v0.1.0
v0.0.0 -> v0.1.0
Summary
Initial release of KuraDB, a read-only RAG database that ingests local files, embeds them, and serves semantic and keyword search to the Agenvoy ecosystem. Storage centers on a single source of truth with an in-memory vector cache rebuilt at startup. Multi-database registry, two-stage retrieval, and a persistent query embedding cache round out the surface.
翻譯
KuraDB 首版發布。一套唯讀 RAG 資料庫服務,匯入本地檔案、產生向量並向 Agenvoy 生態提供語意與關鍵字搜尋。儲存以單一資料源為核心,啟動時重建記憶體向量索引;包含多資料庫註冊、兩階段檢索與持久化 query embedding 快取。Changes
FEAT
- Ingest CSV/XLSX, gate unknown files via UTF-8 sniff, relocate Agenvoy manifests (@pardnchiu) [25511ae]
- Persist query embedding cache in database (@pardnchiu) [c3c7636]
- Add API endpoint to list running and registered databases (@pardnchiu) [ae44e6d]
- Add Markdown and text file parsing with long paragraph splitting (@pardnchiu) [4ea736a]
- Add DB registry CLI and Agenvoy RAG tool registration (@pardnchiu) [991b868]
- Add search API with semantic and keyword retrieval (@pardnchiu) [ea1ce60]
- Add embedding cache preload and reset stale vectors (@pardnchiu) [8e390c4]
- Add background embedding pipeline and persistence support (@pardnchiu) [a65c446]
- Add Office document parsers and indexing integration (@pardnchiu) [e1420db]
- Add env-based workspace setup and snapshot persistence (@pardnchiu) [df28320]
- Add file dismissal timestamps and recursive dismiss sync (@pardnchiu) [61b8e97]
- Add SQLite-backed file storage and PDF ingestion pipeline (@pardnchiu) [ac1bf97]
- Add PDF parsing support during file change detection (@pardnchiu) [7c0d88c]
- Add file polling worker (@pardnchiu) [6ebe1ce]
翻譯
- 新增 CSV/XLSX 匯入,未知副檔名以 UTF-8 sniff 過濾,搬遷 Agenvoy manifest 目錄
- 持久化 query embedding 快取至 SQLite
- 新增列出運行中與已註冊資料庫的 API endpoint
- 新增 Markdown 與純文字檔解析,含長段落保護
- 新增 DB registry CLI 與 Agenvoy RAG tool 註冊
- 新增語意與關鍵字搜尋 API
- 啟動時預載 embedding 快取並重置 stale 向量
- 新增背景 embedding pipeline 與持久化支援
- 新增 Office 文件解析(DOCX / PPTX)與索引整合
- 新增環境變數 workspace 設定與 snapshot 持久化
- 新增檔案 dismiss 時間戳與遞迴 dismiss 同步
- 新增 SQLite 檔案儲存與 PDF 匯入 pipeline
- 新增 PDF 解析支援於檔案變動偵測流程
- 新增檔案輪詢 worker
UPDATE
- Pass vector cache into embedder and reduce embedding dimensions (@pardnchiu) [321756f]
- Add env ignore and file timestamps to parser data (@pardnchiu) [be90942]
翻譯
- 將 vector cache 傳入 embedder 並縮減 embedding 維度
- 加入 env 忽略名單與檔案時間戳至 parser 資料
REFACTOR
- Move module path to github.com/agenvoy/kuradb and bump go-pkg to v0.12.1 (@pardnchiu) [8edf9a0]
- Migrate shared utils and parser code to go-pkg (@pardnchiu) [2993b98]
- Restructure multi-db RAG runtime and search pipeline (@pardnchiu) [f582a9b]
- Rename module path and storage directories to KuraDB (@pardnchiu) [393126d]
- Move database handlers into dedicated package (@pardnchiu) [ee4ff30]
- Extract database dismiss and upsert into dedicated files (@pardnchiu) [a33442b]
翻譯
- 模組路徑遷移至 github.com/agenvoy/kuradb,升級 go-pkg 至 v0.12.1
- 共用 utils 與 parser 抽離至 go-pkg
- 重構多資料庫 RAG runtime 與搜尋 pipeline
- 模組路徑與儲存目錄改名為 KuraDB
- database handler 移至獨立 package
- 拆分 dismiss 與 upsert 至各自檔案
TEST
- Add integration tests for filesystem, parser, openai, database handler packages (@pardnchiu) [a83ebc6]
翻譯
- 新增 filesystem、parser、openai、database handler 等 package 的整合測試
Scope
cmd/app/— FEAT, REFACTORinternal/agenvoy/— FEATinternal/api/— FEATinternal/database/— FEAT, REFACTORinternal/filesystem/— FEATinternal/openai/— FEAT, UPDATEinternal/segmenter/— FEATinternal/vector/— FEAT, UPDATE