Releases: tylevnovik/TextFlow
Releases · tylevnovik/TextFlow
TextFlow Studio v0.2.0
TextFlow Studio v0.2.0 Release Notes / 版本发布说明
中文版说明
🌟 版本亮点 (Highlights)
- 后端驱动的节点 Schema 重构:移除了前端耦合的节点配置,改由 Python 后端节点目录进行动态 Schema 驱动分发,大幅提升节点扩展的灵活性。
- FastAPI 框架迁移:将 Python 侧边栏引擎(Sidecar)的 API 从自定义请求分发器迁移到了更加稳健高效的 FastAPI 架构。
- Fluent UI 工作台集成:依据 Fluent UI 组件规范重构了工作流编辑器,增加了动态的节点编辑面板和槽位。
- Python 引擎架构重构:整合了领域模型(Domain Models)、节点注册表(Registry)和执行调度器(Scheduler),模块化边界更加清晰。
- 性能与测试流程升级:将 Python 引擎测试拆分为
fast和full两套套件,并增加了节点 Schema 和后端目录的一致性契约验证测试。
⚡ 性能优化 (Performance)
- 将 Python 引擎的测试拆分为默认的快速测试集
test:engine(fast) 和耗时较长的完整测试集test:engine:full,日常更改无需再等待繁重的样本生成和集成测试。 - 将内置的官方示范项目转移到了 Python 引擎自带的构建期打包工作区模板中,避免了新用户在首次启动空白工作区时进行在线/实时生成,大幅提升了冷启动速度。
🐛 缺陷修复 (Fixes)
- IncoPat 导入预设修复:调整字段匹配逻辑,使其完美适配例如
标题 (中文)、摘要 (中文)、IPC等真实的专利导出列名,增加了备用专利文本检索,并优化了标点和空格差异下的源字段智能匹配。 - XLSX 导入归一化修复:修复了空单元格在表格中被读取为
nan字符的 Bug,确保必填字段检查时将空单元格正确视为空白值,并优化了日期型年份数据的提取。 - 导入字段编辑器聚焦修复:通过锁定反应(React)映射行中的组件 Keys,解决了复杂导入模式下行内编辑文本框输入时组件意外失去焦点的问题。
English Version Notes
🌟 Highlights
- Backend-Driven Node Schema Refactor: Replaced frontend-coupled node configurations with schemas dynamically served by the Python backend catalog.
- FastAPI Migration: Ported the Python engine sidecar API from custom request dispatchers to a robust FastAPI application structure.
- Fluent UI Workbench Integration: Refactored the workflow workspace, adding dynamic node editing panels and slots using Fluent UI component guidelines.
- Python Engine Restructuring: Consolidated domain models, registry, and execution scheduler to improve modularity and clean architectural boundaries.
- Performance & Testing: Split engine tests into fast and full suites, and added contract validation tests for node schemas and catalog parity.
⚡ Performance
- Split Python engine verification into a default
test:enginefast tier and a heaviertest:engine:fulltier so day-to-day changes do not always pay for bundled sample integration coverage. - Moved the nine official sample projects to a build-time bundled workspace template that ships inside the Python sidecar, avoiding first-run live generation for empty workspaces.
🐛 Fixes
- Fixed IncoPat import presets to match real exported column names such as
标题 (中文)/摘要 (中文)/IPC, added broader patent-text fallbacks, and improved source-field matching across punctuation and spacing variants. - Fixed XLSX import normalization so empty spreadsheet cells no longer become literal
nan, required-field checks correctly treat blank cells as missing, and date-like values now produce a usable document year. - Fixed the advanced import field editor losing focus while typing by keeping mapping-row React keys stable during inline edits.
TextFlow Studio v0.1.1
Highlights
- Promoted the current desktop build to
0.1.1becausev0.1.0already exists on GitHub and points to an older commit. - Added nine backend-built official scenario sample projects based on public datasets, with strict English/Chinese
1:1row balance. - Switched the project model and runtime documentation to the current workflow-only native DAG execution path.
- Kept the result and export surfaces focused on lazy artifact previews, run diff, review queue and experiment matrix.
Fixes
- Fixed project overview startup/rendering slowdown by removing full
run.logsrendering from the overview page. The overview now shows only the latest three run summaries; full history stays in the results page. - Fixed stale live workflow state leaking into the editor after a run completes.
- Fixed run artifact rendering for real artifact handles instead of assuming legacy step summaries.
- Reduced repeated large project refresh/write paths in the desktop store and Python project store.
- Replaced stale demo-style sample data with backend-generated public-source sample caches.
Validation
npm run test --workspace apps/desktopnpm run lintnpm run test:enginenpm run tauri:build --workspace apps/desktop
Release Asset
TextFlow Studio_0.1.1_x64-setup.exe
SHA256: 1E6CB600C7C604B0882E1210C53B96FBFBD4AC080E0C533BCBD109CCE02152A1
v0.1.0
Highlights
- Introduces the workflow editor, native DAG execution foundation, and node registry/plugin architecture.
- Adds bundled dictionaries, sample projects, benchmark tooling, and expanded Python engine coverage.
- Refreshes the documentation set around current project status, architecture, runtime, and development workflows.
Verification
- npm run lint
- npm run test:engine
- npm run tauri:build --workspace apps/desktop
Assets
- SHA256SUMS.txt attached below.
- Windows NSIS installer was built successfully, but uploading the 85 MB binary from this machine hit repeated connection resets.