Phase 4: Features

.gitignore

@@ -34,3 +34,10 @@ cache/
 
 # Build artifacts
 *.log
+
+#Playwright
+playwright-report/
+test-results/
+
+# AI Tools
+beledarian_info.md

AGENTS.md

@@ -0,0 +1,40 @@
+# Praetorian AI Agent Instructions
+
+These instructions help AI coding agents understand the workspace structure, build commands, and the current model/face-detection architecture.
+
+## Key points
+
+- This is a Tauri desktop app with a React frontend in `src/` and a Rust backend in `src-tauri/`.
+- Frontend UI calls backend commands through `src/services/catalog.ts` and state is managed in `src/hooks/useCatalog.ts`.
+- The Rust backend exposes catalog, image, face, and Python-upscaling commands in `src-tauri/src/lib.rs`.
+- The catalog engine is implemented in `src-tauri/src/catalog/manager.rs` with SQLite and image metadata handling.
+- Face detection is implemented under `src-tauri/src/face/` using ONNX Runtime and a RetinaFace model.
+- The Python bridge is implemented under `src-tauri/src/python/` and is used for external upscaling workflows.
+
+## Model / AI conventions
+
+- Model code is primarily in Rust; do not move face-detection logic into the React frontend.
+- The face detection flow is feature-gated with `#[cfg(feature = "face-detection")]` and initializes `retinaface.onnx` in `src-tauri/src/lib.rs`.
+- Use the existing Tauri command patterns when adding or changing model features: backend command -> `src/services/catalog.ts` invoke -> frontend hook/component.
+- Prefer GPU execution providers for ONNX Runtime when available, but maintain fallback behavior.
+
+## Build and run commands
+
+- `npm install` to install the frontend dependencies.
+- `npm run dev` to start Vite in development.
+- `npm run build` to build the frontend (`tsc && vite build`).
+- `npm run tauri` to run the Tauri app.
+- `npm run tauri build` to build the desktop application in `src-tauri/target/release`.
+
+## Documentation links
+
+- Project overview: [README.md](README.md)
+- AI / model requirements: [REQUIREMENTS.md](REQUIREMENTS.md)
+- Initial Prompt: [PROMPT.md](PROMPT.md)
+
+## Guidance for AI agents
+
+- Keep UI and backend responsibilities distinct.
+- Preserve the existing Tauri command and IPC structure.
+- When updating face or model-related behavior, check both Rust backend and TypeScript frontend call sites.
+- Avoid duplicating README or requirements content; link to those docs instead.

PROMPT.md

@@ -18,8 +18,10 @@ Prompt: Implement the libvips background worker to generate thumbnails and smart
 ## Iteration C
 Prompt: "Integrate ONNX Runtime for face detection using GPU acceleration."
 
+
 ---
 
+
 # Full Project Requirements
 
 ## Role
@@ -78,3 +80,147 @@ Please provide:
 4. **Python Bridge Skeleton:** A basic example of how the Rust backend would call a Python script to upscale an image and return the result path.
 
 Start by defining the `Cargo.toml` dependencies required for GPU-accelerated AI and libvips, then outline the core database schema.
+
+
+---
+
+# Design Mockups
+
+```
+Create a high-fidelity desktop application UI mockup for a professional photo management application named "Praetorian."
+
+Design language:
+- Modern, premium creative software
+- Dark mode
+- Inspired by Lightroom Cloud, Capture One, Arc Browser, and Linear
+- Matte charcoal interface (#181818 to #222222)
+- Soft elevation with subtle shadows
+- 8px spacing system
+- Rounded corners (8px)
+- Thin separators instead of heavy borders
+- Clean typography similar to Inter or SF Pro
+- Pixel-perfect alignment
+- Production-quality Figma design
+- Extremely polished UX
+- No futuristic or glass-heavy effects
+
+Window:
+- Native desktop application
+- 16:10 aspect ratio
+- Large 1440p workspace
+- macOS-style window chrome
+- Professional creative application
+
+Layout:
+Left Sidebar
+- Library
+- Collections
+- Favorites
+- Recent Imports
+- External Drives
+- Folder tree
+- Smart Albums
+- Collapsible sections
+- Clean monochrome icons
+
+Center Workspace
+- Primary content area
+- Dense but elegant layout
+- Excellent spacing
+- Strong visual hierarchy
+
+Right Sidebar
+- Context-sensitive inspector
+- Metadata
+- EXIF
+- Camera
+- Lens
+- Histogram
+- Keywords
+- Rating
+- Color labels
+
+The interface should feel like software used by professional photographers managing hundreds of thousands of RAW images.
+
+Generate the "Import Photos" dialog.
+
+The dialog should appear centered over the library.
+
+Dialog layout:
+
+Left Column
+- Connected devices
+- SD cards
+- External SSDs
+- NAS locations
+- Local folders
+- Recently used sources
+
+Center
+- Thumbnail preview grid
+- 30–50 photo thumbnails
+- Portrait photography
+- Every thumbnail is unique
+- Mix of RAW and JPEG indicators
+- Multi-select checkboxes
+- File names
+- Capture dates
+- File sizes
+
+Top Toolbar
+- Breadcrumb navigation
+- Search
+- Filter
+- Sort
+- Select All
+- Number of selected photos
+
+Right Panel
+Import Options
+
+Destination
+- Folder picker
+- Create dated folder
+- Custom folder name
+
+File Handling
+- Copy
+- Move
+- Leave in place
+
+Renaming
+- Rename using template
+- Sequence numbering
+
+Metadata
+- Apply copyright
+- Apply keywords
+- Apply preset
+- Apply star rating
+
+AI Options
+- Generate Smart Keywords
+- Detect Faces
+- Detect Duplicate Photos
+- Estimate Image Quality
+- Build Search Index
+
+Bottom Bar
+- Selected item count
+- Total storage required
+- Cancel button
+- Large blue "Import 248 Photos" button
+
+Visual Details
+- Beautiful thumbnail photography
+- Realistic icons
+- Crisp typography
+- Realistic spacing
+- Professional UI polish
+- Subtle hover states
+- Soft blue accent color
+- Clean monochrome iconography
+- Consistent design system
+
+The mockup should be indistinguishable from a real shipping desktop application screenshot. No placeholder boxes, no lorem ipsum, no wireframe styling.
+```

src-tauri/.cargo/config.toml

@@ -3,6 +3,8 @@ rustflags = ["-C", "target-cpu=native"]
 
 [env]
 SQLX_OFFLINE = "true"
+# vips-sys does not pass VIPS_ROOT/include to bindgen's clang, so we set it here.
+BINDGEN_EXTRA_CLANG_ARGS = { value = "-I'C:/Program Files/vips/vips-dev-8.18/include' -I'C:/Program Files/vips/vips-dev-8.18/include/glib-2.0' -I'C:/Program Files/vips/vips-dev-8.18/lib/glib-2.0/include'", force = true }
 
 [target.x86_64-unknown-linux-gnu]
 linker = "clang"

src-tauri/Cargo.toml

@@ -15,7 +15,7 @@ tauri-build = { version = "2", features = [] }
 
 [dependencies]
 # --- Tauri v2 ---
-tauri = { version = "2" }
+tauri = { version = "2", features = [] }
 tauri-plugin-log = "2"
 tauri-plugin-fs = "2"
 tauri-plugin-dialog = "2"
@@ -27,20 +27,24 @@ serde_json = "1"
 tokio = { version = "1", features = ["full"] }
 
 # --- Database: sqlx with SQLite ---
+# "offline" feature allows compile-time query checking without a live DB.
 sqlx = { version = "0.8", features = ["runtime-tokio-rustls", "sqlite", "chrono", "macros"] }
 
 # --- Image processing ---
-vips = "0.1.0-alpha.5"
+# vips is optional; falls back to the `image` crate when libvips is unavailable.
+vips = { version = "0.1.0-alpha.5", optional = true }
+vips-sys = { version = "0.1.3-beta.2", optional = true }
 image = "0.25"
 
 # --- EXIF / metadata parsing ---
 exif = { package = "kamadak-exif", version = "0.6" }
 
 # --- ONNX Runtime (GPU-accelerated face detection) ---
+# Optional: enable with --features face-detection
 # CUDA support enabled via the "cuda" feature.
 # "load-dynamic" loads CUDA libraries at runtime rather than link time.
-ort = { version = "2.0.0-rc.12", features = ["cuda", "load-dynamic"] }
-ort-sys = "2.0.0-rc.12"
+ort = { version = "2.0.0-rc.12", features = ["cuda", "load-dynamic"], optional = true }
+ort-sys = { version = "2.0.0-rc.12", optional = true }
 
 # --- UUID for hashed cache keys ---
 uuid = { version = "1", features = ["v4"] }
@@ -68,6 +72,8 @@ duct = "0.13"
 [features]
 default = ["custom-protocol"]
 custom-protocol = ["tauri/custom-protocol"]
+vips = ["dep:vips", "dep:vips-sys"]
+face-detection = ["dep:ort", "dep:ort-sys"]
 
 [target.'cfg(not(any(target_os = "android", target_os = "ios")))'.dependencies]
 tauri-plugin-updater = "2"

src-tauri/capabilities/default.json

@@ -0,0 +1,13 @@
+{
+  "$schema": "https://raw.githubusercontent.com/nicegram/tauri/v2.0.0/tooling/cli/src/helpers/json-schemas/capabilities.json",
+  "identifier": "default",
+  "description": "Default capabilities for Praetorian",
+  "windows": ["main"],
+  "permissions": [
+    "core:default",
+    "dialog:allow-open",
+    "dialog:allow-save",
+    "fs:default",
+    "shell:default"
+  ]
+}