvmazeRequests = o.tvmazeRequests + r.tvmazeRequests
138. o.ids = o.ids + r.ids
139. o.newSeasons = o.newSeasons + r.newSeasons
140. o.newEpisodes = o.newEpisodes + r.newEpisodes
141. o.updatedEpisodes = o.updatedEpisodes + r.updatedEpisodes
142. o.updatedSeasons = o.updatedSeasons + r.updatedSeasons
143. o.posters = o.posters + r.posters
144. o.tvmazeEpisodes = o.tvmazeEpisodes + r.tvmazeEpisodes
145. o.tvmazeSeasons = o.tvmazeSeasons + r.tvmazeSeasons
146. o.tvmazeFilled = o.tvmazeFilled + r.tvmazeFilled
147. o.tvmazeSkipped = o.tvmazeSkipped + r.tvmazeSkipped
148. o.errors = o.errors + r.errors
149. return o
150. }
151. static totalsLine = (&t) => {
152. return 'shows=' + t.shows + ' newSeasons=' + t.newSeasons + ' newEpisodes=' + t.newEpisodes + ' updatedEpisodes=' + t.updatedEpisodes
153. + ' updatedSeasons=' + t.updatedSeasons + ' posters=' + t.posters + ' ids=' + t.ids + ' requests=' + t.requests + ' tvmazeRequests=' + t.tvmazeRequests + ' errors=' + t.errors
154. + ' tvmazeEpisodes=' + t.tvmazeEpisodes + ' tvmazeSeasons=' + t.tvmazeSeasons + ' tvmazeFilled=' + t.tvmazeFilled + ' tvmazeSkipped=' + t.tvmazeSkipped
155. }