Add ability to import meeting notes files from shared drives

[hyper-charlie]

Feature Request: Import Meeting Notes Files

Use Case:
User stores meeting notes files on a shared drive and wants to import them into Hyprnote rather than manually copying files.

Desired Behavior:

• Ability to import meeting notes files (e.g., .txt, .md, .docx, or other formats) from local or shared drives into Hyprnote
• Imported files should be accessible within the app alongside native Hyprnote recordings
• Support for batch imports if possible

Current Workflow:
User is manually copying files from shared drive, which is time-consuming and not ideal for team collaboration.

Environment:

• Platform: macOS
• App Version: 1.0.6
• OS: 26.3.0

[devin-ai-integration]

Codebase Exploration — Relevant Context

Existing Import Infrastructure

Hyprnote already has a plugin-based import system that could be extended for this feature:

plugins/importer/ — Tauri Import Plugin

• src/types.rs: Defines ImportSourceKind enum with current sources: Granola, HyprnoteV0Stable, HyprnoteV0Nightly, AsIs. A new variant (e.g., FileImport) would be needed.
• src/ext.rs: Importer struct with run_import() and run_import_dry() methods that convert source data into TinyBase JSON via to_tinybase_json().
• src/sources/mod.rs: Dispatches import based on TransformKind. Currently supports HyprnoteV0, Granola, and AsIs transforms. A new transform for file-based import would plug in here.
• src/sources/as_is.rs: The AsIs source already demonstrates loading a JSON file and converting it to a Collection — this pattern could be adapted for .txt/.md/.docx files.
• src/commands.rs: IPC commands (list_available_sources, run_import, run_import_dry) exposed to the frontend.

crates/importer-core/ — Core Import Data Model

• src/ir.rs: Defines the intermediate representation (Collection) containing Session, Transcript, EnhancedNote, Human, etc. Imported files would need to be mapped into this IR.
• src/output.rs: to_tinybase_json() converts Collection → TinyBase-compatible JSON for merging into the app's store.

apps/desktop/src/store/tinybase/store/importer.ts — Frontend Import Logic

• importData() function that takes parsed data, validates it as [tables, values] format, and merges it into the TinyBase store via store.merge(importStore).
• convertMarkdownToTiptapJson() already handles converting markdown content to TipTap JSON (the internal rich text format), which is exactly what would be needed for .md file import.
• transformImportedTables() transforms sessions.raw_md and enhanced_notes.content fields from markdown to TipTap JSON.

Existing UI for Import

• apps/desktop/src/components/settings/data/index.tsx: Settings > Data page with the import UI. Lists available sources, runs dry import to preview stats, then runs actual import.
• apps/desktop/src/components/settings/data/source-item.tsx: Renders each import source with a "Scan" button.
• apps/desktop/src/components/settings/data/import-preview.tsx: Shows preview stats before confirming import (session count, transcript count, etc.).

Existing File Dialog Usage

The app already uses Tauri's file dialog in a few places:

• apps/desktop/src/components/onboarding/folder-location.tsx: Uses dialog for folder selection.
• apps/desktop/src/components/main/body/sessions/floating/options-menu.tsx: Has file-related operations.
• apps/desktop/src/hooks/useUploadAudio.ts: File picker for audio upload — good reference for implementing a file picker for notes.

Markdown/Content Processing

• apps/desktop/src/store/tinybase/persister/session/load/note.ts: processMdFile() demonstrates how markdown files with frontmatter are parsed and loaded into sessions. Uses md2json() from @hypr/tiptap/shared to convert markdown → TipTap JSON.
• apps/desktop/src/store/tinybase/persister/factories/markdown-dir.ts: Shows the full pattern for reading/writing markdown files from a directory, with frontmatter support.

Export (Inverse of Import)

For reference, the export side exists:

• apps/desktop/src/components/main/body/sessions/outer-header/overflow/export-pdf.tsx: Exports sessions to PDF.
• apps/desktop/src/components/main/body/sessions/outer-header/overflow/export-transcript.tsx: Exports transcripts to VTT.

Recent Changes in Import Area

Recent commits affecting the importer (from git log):

• ae7239878 — bump tauri 2.10.1
• 2ac10250d — fix: preserve speaker hints during data migration
• 5ede53457 — Remove redundant metadata from Word and SpeakerHint schemas
• ae2c87291 — replace importer-interface with importer-core
• a7910627f — refactor importer
• 4ff6838d4 — refactor importer and fix binding issue

The importer was recently refactored, suggesting the architecture is relatively stable for extension.

---

Potential Approach

1. New ImportSourceKind variant (e.g., Files) in plugins/importer/src/types.rs, with a corresponding TransformKind for file-based import.
2. New source handler in plugins/importer/src/sources/ that:
   • Uses Tauri's file dialog (tauri-plugin-dialog) to let users pick files or a directory from a shared drive.
   • Parses .md files using the existing md2json() / frontmatter parsing utilities.
   • Parses .txt files as plain text sessions.
   • For .docx, would need a new dependency (e.g., a Rust crate for docx parsing or a JS library).
   • Maps parsed content into Collection IR.
3. Batch support: The existing Collection IR already supports multiple sessions, transcripts, etc., so batch import is architecturally supported — just need to iterate over multiple files.
4. UI addition: Either extend the Settings > Data page or add a dedicated "Import Files" action (perhaps in the sidebar or empty state). The existing ImportPreview component already handles showing stats before import.
5. Content conversion: The existing convertMarkdownToTiptapJson() in importer.ts already handles markdown → TipTap conversion for imported content. .txt files could be wrapped in a simple paragraph structure.

Considerations

• .docx support would add a non-trivial dependency. Consider whether this is MVP or can be a follow-up.
• Shared drive access: Tauri's file dialog should handle local/network drives on macOS. No special code needed for mounted shared drives — they appear as regular filesystem paths.
• Metadata mapping: Imported files won't have transcript data or participant info — need to decide what metadata to infer (e.g., filename → title, file date → created_at).
• Deduplication: The existing store.merge() approach handles merging, but imported files need unique IDs to avoid conflicts with existing sessions.