fix#14592
### Description
> 🤖 **Note:** The code in this Pull Request were developed with the
assistance of AI, but have been thoroughly reviewed and manually tested.
> I noticed there's a check when opening an issue that asks _"Is your
content generated by AI?"_, so I mention it here in case it's a deal
breaker. If so I understand, you can close the PR, just wanted to share
this in case it's useful anyways.
This PR introduces **Obsidian Vault Import Support** to AFFiNE.
Previously, users migrating from Obsidian had to rely on the generic
Markdown importer, which often resulted in broken cross-links, missing
directory structures, and metadata conflicts because Obsidian relies
heavily on proprietary structures not supported by standard Markdown.
This completely new feature makes migrating to AFFiNE easy.
**Key Features & Implementations:**
1. **Vault (Directory) Selection**
- Utilizes the `openDirectory` blocksuite utility in the import modal to
allow users to select an entire folder directly from their filesystem,
maintaining file context rather than forcing `.zip` uploads.
2. **Wikilink Resolution (Two-Pass Import)**
- Restructured the `importObsidianVault` process into a two-pass
architecture.
- **Pass 1:** Discovers all files, assigns new AFFiNE document IDs, and
maps them efficiently (by title, alias, and filename) into a
high-performance hash map.
- **Pass 2:** Processes the generic markdown AST and correctly maps
custom `[[wikilinks]]` to the actual pre-registered AFFiNE blocksuite
document IDs via `obsidianWikilinkToDeltaMatcher`.
- Safely strips leading emojis from wikilink aliases to prevent
duplicated page icons rendering mid-sentence.
3. **Emoji Metadata & State Fixes**
- Implemented an aggressive, single-pass RegExp to extract multiple
leading/combining emojis (`Emoji_Presentation` / `\ufe0f`) from H1
headers and Frontmatter. Emojis are assigned specifically to the page
icon metadata property and cleanly stripped from the visual document
title.
- Fixed a core mutation bug where the loop iterating over existing
`docMetas` was aggressively overwriting newly minted IDs for the current
import batch. This fully resolves the issue where imported pages
(especially re-imports) were incorrectly flagged as `trashed`.
- Enforces explicit `trash: false` patch instructions.
4. **Syntax Conversion**
- Implemented conversion of Obsidian-style Callouts (`> [!NOTE] Title`)
into native AFFiNE block formats (`> 💡 **Title**`).
- Hardened the `blockquote` parser so that nested structures (like `> -
list items`) are fully preserved instead of discarded.
### UI Changes
- Updated the Import Modal to include the "Import Obsidian Vault" flow
utilizing the native filesystem directory picker.
- Regenerated and synced `i18n-completenesses.json` correctly up to 100%
across all supported locales for the new modal string additions.
### Testing Instructions
1. Navigate to the Workspace sidebar and click "Import".
2. Select "Obsidian" and use the directory picker to define a
comprehensive Vault folder.
3. Validate that cross-links between documents automatically resolve to
their specific AFFiNE instances.
4. Validate documents containing leading Emojis display exactly one
Emoji (in the page icon area), and none duplicated in the actual title
header.
5. Validate Callouts are rendered cleanly and correctly, and no
documents are incorrectly marked as "Trash".
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Import Obsidian vaults with wikilink resolution, emoji/title
preservation, asset handling, and automatic document creation.
* Folder-based imports via a Directory Picker (with hidden-input
fallback) integrated into the import dialog.
* **Localization**
* Added Obsidian import label and tooltip translations.
* **Tests**
* Added end-to-end tests validating Obsidian vault import and asset
handling.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: DarkSky <25152247+darkskygit@users.noreply.github.com>
Co-authored-by: DarkSky <darksky2048@gmail.com>
#### PR Dependency Tree
* **PR #14505** 👈
This tree was auto-generated by
[Charcoal](https://github.com/danerwilliams/charcoal)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Improvements**
* Better image-proxy detection to avoid double-proxying already proxied
images.
* Improved runtime image proxy configuration so images load consistently
across deployments.
* More robust image URL handling for optimized image loading and fewer
redundant requests.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
#### PR Dependency Tree
* **PR #14452** 👈
This tree was auto-generated by
[Charcoal](https://github.com/danerwilliams/charcoal)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Bug Fixes**
* Improved null-safety, dependency tracking, upload validation, and
error logging for more reliable uploads, clipboard, calendar linking,
telemetry, PDF/theme printing, and preview/zoom behavior.
* Tightened handling of all-day calendar events (missing date now
reported).
* **Deprecations**
* Removed deprecated RadioButton and RadioButtonGroup; use RadioGroup.
* **Chores**
* Unified and upgraded linting/config, reorganized imports, and
standardized binary handling for more consistent builds and tooling.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
I used [pdfmake](https://www.npmjs.com/package/pdfmake) to implement an
"export as PDF" feature, and I am happy to share with you!
This should fix#13577, fix#8846, and fix#13959.
A showcase:
[Getting
Started.pdf](https://github.com/user-attachments/files/24013057/Getting.Started.pdf)
Although it might miss rendering some properties currently, it can
evolve in the long run and provide a more native experience for the
users.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
- Experimental "Export to PDF" option added to the export menu (behind a
feature flag)
- PDF export supports headings, paragraphs, lists, code blocks, tables,
images, callouts, linked documents and embedded content
* **Chores**
- Added PDF rendering library and consolidated PDF utilities
- Feature flag introduced to control rollout
* **Tests**
- Comprehensive unit tests added for PDF content rendering logic
<sub>✏️ Tip: You can customize this high-level summary in your review
settings.</sub>
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: DarkSky <darksky2048@gmail.com>
Support importing .docx files, as mentioned in
https://github.com/toeverything/AFFiNE/issues/10154#issuecomment-2655744757
It essentially uses mammoth to convert the docx to html, and then
imports the html with the standard steps.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Import Microsoft Word (.docx) files directly via the import dialog
(creates new documents).
* .docx added as a selectable file type in the file picker and import
options.
* **Localization**
* Added localized labels and tooltips for DOCX import in English,
Simplified Chinese, and Traditional Chinese.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: DarkSky <25152247+darkskygit@users.noreply.github.com>
Co-authored-by: DarkSky <darksky2048@gmail.com>
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
## Summary by CodeRabbit
- **New Features**
- Improved resource management by introducing explicit cleanup for various middleware components, ensuring that resources are properly released when no longer needed.
- **Refactor**
- Updated middleware logic to support cleanup functions, enhancing the stability and performance of the application by preventing potential memory leaks.
- **Chores**
- Enhanced lifecycle management in core systems to automatically dispose of resources when appropriate.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Closes: [BS-3508](https://linear.app/affine-design/issue/BS-3508/google-docs复制内容到affine时自动加粗问题)
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
## Summary by CodeRabbit
- **New Features**
- Improved detection of bold, italic, underline, and strike-through formatting in imported HTML, supporting both tags and inline CSS styles.
- Enhanced handling of inline elements containing block-level children to ensure correct formatting and structure during HTML import.
- Introduced a plugin that converts inline elements with block-level children into block elements, preserving original tag information.
- **Bug Fixes**
- Resolved issues where block-level elements nested inside inline tags could cause incorrect formatting or structure.
- **Tests**
- Added comprehensive test coverage for HTML formatting conversions and plugin behavior to ensure accuracy and reliability.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
## Summary by CodeRabbit
- **New Features**
- Image content now includes file size information in its metadata.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Closes: [BS-3491](https://linear.app/affine-design/issue/BS-3491/支持-callout-导出成-vitepress-文档站支持的-markdown-格式)
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
## Summary by CodeRabbit
- **New Features**
- Added support for exporting callout blocks in both GFM and Admonition markdown styles, including configurable admonition types and custom titles.
- Introduced middleware to customize callout export options for markdown serialization.
- **Tests**
- Enhanced test coverage for callout markdown export, including scenarios with and without export middleware for different styles and content structures.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Closes: [BS-3385](https://linear.app/affine-design/issue/BS-3385/markdown类型的导入，支持media文件和md文件不在同目录的情况)
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
## Summary by CodeRabbit
- **New Features**
- Added utility functions to resolve and normalize image file paths in markdown and HTML imports.
- Introduced middleware to provide full file path context during file import and transformation.
- Added new types for improved asset and file management in zip imports.
- **Refactor**
- Centralized and simplified image processing logic across HTML, Markdown, and Notion HTML adapters for improved maintainability.
- Enhanced type safety and clarity in file and asset handling during zip imports.
- **Tests**
- Added comprehensive tests for image file path resolution utility.
- **Documentation**
- Improved inline code comments explaining file path resolution logic.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
## Summary by CodeRabbit
- **Refactor**
- Improved the internal structure and clarity of ID handling during import processes, leading to more maintainable and modular code. No changes to user-facing functionality.
- **Chores**
- Enhanced type definitions for import events to improve code readability and maintainability. No impact on end-user experience.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Closes: [BS-3486](https://linear.app/affine-design/issue/BS-3486/粘贴从-notion-复制的内容出错)
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
## Summary by CodeRabbit
- **Bug Fixes**
- Improved handling of Notion text segments with empty or invalid style arrays, ensuring plain text and styled text are both processed correctly and preventing errors from malformed input.
- **Tests**
- Added a test case to verify correct conversion of Notion text with empty styles arrays.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Closes: [BS-3358](https://linear.app/affine-design/issue/BS-3358/remark-callout-plugin)
Closes: [BS-3247](https://linear.app/affine-design/issue/BS-3247/callout-markdown-adapter-适配)
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
## Summary by CodeRabbit
- **New Features**
- Added support for callout blocks in Markdown, enabling recognition and conversion of callout syntax (e.g., `[!emoji]`) to and from block structures.
- **Bug Fixes**
- Improved handling to distinguish callout blocks from regular blockquotes and paragraphs during Markdown processing.
- **Tests**
- Introduced comprehensive tests for callout block serialization, deserialization, and plugin behavior to ensure correct Markdown handling.
- **Chores**
- Added a new dependency for Markdown AST traversal.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Closes: [BS-3122](https://linear.app/affine-design/issue/BS-3122/footnote-definition-adapter-适配)
Closes: [BS-3123](https://linear.app/affine-design/issue/BS-3123/几个-block-card-view-适配-footnote-态)
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
## Summary by CodeRabbit
- **New Features**
- Introduced a new citation card component and web element for displaying citations.
- Added support for citation-style rendering in attachment, bookmark, and linked document blocks.
- Enabled citation parsing from footnote definitions in markdown for attachments, bookmarks, and linked docs.
- Added a feature flag to enable or disable citation features.
- Provided new toolbar logic to disable downloads for citation-style attachments.
- **Improvements**
- Updated block models and properties to support citation identifiers.
- Added localization and settings for the citation experimental feature.
- Enhanced markdown adapters to recognize and process citation footnotes.
- Included new constants and styles for citation card display.
- **Bug Fixes**
- Ensured readonly state is respected in block interactions and rendering for citation blocks.
- **Documentation**
- Added exports and effects for new citation components and features.
- **Tests**
- Updated snapshots to include citation-related properties in block data.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
## Summary by CodeRabbit
- **New Features**
- Images and icons in bookmark cards are now loaded through an image proxy for improved reliability and consistency.
- Embed blocks for GitHub, Loom, and YouTube now display banner and creator images via an image proxy service for enhanced image loading.
- **Refactor**
- Simplified backend URL handling and proxy logic for images, resulting in more efficient processing and reduced complexity.
- Consolidated image proxy middleware and services into a shared adapter module for streamlined imports and improved maintainability.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
To close [BS-2870](https://linear.app/affine-design/issue/BS-2870/支持识别-和-[-包裹内容为公式)
## Add Markdown Preprocessor Extension and Enhanced LaTeX Support
### Markdown Preprocessor Extension
This PR introduces a new preprocessor extension for Markdown adapters that allows preprocessing of content before conversion:
Adds a new PreprocessorManager for handling text transformations
Introduces extensible preprocessor interface that supports different processing levels (block/slice/doc)
Integrates preprocessor extension into the existing Markdown adapter workflow
### LaTeX Support Enhancement
Extends LaTeX support to handle both traditional and alternative syntax:
Adds support for backslash LaTeX syntax:
Block math: ```\[...\] ``` alongside existing ```$$...$$```
Inline math: ```\(...\) ``` alongside existing ```$...$```
Implements LaTeX preprocessor to standardize syntax before conversion
Updates tests to cover both syntax variants
### TL;DR
Moved image proxy middleware and adapter extensions to their respective packages and introduced a new spec provider for adapter registration.
### What changed?
- Relocated `defaultImageProxyMiddleware` from blocks to `@blocksuite/affine-block-image`
- Moved `PresentTool` from fragment-frame-panel to block-frame
- Created new adapter extension specs for HTML, Markdown, and Notion HTML
- Introduced a spec provider pattern for adapter registration
- Removed direct transformer references from RootService
- Updated imports across affected files to use new locations
### How to test?
1. Verify image proxy functionality works in exports and imports
2. Test HTML, Markdown, and Notion HTML adapters still function correctly
3. Confirm presentation mode works with the relocated PresentTool
4. Check that all file import/export operations continue to work as expected
### Why make this change?
This reorganization improves code modularity by placing features in their logical packages and introduces a more maintainable pattern for adapter registration. The spec provider pattern makes it easier to manage and extend adapter functionality while reducing coupling between components.
DarkSky