v0.2.18

Feature

• Added optional Windows installer components for a desktop shortcut and supported e-book file associations.

Improvement

• Refreshed the installer banner and welcome artwork with a light Torto visual style.
• Updated publisher, copyright, repository, and installer metadata to TortoTech.
• Preserved the existing installation folder and selected components during major upgrades.

Fix

• Pointed in-app update checks and trusted download validation at the TortoTech release repository.
• Registered e-book launch commands against the installed executable path.

中文更新说明

功能

• Windows 安装器新增可选组件，可选择是否创建桌面快捷方式及关联支持的电子书格式。

改进

• 使用清新的 Torto 视觉风格重新设计安装器横幅与欢迎页侧图。
• 将发布者、版权、仓库及安装器元数据统一更新为 TortoTech。
• 进行大版本升级时保留原有安装路径和安装组件选择。

修复

• 将应用更新检查及可信下载地址校验切换到 TortoTech Release 仓库。
• 使用已安装程序的实际路径注册电子书打开命令。

v0.2.17

Feature

• Added localized update notes that follow the interface language.

Improvement

• Upgraded the egui stack to 0.36 and aligned popup sizing, grid layout, text spacing, and native title-bar theme synchronization.
• Reduced book startup work by reusing shelf metadata and covers, opening directly at saved locations, and avoiding duplicate PDF buffer copies.
• Split oversized single-file books at top-level chapter boundaries to reduce initial pagination work.

Fix

• Fixed semantic text selection and toolbar positioning on the right page in two-page mode.
• Restored mouse-wheel page turns while a text selection toolbar is visible.
• Restored image preview and copy interactions in reflowed PDF OCR content.

中文更新说明

功能

• 新增跟随界面语言展示的本地化更新说明。

改进

• 将 egui 技术栈升级至 0.36，并统一弹窗尺寸、网格布局、文字间距和原生标题栏的主题同步。
• 复用书架元数据与封面、直接从保存位置打开，并避免重复复制 PDF 缓冲区，减少书籍启动阶段的工作量。
• 按顶层章节边界拆分超大的单文件电子书，减少首次分页的计算量。

修复

• 修复双页模式下右侧页面的语义文字选择与工具栏定位。
• 修复选择工具栏显示时鼠标滚轮无法翻页的问题。
• 恢复 PDF OCR 流式布局中的图片预览与复制交互。

v0.2.16

Feature

• Added provider-based PDF document OCR with PaddleOCR and MinerU, cached results, and switchable original/reflow views.
• Added native reflow rendering for OCR images, formulas, and structured tables, including merged cells and pagination.

Improvement

• Added direct multimodal scanned-PDF summaries that send page images to vision models in a single request.
• Made PDF view switching reuse the active reader and persist mode separately, avoiding full-book reopen and OCR cache rewrites.
• Simplified OCR setup with built-in official service endpoints and secure token storage.

Fix

• Fixed OCR mode changes that required extra interaction before content rerendered and corrected hidden-title alignment.
• Fixed missing OCR images, partial formula parse failures, and lost centered caption/formula alignment.

v0.2.15

Feature

• Registered Torto as an available opener for EPUB, MOBI, AZW, AZW3, FB2, FBZ, CBZ, and PDF files on Windows and macOS.
• Automatically imports externally opened books into the shelf while reusing existing managed copies.
• Added direct multimodal chapter summaries for scanned and image-only PDFs with page-level citations.

Improvement

• Adopted compact OpenAI-style citation markers so models can copy complete references instead of constructing fragile Markdown links.
• Unified PDF page-image extraction for contents recognition and AI reading tools with bounded payloads and four-way concurrency.

Fix

• Fixed EPUB images wrapped in inline elements and improved definition-list content parsing for image-heavy books.
• Fixed streaming, malformed, and adjacent AI citations so icons remain clean, clickable, and correctly linked.
• Fixed Markdown table layout and LaTeX text commands containing CJK units such as 4\text{千卡/克}.
• Fixed macOS builds for Finder file-opening integration by isolating the required AppKit FFI behind a safe platform boundary.

Full Changelog: v0.2.14...v0.2.15

v0.2.14

Feature

• Added AI-powered contents recognition for image-based PDFs using a configurable vision model, four concurrent requests, and a focused first-20-page scan.
• Added long-press image selection with a themed overlay and Ctrl/Cmd+C copying of the original image.

Improvement

• Added persistent generated PDF contents with direct application, regeneration, and an editor for hierarchy, titles, and page numbers.
• Refined PDF header alignment and compacted the contents controls with dedicated recognition and edit icons.
• Updated the in-app updater dialog to render Markdown release notes and use consistent action controls.

Fix

• Preserved the existing Windows installation directory during in-app MSI upgrades.
• Fixed generated PDF contents appearing empty after reopening a book.
• Prevented Windows and macOS asset workflows from appending duplicate changelogs to an existing release.

Full Changelog: v0.2.13...v0.2.14

v0.2.13

Feature

• Added viewport-prioritized incremental translation that replaces each completed batch immediately.

Improvement

• Reduced AI context overhead with a compact tool protocol, smaller tool responses, and a concise system prompt.

Fix

• Fixed internal citations by standardizing the link:// protocol, enforcing valid Markdown output, and preserving citation clicks while responses stream.
• Fixed wrapped Markdown tables and selection artifacts around list markers and table cells.
• Fixed long-response text selection so edge dragging scrolls automatically and offscreen endpoints remain selected.

Full Changelog: v0.2.12...v0.2.13

v0.2.12

Feature

• Added secure automatic update checks, verified MSI downloads, and install-and-restart support on Windows.
• Added an About tab that displays the current version and provides separate Check and Update actions.
• Added a desktop shortcut to the Windows installer.

Improvement

• Redesigned the English and Chinese READMEs with clearer feature tables.
• Persisted the selected text-selection mode across books and application restarts.
• Refined text-selection mode labels and aligned About-tab controls with the rest of Settings.
• Improved Windows taskbar branding by supplying an explicit taskbar icon.

Fix

• Fixed semantic-selection toolbar placement on the right page in double-page mode.
• Included sentence- and paragraph-ending punctuation when dragging across multiple words while keeping single-word clicks precise.
• Fixed dynamic Settings buttons that could wrap vertically and expand the AI Providers tab.
• Prevented update-check status text from resizing its button and shifting the Settings layout.

Full Changelog: v0.2.11...v0.2.12

v0.2.11

Feature

• Added a chapter-based Scroll reading mode with continuous vertical reading and previous/next chapter navigation.
• Added an F11 shortcut for toggling borderless fullscreen mode.
• Added Free, Word, Sentence, and Paragraph text-selection modes with both click and drag interactions.
• Added reader menu shortcuts for switching the reading layout and color theme.

Improvement

• Reorganized the reader toolbar around sidebar, bookshelf, and translation actions while preserving hover-to-show and automatic hiding.
• Adjusted the current chapter title to align with the reading column when the toolbar is hidden and center itself when the toolbar is visible.
• Made semantic selection Unicode-aware and able to span logical pages and content fragments.
• Persisted Scroll mode through reader preferences and kept menu changes synchronized with global settings.

Fix

• Fixed chapter navigation controls and end-of-chapter spacing in Scroll mode across different viewport sizes.
• Fixed copied text gaining artificial spaces when a word is split across pagination fragments.
• Fixed semantic click selection being cleared by the existing single-click interaction path.
• Fixed reader toolbar controls and chapter-title visibility during toolbar transitions.

Full Changelog: v0.2.10...v0.2.11

v0.2.10

Feature

• 新增正文图片放大预览，支持滚轮缩放和拖拽平移。
• 新增图片预览与正文选区的 Ctrl+C 快捷复制。

Improvement

• 优化超长 EPUB 目录索引、活动项解析和可见行渲染性能。
• 优化文字划选起点判定，使选择行为更符合字符视觉边界。
• 自研 EPUB/MOBI 与 PDF 元数据解析层，并按需加载 SVG 图标、HTTP 图片和目标平台图形后端，减少无用依赖与分发包冗余。
• 发布说明改为在创建 Release 时根据提交生成，不再维护仓库内固定说明文件。

Fix

• 修复部分包含公共 DOCTYPE、裸 ampersand 或异常 ZIP 目录信息的 EPUB 无法打开的问题。
• 修复超长 EPUB 翻页和目录操作性能较差的问题。
• 修复两端对齐文字行尾选区高亮不完整的问题。
• 修复目录滚动到底部后点击条目产生抖动，同时保留普通条目的定位动画。
• 修复 AI Chat 侧边栏打开后输入框未自动聚焦的问题。

Full Changelog: v0.2.9...v0.2.10

Full Changelog: v0.2.9...v0.2.10

Full Changelog: v0.2.9...v0.2.10

v0.2.9

Torto v0.2.9

本版本补齐桌面端 AI 阅读能力，并改善 Windows 界面渲染与阅读交互。

Feature

• 新增界面字体设置，可从系统字体中选择应用界面与 AI 对话字体。
• 新增界面字号设置，可独立于阅读正文调整应用界面与 AI 对话字号。
• 新增 AI 批注工具，支持创建、更新和删除批注，并在执行前由用户确认。
• 新增 AI 正文改写工具，支持查询、应用和回滚正文改写。
• 补齐 AI 图书元数据、目录、章节内容、选区、全文搜索和批注查询工具。

Improvement

• AI Chat 改为按需调用阅读工具获取正文，不再默认把当前正文整体注入提示词。
• AI 系统提示词和工具语义与 rebook-web 对齐，并保留桌面端阅读上下文。
• 阅读上下文现在区分章节、固定版式页面、目录项和章节内阅读进度。
• PDF 章节内容工具会合并该章节关联的全部页面内容。
• 优化 AI 对话、设置页和工具栏的界面字体缩放与清晰度。
• 优化选择框、单选按钮和焦点框的像素对齐与主题描边。
• 优化小型圆角图标按钮的抗锯齿边缘，避免圆角斜线和暗色 fringe。
• 将鼠标滚轮翻页冷却时间从 180 毫秒缩短为 120 毫秒。

Fix

• 修复 AI Chat 未稳定生成正文引用的问题。
• 修复内部引用仍使用 rebook:/ 前缀的问题，统一改为 link:/。
• 修复部分引用点击后不跳转或不高亮对应正文的问题。
• 修复标题和表格中的引用无法正确响应点击的问题。
• 修复引用显示为中文文本的问题，统一改为图标按钮。
• 修复 PDF 获取章节内容时只返回当前章节第一页的问题。
• 修复系统提示词只显示不准确的当前章节索引的问题。
• 修复界面字体、AI 对话文字和控件边缘发糊的问题。
• 修复全局圆角抗锯齿调整导致小型图标按钮出现角线的问题。

Full Changelog: v0.2.8...v0.2.9