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

lib/queryParser.js

@@ -0,0 +1,85 @@
+// Replicates the query parsing logic shared by api.php and search.php:
+//   - extract a year (last (19|20)\d{2} match, ignoring 1080/2160)
+//   - extract an episode marker (SxxExxx, SxxExx, Sxx, partN, NxN, Exxx)
+//   - choose movie vs tv accordingly
+//   - extract titlein from the bytes before the year/episode
+
+import { FILTER_RE } from './titleFilter.js';
+
+const YEAR_RE = /(19|20)\d{2}/g;
+// Single-pass regex matching the PHP behaviour:
+//   - S/s and E/e and "part" are case-insensitive  ([Ss], [Ee], [Pp]art)
+//   - the lowercase 'x' in NxN, and uppercase 'E' in standalone Exxx, are case-sensitive
+// Greedy left-to-right alternation means "S01E02" is consumed whole, so the
+// trailing "E02" alternative cannot match inside it.
+const EPISODE_RE = /[Ss][0-9]{1,2}.?[Ee][0-9]{1,3}|[Ss][0-9]{2}|[Pp]art\.?[0-9]{1,3}|[0-9]{1,2}x[0-9]{1,3}|E[0-9]{1,3}/g;
+
+// PHP uses byte offsets (substr). To stay byte-faithful, work on the UTF-8 bytes.
+const utf8 = (s) => Buffer.from(s, 'utf8');
+const sliceBytes = (s, end) => utf8(s).slice(0, end).toString('utf8');
+
+function findAll(re, str) {
+  const out = [];
+  re.lastIndex = 0;
+  let m;
+  while ((m = re.exec(str)) !== null) {
+    out.push({ value: m[0], byteOffset: Buffer.byteLength(str.slice(0, m.index), 'utf8') });
+    if (m.index === re.lastIndex) re.lastIndex++;
+  }
+  return out;
+}
+
+function stripFilter(s) {
+  return s.replace(FILTER_RE, '');
+}
+
+export function parseQuery(query) {
+  if (!query) return null;
+
+  let yearin = 0;
+  let yearpos = -1;
+  let titlein = '';
+
+  const years = findAll(YEAR_RE, query).reverse();
+  for (const m of years) {
+    if (m.value === '1080' || m.value === '2160') continue;
+    yearin = parseInt(m.value, 10);
+    yearpos = m.byteOffset;
+    titlein = sliceBytes(query, yearpos);
+    break;
+  }
+
+  let episodein = '';
+  let episodepos = -1;
+  const eps = findAll(EPISODE_RE, query).reverse();
+  for (const m of eps) {
+    episodein = m.value;
+    episodepos = m.byteOffset;
+    break;
+  }
+
+  if (episodein) {
+    if (!yearin) {
+      titlein = sliceBytes(query, episodepos);
+    } else if (episodepos > yearpos) {
+      titlein = sliceBytes(query, yearpos);
+      if (!stripFilter(titlein)) {
+        titlein = sliceBytes(query, episodepos);
+        yearin = 0;
+      }
+    } else {
+      titlein = sliceBytes(query, episodepos);
+      if (!stripFilter(titlein)) {
+        titlein = sliceBytes(query, yearpos);
+        episodein = '';
+      }
+    }
+  }
+
+  if (!yearin && !episodein) {
+    return { error: 'Year or episode not found in query', titlein, yearin, episodein };
+  }
+
+  const type = episodein ? 'tv' : 'movie';
+  return { type, titlein, yearin, episodein };
+}