@cachly-dev/mcp-server 0.5.50 → 0.5.70

package/src/index.ts

@@ -55,7 +55,7 @@ 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 = '0.5.50';
+const CURRENT_VERSION = '0.5.70';
 /**
  * ┌──────────────────────────────────────────────────────────────────────┐
  * │  EMBEDDING PROVIDER — pluggable, client-side first                  │
@@ -434,6 +434,40 @@ const STOPWORDS = new Set([
   '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',
+  // ── Polish — common prepositions, pronouns, particles ─────────────────────
+  'w', 'na', 'z', 'do', 'się', 'że', 'to', 'jest', 'i', 'a', 'nie',
+  'tak', 'jak', 'czy', 'po', 'o', 'ale', 'go', 'mu', 'jej', 'ich',
+  'tego', 'tej', 'te', 'ten', 'ta', 'są', 'był', 'była', 'było', 'byli',
+  'będzie', 'będą', 'ma', 'mam', 'masz', 'mają', 'ze', 'co', 'już',
+  'przez', 'przy', 'za', 'bez', 'nad', 'pod', 'przed', 'po', 'między',
+  'kiedy', 'gdzie', 'dlaczego', 'który', 'która', 'które', 'tylko', 'też',
+  // ── Czech — common words ──────────────────────────────────────────────────
+  'v', 'na', 'z', 'do', 'se', 'že', 'to', 'je', 'a', 'ne', 'pro',
+  'ale', 'jak', 'by', 'byl', 'být', 'jsem', 'jsou', 'má', 'mám',
+  'ten', 'ta', 'ty', 'tato', 'jeho', 'její', 'jejich', 'také', 'jen',
+  'kde', 'kdy', 'proč', 'který', 'která', 'které', 'co', 'při', 'bez',
+  // ── Bengali — common particles, pronouns, verbs ───────────────────────────
+  'এই', 'এটি', 'এটা', 'এর', 'তার', 'তারা', 'আমি', 'তুমি', 'সে', 'আমরা',
+  'এবং', 'কিন্তু', 'বা', 'না', 'হ্যাঁ', 'যে', 'কি', 'কে', 'কী', 'থেকে',
+  'দিয়ে', 'জন্য', 'সাথে', 'মধ্যে', 'উপর', 'নিচে', 'আছে', 'ছিল', 'হয়',
+  'করা', 'করে', 'করেছে', 'হবে', 'হয়েছে', 'একটি', 'একটা', 'অনেক', 'সব',
 ]);
 
 
@@ -450,6 +484,12 @@ const ARABIC_CHAR_RE  = /[\u0600-\u06ff\u0750-\u077f\u08a0-\u08ff\ufb50-\ufdff\u
 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
+// Bengali (Bangla) — U+0980–U+09FF
+const BENGALI_RE      = /[\u0980-\u09ff]/;
 
 // ── Katakana → Romaji conversion table (Hepburn system) ──────────────────────
 // Digraphs must be listed before single chars so they match first.
@@ -611,10 +651,155 @@ function hindiLightStem(word: string): string {
   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\u0902 (īyāṃ)
+  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: string): string {
+  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: string): string {
+  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;
+}
+
+/**
+ * Bengali (Bangla) light stemmer — strips common verbal and nominal suffixes.
+ * Bengali is an Indo-Aryan language with moderate morphological complexity.
+ * Only strips when result is still ≥ 2 chars.
+ *
+ * Covers:
+ *  করছে/করেছে → কর  (progressive/perfect aspect)
+ *  সার্ভারগুলো → সার্ভার  (-গুলো plural)
+ *  সমস্যাটি → সমস্যা  (-টি singular marker)
+ */
+function bengaliLightStem(word: string): string {
+  let w = word;
+  // Plural / collective markers (longest first)
+  if (w.endsWith('\u0997\u09c1\u09b2\u09cb') && w.length > 5) return w.slice(0, -4); // গুলো (-gulo plural)
+  if (w.endsWith('\u0997\u09c1\u09b2\u09bf') && w.length > 5) return w.slice(0, -4); // গুলি (-guli plural)
+  if (w.endsWith('\u09a6\u09c7\u09b0') && w.length > 4) return w.slice(0, -3); // দের (genitive plural)
+  if (w.endsWith('\u09a6\u09bf\u0997\u09c7') && w.length > 5) return w.slice(0, -4); // দিগে
+  // Definiteness / case markers
+  if (w.endsWith('\u099f\u09bf') && w.length > 3) return w.slice(0, -2); // টি (singular definite)
+  if (w.endsWith('\u099f\u09be') && w.length > 3) return w.slice(0, -2); // টা (singular definite)
+  if (w.endsWith('\u0996\u09be\u09a8\u09be') && w.length > 5) return w.slice(0, -4); // খানা
+  // Verbal suffixes
+  if (w.endsWith('\u099b\u09c7') && w.length > 3) return w.slice(0, -2); // ছে (progressive)
+  if (w.endsWith('\u099b\u09bf\u09b2') && w.length > 4) return w.slice(0, -3); // ছিল (past progressive)
+  if (w.endsWith('\u09af\u09be\u09ac\u09c7') && w.length > 5) return w.slice(0, -4); // যাবে (future)
+  if (w.endsWith('\u0995\u09b0\u09be') && w.length > 4) return w.slice(0, -3); // করা (infinitive)
+  if (w.endsWith('\u0995\u09b0\u09c7') && w.length > 4) return w.slice(0, -3); // করে (present)
+  if (w.endsWith('\u09b9\u09df') && w.length > 3) return w.slice(0, -2); // হয় (is/becomes)
+  if (w.endsWith('\u09b9\u09ac\u09c7') && w.length > 4) return w.slice(0, -3); // হবে (will be)
+  return w;
+}
+
+/**
+ * Pre-process text BEFORE tokenization to split code identifiers.
+ * Handles camelCase, PascalCase, snake_case, and kebab-case.
+ *
+ * Examples:
+ *   deployServer       → deploy Server  (then lowercased → deploy server)
+ *   HTMLParser         → HTML Parser
+ *   my_api_key         → my api key
+ *   get-user-by-id     → get user by id (hyphens preserved in regex, spaces also fine)
+ *   BackendServiceImpl → Backend Service Impl
+ */
+function preprocessText(text: string): string {
+  return text
+    // camelCase boundary: lowercase/digit → uppercase
+    .replace(/([a-z0-9])([A-Z])/g, '$1 $2')
+    // Consecutive uppercase → uppercase+lowercase boundary: HTMLParser → HTML Parser
+    .replace(/([A-Z]{2,})([A-Z][a-z])/g, '$1 $2')
+    // snake_case: replace underscores with spaces
+    .replace(/_/g, ' ');
+}
+
 
 // Maps tokens (English, Japanese katakana, Chinese, Korean, Arabic, Hebrew)
 // to their equivalents in other languages. Used to expand tokens at search time
@@ -622,46 +807,104 @@ function hindiLightStem(word: string): string {
 //
 // Keys must be lowercase. CJK values will be bigram-expanded at use time.
 const CROSS_LINGUAL_MAP = new Map<string, string[]>([
+
+  // ── Tech / Ecosystem synonyms (language-agnostic) ─────────────────────────
+  ['javascript',    ['js', 'node', 'nodejs', 'ecmascript']],
+  ['js',            ['javascript', 'node', 'nodejs']],
+  ['typescript',    ['ts', 'javascript', 'js', 'tsc']],
+  ['ts',            ['typescript', 'javascript', 'js']],
+  ['python',        ['py', 'pip', 'django', 'flask', 'fastapi']],
+  ['py',            ['python', 'pip']],
+  ['golang',        ['go', 'gopher', 'goroutine']],
+  ['go',            ['golang', 'goroutine']],
+  ['rust',          ['cargo', 'rustlang', 'crates']],
+  ['cargo',         ['rust', 'rustlang']],
+  ['java',          ['jvm', 'maven', 'gradle', 'spring']],
+  ['kotlin',        ['jvm', 'android', 'coroutine']],
+  ['dotnet',        ['csharp', 'asp', 'nuget', 'aspnet']],
+  ['postgresql',    ['postgres', 'pg', 'psql']],
+  ['postgres',      ['postgresql', 'pg', 'psql']],
+  ['mysql',         ['mariadb', 'sql']],
+  ['mongodb',       ['mongo', 'nosql', 'bson', 'mongoose']],
+  ['mongo',         ['mongodb', 'nosql', 'bson']],
+  ['elasticsearch', ['elastic', 'opensearch', 'lucene', 'kibana']],
+  ['elastic',       ['elasticsearch', 'opensearch']],
+  ['kubernetes',    ['k8s', 'kube', 'kubectl', 'helm', 'k3s']],
+  ['k8s',           ['kubernetes', 'kube', 'kubectl', 'helm']],
+  ['helm',          ['kubernetes', 'k8s', 'chart']],
+  ['terraform',     ['tf', 'hcl', 'iac', 'opentofu']],
+  ['ansible',       ['playbook', 'automation']],
+  ['grafana',       ['dashboard', 'visualization', 'metrics']],
+  ['prometheus',    ['metrics', 'alert', 'scrape', 'grafana']],
+  ['nginx',         ['proxy', 'webserver', 'ingress']],
+  ['traefik',       ['proxy', 'ingress', 'router']],
+  ['rabbitmq',      ['amqp', 'queue', 'broker', 'messaging']],
+  ['kafka',         ['messaging', 'stream', 'broker']],
+  ['grpc',          ['protobuf', 'proto', 'rpc']],
+  ['graphql',       ['gql', 'resolver', 'schema']],
+  ['react',         ['jsx', 'hooks', 'component', 'redux']],
+  ['nextjs',        ['next', 'react', 'ssr', 'vercel']],
+  ['vue',           ['vuejs', 'vite', 'nuxt']],
+  ['angular',       ['ng', 'rxjs', 'typescript']],
+  ['aws',           ['amazon', 'ec2', 's3', 'lambda', 'cloudwatch', 'ecs', 'eks']],
+  ['gcp',           ['google cloud', 'gke', 'bigquery']],
+  ['azure',         ['microsoft cloud', 'aks', 'devops']],
+  ['s3',            ['bucket', 'object storage', 'aws']],
+  ['lambda',        ['serverless', 'function', 'faas', 'aws']],
+  ['jwt',           ['token', 'bearer', 'oauth', 'auth']],
+  ['oauth',         ['oidc', 'auth', 'token', 'sso']],
+  ['webhook',       ['callback', 'event', 'trigger']],
+  ['cron',          ['schedule', 'job', 'timer']],
+  ['yaml',          ['yml', 'config', 'manifest']],
+  ['dotenv',        ['env', 'environment', 'envfile']],
+  ['github',        ['git', 'actions', 'repo', 'ci']],
+  ['gitlab',        ['git', 'pipeline', 'runner']],
+  ['wireguard',     ['vpn', 'tunnel', 'wg']],
+  ['prisma',        ['orm', 'database', 'schema', 'migration']],
+  ['npm',           ['node', 'package', 'registry', 'yarn', 'pnpm']],
+  ['yarn',          ['npm', 'package', 'node']],
+  ['pip',           ['python', 'package', 'pypi']],
+  ['brew',          ['homebrew', 'macos', 'package']],
   // English → all
-  ['deploy',      ['デプロイ', '部署', '배포', 'deployment', 'deploying', 'نشر', 'פריסה', 'استقرار', 'तैनाती', 'bereitstellen', 'bereitstellung', 'déployer', 'déploiement', 'desplegar', 'despliegue', 'distribuire', 'distribuzione', 'implantar', 'implantação']],
-  ['deployment',  ['デプロイ', '部署', '배포', 'deploy', 'نشر', 'פריסה', 'استقرار', 'तैनाती', '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', 'سرور', 'सर्वर']],
+  ['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']],
-  ['error',       ['エラー', '错误', '오류', 'خطأ', 'شגיאה', 'خطا', 'त्रुटि', 'गलती', 'exception', 'err', 'fehler', 'erreur', 'fallo', 'errore', 'erro']],
+  ['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']],
-  ['build',       ['ビルド', '构建', '빌드', 'بناء', 'בנייה', 'ساخت', 'निर्माण', 'bauen', 'construire', 'construir', 'costruire']],
-  ['test',        ['テスト', '测试', '테스트', 'اختبار', 'בדיקה', 'آزمایش', 'परीक्षण', 'testen', 'tester', 'probar', 'testare', 'testar']],
-  ['auth',        ['認証', '认证', '인증', 'authentication', 'login', 'oauth', 'مصادقة', 'אימות', 'احراز هویت', 'प्रमाणीकरण', 'authentifizierung', 'authentification', 'autenticación', 'autenticazione', 'autenticação']],
-  ['authentication', ['auth', 'مصادقة', 'אימות', 'احراز هویت', 'प्रमाणीकरण', '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']],
+  ['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', 'شبکه', 'नेटवर्क', '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', 'حافظه', 'मेमोरी', 'speicher', 'mémoire', 'memoria', 'memória']],
-  ['config',      ['設定', '配置', '설정', 'configuration', 'settings', 'conf', 'إعدادات', 'הגדרות', 'پیکربندی', 'कॉन्फ़िग', 'konfiguration', 'einstellungen', 'configuración', 'configurazione', 'configuração']],
-  ['install',     ['インストール', '安装', '설치', 'تثبيت', 'התקנה', 'نصب', 'इंस्टॉल', 'installieren', 'installer', 'instalar', 'installare']],
-  ['update',      ['アップデート', '更新', '업데이트', 'upgrade', 'تحديث', 'עדכון', 'به‌روزرسانی', 'अपडेट', 'aktualisieren', 'aktualisierung', 'actualizar', 'aggiornare', 'atualizar']],
-  ['log',         ['ログ', '日志', '로그', 'logging', 'سجل', 'יומן', 'ثبت', 'लॉग', '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', 'پشتیبان‌گیری', 'बैकअप', 'sicherung', 'sauvegarde', 'respaldo', 'cópia de segurança']],
-  ['monitor',     ['モニター', '监控', '모니터링', 'monitoring', 'metrics', 'alert', 'مراقبة', 'ניטור', 'نظارت', 'निगरानी', 'überwachung', 'surveiller', 'monitoreo', 'monitorare', 'monitorar']],
-  ['performance', ['パフォーマンス', '性能', '성능', 'latency', 'throughput', 'عملکرد', 'प्रदर्शन', 'leistung', 'performances', 'rendimiento', 'prestazioni', 'desempenho']],
+  ['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']],
@@ -676,10 +919,10 @@ const CROSS_LINGUAL_MAP = new Map<string, string[]>([
   ['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', 'اشکال‌زدایی', 'डीबग', 'debuggen', 'déboguer', 'depurar']],
-  ['crash',       ['クラッシュ', '崩溃', '크래시', 'panic', 'segfault', 'انهيار', 'קריסה', 'خرابی', 'क्रैश', 'absturz', 'panne', 'plantage', 'caída', 'falha']],
+  ['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']],
+  ['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']],
@@ -1114,6 +1357,105 @@ const CROSS_LINGUAL_MAP = new Map<string, string[]>([
   ['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']],
 ]);
 
 /**
@@ -1140,7 +1482,8 @@ function expandCrossLingual(token: string): string[] {
  */
 function tokenize(text: string): string[] {
   const tokens: string[] = [];
-  const lower = text.toLowerCase();
+  // Pre-process: split camelCase/PascalCase identifiers and snake_case before lowercasing
+  const lower = preprocessText(text).toLowerCase();
 
   for (const [seg] of lower.matchAll(SEGMENT_RE)) {
     if (CJK_RE.test(seg)) {
@@ -1187,22 +1530,48 @@ function tokenize(text: string): string[] {
       }
     } 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 (BENGALI_RE.test(seg)) {
+      // Bengali (Bangla) segment — space-separated words, Bengali script
+      const words = seg.trim().split(/[\s\u200c\u200d]+/);
+      for (const w of words) {
+        const bn = w.replace(/[^\u0980-\u09ff0-9]/g, '');
+        if (bn.length <= 1 || STOPWORDS.has(bn)) continue;
+        const stemmed = bengaliLightStem(bn);
+        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 (ğ/ş/ı), Polish (ą/ę/ł/ń/ó/ś/ź/ż/ć),
+      // Czech (č/š/ž/ě/ů/ř), Hungarian (ő/ű), Romanian (ș/ț)
       const words = seg
-        .replace(/[^a-záàâãäåæçéèêëíìîïñóòôõöúùûüýÿßœ0-9:_\-./]/g, ' ')
+        .replace(/[^a-záàâãäåæçéèêëíìîïñóòôõöúùûüýÿßœğşıİąćęłńśźżčšžěůřőűșț0-9:\-./]/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);
+      }
     }
   }
 
@@ -1237,10 +1606,31 @@ function tokenize(text: string): string[] {
         // Devanagari synonym (Hindi)
         const stemmed = hindiLightStem(syn);
         if (stemmed.length > 1 && !STOPWORDS.has(stemmed)) expansions.push(stemmed);
+      } else if (BENGALI_RE.test(syn)) {
+        // Bengali synonym
+        const stemmed = bengaliLightStem(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);
+            }
+          }
+        }
       }
     }
   }
@@ -1268,9 +1658,9 @@ function splitMultiQuery(query: string): string[] {
   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)
+  // 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)\b/i
+    /\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());
 
@@ -1308,6 +1698,7 @@ interface DocEntry {
   tokens: string[];
   tokenFreq: Map<string, number>;  // term → count in this doc
   bigrams: Set<string>;            // "term1|term2" adjacency pairs
+  keyTokens: Set<string>;          // tokens from key/title only (for title boost)
   timestamp?: number;              // epoch ms, extracted from content if present
 }
 
@@ -1427,7 +1818,9 @@ async function keywordSearch(
     }
 
     const timestamp = extractTimestamp(content);
-    docs.push({ key: allKeys[i], content, tokens, tokenFreq, bigrams, timestamp });
+    // Key tokens for title-boost: terms appearing in the Redis key get extra weight
+    const keyTokens = new Set(tokenize(allKeys[i]));
+    docs.push({ key: allKeys[i], content, tokens, tokenFreq, bigrams, keyTokens, timestamp });
     totalTokens += tokens.length;
   }
 
@@ -1535,6 +1928,24 @@ async function keywordSearch(
         }
       }
 
+      // 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;
+      }
+
+      // Key/title boost: if query terms appear in the Redis key (=title), 1.5× boost
+      // Keys often encode the primary topic (e.g. "deploy:api:server"), so key hits are high-precision.
+      if (score > 0) {
+        let keyHits = 0;
+        for (const qt of queryTokens) {
+          if (doc.keyTokens.has(qt)) keyHits++;
+        }
+        if (keyHits > 0) {
+          score *= 1 + 0.5 * (keyHits / queryTokens.length);
+        }
+      }
+
       // Recency boost
       score *= recencyBoost(doc.timestamp);