ruvector 0.2.16 → 0.2.17

package/dist/core/onnx-llm.js

@@ -0,0 +1,430 @@
+"use strict";
+/**
+ * ONNX LLM Text Generation for RuVector
+ *
+ * Provides real local LLM inference using ONNX Runtime via transformers.js
+ * Supports small models that run efficiently on CPU:
+ * - SmolLM 135M - Smallest, fast (~135MB)
+ * - SmolLM 360M - Better quality (~360MB)
+ * - TinyLlama 1.1B - Best small model quality (~1GB quantized)
+ * - Qwen2.5 0.5B - Good balance (~500MB)
+ *
+ * Features:
+ * - Automatic model downloading and caching
+ * - Quantized INT4/INT8 models for efficiency
+ * - Streaming generation support
+ * - Temperature, top-k, top-p sampling
+ * - KV cache for efficient multi-turn conversations
+ */
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.OnnxLLM = exports.AVAILABLE_MODELS = void 0;
+exports.isTransformersAvailable = isTransformersAvailable;
+exports.initOnnxLLM = initOnnxLLM;
+exports.generate = generate;
+exports.generateStream = generateStream;
+exports.chat = chat;
+exports.getModelInfo = getModelInfo;
+exports.unload = unload;
+const path = __importStar(require("path"));
+const fs = __importStar(require("fs"));
+// Force native dynamic import (avoids TypeScript transpiling to require)
+// eslint-disable-next-line @typescript-eslint/no-implied-eval
+const dynamicImport = new Function('specifier', 'return import(specifier)');
+// ============================================================================
+// Available Models
+// ============================================================================
+exports.AVAILABLE_MODELS = {
+    // =========================================================================
+    // TRM - Tiny Random Models (smallest, fastest)
+    // =========================================================================
+    'trm-tinystories': {
+        id: 'Xenova/TinyStories-33M',
+        name: 'TinyStories 33M (TRM)',
+        size: '~65MB',
+        description: 'Ultra-tiny model for stories and basic generation',
+        contextLength: 512,
+    },
+    'trm-gpt2-tiny': {
+        id: 'Xenova/gpt2',
+        name: 'GPT-2 124M (TRM)',
+        size: '~250MB',
+        description: 'Classic GPT-2 tiny for general text',
+        contextLength: 1024,
+    },
+    'trm-distilgpt2': {
+        id: 'Xenova/distilgpt2',
+        name: 'DistilGPT-2 (TRM)',
+        size: '~82MB',
+        description: 'Distilled GPT-2, fastest general model',
+        contextLength: 1024,
+    },
+    // =========================================================================
+    // SmolLM - Smallest production-ready models
+    // =========================================================================
+    'smollm-135m': {
+        id: 'HuggingFaceTB/SmolLM-135M-Instruct',
+        name: 'SmolLM 135M',
+        size: '~135MB',
+        description: 'Smallest instruct model, very fast',
+        contextLength: 2048,
+    },
+    'smollm-360m': {
+        id: 'HuggingFaceTB/SmolLM-360M-Instruct',
+        name: 'SmolLM 360M',
+        size: '~360MB',
+        description: 'Small model, fast, better quality',
+        contextLength: 2048,
+    },
+    'smollm2-135m': {
+        id: 'HuggingFaceTB/SmolLM2-135M-Instruct',
+        name: 'SmolLM2 135M',
+        size: '~135MB',
+        description: 'Latest SmolLM v2, improved capabilities',
+        contextLength: 2048,
+    },
+    'smollm2-360m': {
+        id: 'HuggingFaceTB/SmolLM2-360M-Instruct',
+        name: 'SmolLM2 360M',
+        size: '~360MB',
+        description: 'Latest SmolLM v2, better quality',
+        contextLength: 2048,
+    },
+    // =========================================================================
+    // Qwen - Chinese/English bilingual models
+    // =========================================================================
+    'qwen2.5-0.5b': {
+        id: 'Qwen/Qwen2.5-0.5B-Instruct',
+        name: 'Qwen2.5 0.5B',
+        size: '~300MB quantized',
+        description: 'Good balance of speed and quality, multilingual',
+        contextLength: 4096,
+    },
+    // =========================================================================
+    // TinyLlama - Llama architecture in tiny form
+    // =========================================================================
+    'tinyllama': {
+        id: 'TinyLlama/TinyLlama-1.1B-Chat-v1.0',
+        name: 'TinyLlama 1.1B',
+        size: '~600MB quantized',
+        description: 'Best small model quality, slower',
+        contextLength: 2048,
+    },
+    // =========================================================================
+    // Code-specialized models
+    // =========================================================================
+    'codegemma-2b': {
+        id: 'google/codegemma-2b',
+        name: 'CodeGemma 2B',
+        size: '~1GB quantized',
+        description: 'Code generation specialist',
+        contextLength: 8192,
+    },
+    'deepseek-coder-1.3b': {
+        id: 'deepseek-ai/deepseek-coder-1.3b-instruct',
+        name: 'DeepSeek Coder 1.3B',
+        size: '~700MB quantized',
+        description: 'Excellent for code tasks',
+        contextLength: 4096,
+    },
+    // =========================================================================
+    // Phi models - Microsoft's tiny powerhouses
+    // =========================================================================
+    'phi-2': {
+        id: 'microsoft/phi-2',
+        name: 'Phi-2 2.7B',
+        size: '~1.5GB quantized',
+        description: 'High quality small model',
+        contextLength: 2048,
+    },
+    'phi-3-mini': {
+        id: 'microsoft/Phi-3-mini-4k-instruct',
+        name: 'Phi-3 Mini',
+        size: '~2GB quantized',
+        description: 'Best quality tiny model',
+        contextLength: 4096,
+    },
+};
+// ============================================================================
+// ONNX LLM Generator
+// ============================================================================
+let pipeline = null;
+let transformers = null;
+let loadedModel = null;
+let loadPromise = null;
+let loadError = null;
+/**
+ * Check if transformers.js is available
+ */
+async function isTransformersAvailable() {
+    try {
+        await dynamicImport('@xenova/transformers');
+        return true;
+    }
+    catch {
+        return false;
+    }
+}
+/**
+ * Initialize the ONNX LLM with specified model
+ */
+async function initOnnxLLM(config = {}) {
+    if (pipeline && loadedModel === config.modelId) {
+        return true;
+    }
+    if (loadError)
+        throw loadError;
+    if (loadPromise) {
+        await loadPromise;
+        return pipeline !== null;
+    }
+    const modelId = config.modelId || 'HuggingFaceTB/SmolLM-135M-Instruct';
+    loadPromise = (async () => {
+        try {
+            console.error(`Loading ONNX LLM: ${modelId}...`);
+            // Import transformers.js
+            transformers = await dynamicImport('@xenova/transformers');
+            const { pipeline: createPipeline, env } = transformers;
+            // Configure cache directory
+            if (config.cacheDir) {
+                env.cacheDir = config.cacheDir;
+            }
+            else {
+                env.cacheDir = path.join(process.env.HOME || '/tmp', '.ruvector', 'models', 'onnx-llm');
+            }
+            // Ensure cache directory exists
+            if (!fs.existsSync(env.cacheDir)) {
+                fs.mkdirSync(env.cacheDir, { recursive: true });
+            }
+            // Disable remote model fetching warnings
+            env.allowRemoteModels = true;
+            env.allowLocalModels = true;
+            // Create text generation pipeline
+            console.error(`Downloading model (first run may take a while)...`);
+            pipeline = await createPipeline('text-generation', modelId, {
+                quantized: config.quantized !== false,
+                device: config.device || 'cpu',
+            });
+            loadedModel = modelId;
+            console.error(`ONNX LLM ready: ${modelId}`);
+        }
+        catch (e) {
+            loadError = new Error(`Failed to initialize ONNX LLM: ${e.message}`);
+            throw loadError;
+        }
+    })();
+    await loadPromise;
+    return pipeline !== null;
+}
+/**
+ * Generate text using ONNX LLM
+ */
+async function generate(prompt, config = {}) {
+    if (!pipeline) {
+        await initOnnxLLM();
+    }
+    if (!pipeline) {
+        throw new Error('ONNX LLM not initialized');
+    }
+    const start = performance.now();
+    // Build the input text (apply chat template if needed)
+    let inputText = prompt;
+    if (config.systemPrompt) {
+        // Apply simple chat format
+        inputText = `<|system|>\n${config.systemPrompt}<|end|>\n<|user|>\n${prompt}<|end|>\n<|assistant|>\n`;
+    }
+    // Generate
+    const outputs = await pipeline(inputText, {
+        max_new_tokens: config.maxNewTokens || 128,
+        temperature: config.temperature || 0.7,
+        top_p: config.topP || 0.9,
+        top_k: config.topK || 50,
+        repetition_penalty: config.repetitionPenalty || 1.1,
+        do_sample: (config.temperature || 0.7) > 0,
+        return_full_text: false,
+    });
+    const timeMs = performance.now() - start;
+    const generatedText = outputs[0]?.generated_text || '';
+    // Estimate tokens (rough approximation)
+    const tokensGenerated = Math.ceil(generatedText.split(/\s+/).length * 1.3);
+    return {
+        text: generatedText.trim(),
+        tokensGenerated,
+        timeMs,
+        tokensPerSecond: tokensGenerated / (timeMs / 1000),
+        model: loadedModel || 'unknown',
+        cached: true,
+    };
+}
+/**
+ * Generate with streaming (token by token)
+ */
+async function generateStream(prompt, config = {}) {
+    if (!pipeline) {
+        await initOnnxLLM();
+    }
+    if (!pipeline) {
+        throw new Error('ONNX LLM not initialized');
+    }
+    const start = performance.now();
+    let fullText = '';
+    let tokenCount = 0;
+    // Build input text
+    let inputText = prompt;
+    if (config.systemPrompt) {
+        inputText = `<|system|>\n${config.systemPrompt}<|end|>\n<|user|>\n${prompt}<|end|>\n<|assistant|>\n`;
+    }
+    // Create streamer
+    const { TextStreamer } = transformers;
+    const streamer = new TextStreamer(pipeline.tokenizer, {
+        skip_prompt: true,
+        callback_function: (text) => {
+            fullText += text;
+            tokenCount++;
+            if (config.onToken) {
+                config.onToken(text);
+            }
+        },
+    });
+    // Generate with streamer
+    await pipeline(inputText, {
+        max_new_tokens: config.maxNewTokens || 128,
+        temperature: config.temperature || 0.7,
+        top_p: config.topP || 0.9,
+        top_k: config.topK || 50,
+        repetition_penalty: config.repetitionPenalty || 1.1,
+        do_sample: (config.temperature || 0.7) > 0,
+        streamer,
+    });
+    const timeMs = performance.now() - start;
+    // Return generator that yields the collected text
+    async function* generator() {
+        yield fullText;
+        return {
+            text: fullText.trim(),
+            tokensGenerated: tokenCount,
+            timeMs,
+            tokensPerSecond: tokenCount / (timeMs / 1000),
+            model: loadedModel || 'unknown',
+            cached: true,
+        };
+    }
+    return generator();
+}
+/**
+ * Chat completion with conversation history
+ */
+async function chat(messages, config = {}) {
+    if (!pipeline) {
+        await initOnnxLLM();
+    }
+    if (!pipeline) {
+        throw new Error('ONNX LLM not initialized');
+    }
+    // Build conversation text from messages
+    let conversationText = '';
+    for (const msg of messages) {
+        if (msg.role === 'system') {
+            conversationText += `<|system|>\n${msg.content}<|end|>\n`;
+        }
+        else if (msg.role === 'user') {
+            conversationText += `<|user|>\n${msg.content}<|end|>\n`;
+        }
+        else if (msg.role === 'assistant') {
+            conversationText += `<|assistant|>\n${msg.content}<|end|>\n`;
+        }
+    }
+    conversationText += '<|assistant|>\n';
+    return generate(conversationText, { ...config, systemPrompt: undefined });
+}
+/**
+ * Get model information
+ */
+function getModelInfo() {
+    return {
+        model: loadedModel,
+        ready: pipeline !== null,
+        availableModels: exports.AVAILABLE_MODELS,
+    };
+}
+/**
+ * Unload the current model to free memory
+ */
+async function unload() {
+    if (pipeline) {
+        // Note: transformers.js doesn't have explicit dispose, but we can null the reference
+        pipeline = null;
+        loadedModel = null;
+        loadPromise = null;
+        loadError = null;
+    }
+}
+// ============================================================================
+// Class wrapper for OOP usage
+// ============================================================================
+class OnnxLLM {
+    constructor(config = {}) {
+        this.initialized = false;
+        this.config = config;
+    }
+    async init() {
+        if (this.initialized)
+            return true;
+        this.initialized = await initOnnxLLM(this.config);
+        return this.initialized;
+    }
+    async generate(prompt, config) {
+        if (!this.initialized)
+            await this.init();
+        return generate(prompt, config);
+    }
+    async chat(messages, config) {
+        if (!this.initialized)
+            await this.init();
+        return chat(messages, config);
+    }
+    async unload() {
+        await unload();
+        this.initialized = false;
+    }
+    get ready() {
+        return this.initialized;
+    }
+    get model() {
+        return loadedModel;
+    }
+}
+exports.OnnxLLM = OnnxLLM;
+exports.default = OnnxLLM;

package/dist/core/sona-wrapper.d.ts

@@ -194,6 +194,11 @@ export declare class SonaEngine {
      * @returns Statistics object
      */
     getStats(): SonaStats;
+    /**
+     * Save engine state to JSON string for persistence across restarts (fixes #274)
+     * @returns State JSON that can be stored to disk/database
+     */
+    saveState(): string;
     /**
      * Enable or disable the engine
      * @param enabled Whether to enable

package/dist/core/sona-wrapper.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"sona-wrapper.d.ts","sourceRoot":"","sources":["../../src/core/sona-wrapper.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAMH,sEAAsE;AACtE,MAAM,MAAM,UAAU,GAAG,MAAM,EAAE,GAAG,YAAY,GAAG,YAAY,CAAC;AAEhE,iCAAiC;AACjC,MAAM,WAAW,UAAU;IACzB,uCAAuC;IACvC,SAAS,EAAE,MAAM,CAAC;IAClB,kDAAkD;IAClD,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,wCAAwC;IACxC,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,kCAAkC;IAClC,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,gDAAgD;IAChD,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,gDAAgD;IAChD,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,kDAAkD;IAClD,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,+CAA+C;IAC/C,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,kDAAkD;IAClD,kBAAkB,CAAC,EAAE,MAAM,CAAC;IAC5B,qEAAqE;IACrE,oBAAoB,CAAC,EAAE,MAAM,CAAC;IAC9B,oDAAoD;IACpD,gBAAgB,CAAC,EAAE,MAAM,CAAC;IAC1B,gDAAgD;IAChD,UAAU,CAAC,EAAE,OAAO,CAAC;CACtB;AAED,yCAAyC;AACzC,MAAM,WAAW,cAAc;IAC7B,yBAAyB;IACzB,EAAE,EAAE,MAAM,CAAC;IACX,iCAAiC;IACjC,QAAQ,EAAE,MAAM,EAAE,CAAC;IACnB,wCAAwC;IACxC,WAAW,EAAE,MAAM,CAAC;IACpB,mCAAmC;IACnC,WAAW,EAAE,MAAM,CAAC;IACpB,6CAA6C;IAC7C,UAAU,EAAE,MAAM,CAAC;IACnB,yBAAyB;IACzB,SAAS,EAAE,MAAM,CAAC;IAClB,4BAA4B;IAC5B,YAAY,EAAE,MAAM,CAAC;IACrB,yBAAyB;IACzB,WAAW,EAAE,MAAM,CAAC;IACpB,mBAAmB;IACnB,WAAW,EAAE,MAAM,CAAC;CACrB;AAED,6BAA6B;AAC7B,MAAM,WAAW,SAAS;IACxB,oBAAoB,EAAE,MAAM,CAAC;IAC7B,eAAe,EAAE,MAAM,CAAC;IACxB,gBAAgB,EAAE,MAAM,CAAC;IACzB,eAAe,EAAE,MAAM,CAAC;IACxB,iBAAiB,EAAE,MAAM,CAAC;IAC1B,iBAAiB,EAAE,MAAM,CAAC;CAC3B;AAoCD,iCAAiC;AACjC,wBAAgB,eAAe,IAAI,OAAO,CAOzC;AAMD;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA+BG;AACH,qBAAa,UAAU;IACrB,OAAO,CAAC,OAAO,CAAM;IAErB;;;OAGG;gBACS,SAAS,EAAE,MAAM;IAK7B;;;OAGG;IACH,MAAM,CAAC,UAAU,CAAC,MAAM,EAAE,UAAU,GAAG,UAAU;IAYjD;;;;OAIG;IACH,eAAe,CAAC,cAAc,EAAE,UAAU,GAAG,MAAM;IAInD;;;;;;OAMG;IACH,OAAO,CACL,YAAY,EAAE,MAAM,EACpB,WAAW,EAAE,UAAU,EACvB,gBAAgB,EAAE,UAAU,EAC5B,MAAM,EAAE,MAAM,GACb,IAAI;IASP;;OAEG;IACH,iBAAiB,CACf,YAAY,EAAE,MAAM,EACpB,WAAW,EAAE,UAAU,EACvB,gBAAgB,EAAE,UAAU,EAC5B,MAAM,EAAE,MAAM,GACb,IAAI;IAIP;;;;OAIG;IACH,QAAQ,CAAC,YAAY,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,GAAG,IAAI;IAInD;;;;OAIG;IACH,UAAU,CAAC,YAAY,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,GAAG,IAAI;IAIzD;;;;OAIG;IACH,aAAa,CAAC,YAAY,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,GAAG,IAAI;IAQ1D;;;;OAIG;IACH,cAAc,CAAC,KAAK,EAAE,UAAU,GAAG,MAAM,EAAE;IAI3C;;;;;OAKG;IACH,aAAa,CAAC,QAAQ,EAAE,MAAM,EAAE,KAAK,EAAE,UAAU,GAAG,MAAM,EAAE;IAQ5D;;;;OAIG;IACH,IAAI,IAAI,MAAM,GAAG,IAAI;IAIrB;;;OAGG;IACH,UAAU,IAAI,MAAM;IAIpB;;OAEG;IACH,KAAK,IAAI,IAAI;IAQb;;;;;OAKG;IACH,YAAY,CAAC,cAAc,EAAE,UAAU,EAAE,CAAC,EAAE,MAAM,GAAG,cAAc,EAAE;IAQrE;;;OAGG;IACH,QAAQ,IAAI,SAAS;IAKrB;;;OAGG;IACH,UAAU,CAAC,OAAO,EAAE,OAAO,GAAG,IAAI;IAIlC;;OAEG;IACH,SAAS,IAAI,OAAO;CAGrB;AAMD;;GAEG;AACH,eAAO,MAAM,IAAI;;;CAGhB,CAAC;AAEF,eAAe,IAAI,CAAC"}
+{"version":3,"file":"sona-wrapper.d.ts","sourceRoot":"","sources":["../../src/core/sona-wrapper.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAMH,sEAAsE;AACtE,MAAM,MAAM,UAAU,GAAG,MAAM,EAAE,GAAG,YAAY,GAAG,YAAY,CAAC;AAEhE,iCAAiC;AACjC,MAAM,WAAW,UAAU;IACzB,uCAAuC;IACvC,SAAS,EAAE,MAAM,CAAC;IAClB,kDAAkD;IAClD,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,wCAAwC;IACxC,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,kCAAkC;IAClC,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,gDAAgD;IAChD,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,gDAAgD;IAChD,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,kDAAkD;IAClD,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,+CAA+C;IAC/C,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,kDAAkD;IAClD,kBAAkB,CAAC,EAAE,MAAM,CAAC;IAC5B,qEAAqE;IACrE,oBAAoB,CAAC,EAAE,MAAM,CAAC;IAC9B,oDAAoD;IACpD,gBAAgB,CAAC,EAAE,MAAM,CAAC;IAC1B,gDAAgD;IAChD,UAAU,CAAC,EAAE,OAAO,CAAC;CACtB;AAED,yCAAyC;AACzC,MAAM,WAAW,cAAc;IAC7B,yBAAyB;IACzB,EAAE,EAAE,MAAM,CAAC;IACX,iCAAiC;IACjC,QAAQ,EAAE,MAAM,EAAE,CAAC;IACnB,wCAAwC;IACxC,WAAW,EAAE,MAAM,CAAC;IACpB,mCAAmC;IACnC,WAAW,EAAE,MAAM,CAAC;IACpB,6CAA6C;IAC7C,UAAU,EAAE,MAAM,CAAC;IACnB,yBAAyB;IACzB,SAAS,EAAE,MAAM,CAAC;IAClB,4BAA4B;IAC5B,YAAY,EAAE,MAAM,CAAC;IACrB,yBAAyB;IACzB,WAAW,EAAE,MAAM,CAAC;IACpB,mBAAmB;IACnB,WAAW,EAAE,MAAM,CAAC;CACrB;AAED,6BAA6B;AAC7B,MAAM,WAAW,SAAS;IACxB,oBAAoB,EAAE,MAAM,CAAC;IAC7B,eAAe,EAAE,MAAM,CAAC;IACxB,gBAAgB,EAAE,MAAM,CAAC;IACzB,eAAe,EAAE,MAAM,CAAC;IACxB,iBAAiB,EAAE,MAAM,CAAC;IAC1B,iBAAiB,EAAE,MAAM,CAAC;CAC3B;AAoCD,iCAAiC;AACjC,wBAAgB,eAAe,IAAI,OAAO,CAOzC;AAMD;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA+BG;AACH,qBAAa,UAAU;IACrB,OAAO,CAAC,OAAO,CAAM;IAErB;;;OAGG;gBACS,SAAS,EAAE,MAAM;IAK7B;;;OAGG;IACH,MAAM,CAAC,UAAU,CAAC,MAAM,EAAE,UAAU,GAAG,UAAU;IAYjD;;;;OAIG;IACH,eAAe,CAAC,cAAc,EAAE,UAAU,GAAG,MAAM;IAInD;;;;;;OAMG;IACH,OAAO,CACL,YAAY,EAAE,MAAM,EACpB,WAAW,EAAE,UAAU,EACvB,gBAAgB,EAAE,UAAU,EAC5B,MAAM,EAAE,MAAM,GACb,IAAI;IASP;;OAEG;IACH,iBAAiB,CACf,YAAY,EAAE,MAAM,EACpB,WAAW,EAAE,UAAU,EACvB,gBAAgB,EAAE,UAAU,EAC5B,MAAM,EAAE,MAAM,GACb,IAAI;IAIP;;;;OAIG;IACH,QAAQ,CAAC,YAAY,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,GAAG,IAAI;IAInD;;;;OAIG;IACH,UAAU,CAAC,YAAY,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,GAAG,IAAI;IAIzD;;;;OAIG;IACH,aAAa,CAAC,YAAY,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,GAAG,IAAI;IAQ1D;;;;OAIG;IACH,cAAc,CAAC,KAAK,EAAE,UAAU,GAAG,MAAM,EAAE;IAI3C;;;;;OAKG;IACH,aAAa,CAAC,QAAQ,EAAE,MAAM,EAAE,KAAK,EAAE,UAAU,GAAG,MAAM,EAAE;IAQ5D;;;;OAIG;IACH,IAAI,IAAI,MAAM,GAAG,IAAI;IAIrB;;;OAGG;IACH,UAAU,IAAI,MAAM;IAIpB;;OAEG;IACH,KAAK,IAAI,IAAI;IAQb;;;;;OAKG;IACH,YAAY,CAAC,cAAc,EAAE,UAAU,EAAE,CAAC,EAAE,MAAM,GAAG,cAAc,EAAE;IAQrE;;;OAGG;IACH,QAAQ,IAAI,SAAS;IAcrB;;;OAGG;IACH,SAAS,IAAI,MAAM;IAInB;;;OAGG;IACH,UAAU,CAAC,OAAO,EAAE,OAAO,GAAG,IAAI;IAIlC;;OAEG;IACH,SAAS,IAAI,OAAO;CAGrB;AAMD;;GAEG;AACH,eAAO,MAAM,IAAI;;;CAGhB,CAAC;AAEF,eAAe,IAAI,CAAC"}

package/dist/core/sona-wrapper.js

@@ -228,7 +228,23 @@ class SonaEngine {
      */
     getStats() {
         const statsJson = this._native.getStats();
-        return JSON.parse(statsJson);
+        const raw = JSON.parse(statsJson);
+        // Map snake_case Rust fields to camelCase TypeScript (fixes #273)
+        return {
+            trajectoriesRecorded: raw.trajectories_recorded ?? raw.trajectoriesRecorded ?? (raw.trajectories_buffered ?? 0) + (raw.trajectories_dropped ?? 0),
+            patternsLearned: raw.patterns_learned ?? raw.patternsLearned ?? raw.patterns_stored ?? 0,
+            microLoraUpdates: raw.micro_lora_updates ?? raw.microLoraUpdates ?? 0,
+            baseLoraUpdates: raw.base_lora_updates ?? raw.baseLoraUpdates ?? 0,
+            ewcConsolidations: raw.ewc_tasks ?? raw.ewcConsolidations ?? 0,
+            avgLearningTimeMs: raw.avg_learning_time_ms ?? raw.avgLearningTimeMs ?? 0,
+        };
+    }
+    /**
+     * Save engine state to JSON string for persistence across restarts (fixes #274)
+     * @returns State JSON that can be stored to disk/database
+     */
+    saveState() {
+        return this._native.saveState();
     }
     /**
      * Enable or disable the engine

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ruvector",
-  "version": "0.2.16",
+  "version": "0.2.17",
   "description": "High-performance vector database for Node.js with automatic native/WASM fallback",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",
@@ -86,8 +86,12 @@
     "@ruvector/router": ">=0.1.0"
   },
   "peerDependenciesMeta": {
-    "@ruvector/ruvllm": { "optional": true },
-    "@ruvector/router": { "optional": true }
+    "@ruvector/ruvllm": {
+      "optional": true
+    },
+    "@ruvector/router": {
+      "optional": true
+    }
   },
   "files": [
     "bin/",

package/dist/core/core/agentdb-fast.d.ts

@@ -1,148 +0,0 @@
-/**
- * AgentDB Fast - High-performance in-process alternative to AgentDB CLI
- *
- * The AgentDB CLI has ~2.3s startup overhead due to npx initialization.
- * This module provides 50-200x faster operations by using in-process calls.
- *
- * Features:
- * - In-memory episode storage with LRU eviction
- * - Vector similarity search using @ruvector/core
- * - Compatible API with AgentDB's episode/trajectory interfaces
- */
-/**
- * Episode entry for trajectory storage
- */
-export interface Episode {
-    id: string;
-    state: number[];
-    action: string | number;
-    reward: number;
-    nextState: number[];
-    done: boolean;
-    metadata?: Record<string, any>;
-    timestamp?: number;
-}
-/**
- * Trajectory (sequence of episodes)
- */
-export interface Trajectory {
-    id: string;
-    episodes: Episode[];
-    totalReward: number;
-    metadata?: Record<string, any>;
-}
-/**
- * Search result for episode queries
- */
-export interface EpisodeSearchResult {
-    episode: Episode;
-    similarity: number;
-    trajectoryId?: string;
-}
-/**
- * Fast in-memory AgentDB implementation
- */
-export declare class FastAgentDB {
-    private episodes;
-    private trajectories;
-    private vectorDb;
-    private dimensions;
-    private maxEpisodes;
-    private episodeOrder;
-    /**
-     * Create a new FastAgentDB instance
-     *
-     * @param dimensions - Vector dimensions for state embeddings
-     * @param maxEpisodes - Maximum episodes to store (LRU eviction)
-     */
-    constructor(dimensions?: number, maxEpisodes?: number);
-    /**
-     * Initialize the vector database
-     */
-    private initVectorDb;
-    /**
-     * Store an episode
-     *
-     * @param episode - Episode to store
-     * @returns Episode ID
-     */
-    storeEpisode(episode: Omit<Episode, 'id'> & {
-        id?: string;
-    }): Promise<string>;
-    /**
-     * Store multiple episodes in batch
-     */
-    storeEpisodes(episodes: (Omit<Episode, 'id'> & {
-        id?: string;
-    })[]): Promise<string[]>;
-    /**
-     * Retrieve an episode by ID
-     */
-    getEpisode(id: string): Promise<Episode | null>;
-    /**
-     * Search for similar episodes by state
-     *
-     * @param queryState - State vector to search for
-     * @param k - Number of results to return
-     * @returns Similar episodes sorted by similarity
-     */
-    searchByState(queryState: number[] | Float32Array, k?: number): Promise<EpisodeSearchResult[]>;
-    /**
-     * Fallback similarity search using brute-force cosine similarity
-     */
-    private fallbackSearch;
-    /**
-     * Compute cosine similarity between two vectors
-     */
-    private cosineSimilarity;
-    /**
-     * Store a trajectory (sequence of episodes)
-     */
-    storeTrajectory(episodes: (Omit<Episode, 'id'> & {
-        id?: string;
-    })[], metadata?: Record<string, any>): Promise<string>;
-    /**
-     * Get a trajectory by ID
-     */
-    getTrajectory(id: string): Promise<Trajectory | null>;
-    /**
-     * Get top trajectories by total reward
-     */
-    getTopTrajectories(k?: number): Promise<Trajectory[]>;
-    /**
-     * Sample random episodes (for experience replay)
-     */
-    sampleEpisodes(n: number): Promise<Episode[]>;
-    /**
-     * Get database statistics
-     */
-    getStats(): {
-        episodeCount: number;
-        trajectoryCount: number;
-        dimensions: number;
-        maxEpisodes: number;
-        vectorDbAvailable: boolean;
-    };
-    /**
-     * Clear all data
-     */
-    clear(): void;
-    /**
-     * Generate a unique ID
-     */
-    private generateId;
-}
-/**
- * Create a fast AgentDB instance
- */
-export declare function createFastAgentDB(dimensions?: number, maxEpisodes?: number): FastAgentDB;
-/**
- * Get the default FastAgentDB instance
- */
-export declare function getDefaultAgentDB(): FastAgentDB;
-declare const _default: {
-    FastAgentDB: typeof FastAgentDB;
-    createFastAgentDB: typeof createFastAgentDB;
-    getDefaultAgentDB: typeof getDefaultAgentDB;
-};
-export default _default;