@easbot/note 0.1.11

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 houjallen
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.en.md

@@ -0,0 +1,211 @@
+# @easbot/note
+
+> Note Knowledge Base - Unstructured Memory and Document Knowledge Base Management System
+
+## Introduction
+
+Note Knowledge Base is a TypeScript/Node.js-based unstructured memory and document knowledge base management system, supporting hybrid search (vector similarity + full-text search + graph relationship reasoning) and LLM-driven knowledge ingestion.
+
+## Features
+
+- **Document Ingestion**: Markdown and plain text parsing, intelligent chunking and embedding generation
+- **Hybrid Search**: Vector similarity + FTS5 full-text search + graph relationship reasoning with configurable weights
+- **Graph Queries**: SQL interface for querying nodes and edges, path finding support
+- **Entity Management**: Complete CRUD operations for nodes and edges
+- **Model Injection**: Support for AI SDK-compliant LLM and Embedding models
+- **Status Monitoring**: Knowledge base status and health check
+- **Intelligent Chunking**: Markdown-aware chunking that preserves code blocks and inline code
+
+## Installation
+
+```bash
+npm install @easbot/note
+# or
+pnpm add @easbot/note
+```
+
+## Quick Start
+
+### Basic Usage
+
+```typescript
+import { NoteKnowledge } from '@easbot/note';
+
+// Create knowledge base instance
+const kb = new NoteKnowledge({
+  storagePath: './data/knowledge.db',
+  vectorDims: 768,
+  chunkSize: 1000,
+  chunkOverlap: 200,
+  embedding: yourEmbeddingModel, // AI SDK EmbeddingModelV2
+  llm: yourLLM,                  // Optional, for entity extraction
+  hybridSearch: {
+    vectorWeight: 0.7,
+    ftsWeight: 0.3,
+    graphWeight: 0.0,
+    rerankTopK: 10,
+  },
+});
+
+// Initialize
+await kb.initialize();
+
+// Ingest document
+const result = await kb.ingestDocument('./docs/example.md');
+console.log(`Created ${result.chunksCreated} chunks`);
+
+// Search
+const results = await kb.search('query content', { limit: 10 });
+for (const r of results) {
+  console.log(`[${r.score}] ${r.content}`);
+}
+
+// Close connection
+await kb.close();
+```
+
+### Graph Queries
+
+```typescript
+// Create nodes
+const node1 = await kb.createNode('EntityA', 'Type1', { prop: 'value' });
+const node2 = await kb.createNode('EntityB', 'Type1');
+
+// Create edge
+const edge = await kb.createEdge(node1.id, node2.id, 'RELATED_TO');
+
+// Find path
+const paths = await kb.findPath('EntityA', 'EntityB', 3);
+
+// SQL query
+const nodes = await kb.queryNodes('SELECT * FROM nodes WHERE type = ?', ['Type1']);
+```
+
+## API Documentation
+
+### NoteKnowledge
+
+#### Constructor
+
+```typescript
+constructor(config: NoteKnowledgeConfig)
+```
+
+Configuration options:
+- `storagePath`: Database storage path
+- `vectorDims`: Vector dimensions (default 768)
+- `chunkSize`: Chunk size (default 1000)
+- `chunkOverlap`: Chunk overlap (default 200)
+- `embedding`: Embedding model (AI SDK EmbeddingModelV2)
+- `llm`: LLM model (optional, for summary generation)
+- `graphLlm`: Graph LLM (optional, for entity extraction)
+- `rerankLlm`: Rerank LLM (optional)
+- `hybridSearch`: Hybrid search weight configuration
+
+#### Document Operations
+
+- `initialize()`: Initialize knowledge base
+- `ingestDocument(filePath)`: Ingest single document
+- `ingestDocuments(filePaths, progress?)`: Batch ingest documents
+- `removeDocument(filePath)`: Remove document
+
+#### Search Operations
+
+- `search(query, options?)`: Hybrid search
+
+#### Graph Operations
+
+- `queryNodes(sql, params?)`: SQL query for nodes
+- `queryEdges(sql, params?)`: SQL query for edges
+- `findPath(startNode, endNode, maxDepth?)`: Find path
+- `createNode(name, type, properties?)`: Create node
+- `updateNode(id, properties)`: Update node
+- `deleteNode(id)`: Delete node
+- `createEdge(source, target, relation, properties?)`: Create edge
+- `deleteEdge(id)`: Delete edge
+
+#### Status and Lifecycle
+
+- `getStatus()`: Get knowledge base status
+- `healthCheck()`: Health check
+- `close()`: Close connection
+
+## Data Models
+
+### Document
+
+```typescript
+interface Document {
+  id: number;
+  path: string;
+  title?: string;
+  contentHash: string;
+  summary?: string;
+  metadata?: Record<string, unknown>;
+  lastModified: number;
+}
+```
+
+### Node
+
+```typescript
+interface Node {
+  id: number;
+  name: string;
+  type: string;
+  properties?: Record<string, unknown>;
+  createdAt: number;
+  updatedAt: number;
+}
+```
+
+### Edge
+
+```typescript
+interface Edge {
+  id: number;
+  source: number;
+  target: number;
+  relation: string;
+  properties?: Record<string, unknown>;
+  createdAt: number;
+}
+```
+
+### SearchResult
+
+```typescript
+interface SearchResult {
+  id: number;
+  content: string;
+  score: number;
+  path: string;
+  startLine: number;
+  endLine: number;
+  documentId: number;
+}
+```
+
+## Hybrid Search
+
+The system supports three search methods:
+
+1. **Vector Search**: Cosine similarity based on embedding vectors
+2. **Full-Text Search**: SQLite FTS5 with porter stemmer and unicode61 tokenizer
+3. **Graph Search**: Graph traversal based on entity relationships
+
+Results are merged using weighted score combination with configurable weights for each method.
+
+## Dependencies
+
+- `better-sqlite3`: SQLite database
+- `@easbot/utils`: Utility library (Markdown parsing, intelligent chunking)
+- `ai`: AI SDK
+
+## License
+
+MIT
+
+## Author
+
+houjallen

package/README.md

@@ -0,0 +1,211 @@
+# @easbot/note
+
+> Note Knowledge Base - 非结构化记忆和文档知识库管理系统
+
+## 简介
+
+Note Knowledge Base 是一个基于 TypeScript/Node.js 的非结构化记忆和文档知识库管理系统，支持混合搜索（向量相似度 + 全文搜索 + 图关系推理）和 LLM 驱动的知识摄取。
+
+## 特性
+
+- **文档摄取**：支持 Markdown、纯文本解析，智能分块和嵌入生成
+- **混合搜索**：向量相似度 + FTS5 全文搜索 + 图关系推理，可配置权重
+- **图查询**：SQL 接口查询节点和边，支持路径查找
+- **实体管理**：节点和边的完整 CRUD 操作
+- **模型注入**：支持 AI SDK 规范的 LLM 和 Embedding 模型
+- **状态监控**：提供知识库状态和健康检查
+- **智能分块**：使用 Markdown 智能分块，保护代码块和内联代码
+
+## 安装
+
+```bash
+npm install @easbot/note
+# 或
+pnpm add @easbot/note
+```
+
+## 快速开始
+
+### 基本使用
+
+```typescript
+import { NoteKnowledge } from '@easbot/note';
+
+// 创建知识库实例
+const kb = new NoteKnowledge({
+  storagePath: './data/knowledge.db',
+  vectorDims: 768,
+  chunkSize: 1000,
+  chunkOverlap: 200,
+  embedding: yourEmbeddingModel, // AI SDK EmbeddingModelV2
+  llm: yourLLM,                  // 可选，用于实体提取
+  hybridSearch: {
+    vectorWeight: 0.7,
+    ftsWeight: 0.3,
+    graphWeight: 0.0,
+    rerankTopK: 10,
+  },
+});
+
+// 初始化
+await kb.initialize();
+
+// 摄取文档
+const result = await kb.ingestDocument('./docs/example.md');
+console.log(`创建了 ${result.chunksCreated} 个块`);
+
+// 搜索
+const results = await kb.search('查询内容', { limit: 10 });
+for (const r of results) {
+  console.log(`[${r.score}] ${r.content}`);
+}
+
+// 关闭连接
+await kb.close();
+```
+
+### 图查询
+
+```typescript
+// 创建节点
+const node1 = await kb.createNode('实体A', '类型1', { prop: 'value' });
+const node2 = await kb.createNode('实体B', '类型1');
+
+// 创建边
+const edge = await kb.createEdge(node1.id, node2.id, 'RELATED_TO');
+
+// 查找路径
+const paths = await kb.findPath('实体A', '实体B', 3);
+
+// SQL 查询
+const nodes = await kb.queryNodes('SELECT * FROM nodes WHERE type = ?', ['类型1']);
+```
+
+## API 文档
+
+### NoteKnowledge
+
+#### 构造函数
+
+```typescript
+constructor(config: NoteKnowledgeConfig)
+```
+
+配置选项：
+- `storagePath`: 数据库存储路径
+- `vectorDims`: 向量维度（默认 768）
+- `chunkSize`: 分块大小（默认 1000）
+- `chunkOverlap`: 分块重叠（默认 200）
+- `embedding`: Embedding 模型（AI SDK EmbeddingModelV2）
+- `llm`: LLM 模型（可选，用于摘要生成）
+- `graphLlm`: 图 LLM（可选，用于实体提取）
+- `rerankLlm`: 重排序 LLM（可选）
+- `hybridSearch`: 混合搜索权重配置
+
+#### 文档操作
+
+- `initialize()`: 初始化知识库
+- `ingestDocument(filePath)`: 摄取单个文档
+- `ingestDocuments(filePaths, progress?)`: 批量摄取文档
+- `removeDocument(filePath)`: 删除文档
+
+#### 搜索操作
+
+- `search(query, options?)`: 混合搜索
+
+#### 图操作
+
+- `queryNodes(sql, params?)`: SQL 查询节点
+- `queryEdges(sql, params?)`: SQL 查询边
+- `findPath(startNode, endNode, maxDepth?)`: 查找路径
+- `createNode(name, type, properties?)`: 创建节点
+- `updateNode(id, properties)`: 更新节点
+- `deleteNode(id)`: 删除节点
+- `createEdge(source, target, relation, properties?)`: 创建边
+- `deleteEdge(id)`: 删除边
+
+#### 状态和生命周期
+
+- `getStatus()`: 获取知识库状态
+- `healthCheck()`: 健康检查
+- `close()`: 关闭连接
+
+## 数据模型
+
+### 文档 (Document)
+
+```typescript
+interface Document {
+  id: number;
+  path: string;
+  title?: string;
+  contentHash: string;
+  summary?: string;
+  metadata?: Record<string, unknown>;
+  lastModified: number;
+}
+```
+
+### 节点 (Node)
+
+```typescript
+interface Node {
+  id: number;
+  name: string;
+  type: string;
+  properties?: Record<string, unknown>;
+  createdAt: number;
+  updatedAt: number;
+}
+```
+
+### 边 (Edge)
+
+```typescript
+interface Edge {
+  id: number;
+  source: number;
+  target: number;
+  relation: string;
+  properties?: Record<string, unknown>;
+  createdAt: number;
+}
+```
+
+### 搜索结果 (SearchResult)
+
+```typescript
+interface SearchResult {
+  id: number;
+  content: string;
+  score: number;
+  path: string;
+  startLine: number;
+  endLine: number;
+  documentId: number;
+}
+```
+
+## 混合搜索
+
+系统支持三种搜索方式的混合：
+
+1. **向量搜索**：基于嵌入向量的余弦相似度
+2. **全文搜索**：SQLite FTS5，支持 porter 词干提取和 unicode61 分词
+3. **图搜索**：基于实体关系的图遍历
+
+结果合并使用加权分数组合，可配置各搜索方式的权重。
+
+## 依赖
+
+- `better-sqlite3`: SQLite 数据库
+- `@easbot/utils`: 工具库（Markdown 解析、智能分块）
+- `ai`: AI SDK
+
+## 许可证
+
+MIT
+
+## 作者
+
+houjallen