@cachly-dev/mcp-server 0.5.49 → 0.5.60

package/dist/index.js

@@ -47,7 +47,7 @@ import { Redis } from 'ioredis';
 const API_URL = process.env.CACHLY_API_URL ?? 'https://api.cachly.dev';
 const JWT = process.env.CACHLY_JWT ?? '';
 const EMBED_MODEL = process.env.CACHLY_EMBED_MODEL ?? '';
-const CURRENT_VERSION = '0.5.49';
+const CURRENT_VERSION = '0.5.60';
 /**
  * ┌──────────────────────────────────────────────────────────────────────┐
  * │  EMBEDDING PROVIDER — pluggable, client-side first                  │
@@ -403,6 +403,40 @@ const STOPWORDS = new Set([
     'רק', 'עוד', 'יש', 'אין', 'מה', 'זה', 'זאת', 'אלה', 'כל', 'כן',
     'לו', 'לה', 'להם', 'בו', 'בה', 'בהם', 'שם', 'כך', 'כן', 'מי',
     'אשר', 'אחרי', 'לפני', 'תחת', 'בין', 'מאז', 'עד', 'כמו', 'אז',
+    // ── Farsi/Persian — common function words, prepositions, pronouns ──
+    'در', 'از', 'به', 'با', 'که', 'این', 'آن', 'را', 'است',
+    'بود', 'باشد', 'شد', 'شده', 'می', 'نه', 'نیست', 'هم', 'هر',
+    'یک', 'اما', 'و', 'یا', 'تا', 'اگر', 'برای', 'چون', 'چه',
+    'کجا', 'کی', 'چطور', 'وقتی', 'بعد', 'قبل', 'مثل', 'همه', 'بعضی',
+    'هیچ', 'آیا', 'خود', 'چند', 'دیگر', 'هنوز', 'همان', 'آنها', 'اینها',
+    'ما', 'شما', 'آنان', 'من', 'تو', 'او', 'آنان', 'ایشان',
+    'روی', 'زیر', 'بین', 'پیش', 'پس', 'طرف', 'داخل', 'خارج', 'کنار',
+    'همچنین', 'مگر', 'ولی', 'وگرنه', 'چرا', 'چگونه', 'کدام',
+    // ── Hindi — common particles, postpositions, pronouns, auxiliary verbs ──
+    'है', 'में', 'से', 'को', 'का', 'की', 'के', 'पर', 'और', 'या',
+    'नहीं', 'यह', 'वह', 'एक', 'इस', 'उस', 'भी', 'हो', 'था', 'थी',
+    'थे', 'हैं', 'हूँ', 'लिए', 'तक', 'साथ', 'बाद', 'पहले', 'जो',
+    'जब', 'कैसे', 'क्यों', 'क्या', 'कहाँ', 'कौन', 'हम', 'आप', 'वे',
+    'मैं', 'तुम', 'उन', 'इन', 'ने', 'बहुत', 'सब', 'कुछ', 'फिर',
+    'अब', 'तो', 'ही', 'तरह', 'जैसे', 'करना', 'होना', 'रहा', 'रही',
+    'रहे', 'गया', 'गई', 'गए', 'किया', 'किए', 'कर', 'हुआ', 'हुई',
+    // ── Russian — common prepositions, pronouns, auxiliary verbs ───────────────
+    'в', 'на', 'не', 'с', 'и', 'а', 'но', 'по', 'за', 'из', 'от', 'до',
+    'к', 'у', 'о', 'об', 'во', 'при', 'под', 'над', 'без', 'для',
+    'что', 'как', 'это', 'все', 'так', 'уже', 'или', 'же', 'ли',
+    'если', 'то', 'да', 'нет', 'был', 'была', 'были', 'быть', 'есть',
+    'его', 'её', 'их', 'этот', 'эта', 'эти', 'который', 'которая',
+    'которые', 'который', 'мой', 'моя', 'мои', 'ваш', 'ваша', 'ваши',
+    'он', 'она', 'они', 'мы', 'вы', 'я', 'ты', 'тот', 'та', 'те',
+    'очень', 'тоже', 'также', 'когда', 'где', 'как', 'почему', 'зачем',
+    'сейчас', 'здесь', 'там', 'потому', 'поэтому', 'чтобы', 'тут',
+    // ── Turkish — common particles, postpositions, copulas ─────────────────────
+    'bir', 'bu', 'şu', 'o', 've', 'ile', 'de', 'da', 'için', 'gibi',
+    'ama', 'fakat', 'ancak', 'ya', 'ne', 'ki', 'mi', 'mı', 'mu', 'mü',
+    'var', 'yok', 'olan', 'oldu', 'olur', 'olarak', 'ise', 'daha',
+    'çok', 'en', 'her', 'hiç', 'bazı', 'tüm', 'bütün', 'hem', 'veya',
+    'bu', 'şu', 'ben', 'sen', 'biz', 'siz', 'onlar', 'benim', 'senin',
+    'nasıl', 'neden', 'niçin', 'nerede', 'ne', 'hangi', 'kaç',
 ]);
 // CJK Unicode ranges used for bigram extraction
 const CJK_RE = /[\u3040-\u30ff\u3400-\u4dbf\u4e00-\u9fff\uf900-\ufaff\uac00-\ud7af]/;
@@ -412,7 +446,14 @@ const RTL_RE = /[\u0590-\u05ff\u0600-\u06ff\u0750-\u077f\u08a0-\u08ff\ufb1d-\ufb
 // Distinguish Arabic from Hebrew within RTL segments
 const HEBREW_CHAR_RE = /[\u0590-\u05ff\ufb1d-\ufb4f]/;
 const ARABIC_CHAR_RE = /[\u0600-\u06ff\u0750-\u077f\u08a0-\u08ff\ufb50-\ufdff\ufe70-\ufeff]/;
-;
+// Persian/Farsi: exclusive chars not in Arabic (U+067E=پ, U+0686=چ, U+0698=ژ, U+06AF=گ, U+06CC=ی)
+const FARSI_CHAR_RE = /[\u067e\u0686\u0698\u06af\u06cc]/;
+// Devanagari (Hindi, Sanskrit, Marathi …) — U+0900–U+097F + extended
+const DEVANAGARI_RE = /[\u0900-\u097f]/;
+// Cyrillic (Russian, Bulgarian, Ukrainian, Serbian …) — U+0400–U+04FF
+const CYRILLIC_RE = /[\u0400-\u04ff]/;
+// Turkish uses Latin alphabet but has unique chars (ğ, ş, ı, ö, ü, ç) — detected by these
+const TURKISH_CHAR_RE = /[\u011f\u015f\u0131\u0130\u00e7]/; // ğ ş ı İ ç
 // ── Katakana → Romaji conversion table (Hepburn system) ──────────────────────
 // Digraphs must be listed before single chars so they match first.
 const KATA_ROMAJI_TABLE = [
@@ -540,7 +581,205 @@ function hebrewLightStem(word) {
     }
     return result;
 }
-// ── Cross-lingual tech term dictionary ────────────────────────────────────────
+/**
+ * Farsi/Persian light stemmer — handles Persian morphology.
+ * Strips:
+ *   \u0647\u0627 / \u0647\u0627\u06cc (ha/haye) — plural suffixes:  \u0633\u0631\u0648\u0631\u0647\u0627 \u2192 \u0633\u0631\u0648\u0631
+ *   \u0645\u06cc\u200c / \u0645\u06cc   (mi-)    — present-tense prefix: \u0645\u06cc\u200c\u06a9\u0646\u062f \u2192 \u06a9\u0646\u062f
+ *   \u0646\u0645\u06cc\u200c             (nami-)  — negated present:  \u0646\u0645\u06cc\u200c\u0634\u0648\u062f \u2192 \u0634\u0648\u062f
+ * Only strips when the result is still \u2265 3 chars.
+ */
+function farsiLightStem(word) {
+    let result = word;
+    // Suffixes first (longest first)
+    if (result.endsWith('\u0647\u0627\u06cc') && result.length > 5)
+        result = result.slice(0, -3);
+    else if (result.endsWith('\u0647\u0627') && result.length > 4)
+        result = result.slice(0, -2);
+    // Prefixes
+    if (result.startsWith('\u0646\u0645\u06cc\u200c') && result.length > 5)
+        result = result.slice(4);
+    else if (result.startsWith('\u0645\u06cc\u200c') && result.length > 4)
+        result = result.slice(3);
+    else if (result.startsWith('\u0645\u06cc') && result.length > 4)
+        result = result.slice(2);
+    return result;
+}
+/**
+ * Hindi light stemmer — strips the most common inflectional suffixes.
+ * Handles basic verb forms and oblique plural markers.
+ * Avoids over-stemming loanwords (most tech terms in Hindi are English loanwords).
+ * Only strips when result is still \u2265 3 chars.
+ */
+function hindiLightStem(word) {
+    const result = word;
+    // Infinitive / verb suffixes (longest first to avoid partial match)
+    if (result.endsWith('\u0928\u093e') && result.length > 4)
+        return result.slice(0, -2); // \u0928\u093e (nā) infinitive
+    if (result.endsWith('\u0928\u0947') && result.length > 4)
+        return result.slice(0, -2); // \u0928\u0947 (ne) ergative
+    if (result.endsWith('\u0928\u0940') && result.length > 4)
+        return result.slice(0, -2); // \u0928\u0940 (nī) fem infinitive
+    if (result.endsWith('\u0924\u093e') && result.length > 4)
+        return result.slice(0, -2); // \u0924\u093e (tā) m present participle
+    if (result.endsWith('\u0924\u0940') && result.length > 4)
+        return result.slice(0, -2); // \u0924\u0940 (tī) f present participle
+    if (result.endsWith('\u0915\u0930') && result.length > 4)
+        return result.slice(0, -2); // \u0915\u0930 (kar) conjunctive
+    if (result.endsWith('\u0913\u0902') && result.length > 4)
+        return result.slice(0, -2); // \u0913\u0902 (oṃ) oblique plural
+    if (result.endsWith('\u0907\u092f\u093e\u0902') && result.length > 5)
+        return result.slice(0, -4); // \u0907\u092f\u093e\u0902 (iyāṃ) f plural
+    if (result.endsWith('\u0940\u092f\u093e\u0902') && result.length > 5)
+        return result.slice(0, -4); // \u0940\u092f\u093e\u0102 (īyāṃ)
+    return result;
+}
+/**
+ * Russian light stemmer — strips common inflectional suffixes.
+ * Handles the most frequent verb/noun/adjective endings to improve recall.
+ * Only strips when result is still ≥ 3 chars.
+ *
+ * Covers:
+ *  ошибки/ошибка → ошибк  (noun plural/genitive)
+ *  развёртывание → развёртыва  (gerund → stem)
+ *  установить → установ  (infinitive)
+ *  настройки → настройк  (genitive plural)
+ */
+function russianLightStem(word) {
+    let w = word;
+    // Longest first to avoid partial stripping
+    // Verb infinitives / gerunds
+    if (w.endsWith('ывание') && w.length > 7)
+        return w.slice(0, -6);
+    if (w.endsWith('ивание') && w.length > 7)
+        return w.slice(0, -6);
+    if (w.endsWith('ование') && w.length > 7)
+        return w.slice(0, -6);
+    if (w.endsWith('вание') && w.length > 6)
+        return w.slice(0, -5);
+    if (w.endsWith('ение') && w.length > 5)
+        return w.slice(0, -4);
+    if (w.endsWith('ание') && w.length > 5)
+        return w.slice(0, -4);
+    if (w.endsWith('ить') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('ать') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('еть') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('уть') && w.length > 4)
+        return w.slice(0, -3);
+    // Noun plural/genitive
+    if (w.endsWith('ки') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ги') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ов') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ев') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ей') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ий') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ый') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ая') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ое') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ую') && w.length > 4)
+        return w.slice(0, -2);
+    // Verb present tense
+    if (w.endsWith('ет') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ют') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ут') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ят') && w.length > 4)
+        return w.slice(0, -2);
+    return w;
+}
+/**
+ * Turkish light stemmer — strips common agglutinative suffixes.
+ * Turkish is highly agglutinative; this covers the most common inflectional ends.
+ *
+ * Covers:
+ *  hatalar → hata  (-lar/-ler plural)
+ *  sunucuda → sunucu  (-da/-de locative)
+ *  dağıtımı → dağıtım  (-ı/-i/-u/-ü accusative)
+ *  yüklemek → yükle  (-mek/-mak infinitive)
+ */
+function turkishLightStem(word) {
+    let w = word;
+    // Suffixes longest-first
+    if (w.endsWith('lardaki') && w.length > 7)
+        return w.slice(0, -7);
+    if (w.endsWith('lerdeki') && w.length > 7)
+        return w.slice(0, -7);
+    if (w.endsWith('ların') && w.length > 6)
+        return w.slice(0, -5);
+    if (w.endsWith('lerin') && w.length > 6)
+        return w.slice(0, -5);
+    if (w.endsWith('larda') && w.length > 6)
+        return w.slice(0, -5);
+    if (w.endsWith('lerde') && w.length > 6)
+        return w.slice(0, -5);
+    if (w.endsWith('ları') && w.length > 5)
+        return w.slice(0, -4);
+    if (w.endsWith('leri') && w.length > 5)
+        return w.slice(0, -4);
+    if (w.endsWith('mek') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('mak') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('ler') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('lar') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('nın') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('nin') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('nun') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('nün') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('da') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('de') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ta') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('te') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('dan') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('den') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('tan') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('ten') && w.length > 4)
+        return w.slice(0, -3);
+    if (w.endsWith('ın') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('in') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('un') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ün') && w.length > 4)
+        return w.slice(0, -2);
+    if (w.endsWith('ı') && w.length > 4)
+        return w.slice(0, -1);
+    if (w.endsWith('i') && w.length > 4)
+        return w.slice(0, -1);
+    if (w.endsWith('u') && w.length > 4)
+        return w.slice(0, -1);
+    if (w.endsWith('ü') && w.length > 4)
+        return w.slice(0, -1);
+    return w;
+}
 // Maps tokens (English, Japanese katakana, Chinese, Korean, Arabic, Hebrew)
 // to their equivalents in other languages. Used to expand tokens at search time
 // so "deploy" finds documents containing "デプロイ" / "部署" / "배포", and vice versa.
@@ -548,46 +787,46 @@ function hebrewLightStem(word) {
 // Keys must be lowercase. CJK values will be bigram-expanded at use time.
 const CROSS_LINGUAL_MAP = new Map([
     // English → all
-    ['deploy', ['デプロイ', '部署', '배포', 'deployment', 'deploying', 'نشر', 'פריסה', 'bereitstellen', 'bereitstellung', 'déployer', 'déploiement', 'desplegar', 'despliegue', 'distribuire', 'distribuzione', 'implantar', 'implantação']],
-    ['deployment', ['デプロイ', '部署', '배포', 'deploy', 'نشر', 'פריסה', 'bereitstellung', 'déploiement', 'despliegue', 'distribuzione', 'implantação']],
+    ['deploy', ['デプロイ', '部署', '배포', 'deployment', 'deploying', 'نشر', 'פריסה', 'استقرار', 'तैनाती', 'bereitstellen', 'bereitstellung', 'déployer', 'déploiement', 'desplegar', 'despliegue', 'distribuire', 'distribuzione', 'implantar', 'implantação', 'развёртывание', 'развертывание', 'dağıtım', 'dağıtmak']],
+    ['deployment', ['デプロイ', '部署', '배포', 'deploy', 'نشر', 'פריסה', 'استقرار', 'तैनाती', 'bereitstellung', 'déploiement', 'despliegue', 'distribuzione', 'implantação', 'развёртывание', 'dağıtım']],
     ['container', ['コンテナ', '容器', '컨테이너', 'docker']],
-    ['server', ['サーバー', 'サーバ', '服务器', '서버', 'سيرفر', 'שרת']],
-    ['database', ['データベース', '数据库', '데이터베이스', 'db', 'قاعدة البيانات', 'datenbank', 'base de données', 'base de datos', 'banco de dados']],
-    ['cache', ['キャッシュ', '缓存', '캐시', 'كاش', 'מטמון']],
-    ['error', ['エラー', '错误', '오류', 'خطأ', 'שגיאה', 'exception', 'err', 'fehler', 'erreur', 'fallo', 'errore', 'erro']],
-    ['bug', ['バグ', '缺陷', '버그', 'issue', 'defect', 'خلل']],
-    ['fix', ['修正', '修复', '수정', 'bugfix', 'patch', 'hotfix', 'إصلاح', 'תיקון', 'beheben', 'behoben', 'réparer', 'correction', 'arreglar', 'corrección', 'correggere', 'corrigir']],
-    ['build', ['ビルド', '构建', '빌드', 'بناء', 'בנייה', 'bauen', 'construire', 'construir', 'costruire']],
-    ['test', ['テスト', '测试', '테스트', 'اختبار', 'בדיקה', 'testen', 'tester', 'probar', 'testare', 'testar']],
-    ['auth', ['認証', '认证', '인증', 'authentication', 'login', 'oauth', 'مصادقة', 'אימות', 'authentifizierung', 'authentification', 'autenticación', 'autenticazione', 'autenticação']],
-    ['authentication', ['auth', 'مصادقة', 'אימות', '認証', '认证', '인증', 'login', 'oauth']],
-    ['environment', ['環境', '环境', '환경', 'env', 'بيئة', 'סביבה', 'umgebung', 'environnement', 'entorno', 'ambiente']],
-    ['secret', ['key', 'token', 'password', 'jwt', 'مفتاح', 'מפתח', 'geheimnis', 'schlüssel', 'clave', 'clé', 'chiave', 'chave']],
-    ['problem', ['error', 'issue', 'bug', 'failure', 'مشكلة', 'بعيا', '問題', '問題', '오류', 'fehler', 'problème', 'problema']],
+    ['server', ['サーバー', 'サーバ', '服务器', '서버', 'سيرفر', 'שרת', 'سرور', 'सर्वर', 'сервер', 'sunucu']],
+    ['database', ['データベース', '数据库', '데이터베이스', 'db', 'قاعدة البيانات', 'datenbank', 'پایگاه داده', 'डेटाबेس', 'base de données', 'base de datos', 'banco de dados']],
+    ['cache', ['キャッシュ', '缓存', '캐시', 'كاش', 'מטמון', 'کش', 'कैश', 'кэш', 'önbellek']],
+    ['error', ['エラー', '错误', '오류', 'خطأ', 'שגיאה', 'خطا', 'त्रुटि', 'गलती', 'exception', 'err', 'fehler', 'erreur', 'fallo', 'errore', 'erro', 'ошибка', 'hata', 'hatalı']],
+    ['bug', ['バグ', '缺陷', '버그', 'issue', 'defect', 'خلل', 'اشکال', 'बग']],
+    ['fix', ['修正', '修复', '수정', 'bugfix', 'patch', 'hotfix', 'إصلاح', 'תיקון', 'رفع', 'सुधार', 'beheben', 'behoben', 'réparer', 'correction', 'arreglar', 'corrección', 'correggere', 'corrigir', 'исправить', 'исправление', 'düzeltmek', 'düzeltme']],
+    ['build', ['ビルド', '构建', '빌드', 'بناء', 'בנייה', 'ساخت', 'निर्माण', 'bauen', 'construire', 'construir', 'costruire', 'сборка', 'собрать', 'derleme', 'derlemek']],
+    ['test', ['テスト', '测试', '테스트', 'اختبار', 'בדיקה', 'آزمایش', 'परीक्षण', 'testen', 'tester', 'probar', 'testare', 'testar', 'тест', 'тестирование', 'test', 'testlemek']],
+    ['auth', ['認証', '认证', '인증', 'authentication', 'login', 'oauth', 'مصادقة', 'אימות', 'احراز هویت', 'प्रमाणीकरण', 'authentifizierung', 'authentification', 'autenticación', 'autenticazione', 'autenticação', 'аутентификация', 'авторизация', 'kimlik doğrulama', 'yetkilendirme']],
+    ['authentication', ['auth', 'مصادقة', 'אימות', 'احراز هویت', 'प्रमाणीकरण', '認証', '认证', '인증', 'login', 'oauth', 'аутентификация', 'авторизация', 'kimlik doğrulama']],
+    ['environment', ['環境', '环境', '환경', 'env', 'بيئة', 'סביבה', 'umgebung', 'environnement', 'entorno', 'ambiente', 'среда', 'окружение', 'ortam']],
+    ['secret', ['key', 'token', 'password', 'jwt', 'مفتاح', 'מפתח', 'geheimnis', 'schlüssel', 'clave', 'clé', 'chiave', 'chave', 'секрет', 'ключ', 'gizli', 'anahtar']],
+    ['problem', ['error', 'issue', 'bug', 'failure', 'مشكلة', '問題', '오류', 'fehler', 'problème', 'problema', 'проблема', 'ошибка', 'sorun', 'problem']],
     ['kubernetes', ['クベルネテス', 'k8s', 'kube']],
-    ['network', ['ネットワーク', '网络', '네트워크', 'شبكة', 'רשת', 'netzwerk', 'réseau', 'red', 'rete', 'rede']],
+    ['network', ['ネットワーク', '网络', '네트워크', 'شبكة', 'רשת', 'شبکه', 'नेटवर्क', 'netzwerk', 'réseau', 'red', 'rete', 'rede', 'сеть', 'ağ']],
     ['timeout', ['タイムアウト', '超时', '타임아웃', 'مهلة']],
-    ['memory', ['メモリ', '内存', '메모리', 'ram', 'ذاكرة', 'זיכרון', 'speicher', 'mémoire', 'memoria', 'memória']],
-    ['config', ['設定', '配置', '설정', 'configuration', 'settings', 'conf', 'إعدادات', 'הגדרות', 'konfiguration', 'einstellungen', 'configuración', 'configurazione', 'configuração']],
-    ['install', ['インストール', '安装', '설치', 'تثبيت', 'התקנה', 'installieren', 'installer', 'instalar', 'installare']],
-    ['update', ['アップデート', '更新', '업데이트', 'upgrade', 'تحديث', 'עדכון', 'aktualisieren', 'aktualisierung', 'actualizar', 'aggiornare', 'atualizar']],
-    ['log', ['ログ', '日志', '로그', 'logging', 'سجل', 'יומן', 'protokoll', 'protokolle', 'journal', 'registro']],
+    ['memory', ['メモリ', '内存', '메모리', 'ram', 'ذاكرة', 'זיכרון', 'حافظه', 'मेमोरी', 'speicher', 'mémoire', 'memoria', 'memória', 'память', 'bellek', 'hafıza']],
+    ['config', ['設定', '配置', '설정', 'configuration', 'settings', 'conf', 'إعدادات', 'הגדרות', 'پیکربندی', 'कॉन्फ़िग', 'konfiguration', 'einstellungen', 'configuración', 'configurazione', 'configuração', 'конфигурация', 'настройки', 'yapılandırma', 'ayarlar']],
+    ['install', ['インストール', '安装', '설치', 'تثبيت', 'התקנה', 'نصب', 'इंस्टॉल', 'installieren', 'installer', 'instalar', 'installare', 'установить', 'установка', 'yüklemek', 'kurulum']],
+    ['update', ['アップデート', '更新', '업데이트', 'upgrade', 'تحديث', 'עדכון', 'به‌روزرسانی', 'अपडेट', 'aktualisieren', 'aktualisierung', 'actualizar', 'aggiornare', 'atualizar', 'обновление', 'обновить', 'güncelleme']],
+    ['log', ['ログ', '日志', '로그', 'logging', 'سجل', 'יומן', 'ثبت', 'लॉग', 'protokoll', 'protokolle', 'journal', 'registro', 'журнал', 'лог', 'günlük']],
     ['port', ['ポート', '端口', '포트', 'منفذ', 'פורט']],
     ['file', ['ファイル', '文件', '파일', 'ملف', 'קובץ']],
     ['image', ['イメージ', '镜像', '이미지', 'صورة', 'תמונה', 'abbild', 'imagen', 'imagem']],
     ['volume', ['ボリューム', '卷', '볼륨']],
     ['cluster', ['クラスター', '集群', '클러스터']],
     ['node', ['ノード', '节点', '노드']],
-    ['service', ['サービス', '服务', '서비스', 'svc', 'خدمة', 'שירות', 'dienst', 'servicio', 'servizio', 'serviço']],
+    ['service', ['サービス', '服务', '서비스', 'svc', 'خدمة', 'שירות', 'dienst', 'servicio', 'servizio', 'serviço', 'сервис', 'служба', 'servis', 'hizmet']],
     ['certificate', ['証明書', '证书', '인증서', 'cert', 'ssl', 'tls', 'شهادة', 'תעודה']],
     ['password', ['パスワード', '密码', '비밀번호', 'passwd', 'pwd', 'كلمة المرور', 'סיסמה']],
     ['token', ['トークン', '令牌', '토큰', 'jwt', 'secret', 'رمز', 'אסימון']],
     ['health', ['ヘルス', '健康', '헬스', 'healthcheck', 'probe', 'صحة']],
     ['migration', ['マイグレーション', '迁移', '마이그레이션', 'migrate', 'هجرة']],
-    ['backup', ['バックアップ', '备份', '백업', 'نسخ احتياطي', 'גיבוי', 'sicherung', 'sauvegarde', 'respaldo', 'cópia de segurança']],
-    ['monitor', ['モニター', '监控', '모니터링', 'monitoring', 'metrics', 'alert', 'مراقبة', 'ניטור', 'überwachung', 'surveiller', 'monitoreo', 'monitorare', 'monitorar']],
-    ['performance', ['パフォーマンス', '性能', '성능', 'latency', 'throughput', 'leistung', 'performances', 'rendimiento', 'prestazioni', 'desempenho']],
-    ['connection', ['接続', '连接', '연결', 'conn', 'socket', 'اتصال', 'חיבור', 'verbindung', 'connexion', 'conexión', 'connessione', 'conexão']],
+    ['backup', ['バックアップ', '备份', '백업', 'نسخ احتياطي', 'גיבוי', 'پشتیبان‌گیری', 'बैकअप', 'sicherung', 'sauvegarde', 'respaldo', 'cópia de segurança', 'резервная копия', 'резерв', 'yedekleme']],
+    ['monitor', ['モニター', '监控', '모니터링', 'monitoring', 'metrics', 'alert', 'مراقبة', 'ניטור', 'نظارت', 'निगरानी', 'überwachung', 'surveiller', 'monitoreo', 'monitorare', 'monitorar', 'мониторинг', 'наблюдение', 'izleme']],
+    ['performance', ['パフォーマンス', '性能', '성능', 'latency', 'throughput', 'عملکرد', 'प्रदर्शन', 'leistung', 'performances', 'rendimiento', 'prestazioni', 'desempenho', 'производительность', 'быстродействие', 'performans']],
+    ['connection', ['接続', '连接', '연결', 'conn', 'socket', 'اتصال', 'חיבור', 'اتصال', 'कनेक्शन', 'verbindung', 'connexion', 'conexión', 'connessione', 'conexão']],
     ['queue', ['キュー', '队列', '큐', 'طابور', 'warteschlange', 'file d\'attente', 'cola', 'coda', 'fila']],
     ['redis', ['レディス', '레디스']],
     ['nginx', ['エンジンエックス']],
@@ -601,10 +840,10 @@ const CROSS_LINGUAL_MAP = new Map([
     ['git', ['version control', 'repo', 'repository', 'commit', 'push', 'pull']],
     ['ci', ['pipeline', 'github actions', 'gitlab', 'jenkins', 'build']],
     ['cd', ['deploy', 'deployment', 'release']],
-    ['debug', ['デバッグ', '调试', '디버그', 'debugging', 'breakpoint', 'تصحيح', 'איתור באגים', 'debuggen', 'déboguer', 'depurar']],
-    ['crash', ['クラッシュ', '崩溃', '크래시', 'panic', 'segfault', 'انهيار', 'קריסה', 'absturz', 'panne', 'plantage', 'caída', 'falha']],
-    ['restart', ['再起動', '重启', '재시작', 'reboot', 'إعادة تشغيل', 'הפעלה מחדש', 'neustart', 'redémarrer', 'reiniciar', 'riavviare']],
-    ['permission', ['権限', '权限', '권한', 'access', 'acl', 'chmod', 'صلاحية', 'הרשאה', 'berechtigung', 'berechtigungen', 'permiso', 'permesso', 'permissão']],
+    ['debug', ['デバッグ', '调试', '디버그', 'debugging', 'breakpoint', 'تصحيح', 'איתור באגים', 'اشکال‌زدایی', 'डीबग', 'debuggen', 'déboguer', 'depurar', 'отладка', 'дебаг', 'hata ayıklama']],
+    ['crash', ['クラッシュ', '崩溃', '크래시', 'panic', 'segfault', 'انهيار', 'קריסה', 'خرابی', 'क्रैश', 'absturz', 'panne', 'plantage', 'caída', 'falha', 'сбой', 'крэш', 'çökme', 'çöküş']],
+    ['restart', ['再起動', '重启', '재시작', 'reboot', 'إعادة تشغيل', 'הפעלה מחדש', 'راه‌اندازی مجدد', 'पुنरारंभ', 'neustart', 'redémarrer', 'reiniciar', 'riavviare']],
+    ['permission', ['権限', '权限', '권한', 'access', 'acl', 'chmod', 'صلاحية', 'הרשאה', 'مجوز', 'अनुमति', 'berechtigung', 'berechtigungen', 'permiso', 'permesso', 'permissão', 'разрешение', 'доступ', 'izin', 'yetki']],
     // Japanese katakana → English
     ['デプロイ', ['deploy', 'deployment']],
     ['コンテナ', ['container', 'docker']],
@@ -790,6 +1029,67 @@ const CROSS_LINGUAL_MAP = new Map([
     ['ייצור', ['production', 'prod']],
     ['סודי', ['secret', 'key', 'token']],
     ['מפתח סודי', ['secret', 'key']],
+    // ── Farsi/Persian → English ───────────────────────────────────────────────
+    ['استقرار', ['deploy', 'deployment']],
+    ['خطا', ['error', 'err']], // Farsi: خطا  vs Arabic: خطأ
+    ['اشکال', ['bug', 'issue', 'error']],
+    ['رفع اشکال', ['fix', 'debug', 'debugging']],
+    ['ساخت', ['build']],
+    ['آزمایش', ['test']],
+    ['احراز هویت', ['auth', 'authentication']],
+    ['پیکربندی', ['config', 'configuration']],
+    ['نصب', ['install']],
+    ['به‌روزرسانی', ['update', 'upgrade']],
+    ['ثبت', ['log']],
+    ['سرور', ['server']],
+    ['پایگاه داده', ['database', 'db']],
+    ['شبکه', ['network']], // Farsi شبکه vs Arabic شبكة
+    ['حافظه', ['memory', 'ram']], // Farsi حافظه vs Arabic ذاكرة
+    ['کش', ['cache']],
+    ['اشکال‌زدایی', ['debug', 'debugging']],
+    ['خرابی', ['crash', 'failure', 'error']],
+    ['راه‌اندازی مجدد', ['restart', 'reboot']],
+    ['مجوز', ['permission', 'access']],
+    ['پشتیبان‌گیری', ['backup']],
+    ['عملکرد', ['performance']],
+    ['مشکل', ['problem', 'issue', 'error']],
+    ['سرویس', ['service']],
+    ['کلید', ['key', 'token', 'secret']],
+    ['محیط', ['environment', 'env']],
+    ['تولید', ['production', 'prod']],
+    ['رمز', ['secret', 'key', 'token']],
+    ['اتصال', ['connection', 'conn']],
+    // ── Hindi (Devanagari) → English ──────────────────────────────────────────
+    ['तैनाती', ['deploy', 'deployment']],
+    ['त्रुटि', ['error', 'err']],
+    ['गलती', ['error', 'bug', 'issue']],
+    ['सुधार', ['fix', 'bugfix']],
+    ['निर्माण', ['build']],
+    ['परीक्षण', ['test']],
+    ['प्रमाणीकरण', ['auth', 'authentication']],
+    ['विन्यास', ['config', 'configuration']],
+    ['कॉन्फ़िगरेशन', ['config', 'configuration']],
+    ['इंस्टॉल', ['install']],
+    ['अपडेट', ['update', 'upgrade']],
+    ['लॉग', ['log']],
+    ['सर्वर', ['server']],
+    ['डेटाबेस', ['database', 'db']],
+    ['नेटवर्क', ['network']],
+    ['मेमोरी', ['memory', 'ram']],
+    ['कैश', ['cache']],
+    ['डीबग', ['debug', 'debugging']],
+    ['क्रैश', ['crash', 'panic']],
+    ['पुनरारंभ', ['restart', 'reboot']],
+    ['अनुमति', ['permission', 'access']],
+    ['बैकअप', ['backup']],
+    ['प्रदर्शन', ['performance']],
+    ['समस्या', ['problem', 'issue', 'error']],
+    ['सेवा', ['service']],
+    ['कनेक्शन', ['connection', 'conn']],
+    ['पासवर्ड', ['password']],
+    ['टोकन', ['token', 'secret']],
+    ['वातावरण', ['environment', 'env']],
+    ['उत्पादन', ['production', 'prod']],
     // ── European languages → English ─────────────────────────────────────────
     // German (Deutsch) → English
     ['fehler', ['error', 'err']],
@@ -971,6 +1271,103 @@ const CROSS_LINGUAL_MAP = new Map([
     ['ambiente', ['environment', 'env']],
     ['fila', ['queue']],
     ['imagem', ['image']],
+    // ── Russian (Cyrillic) → English ──────────────────────────────────────────
+    ['ошибка', ['error', 'err']],
+    ['ошибки', ['error', 'err']],
+    ['сбой', ['crash', 'failure', 'error']],
+    ['крэш', ['crash', 'panic']],
+    ['баг', ['bug', 'issue']],
+    ['исправить', ['fix', 'bugfix']],
+    ['исправление', ['fix', 'patch']],
+    ['сборка', ['build']],
+    ['собрать', ['build']],
+    ['тест', ['test']],
+    ['тестирование', ['test', 'testing']],
+    ['аутентификация', ['auth', 'authentication']],
+    ['авторизация', ['auth', 'authorization']],
+    ['конфигурация', ['config', 'configuration']],
+    ['настройки', ['config', 'settings']],
+    ['установить', ['install']],
+    ['установка', ['install']],
+    ['обновление', ['update', 'upgrade']],
+    ['обновить', ['update', 'upgrade']],
+    ['журнал', ['log']],
+    ['лог', ['log', 'logs']],
+    ['сервер', ['server']],
+    ['база данных', ['database', 'db']],
+    ['сеть', ['network']],
+    ['память', ['memory', 'ram']],
+    ['кэш', ['cache']],
+    ['отладка', ['debug', 'debugging']],
+    ['перезапуск', ['restart', 'reboot']],
+    ['перезагрузка', ['restart', 'reboot']],
+    ['разрешение', ['permission', 'access']],
+    ['доступ', ['permission', 'access']],
+    ['резерв', ['backup']],
+    ['производительность', ['performance']],
+    ['проблема', ['problem', 'issue', 'error']],
+    ['сервис', ['service']],
+    ['служба', ['service']],
+    ['подключение', ['connection', 'conn']],
+    ['соединение', ['connection', 'conn']],
+    ['развёртывание', ['deploy', 'deployment']],
+    ['развертывание', ['deploy', 'deployment']],
+    ['контейнер', ['container', 'docker']],
+    ['образ', ['image']],
+    ['кластер', ['cluster']],
+    ['узел', ['node']],
+    ['секрет', ['secret', 'key', 'token']],
+    ['ключ', ['key', 'token', 'secret']],
+    ['пароль', ['password']],
+    ['токен', ['token', 'secret']],
+    ['мониторинг', ['monitor', 'monitoring']],
+    ['наблюдение', ['monitor', 'monitoring']],
+    ['мигрировать', ['migrate', 'migration']],
+    ['миграция', ['migrate', 'migration']],
+    // ── Turkish (Latin + special chars ğ/ş/ı) → English ─────────────────────
+    ['hata', ['error', 'err']],
+    ['hatalı', ['error', 'err']],
+    ['çökme', ['crash', 'panic']],
+    ['çöküş', ['crash', 'failure']],
+    ['düzeltmek', ['fix', 'bugfix']],
+    ['düzeltme', ['fix', 'patch']],
+    ['derleme', ['build']],
+    ['derlemek', ['build']],
+    ['test', ['test']],
+    ['testlemek', ['test', 'testing']],
+    ['kimlik doğrulama', ['auth', 'authentication']],
+    ['yetkilendirme', ['auth', 'authorization']],
+    ['yapılandırma', ['config', 'configuration']],
+    ['ayarlar', ['config', 'settings']],
+    ['yüklemek', ['install']],
+    ['kurulum', ['install']],
+    ['güncelleme', ['update', 'upgrade']],
+    ['günlük', ['log']],
+    ['sunucu', ['server']],
+    ['veritabanı', ['database', 'db']],
+    ['ağ', ['network']],
+    ['bellek', ['memory', 'ram']],
+    ['önbellek', ['cache']],
+    ['hata ayıklama', ['debug', 'debugging']],
+    ['yeniden başlatma', ['restart', 'reboot']],
+    ['izin', ['permission', 'access']],
+    ['yetki', ['permission', 'access', 'acl']],
+    ['yedekleme', ['backup']],
+    ['performans', ['performance']],
+    ['sorun', ['problem', 'issue', 'error']],
+    ['servis', ['service']],
+    ['hizmet', ['service']],
+    ['bağlantı', ['connection', 'conn']],
+    ['dağıtım', ['deploy', 'deployment']],
+    ['dağıtmak', ['deploy']],
+    ['konteyner', ['container', 'docker']],
+    ['küme', ['cluster']],
+    ['düğüm', ['node']],
+    ['gizli', ['secret', 'key']],
+    ['anahtar', ['key', 'token', 'secret']],
+    ['şifre', ['password']],
+    ['izleme', ['monitor', 'monitoring']],
+    ['göç', ['migrate', 'migration']],
 ]);
 /**
  * Expand a token to its cross-lingual synonyms.
@@ -987,9 +1384,11 @@ function expandCrossLingual(token) {
  * Features:
  *  • CJK → character bigrams
  *  • Katakana → additionally emits Hepburn romaji tokens (enables romaji queries)
- *  • Arabic → word tokenization + iterative prefix stemming (al-, wa-, bi-, li-, fa-)
- *  • Hebrew → word tokenization + iterative prefix stemming (ha-, ve-, be-, le-, mi-, ke-)
- *  • All scripts → cross-lingual synonym expansion (EN↔JA↔ZH↔KO↔AR↔HE↔DE↔FR↔ES↔IT↔PT)
+ *  • Arabic  → word tokenization + iterative prefix stemming (ال/و/ب/ل/ف/ك)
+ *  • Hebrew  → word tokenization + iterative prefix stemming (ה/ו/ב/ל/מ/כ)
+ *  • Farsi   → word tokenization + suffix/prefix stemming (ها/های plural, می prefix)
+ *  • Hindi   → Devanagari word tokenization + light suffix stemming
+ *  • All scripts → cross-lingual synonym expansion (EN↔JA↔ZH↔KO↔AR↔HE↔FA↔HI↔DE↔FR↔ES↔IT↔PT)
  */
 function tokenize(text) {
     const tokens = [];
@@ -1028,26 +1427,65 @@ function tokenize(text) {
             }
         }
         else if (RTL_RE.test(seg)) {
-            // RTL segment — split on spaces, apply language-specific prefix stemming
-            // Arabic: strips ال/و/ب/ل/ف/ك prefixes iteratively
-            // Hebrew: strips ה/ו/ב/ל/מ/כ prefixes iteratively
-            const isHebrew = HEBREW_CHAR_RE.test(seg);
+            // RTL segment — 3-way branch: Farsi / Hebrew / Arabic
             const words = seg.trim().split(/\s+/);
+            let stemFn;
+            if (FARSI_CHAR_RE.test(seg))
+                stemFn = farsiLightStem;
+            else if (HEBREW_CHAR_RE.test(seg))
+                stemFn = hebrewLightStem;
+            else
+                stemFn = arabicLightStem;
             for (const w of words) {
                 if (w.length <= 1 || STOPWORDS.has(w))
                     continue;
-                const stemmed = isHebrew ? hebrewLightStem(w) : arabicLightStem(w);
+                const stemmed = stemFn(w);
+                if (stemmed.length > 1 && !STOPWORDS.has(stemmed))
+                    tokens.push(stemmed);
+            }
+        }
+        else if (DEVANAGARI_RE.test(seg)) {
+            // Devanagari segment (Hindi / Marathi) — naturally space-separated
+            // Words don't need bigram expansion; stems reduce inflected forms.
+            const words = seg.trim().split(/[\s\u200c\u200d]+/); // handle zero-width joiners
+            for (const w of words) {
+                // Keep Devanagari chars + digits; discard pure Latin portions (e.g. embedded numbers)
+                const deva = w.replace(/[^\u0900-\u097f0-9]/g, '');
+                if (deva.length <= 1 || STOPWORDS.has(deva))
+                    continue;
+                const stemmed = hindiLightStem(deva);
+                if (stemmed.length > 1 && !STOPWORDS.has(stemmed))
+                    tokens.push(stemmed);
+            }
+        }
+        else if (CYRILLIC_RE.test(seg)) {
+            // Cyrillic segment (Russian, Bulgarian, Ukrainian, Serbian …)
+            const words = seg.trim().split(/\s+/);
+            for (const w of words) {
+                const cyr = w.replace(/[^\u0400-\u04ff]/g, '');
+                if (cyr.length <= 1 || STOPWORDS.has(cyr))
+                    continue;
+                const stemmed = russianLightStem(cyr);
                 if (stemmed.length > 1 && !STOPWORDS.has(stemmed))
                     tokens.push(stemmed);
             }
         }
         else {
-            // Latin / other — existing logic preserved exactly
+            // Latin / other (includes Turkish with special chars ğ/ş/ı)
             const words = seg
-                .replace(/[^a-záàâãäåæçéèêëíìîïñóòôõöúùûüýÿßœ0-9:_\-./]/g, ' ')
+                .replace(/[^a-záàâãäåæçéèêëíìîïñóòôõöúùûüýÿßœğşıİ0-9:_\-./]/g, ' ')
                 .split(/\s+/)
                 .filter(w => w.length > 1 && !STOPWORDS.has(w));
-            tokens.push(...words);
+            // Turkish: apply light stemmer to words with Turkish-specific chars
+            for (const w of words) {
+                if (TURKISH_CHAR_RE.test(w)) {
+                    const stemmed = turkishLightStem(w);
+                    if (stemmed !== w && stemmed.length > 1 && !STOPWORDS.has(stemmed)) {
+                        tokens.push(stemmed);
+                    }
+                }
+                tokens.push(w);
+            }
         }
     }
     // ── Cross-lingual expansion ──────────────────────────────────────────────
@@ -1075,15 +1513,44 @@ function tokenize(text) {
                 }
             }
             else if (RTL_RE.test(syn)) {
-                // RTL synonym
-                const stemmed = arabicLightStem(syn);
+                // RTL synonym — apply correct stemmer per language
+                const rtlStem = FARSI_CHAR_RE.test(syn) ? farsiLightStem
+                    : HEBREW_CHAR_RE.test(syn) ? hebrewLightStem
+                        : arabicLightStem;
+                const stemmed = rtlStem(syn);
+                if (stemmed.length > 1 && !STOPWORDS.has(stemmed))
+                    expansions.push(stemmed);
+            }
+            else if (DEVANAGARI_RE.test(syn)) {
+                // Devanagari synonym (Hindi)
+                const stemmed = hindiLightStem(syn);
+                if (stemmed.length > 1 && !STOPWORDS.has(stemmed))
+                    expansions.push(stemmed);
+            }
+            else if (CYRILLIC_RE.test(syn)) {
+                // Cyrillic synonym (Russian)
+                const cyr = syn.replace(/[^\u0400-\u04ff]/g, '');
+                const stemmed = russianLightStem(cyr);
                 if (stemmed.length > 1 && !STOPWORDS.has(stemmed))
                     expansions.push(stemmed);
             }
             else {
-                // Latin synonym — split in case of multi-word (e.g. "load balancer")
-                const parts = syn.split(/\s+/).filter(w => w.length > 1 && !STOPWORDS.has(w));
-                expansions.push(...parts);
+                // Latin synonym (includes Turkish) — split in case of multi-word
+                const parts = syn.split(/\s+/).filter(w => {
+                    if (w.length <= 1 || STOPWORDS.has(w))
+                        return false;
+                    return true;
+                });
+                for (const p of parts) {
+                    expansions.push(p);
+                    // Apply Turkish stemmer if it has Turkish-specific characters
+                    if (TURKISH_CHAR_RE.test(p)) {
+                        const stemmed = turkishLightStem(p);
+                        if (stemmed !== p && stemmed.length > 1 && !STOPWORDS.has(stemmed)) {
+                            expansions.push(stemmed);
+                        }
+                    }
+                }
             }
         }
     }
@@ -1109,8 +1576,8 @@ function splitMultiQuery(query) {
     const semiParts = query.split(/[;\n]+/).filter(s => s.trim().length > 2);
     if (semiParts.length >= 2)
         return semiParts.map(s => s.trim());
-    // Conjunctions (EN + DE + FR + ES)
-    const conjParts = query.split(/\b(?:and also|also noch|außerdem|plus|additionally|furthermore|de plus|además|inoltre|além disso)\b/i).filter(s => s.trim().length > 2);
+    // Conjunctions (EN + DE + FR + ES + IT + PT + RU + TR)
+    const conjParts = query.split(/\b(?:and also|also noch|außerdem|plus|additionally|furthermore|de plus|además|inoltre|além disso|а также|и также|кроме того|плюс|ayrıca|bunun yanı sıra|همچنین|وأيضاً|وكذلك|وهمچنین|और भी|इसके अलावा)\b/i).filter(s => s.trim().length > 2);
     if (conjParts.length >= 2)
         return conjParts.map(s => s.trim());
     // Comma-separated with 3+ parts (likely a list)
@@ -1340,6 +1807,11 @@ async function keywordSearch(redis, patterns, query, topK = 10) {
                     score *= 1 + 0.5 * (bigramHits / queryBigrams.size);
                 }
             }
+            // Phrase-match boost: if the raw (lowercased) query appears verbatim in content, 2× boost
+            // This rewards docs where the exact phrase exists (vs scattered tokens).
+            if (sq.length >= 4 && doc.content.toLowerCase().includes(sq.toLowerCase())) {
+                score *= 2.0;
+            }
             // Recency boost
             score *= recencyBoost(doc.timestamp);
             if (score > 0) {