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

cron/buildSearch.js

@@ -0,0 +1,133 @@
+// Port of tmdbintegral/search.php
+// Builds the chunked search database files (searchmovieN.json / searchtvN.json).
+//
+// Each entry has the same positional shape as the PHP version:
+//   [TMDB, TITLE, ENGLISHTITLE, ORIGINALTITLE,
+//    FILTEREDTITLE, FILTEREDENGLISHTITLE, FILTEREDORIGINALTITLE,
+//    YEARS[], POPULARITY]
+// so the runtime search worker can use the same indices.
+
+import { createReadStream, existsSync, readFileSync } from 'node:fs';
+import { writeFile } from 'node:fs/promises';
+import { createInterface } from 'node:readline';
+import { join } from 'node:path';
+import { TMDBINTEGRAL_DIR, NB_SEARCH_PARTS } from '../config.js';
+import { entryPath } from '../lib/paths.js';
+import { filterTitle } from '../lib/titleFilter.js';
+import { mbStrlen } from '../lib/mbLevenshtein.js';
+
+function lower(s) { return s.toLocaleLowerCase(); }
+
+function extractEnglishTitle(detail, type) {
+  const tr = detail?.translations?.translations;
+  if (!Array.isArray(tr)) return '';
+  for (const t of tr) {
+    if (t.iso_639_1 === 'en') {
+      return type === 'movie' ? (t.data?.title || '') : (t.data?.name || '');
+    }
+  }
+  return '';
+}
+
+function buildEntry(masterObj, detail, type) {
+  const tmdb = masterObj.id;
+  const popularity = parseFloat(masterObj.popularity) || 0;
+
+  let title, originalTitle, englishTitle;
+  const years = [];
+
+  if (type === 'movie') {
+    const date = String(detail.release_date || '').split('-');
+    years.push(parseInt(date[0], 10) || 0);
+    title = detail.title || '';
+    originalTitle = detail.original_title || '';
+    englishTitle = extractEnglishTitle(detail, 'movie');
+  } else {
+    const date = String(detail.first_air_date || '').split('-');
+    years.push(parseInt(date[0], 10) || 0);
+    title = detail.name || '';
+    originalTitle = detail.original_name || '';
+    englishTitle = extractEnglishTitle(detail, 'tv');
+    if (Array.isArray(detail.seasons)) {
+      for (const s of detail.seasons) {
+        const sd = String(s.air_date || '').split('-');
+        const sy = parseInt(sd[0], 10);
+        if (sy) years.push(sy);
+      }
+    }
+  }
+
+  if (!years[0]) return null;
+
+  let ft = filterTitle(title);
+  let fe = filterTitle(englishTitle);
+  let fo = filterTitle(originalTitle);
+
+  if (!ft && !fe && !fo) return null;
+
+  if (ft && mbStrlen(ft) / mbStrlen(title) < 0.5) ft = '';
+  if (fe && mbStrlen(fe) / mbStrlen(englishTitle) < 0.5) fe = '';
+  if (fo && mbStrlen(fo) / mbStrlen(originalTitle) < 0.5) fo = '';
+
+  // Dedupe years preserving order (PHP array_values(array_unique($years)))
+  const seen = new Set();
+  const uniqYears = [];
+  for (const y of years) {
+    if (!seen.has(y)) { seen.add(y); uniqYears.push(y); }
+  }
+
+  return [
+    tmdb,
+    title,
+    englishTitle,
+    originalTitle,
+    lower(ft),
+    lower(fe),
+    lower(fo),
+    uniqYears,
+    popularity,
+  ];
+}
+
+export async function buildSearch(type, nbParts = NB_SEARCH_PARTS) {
+  const indexFile = join(TMDBINTEGRAL_DIR, `${type}.json`);
+  const database = [];
+
+  const stream = createReadStream(indexFile, { encoding: 'utf8' });
+  const rl = createInterface({ input: stream, crlfDelay: Infinity });
+
+  for await (const line of rl) {
+    if (!line) continue;
+    let masterObj;
+    try { masterObj = JSON.parse(line); } catch { continue; }
+    const path = entryPath(type, masterObj.id);
+    if (!existsSync(path)) continue;
+    let detail;
+    try { detail = JSON.parse(readFileSync(path, 'utf8')); } catch { continue; }
+    const entry = buildEntry(masterObj, detail, type);
+    if (entry) database.push(entry);
+  }
+
+  const partSize = Math.ceil(database.length / nbParts);
+  const writes = [];
+  for (let p = 0; p < nbParts; p++) {
+    const chunk = database.slice(p * partSize, (p + 1) * partSize);
+    const out = join(TMDBINTEGRAL_DIR, `search${type}${p}.json`);
+    console.log(`Writing ${chunk.length} entries to search${type}${p}.json`);
+    writes.push(writeFile(out, JSON.stringify(chunk)));
+  }
+  await Promise.all(writes);
+}
+
+if (import.meta.url === `file://${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/buildSearch.js movie|tv [nbParts]');
+    process.exit(1);
+  }
+  buildSearch(type, nb).catch((err) => {
+    console.error(err);
+    process.exit(1);
+  });
+}