Portage complet PHP/Bash vers Node.js (Fastify + worker_threads)

cron/ambiguity.js

@@ -0,0 +1,104 @@
+// Port of tmdbintegral/ambiguity.php
+// Detects pairs of distinct TMDb ids whose filtered titles collide and whose
+// years are within YEARTOLERANCE.
+
+import { readFileSync } from 'node:fs';
+import { writeFile } from 'node:fs/promises';
+import { join } from 'node:path';
+import { TMDBINTEGRAL_DIR, NB_SEARCH_PARTS } from '../config.js';
+
+const TMDB = 0;
+const FILTEREDTITLE = 4;
+const FILTEREDENGLISHTITLE = 5;
+const FILTEREDORIGINALTITLE = 6;
+const YEAR = 7;
+
+export async function buildAmbiguity(type, nbParts = NB_SEARCH_PARTS) {
+  const yearTolerance = type === 'tv' ? 200 : 1;
+  const out = join(TMDBINTEGRAL_DIR, `ambiguity${type}.csv`);
+
+  const database = [];
+  for (let p = 0; p < nbParts; p++) {
+    const file = join(TMDBINTEGRAL_DIR, `search${type}${p}.json`);
+    const chunk = JSON.parse(readFileSync(file, 'utf8'));
+    for (const e of chunk) database.push(e);
+  }
+
+  const tmdbs = [];
+  const filteredTitles = [];
+  const languages = [];
+  const years = [];
+  for (const db of database) {
+    const fr = db[FILTEREDTITLE];
+    const en = db[FILTEREDENGLISHTITLE];
+    const vo = db[FILTEREDORIGINALTITLE];
+    if (fr) { tmdbs.push(db[TMDB]); filteredTitles.push(fr); years.push(db[YEAR][0]); languages.push('FR'); }
+    if (en) { tmdbs.push(db[TMDB]); filteredTitles.push(en); years.push(db[YEAR][0]); languages.push('EN'); }
+    if (vo) { tmdbs.push(db[TMDB]); filteredTitles.push(vo); years.push(db[YEAR][0]); languages.push('VO'); }
+  }
+
+  // PHP: array_multisort(filteredtitles, years, tmdbs, languages)
+  // Sort indices by (filteredTitle ASC, year ASC, tmdb ASC, language ASC).
+  const idx = filteredTitles.map((_, i) => i);
+  idx.sort((a, b) => {
+    if (filteredTitles[a] < filteredTitles[b]) return -1;
+    if (filteredTitles[a] > filteredTitles[b]) return 1;
+    if (years[a] !== years[b]) return years[a] - years[b];
+    if (tmdbs[a] !== tmdbs[b]) return tmdbs[a] - tmdbs[b];
+    if (languages[a] < languages[b]) return -1;
+    if (languages[a] > languages[b]) return 1;
+    return 0;
+  });
+
+  const sortedTmdbs = idx.map((i) => tmdbs[i]);
+  const sortedFiltered = idx.map((i) => filteredTitles[i]);
+  const sortedYears = idx.map((i) => years[i]);
+  const sortedLanguages = idx.map((i) => languages[i]);
+
+  let oldTmdb = 0;
+  let nbTmdbs = 0;
+  let oldFiltered = '';
+  let ambiguities = [];
+  const lines = [];
+
+  const flush = () => {
+    if (nbTmdbs >= 2) {
+      for (const a1 of ambiguities) {
+        for (const a2 of ambiguities) {
+          if (a1[0] !== a2[0] && Math.abs(a1[1] - a2[1]) <= yearTolerance) {
+            lines.push(`${a1[0]};${a1[2]};${a2[0]};${a2[2]}`);
+          }
+        }
+      }
+    }
+    ambiguities = [];
+    nbTmdbs = 0;
+  };
+
+  for (let i = 0; i < sortedFiltered.length; i++) {
+    if (sortedTmdbs[i] !== oldTmdb) nbTmdbs++;
+    oldTmdb = sortedTmdbs[i];
+
+    if (sortedFiltered[i] !== oldFiltered) {
+      flush();
+    }
+    oldFiltered = sortedFiltered[i];
+    ambiguities.push([sortedTmdbs[i], sortedYears[i], sortedLanguages[i]]);
+  }
+  flush();
+
+  await writeFile(out, lines.length ? lines.join('\n') + '\n' : '');
+}
+
+if (import.meta.url === `file://${process.argv[1]}`) {
+  const type = process.argv[2];
+  const nb = parseInt(process.argv[3] || String(NB_SEARCH_PARTS), 10);
+  if (type !== 'movie' && type !== 'tv') {
+    console.error('Usage: node cron/ambiguity.js movie|tv [nbParts]');
+    process.exit(1);
+  }
+  buildAmbiguity(type, nb).catch((err) => {
+    console.error(err);
+    process.exit(1);
+  });
+}