Skip to content

TrayClip v0.3.2

Choose a tag to compare

View all tags
@github-actions github-actions released this 07 Nov 16:20
· 2 commits to main since this release
v0.3.2
a06f675

TrayClip v0.3.2

本次更新新增离线 OCR 文字识别，采用按需下载的独立引擎，兼顾主安装包体积与离线使用体验；同时修复 Windows 卸载数据清理问题，并完善多平台构建流程。

新增功能

  • 图片文字识别：支持从剪贴板图片中提取文字，识别结果可编辑、复制。
  • 本地离线处理：引擎和模型准备完成后，识别过程无需联网，图片不上传至云端。
  • 按需安装 OCR：首次使用时下载独立引擎和模型，不将推理运行库捆绑进主安装包。
  • 离线导入引擎：支持从 GitHub Releases 手动下载 OCR 引擎 ZIP 后导入，方便自动下载不畅时使用。
  • 下载与完整性校验：提供下载阶段提示、失败重试，并校验引擎和模型文件的完整性。

优化调整

  • OCR 引擎和模型版本由 TrayClip 版本统一管理，避免自动追随上游更新带来的兼容性问题。
  • 减少重复校验逻辑和无用状态字段，提升代码及测试脚本的可维护性。

问题修复

  • 修复 Windows 卸载时勾选“清理应用数据”后，安装目录中的历史记录、图片及 OCR 缓存仍被保留的问题。
  • 未勾选清理时继续保留数据，自动更新不主动清理数据。
  • 修复 Linux、macOS 测试构建中 Windows 专用函数导入失败的问题。
  • 修复 CI 中前端产物缺失、Windows 链接器冲突，以及 OCR manifest 路径格式不兼容导致的构建失败。

使用与升级说明

  • OCR 首次使用需要下载引擎和模型，耗时取决于网络状况；准备完成后可离线使用。
  • OCR 引擎通过 GitHub Releases 分发，也可下载对应平台的引擎 ZIP 后手动导入。
  • Windows 清理修复的生效范围：首次从旧版升级时仍会调用旧版卸载器，清理选项可能依然无效。安装本版本后，后续调用新版卸载器时该修复才会生效。

TrayClip v0.3.2

This update introduces offline OCR (Optical Character Recognition) capabilities. By utilizing an on-demand, separately downloaded engine, we maintain a compact main installation package while enabling offline functionality. Additionally, this release addresses data cleanup issues during uninstallation on Windows and refines the multi-platform build process.

New Features

  • Image-to-Text (OCR): Extract text from images stored in the clipboard; the recognized text can be edited and copied.
  • Local Offline Processing: Once the engine and models are set up, recognition occurs entirely offline—no images are uploaded to the cloud.
  • On-Demand OCR Installation: The OCR engine and models are downloaded only when first needed, keeping the inference runtime libraries out of the main installation package.
  • Offline Engine Import: Users can manually download the OCR engine ZIP file from GitHub Releases and import it—a useful fallback if automatic downloading fails.
  • Download Management & Integrity Verification: Includes progress notifications, retry logic for failed downloads, and integrity checks for engine and model files.

Improvements & Adjustments

  • OCR engine and model versions are now managed in sync with the TrayClip version, preventing compatibility issues that could arise from automatically tracking upstream updates.
  • Reduced redundant verification logic and unnecessary state fields, improving the maintainability of the codebase and test scripts.

Bug Fixes

  • Fixed an issue on Windows where historical records, images, and OCR cache files remained in the installation directory even when the "Clean up application data" option was selected during uninstallation.

  • Data is now preserved if the cleanup option is not selected; automatic updates will not trigger data deletion.

  • Resolved import failures for Windows-specific functions during Linux and macOS test builds.

  • Fixed build failures caused by missing frontend artifacts in the CI pipeline, Windows linker conflicts, and incompatible path formatting in the OCR manifest. ### Usage and Upgrade Notes

  • Upon first use, the OCR feature requires downloading an engine and a model; the time required depends on your network conditions. Once set up, it can be used offline.

  • The OCR engine is distributed via GitHub Releases; alternatively, you can download the engine ZIP file for your specific platform and import it manually.

  • Scope of Windows Cleanup/Repair Fixes: When upgrading from an older version for the first time, the legacy uninstaller will still be invoked, meaning cleanup options may remain ineffective. The fix will only take effect during subsequent uninstallation processes, once the new version's uninstaller is used.

Assets 16
Loading