@cachly-dev/mcp-server 0.5.43 → 0.5.45

package/dist/__tests__/namespace.test.js

@@ -0,0 +1,114 @@
+import { describe, it, expect } from 'vitest';
+import { detectNamespace } from '../namespace.js';
+// ── Namespace Auto-Detection ───────────────────────────────────────────────
+describe('detectNamespace', () => {
+    // ── Code ────────────────────────────────────────────────────────────────────
+    describe('code detection', () => {
+        it.each([
+            ['python def', 'def process(data): return data * 2'],
+            ['js const arrow', 'const handler = () => response.send(200)'],
+            ['ts class', 'class UserService { constructor(private repo: Repo) {} }'],
+            ['import statement', 'import { useState } from "react"'],
+            ['shebang line', '#!/usr/bin/env python3\nprint("hello")'],
+            ['go func', 'func main() { fmt.Println("hello") }'],
+            ['cpp include', '#include <iostream>\nint main() { return 0; }'],
+            ['interface block', 'interface ICache { get(key: string): void; }'],
+            ['struct block', 'struct Config { host string }'],
+            ['async def', 'async def fetch_data(url: str) -> dict:'],
+            ['lambda', 'fn = lambda x: x * 2'],
+            ['package decl', 'package com.example.service;'],
+        ])('detects code: %s', (_, prompt) => {
+            expect(detectNamespace(prompt)).toBe('cachly:sem:code');
+        });
+    });
+    // ── Translation ─────────────────────────────────────────────────────────────
+    describe('translation detection', () => {
+        it.each([
+            ['translate', 'translate this paragraph to Spanish'],
+            ['übersetze', 'übersetze diesen Text bitte'],
+            ['auf deutsch', 'Schreib das auf deutsch'],
+            ['in english', 'Please write this in english'],
+            ['ins deutsche', 'Bitte übersetze ins Deutsche'],
+            ['traduce', 'traduce este texto al alemán'],
+            ['traduis', 'traduis ce texte en anglais'],
+            ['vertaal', 'vertaal dit naar het Nederlands'],
+        ])('detects translation: %s', (_, prompt) => {
+            expect(detectNamespace(prompt)).toBe('cachly:sem:translation');
+        });
+    });
+    // ── Summary ──────────────────────────────────────────────────────────────────
+    describe('summary detection', () => {
+        it.each([
+            ['summarize', 'summarize this article for me'],
+            ['summarise', 'summarise the key points'],
+            ['tl;dr', 'tl;dr of the following blog post:'],
+            ['tldr', 'tldr of this document'],
+            ['key points', 'what are the key points?'],
+            ['zusammenfass', 'fasse zusammen worum es geht'],
+            ['give me a brief', 'give me a brief overview'],
+            ['in a nutshell', 'explain machine learning in a nutshell'],
+            ['stichpunkte', 'Gib mir die wichtigsten Stichpunkte'],
+        ])('detects summary: %s', (_, prompt) => {
+            expect(detectNamespace(prompt)).toBe('cachly:sem:summary');
+        });
+    });
+    // ── Q&A ──────────────────────────────────────────────────────────────────────
+    describe('qa detection', () => {
+        it.each([
+            ['what is', 'what is the capital of France?'],
+            ['how does', 'how does photosynthesis work?'],
+            ['why is', 'why is the sky blue?'],
+            ['who invented', 'who invented the internet?'],
+            ['where is', 'where is Mount Everest located?'],
+            ['when was', 'when was the Eiffel Tower built?'],
+            ['can you', 'can you explain recursion?'],
+            ['is redis', 'is Redis single-threaded?'],
+            ['does postgres', 'does postgres support JSON natively?'],
+            ['wer ist', 'wer ist der aktuelle Bundeskanzler?'],
+            ['wie funktioniert', 'wie funktioniert ein JWT Token?'],
+            ['trailing ?', 'Redis vs Memcached?'],
+        ])('detects q&a: %s', (_, prompt) => {
+            expect(detectNamespace(prompt)).toBe('cachly:sem:qa');
+        });
+    });
+    // ── Creative ─────────────────────────────────────────────────────────────────
+    describe('creative fallback', () => {
+        it.each([
+            ['poem request', 'Write a short poem about autumn leaves'],
+            ['story request', 'Tell me a fantasy story about dragons'],
+            ['product copy', 'Generate a product description for running shoes'],
+            ['greeting', 'Good morning, how are you today'],
+            ['general task', 'Help me brainstorm names for my startup'],
+        ])('falls back to creative: %s', (_, prompt) => {
+            expect(detectNamespace(prompt)).toBe('cachly:sem:creative');
+        });
+    });
+    // ── Edge cases ────────────────────────────────────────────────────────────────
+    describe('edge cases', () => {
+        it('is case-insensitive', () => {
+            expect(detectNamespace('CONST x = 1')).toBe('cachly:sem:code');
+            expect(detectNamespace('TRANSLATE THIS')).toBe('cachly:sem:translation');
+            expect(detectNamespace('SUMMARIZE PLEASE')).toBe('cachly:sem:summary');
+            expect(detectNamespace('WHAT IS GRAVITY?')).toBe('cachly:sem:qa');
+        });
+        it('trims surrounding whitespace', () => {
+            expect(detectNamespace('   what is Redis?   ')).toBe('cachly:sem:qa');
+            expect(detectNamespace('  const x = 1  ')).toBe('cachly:sem:code');
+        });
+        it('code keyword takes priority over translation', () => {
+            // Both "translate" and "def" present – code wins (checked first)
+            expect(detectNamespace('translate this function: def foo(): pass')).toBe('cachly:sem:code');
+        });
+        it('code keyword takes priority over summary', () => {
+            expect(detectNamespace('summarize this class MyService { }'))
+                .toBe('cachly:sem:code');
+        });
+        it('handles empty string as creative', () => {
+            expect(detectNamespace('')).toBe('cachly:sem:creative');
+        });
+        it('a single ? at end is qa', () => {
+            expect(detectNamespace('Interesting concept?')).toBe('cachly:sem:qa');
+        });
+    });
+});
+//# sourceMappingURL=namespace.test.js.map

package/dist/__tests__/namespace.test.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"namespace.test.js","sourceRoot":"","sources":["../../src/__tests__/namespace.test.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,EAAE,EAAE,MAAM,EAAE,MAAM,QAAQ,CAAC;AAC9C,OAAO,EAAE,eAAe,EAAE,MAAM,iBAAiB,CAAC;AAElD,8EAA8E;AAE9E,QAAQ,CAAC,iBAAiB,EAAE,GAAG,EAAE;IAC/B,+EAA+E;IAC/E,QAAQ,CAAC,gBAAgB,EAAE,GAAG,EAAE;QAC9B,EAAE,CAAC,IAAI,CAAC;YACN,CAAC,YAAY,EAAS,oCAAoC,CAAC;YAC3D,CAAC,gBAAgB,EAAK,0CAA0C,CAAC;YACjE,CAAC,UAAU,EAAW,0DAA0D,CAAC;YACjF,CAAC,kBAAkB,EAAG,kCAAkC,CAAC;YACzD,CAAC,cAAc,EAAO,wCAAwC,CAAC;YAC/D,CAAC,SAAS,EAAY,sCAAsC,CAAC;YAC7D,CAAC,aAAa,EAAQ,+CAA+C,CAAC;YACtE,CAAC,iBAAiB,EAAI,8CAA8C,CAAC;YACrE,CAAC,cAAc,EAAO,+BAA+B,CAAC;YACtD,CAAC,WAAW,EAAU,yCAAyC,CAAC;YAChE,CAAC,QAAQ,EAAa,sBAAsB,CAAC;YAC7C,CAAC,cAAc,EAAO,8BAA8B,CAAC;SACtD,CAAC,CAAC,kBAAkB,EAAE,CAAC,CAAC,EAAE,MAAM,EAAE,EAAE;YACnC,MAAM,CAAC,eAAe,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,iBAAiB,CAAC,CAAC;QAC1D,CAAC,CAAC,CAAC;IACL,CAAC,CAAC,CAAC;IAEH,+EAA+E;IAC/E,QAAQ,CAAC,uBAAuB,EAAE,GAAG,EAAE;QACrC,EAAE,CAAC,IAAI,CAAC;YACN,CAAC,WAAW,EAAU,qCAAqC,CAAC;YAC5D,CAAC,WAAW,EAAU,6BAA6B,CAAC;YACpD,CAAC,aAAa,EAAQ,yBAAyB,CAAC;YAChD,CAAC,YAAY,EAAS,8BAA8B,CAAC;YACrD,CAAC,cAAc,EAAO,8BAA8B,CAAC;YACrD,CAAC,SAAS,EAAY,8BAA8B,CAAC;YACrD,CAAC,SAAS,EAAY,6BAA6B,CAAC;YACpD,CAAC,SAAS,EAAY,iCAAiC,CAAC;SACzD,CAAC,CAAC,yBAAyB,EAAE,CAAC,CAAC,EAAE,MAAM,EAAE,EAAE;YAC1C,MAAM,CAAC,eAAe,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,wBAAwB,CAAC,CAAC;QACjE,CAAC,CAAC,CAAC;IACL,CAAC,CAAC,CAAC;IAEH,gFAAgF;IAChF,QAAQ,CAAC,mBAAmB,EAAE,GAAG,EAAE;QACjC,EAAE,CAAC,IAAI,CAAC;YACN,CAAC,WAAW,EAAU,+BAA+B,CAAC;YACtD,CAAC,WAAW,EAAU,0BAA0B,CAAC;YACjD,CAAC,OAAO,EAAc,mCAAmC,CAAC;YAC1D,CAAC,MAAM,EAAe,uBAAuB,CAAC;YAC9C,CAAC,YAAY,EAAS,0BAA0B,CAAC;YACjD,CAAC,cAAc,EAAO,8BAA8B,CAAC;YACrD,CAAC,iBAAiB,EAAI,0BAA0B,CAAC;YACjD,CAAC,eAAe,EAAM,wCAAwC,CAAC;YAC/D,CAAC,aAAa,EAAQ,qCAAqC,CAAC;SAC7D,CAAC,CAAC,qBAAqB,EAAE,CAAC,CAAC,EAAE,MAAM,EAAE,EAAE;YACtC,MAAM,CAAC,eAAe,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,oBAAoB,CAAC,CAAC;QAC7D,CAAC,CAAC,CAAC;IACL,CAAC,CAAC,CAAC;IAEH,gFAAgF;IAChF,QAAQ,CAAC,cAAc,EAAE,GAAG,EAAE;QAC5B,EAAE,CAAC,IAAI,CAAC;YACN,CAAC,SAAS,EAAY,gCAAgC,CAAC;YACvD,CAAC,UAAU,EAAW,+BAA+B,CAAC;YACtD,CAAC,QAAQ,EAAa,sBAAsB,CAAC;YAC7C,CAAC,cAAc,EAAO,4BAA4B,CAAC;YACnD,CAAC,UAAU,EAAW,iCAAiC,CAAC;YACxD,CAAC,UAAU,EAAW,kCAAkC,CAAC;YACzD,CAAC,SAAS,EAAY,4BAA4B,CAAC;YACnD,CAAC,UAAU,EAAW,2BAA2B,CAAC;YAClD,CAAC,eAAe,EAAM,sCAAsC,CAAC;YAC7D,CAAC,SAAS,EAAY,qCAAqC,CAAC;YAC5D,CAAC,kBAAkB,EAAG,iCAAiC,CAAC;YACxD,CAAC,YAAY,EAAS,qBAAqB,CAAC;SAC7C,CAAC,CAAC,iBAAiB,EAAE,CAAC,CAAC,EAAE,MAAM,EAAE,EAAE;YAClC,MAAM,CAAC,eAAe,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC;QACxD,CAAC,CAAC,CAAC;IACL,CAAC,CAAC,CAAC;IAEH,gFAAgF;IAChF,QAAQ,CAAC,mBAAmB,EAAE,GAAG,EAAE;QACjC,EAAE,CAAC,IAAI,CAAC;YACN,CAAC,cAAc,EAAM,wCAAwC,CAAC;YAC9D,CAAC,eAAe,EAAK,uCAAuC,CAAC;YAC7D,CAAC,cAAc,EAAM,kDAAkD,CAAC;YACxE,CAAC,UAAU,EAAU,iCAAiC,CAAC;YACvD,CAAC,cAAc,EAAM,yCAAyC,CAAC;SAChE,CAAC,CAAC,4BAA4B,EAAE,CAAC,CAAC,EAAE,MAAM,EAAE,EAAE;YAC7C,MAAM,CAAC,eAAe,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAC;QAC9D,CAAC,CAAC,CAAC;IACL,CAAC,CAAC,CAAC;IAEH,iFAAiF;IACjF,QAAQ,CAAC,YAAY,EAAE,GAAG,EAAE;QAC1B,EAAE,CAAC,qBAAqB,EAAE,GAAG,EAAE;YAC7B,MAAM,CAAC,eAAe,CAAC,aAAa,CAAC,CAAC,CAAC,IAAI,CAAC,iBAAiB,CAAC,CAAC;YAC/D,MAAM,CAAC,eAAe,CAAC,gBAAgB,CAAC,CAAC,CAAC,IAAI,CAAC,wBAAwB,CAAC,CAAC;YACzE,MAAM,CAAC,eAAe,CAAC,kBAAkB,CAAC,CAAC,CAAC,IAAI,CAAC,oBAAoB,CAAC,CAAC;YACvE,MAAM,CAAC,eAAe,CAAC,kBAAkB,CAAC,CAAC,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC;QACpE,CAAC,CAAC,CAAC;QAEH,EAAE,CAAC,8BAA8B,EAAE,GAAG,EAAE;YACtC,MAAM,CAAC,eAAe,CAAC,sBAAsB,CAAC,CAAC,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC;YACtE,MAAM,CAAC,eAAe,CAAC,iBAAiB,CAAC,CAAC,CAAC,IAAI,CAAC,iBAAiB,CAAC,CAAC;QACrE,CAAC,CAAC,CAAC;QAEH,EAAE,CAAC,8CAA8C,EAAE,GAAG,EAAE;YACtD,iEAAiE;YACjE,MAAM,CAAC,eAAe,CAAC,0CAA0C,CAAC,CAAC,CAAC,IAAI,CAAC,iBAAiB,CAAC,CAAC;QAC9F,CAAC,CAAC,CAAC;QAEH,EAAE,CAAC,0CAA0C,EAAE,GAAG,EAAE;YAClD,MAAM,CAAC,eAAe,CAAC,oCAAoC,CAAC,CAAC;iBAC1D,IAAI,CAAC,iBAAiB,CAAC,CAAC;QAC7B,CAAC,CAAC,CAAC;QAEH,EAAE,CAAC,kCAAkC,EAAE,GAAG,EAAE;YAC1C,MAAM,CAAC,eAAe,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAC;QAC1D,CAAC,CAAC,CAAC;QAEH,EAAE,CAAC,yBAAyB,EAAE,GAAG,EAAE;YACjC,MAAM,CAAC,eAAe,CAAC,sBAAsB,CAAC,CAAC,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC;QACxE,CAAC,CAAC,CAAC;IACL,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC"}

package/dist/index.d.ts

@@ -1,11 +1,39 @@
 #!/usr/bin/env node
 /**
  * Stopwords — filtered out during tokenization.
- * Covers: English, German, French, Spanish, Italian, Portuguese.
+ * Covers: English, German, French, Spanish, Italian, Portuguese,
+ *         Chinese (Simplified + Traditional), Japanese, Korean.
  * Keeps the index small and scores meaningful.
  */
 declare const STOPWORDS: Set<string>;
-/** Tokenize text into meaningful keywords. Handles unicode (accented chars). */
+/**
+ * Convert a katakana string to Hepburn romaji.
+ * Handles digraphs, geminate consonants (ッ), and long vowel marks (ー).
+ */
+declare function katakanaToRomaji(kata: string): string;
+/**
+ * Arabic light stemmer — strips definite article and common prefix particles.
+ * Handles: ال (al-), و (wa-), ب (bi-), ل (li-), ف (fa-), ك (ka-).
+ * Only strips when the result is still ≥ 3 chars to avoid over-stemming.
+ */
+declare function arabicLightStem(word: string): string;
+declare const CROSS_LINGUAL_MAP: Map<string, string[]>;
+/**
+ * Expand a token to its cross-lingual synonyms.
+ * Returns synonyms to be added to the token stream (pre-tokenized).
+ * CJK synonyms are NOT pre-bigram'd here — caller handles that.
+ */
+declare function expandCrossLingual(token: string): string[];
+/**
+ * Tokenize text into meaningful keywords.
+ * Handles accented Latin, RTL (Arabic, Hebrew), and CJK (Chinese, Japanese, Korean).
+ *
+ * Features:
+ *  • CJK → character bigrams
+ *  • Katakana → additionally emits Hepburn romaji tokens (enables romaji queries)
+ *  • Arabic/Hebrew → word tokenization with light Arabic stemming
+ *  • All scripts → cross-lingual synonym expansion (EN↔JA↔ZH↔KO↔AR↔HE)
+ */
 declare function tokenize(text: string): string[];
 /**
  * Split a multi-topic query into sub-queries.
@@ -28,4 +56,4 @@ declare function levenshtein(a: string, b: string): number;
 declare function extractTimestamp(content: string): number | undefined;
 /** Recency multiplier: 1.0 for now, 0.5 after half-life days, exponential decay. */
 declare function recencyBoost(timestampMs: number | undefined): number;
-export { tokenize, splitMultiQuery, levenshtein, recencyBoost, extractTimestamp, STOPWORDS };
+export { tokenize, splitMultiQuery, levenshtein, recencyBoost, extractTimestamp, STOPWORDS, katakanaToRomaji, arabicLightStem, expandCrossLingual, CROSS_LINGUAL_MAP };