feat: multi projects

Author: hpbyte

README.md

@@ -23,7 +23,7 @@ Edit your `claude_mcp_settings.json` file:
   "mcpServers": {
     "h-codex": {
       "command": "npx",
-      "args": ["@h-codex/mcp"],
+      "args": ["@hpbyte/h-codex-mcp"],
       "env": {
         "OPENAI_API_KEY": "your_openai_api_key_here",
         "DB_CONNECTION_STRING": "postgresql://postgres:password@localhost:5432/h-codex"

package.json

@@ -1,15 +1,15 @@
 {
   "name": "h-codex",
-  "version": "0.1.0",
+  "version": "0.1.4",
   "description": "A semantic code search tool for intelligent, cross-repo context retrieval.",
   "author": "Htoo Pyae Lwin",
   "license": "MIT",
   "type": "module",
   "scripts": {
     "build": "pnpm -r run build",
-    "dev": "pnpm --filter @h-codex/core run dev",
-    "db:generate": "pnpm --filter @h-codex/core run db:generate",
-    "db:migrate": "pnpm --filter @h-codex/core run db:migrate",
+    "dev": "pnpm --filter @hpbyte/h-codex-core run dev",
+    "db:generate": "pnpm --filter @hpbyte/h-codex-core run db:generate",
+    "db:migrate": "pnpm --filter @hpbyte/h-codex-core run db:migrate",
     "lint:fix": "eslint packages --ext .ts --fix",
     "format": "prettier --write \"packages/**/*.ts\""
   },

packages/core/README.md

@@ -1,4 +1,4 @@
-# @h-codex/core
+# @hpbyte/h-codex-core
 
 Core package for h-codex semantic code indexing and search.
 
@@ -14,7 +14,7 @@ Core package for h-codex semantic code indexing and search.
 ### Installation
 
 ```bash
-pnpm add @h-codex/core
+pnpm add @hpbyte/h-codex-core
 ```
 
 ### Environment Setup
@@ -30,7 +30,7 @@ DB_CONNECTION_STRING=postgresql://postgres:password@localhost:5432/h-codex
 ### Usage Example
 
 ```typescript
-import { indexer, semanticSearch } from '@h-codex/core'
+import { indexer, semanticSearch } from '@hpbyte/h-codex-core'
 
 // Index a codebase
 const indexResult = await indexer.index('./path/to/codebase')

packages/core/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hpbyte/h-codex-core",
-  "version": "0.1.0",
+  "version": "0.1.4",
   "description": "Core indexing and search functionality for h-codex",
   "author": "Htoo Pyae Lwin",
   "license": "MIT",
@@ -40,5 +40,11 @@
   },
   "publishConfig": {
     "access": "public"
-  }
+  },
+  "keywords": [
+    "code-intelligence",
+    "embeddings",
+    "ast",
+    "tree-sitter"
+  ]
 }

packages/core/src/config/database.ts

@@ -1,4 +1,4 @@
-import { getEnvVar } from './helper'
+import { getEnvVar } from '../utils'
 
 export const dbConnectionString = getEnvVar(
   'DB_CONNECTION_STRING',

packages/core/src/config/embedding.ts

@@ -1,4 +1,4 @@
-import { getEnvVar, getEnvNumber } from './helper'
+import { getEnvVar, getEnvNumber } from '../utils'
 
 interface EmbeddingsConfig {
   apiKey: string

packages/core/src/index.ts

@@ -1,2 +1,3 @@
 export * from './ingestion'
 export * from './search'
+export * from './types'

packages/core/src/ingestion/chunker/index.ts

@@ -11,16 +11,18 @@ export class CSTChunker {
   private coalesce = 50
   private filePath = ''
   private fileContent = ''
+  private projectId = ''
   private language: SupportedLanguage | null = null
 
-  async processFile(filePath: string): Promise<CodeChunkInsert[]> {
+  async processFile(filePath: string, projectId: string): Promise<CodeChunkInsert[]> {
     try {
       const language = this.detectLanguage(filePath)
       const cstRootNode = await treeSitterParser.parseFile(filePath, language)
 
       this.filePath = filePath
       this.language = language
       this.fileContent = cstRootNode.text
+      this.projectId = projectId
 
       const chunks = this.chunkNode(cstRootNode)
       const processedChunks = this.processChunks(chunks)
@@ -123,7 +125,8 @@ export class CSTChunker {
     return {
       content,
       filePath: this.filePath,
-      startLine: chunk.start + 1, // tree-sitter index starts from 0, make it human-readable
+      projectId: this.projectId,
+      startLine: chunk.start + 1, // tree-sitter index starts from 0
       endLine: chunk.end + 1,
       nodeType: chunk.nodeType,
       language: this.language!,

packages/core/src/ingestion/chunker/tree-sitter-parser.ts

@@ -18,8 +18,6 @@ export class TreeSitterParser {
     const content = await fs.readFile(filePath, 'utf-8')
     const tree = this.parser.parse(content)
 
-    console.log('total content length: ', content.length)
-
     return tree.rootNode
   }
 

packages/core/src/ingestion/indexer/index.ts

@@ -1,20 +1,17 @@
 import { chunker } from '../chunker'
 import { embedder } from '../embedder'
 import { fileExplorer } from '../explorer'
-import { repository } from '../../storage/repository'
+import { chunkEmbeddingsRepository } from '../../storage/chunk-embeddings.repository'
+import { projectsRepository } from '../../storage/projects.repository'
 
 export class Indexer {
-  async index(path: string): Promise<{
-    totalFiles: number
-    processedFiles: number
-    failedFiles: number
-    errors: Array<{ file: string; error: string }>
-  }> {
-    console.log(`Starting indexing of folder: ${path}`)
-
+  async index(path: string) {
     try {
       const files = await fileExplorer.discover(path)
-      console.log(`Found ${files.length} files to process`)
+
+      const projectName = path.split('/').pop() || 'untitled-project'
+
+      const project = await projectsRepository.getOrCreate(projectName, path)
 
       if (files.length === 0) {
         return {
@@ -26,16 +23,12 @@ export class Indexer {
       }
 
       const { results, errors } = await fileExplorer.processInBatches(files, (filePath: string) =>
-        this.indexFile(filePath)
+        this.indexFile(filePath, project.id)
       )
 
       const processedFiles = results.filter(result => result !== null).length
       const failedFiles = errors.length
 
-      console.log(
-        `Indexing completed: ${processedFiles} files processed, ${failedFiles} files failed`
-      )
-
       return {
         totalFiles: files.length,
         processedFiles,
@@ -46,22 +39,23 @@ export class Indexer {
         })),
       }
     } catch (error) {
-      console.error('Error during folder indexing:', error)
       throw new Error(`Failed to index folder ${path}: ${error}`)
     }
   }
 
-  async indexFile(filePath: string) {
+  async indexFile(filePath: string, projectId: string) {
     try {
-      const chunks = await chunker.processFile(filePath)
-      const newCodeChunks = await repository.insertCodeChunks(chunks)
+      const chunks = await chunker.processFile(filePath, projectId)
+      const newCodeChunks = await chunkEmbeddingsRepository.insertCodeChunks(chunks)
 
       const embeddings = await embedder.generate(newCodeChunks)
-      const newEmbeddings = await repository.insertEmbeddings(newCodeChunks, embeddings)
+      const newEmbeddings = await chunkEmbeddingsRepository.insertEmbeddings(
+        newCodeChunks,
+        embeddings
+      )
 
       return { newCodeChunks, newEmbeddings }
     } catch (error) {
-      console.error('Error indexing file:', error)
       throw new Error(`Failed to index file ${filePath}: ${error}`)
     }
   }