…per (#3239)

Completes the platform matrix promised in #3232.  Linux (evdev) landed
in #3238; this PR adds the Windows (hidapi) and macOS (IOKit) sides.
The mapper dialog drops its Linux guard and now builds everywhere.

Architecture
------------
All three backends expose the same Qt signal contract:
    tuneSteps(int)               - rotary delta
    buttonEvent(sig, action)     - canonical Linux-keycode-name strings
    connectionChanged(bool, name)

A new typedef `UlanziDialBackend` (src/core/UlanziDialBackend.h)
resolves to the right concrete class per platform.  Mapper dialog and
MainWindow dispatcher use only the alias and have no platform switches
of their own.

Windows (UlanziDialWindowsManager)
----------------------------------
- hidapi (already a project dep) enumerates HID devices, matches by
  product_string substring "Ulanzi Dial"
- Opens every matching interface (the dial enumerates as multiple
  HID interfaces on Windows: Keyboard + Consumer Control, possibly
  Mouse) and polls each via QTimer at 5ms intervals
- HID report parsing: standard 8-byte keyboard report (modifier mask
  + up-to-6-keys array, diffed across frames for transitions) and
  Consumer Control reports (2-byte usage code)
- HID usage codes mapped to canonical Linux keycode integers so the
  signature strings (`KEY_PLAYPAUSE`, `Ctrl+V`, ...) match Linux
  exactly and `kPillSpecs` is unchanged
- Chord assembly state machine identical to the Linux backend
- Hot-plug via 3s rescan timer
- LIMITATION: no exclusive-claim equivalent yet.  The dial's keystrokes
  still reach the focused window because hidapi doesn't suppress OS
  input routing.  RawInput + RIDEV_NOLEGACY is the proper fix; tracked
  for a follow-up

macOS (UlanziDialMacOSManager)
------------------------------
- IOKit HID Manager with device-matching dictionary
  ({ kIOHIDProductKey: "Ulanzi Dial" })
- `IOHIDManagerOpen` with `kIOHIDOptionsTypeSeizeDevice` -- the
  documented macOS exclusive-claim mechanism.  Equivalent to Linux
  EVIOCGRAB: when seized, the dial's events stop reaching the OS
  keyboard stack, so its media keys don't leak to the focused window
- Event-driven via `IOHIDManagerRegisterInputValueCallback`; no polling
- Same usage-page-to-Linux-keycode mapping and chord assembly as
  Windows backend
- Scheduled on `CFRunLoopGetMain()` (IOKit requires a CFRunLoop, and
  only the main thread has one integrated with Qt's event loop on
  macOS).  MainWindow conditionally skips `moveToThread` on macOS to
  match this constraint
- Hot-plug native via IOHIDManager device-matching/removal callbacks

Build wiring (CMakeLists.txt)
-----------------------------
- UlanziDialBackend.h + UlanziDialMapperDialog.{cpp,h}: unconditional
- EvdevEncoderManager.{cpp,h}: UNIX AND NOT APPLE (Linux)
- UlanziDialWindowsManager.{cpp,h}: WIN32
- UlanziDialMacOSManager.{cpp,h}: APPLE
- macOS links -framework IOKit and -framework CoreFoundation

Tested
------
- Linux: compiles + runs as before (Linux behaviour unchanged; same
  EvdevEncoderManager underneath the alias)
- Windows + macOS: compile validation gated by CI (check-windows from
  #3210 and check-macos from #3223).  Live testing on hardware happens
  in the user's environment

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

## Summary

Adds **`plugins/ulanzi-aethersdr/`** as the third Stream-Deck-style
sibling alongside `elgato-aethersdr` and `streamcontroller-aethersdr`.
Drives AetherSDR over TCI WebSocket from Ulanzi-Studio-compatible macro
keypads / dials.

- **22 actions**, as both keypad buttons and rotary (Encoder) knobs:
- **Encoder (dial):** VFO Tune, plus rotary **AF Gain / RF Power / Mic
Gain / Filter Width / Squelch Level** (rotate to adjust, press+rotate =
coarse). VFO dial press toggles MOX; level-knob press is a deliberate
no-op so a knob can't key TX.
  - **TX:** MOX Toggle, TUNE / ATU, RIT Toggle.
  - **Modes:** Mode Cycle, plus direct USB / LSB / CW / DIGU.
  - **Bands:** Band Up, Band Down.
  - **Gain (keypad):** AF / RF / Mic ±5 per press.
- TCI command vocabulary kept faithful to `elgato-aethersdr` so both
behave identically against the same radio.
- Iconography generated programmatically from a single config
(`scripts/Generate-Icons.ps1`, PowerShell + `System.Drawing`, no
npm/Node deps) → 196×196 PNGs matching Ulanzi Studio's marketplace icon
convention. Currently Windows-only; a cross-platform port (ImageMagick /
SkiaSharp) is future work.
- 5 coordinated shack-companion launcher tiles (AetherSDR / TCI Monitor
/ ShackLog / IQ Capture / aether-pad) for pairing with Studio's built-in
`System → Open` action.
- Bundled D100H default profile
(`profiles/aethersdr-d100h-default.ulanziDeckProfile`) — one-click
Studio import.
- **License: GPL-3.0-or-later**, matching the parent project. The
vendored Ulanzi SDK in `libs/common-node/` and `libs/common-html/` keeps
its own license.

The change is fully self-contained under `plugins/ulanzi-aethersdr/` —
no C++, CMake, or CI changes.

## Status

- **D100H / KEHWIN Dial_Lite** (6 keys + 1 dial, BLE HOGP): smoke-tested
end-to-end; MOX, AF, RF and **mic gain all confirmed live** against the
radio. (I run this unit on CAD duty now, but it remains a fully
supported radio surface for anyone who wants it.)
- **D200 / Zkswe "ulanzi"** (USB): used for the gain-seeding and
coexistence verification below.
- **D200H** (14 keys): plugin code complete, profile layout in progress.
- **D200X** (14 keys + 3 knobs): unit arriving shortly; profile follows
once on the bench.

### Gain seeding (fixes first-press jump)

On TCI connect the plugin issues GET `volume;` / `drive;` / `mic_level;`
so the AF / RF / mic local mirrors start from the radio's real values.
Without this the first ±5 press jumped from the hard-coded defaults (50
/ 100 / 50). AE's init burst does not emit `volume:`, so this pairs with
the server-side init-burst fix in **#3245** — the GET-on-connect makes
the plugin robust regardless of which AE build it talks to. Verified
live 2026-05-28: all three step smoothly with no jump.

### Mic gain

`mic_level:` is an AE extension (not in the published TCI spec) but it
**is honoured** — verified live 2026-05-28 to actually drive TX mic
gain.

## Coexistence with the native UlanziDial backend

AetherSDR `main` now ships a native UlanziDial HID backend (**#3238**
Linux evdev, **#3239** Windows hidapi + macOS IOKit). This plugin and
that backend do **not** collide — verified live 2026-05-28:

- The native Windows matcher keeps only HID devices whose hidapi
`product_string` contains the case-sensitive substring `"Ulanzi Dial"`
(`src/core/UlanziDialWindowsManager.cpp`, `kProductMatch`).
- `hid.enumerate()` on this machine reports: D100H → product `Dial_Lite`
(manuf `KEHWIN`, VID `0xFFF1` / PID `0x0082`, BLE HOGP); D200 → product
`ulanzi` (manuf `Zkswe`, VID `0x2207` / PID `0x0019`, USB). **Neither
contains `"Ulanzi Dial"`**, so the native backend claims neither device
— both stay free for this plugin via Ulanzi Studio.
- Tested against a clean build of this branch (rebased on `main`,
`-DHAVE_HIDAPI`, native backend compiled in) with both Ulanzi units
plugged in: no double-claim, no conflict.
- Trap for reviewers: Windows Device Manager may display the D100H's PnP
*FriendlyName* as "Ulanzi Dial", but that is a **different field** from
the hidapi `product_string` the matcher reads (which is `Dial_Lite`).

## Repository relationship

Independent public-facing releases / Ulanzi Marketplace path lives at
**[nigelfenton/aethersdr-ulanzi-plugin](https://github.com/nigelfenton/aethersdr-ulanzi-plugin)**.
Both repos stay alive — the standalone is canonical for end-user
downloads; this in-tree copy is bundled context for AetherSDR
developers. The README inside the plugin documents the relationship
explicitly and is the authoritative description of the plugin.

## Test plan

- [x] Plugin source loads in Ulanzi Studio without errors (D100H,
Windows)
- [x] TCI WebSocket connects to AetherSDR on `ws://127.0.0.1:40001`
- [x] MOX action flips radio state (`trx:0,<bool>;`) on D100H
- [x] AF / RF / mic gain seed from the radio's real values on connect —
no first-press jump (2026-05-28)
- [x] `mic_level:` verified to actually drive TX mic gain on AE
(2026-05-28)
- [x] Coexists with the native UlanziDial backend — neither Ulanzi unit
matches `kProductMatch` (2026-05-28)
- [x] D100H default profile imports cleanly into a fresh Studio
- [x] Icon generator (`scripts/Generate-Icons.ps1`) regenerates the PNGs
idempotently
- [ ] D200H full-keypad layout validated (in progress)
- [ ] D200X layout validated (waiting on hardware)
- [ ] macOS Studio compatibility re-confirmed against this in-repo path

## Out of scope / future work

- Per-app sub-profiles (TCI Monitor / ShackLog / IQ Capture /
aether-pad) using Studio's built-in `System.Hotkey` — will land here
once each app's hotkey table is known.
- Cross-platform icon generator (port `Generate-Icons.ps1` to
ImageMagick / SkiaSharp so macOS/Linux contributors can regenerate
without `pwsh`).
- Marketplace submission (independent of this PR).
- A properly-stateful slice-cycle action (the earlier local-index
version was removed in v0.1.7; a future one should be driven by AE's
slice-list event rather than a local counter).
- Broadening the native backend's device match so it also recognises
OEM-rebranded Ulanzi units (`Dial_Lite` / `ulanzi`) — raised separately
on #3239.

---

<sub>_PR description refreshed to match v0.1.7 (rotary knobs,
slice-cycle removed, `Generate-Icons.ps1`, GPL-3.0-or-later). The
in-plugin README remains the authoritative source of truth._</sub>

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-authored-by: jensenpat <patjensen@gmail.com>

…urator (aethersdr#3238)

First-platform delivery of aethersdr#3232 (cross-platform BT-HID encoder
support).
Linux side complete; Windows/macOS land in follow-up PRs.

Backend (EvdevEncoderManager, Linux-only):
- Scans /dev/input/event* for "Ulanzi Dial Keyboard" sub-device
- Opens with EVIOCGRAB to claim it exclusively so the dial's media-key
  emissions (KEY_PREVIOUSSONG/NEXTSONG/PLAYPAUSE/MUTE/Ctrl+chords) don't
  leak to the focused window
- Event-driven via QSocketNotifier; no polling
- Chord assembly state machine decodes Ctrl+key + compound chord events
- Hot-plug via QFileSystemWatcher on /dev/input/

UI (UlanziDialMapperDialog, fixed 640x640):
- Product image of the dial in the centre, action-combo "pills"
  positioned around it at each physical control
- Each pill is a QComboBox listing every registered ShortcutManager
  action + every discrete-type MIDI param (shortcut:<id> / midi:<id>
  tagged so MainWindow dispatcher can route correctly)
- Signature-to-pill mapping is firmware-property (hardcoded in
  kPillSpecs); the function assigned to each pill is user-configurable
  via the combo and persists to AppSettings under
  UlanziDial/action/<pillId>

MainWindow integration:
- Instantiates the manager on the existing ExtControllers thread
- Rotary tuneSteps coalesces into the active slice tune step
  (same pattern as HidEncoderManager from aethersdr#3171)
- buttonEvent dispatcher parses the prefix and invokes the matching
  ShortcutManager handler or MIDI param setter; falls back to the
  per-pill defaultAction so first-launch bindings work without the
  user ever opening the mapper

Files:
- src/core/EvdevEncoderManager.{h,cpp}   - new, Linux-only
- src/gui/UlanziDialMapperDialog.{h,cpp} - new, Linux-only
- resources/images/ulanzi-dial.png       - product image
- src/gui/MainWindow.{h,cpp}             - wiring + menu entry
- CMakeLists.txt - gate new sources on UNIX AND NOT APPLE

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

…per (aethersdr#3239)

Completes the platform matrix promised in aethersdr#3232.  Linux (evdev) landed
in aethersdr#3238; this PR adds the Windows (hidapi) and macOS (IOKit) sides.
The mapper dialog drops its Linux guard and now builds everywhere.

Architecture
------------
All three backends expose the same Qt signal contract:
    tuneSteps(int)               - rotary delta
    buttonEvent(sig, action)     - canonical Linux-keycode-name strings
    connectionChanged(bool, name)

A new typedef `UlanziDialBackend` (src/core/UlanziDialBackend.h)
resolves to the right concrete class per platform.  Mapper dialog and
MainWindow dispatcher use only the alias and have no platform switches
of their own.

Windows (UlanziDialWindowsManager)
----------------------------------
- hidapi (already a project dep) enumerates HID devices, matches by
  product_string substring "Ulanzi Dial"
- Opens every matching interface (the dial enumerates as multiple
  HID interfaces on Windows: Keyboard + Consumer Control, possibly
  Mouse) and polls each via QTimer at 5ms intervals
- HID report parsing: standard 8-byte keyboard report (modifier mask
  + up-to-6-keys array, diffed across frames for transitions) and
  Consumer Control reports (2-byte usage code)
- HID usage codes mapped to canonical Linux keycode integers so the
  signature strings (`KEY_PLAYPAUSE`, `Ctrl+V`, ...) match Linux
  exactly and `kPillSpecs` is unchanged
- Chord assembly state machine identical to the Linux backend
- Hot-plug via 3s rescan timer
- LIMITATION: no exclusive-claim equivalent yet.  The dial's keystrokes
  still reach the focused window because hidapi doesn't suppress OS
  input routing.  RawInput + RIDEV_NOLEGACY is the proper fix; tracked
  for a follow-up

macOS (UlanziDialMacOSManager)
------------------------------
- IOKit HID Manager with device-matching dictionary
  ({ kIOHIDProductKey: "Ulanzi Dial" })
- `IOHIDManagerOpen` with `kIOHIDOptionsTypeSeizeDevice` -- the
  documented macOS exclusive-claim mechanism.  Equivalent to Linux
  EVIOCGRAB: when seized, the dial's events stop reaching the OS
  keyboard stack, so its media keys don't leak to the focused window
- Event-driven via `IOHIDManagerRegisterInputValueCallback`; no polling
- Same usage-page-to-Linux-keycode mapping and chord assembly as
  Windows backend
- Scheduled on `CFRunLoopGetMain()` (IOKit requires a CFRunLoop, and
  only the main thread has one integrated with Qt's event loop on
  macOS).  MainWindow conditionally skips `moveToThread` on macOS to
  match this constraint
- Hot-plug native via IOHIDManager device-matching/removal callbacks

Build wiring (CMakeLists.txt)
-----------------------------
- UlanziDialBackend.h + UlanziDialMapperDialog.{cpp,h}: unconditional
- EvdevEncoderManager.{cpp,h}: UNIX AND NOT APPLE (Linux)
- UlanziDialWindowsManager.{cpp,h}: WIN32
- UlanziDialMacOSManager.{cpp,h}: APPLE
- macOS links -framework IOKit and -framework CoreFoundation

Tested
------
- Linux: compiles + runs as before (Linux behaviour unchanged; same
  EvdevEncoderManager underneath the alias)
- Windows + macOS: compile validation gated by CI (check-windows from
  aethersdr#3210 and check-macos from aethersdr#3223).  Live testing on hardware happens
  in the user's environment

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

…r#3227)

## Summary

Adds **`plugins/ulanzi-aethersdr/`** as the third Stream-Deck-style
sibling alongside `elgato-aethersdr` and `streamcontroller-aethersdr`.
Drives AetherSDR over TCI WebSocket from Ulanzi-Studio-compatible macro
keypads / dials.

- **22 actions**, as both keypad buttons and rotary (Encoder) knobs:
- **Encoder (dial):** VFO Tune, plus rotary **AF Gain / RF Power / Mic
Gain / Filter Width / Squelch Level** (rotate to adjust, press+rotate =
coarse). VFO dial press toggles MOX; level-knob press is a deliberate
no-op so a knob can't key TX.
  - **TX:** MOX Toggle, TUNE / ATU, RIT Toggle.
  - **Modes:** Mode Cycle, plus direct USB / LSB / CW / DIGU.
  - **Bands:** Band Up, Band Down.
  - **Gain (keypad):** AF / RF / Mic ±5 per press.
- TCI command vocabulary kept faithful to `elgato-aethersdr` so both
behave identically against the same radio.
- Iconography generated programmatically from a single config
(`scripts/Generate-Icons.ps1`, PowerShell + `System.Drawing`, no
npm/Node deps) → 196×196 PNGs matching Ulanzi Studio's marketplace icon
convention. Currently Windows-only; a cross-platform port (ImageMagick /
SkiaSharp) is future work.
- 5 coordinated shack-companion launcher tiles (AetherSDR / TCI Monitor
/ ShackLog / IQ Capture / aether-pad) for pairing with Studio's built-in
`System → Open` action.
- Bundled D100H default profile
(`profiles/aethersdr-d100h-default.ulanziDeckProfile`) — one-click
Studio import.
- **License: GPL-3.0-or-later**, matching the parent project. The
vendored Ulanzi SDK in `libs/common-node/` and `libs/common-html/` keeps
its own license.

The change is fully self-contained under `plugins/ulanzi-aethersdr/` —
no C++, CMake, or CI changes.

## Status

- **D100H / KEHWIN Dial_Lite** (6 keys + 1 dial, BLE HOGP): smoke-tested
end-to-end; MOX, AF, RF and **mic gain all confirmed live** against the
radio. (I run this unit on CAD duty now, but it remains a fully
supported radio surface for anyone who wants it.)
- **D200 / Zkswe "ulanzi"** (USB): used for the gain-seeding and
coexistence verification below.
- **D200H** (14 keys): plugin code complete, profile layout in progress.
- **D200X** (14 keys + 3 knobs): unit arriving shortly; profile follows
once on the bench.

### Gain seeding (fixes first-press jump)

On TCI connect the plugin issues GET `volume;` / `drive;` / `mic_level;`
so the AF / RF / mic local mirrors start from the radio's real values.
Without this the first ±5 press jumped from the hard-coded defaults (50
/ 100 / 50). AE's init burst does not emit `volume:`, so this pairs with
the server-side init-burst fix in **aethersdr#3245** — the GET-on-connect makes
the plugin robust regardless of which AE build it talks to. Verified
live 2026-05-28: all three step smoothly with no jump.

### Mic gain

`mic_level:` is an AE extension (not in the published TCI spec) but it
**is honoured** — verified live 2026-05-28 to actually drive TX mic
gain.

## Coexistence with the native UlanziDial backend

AetherSDR `main` now ships a native UlanziDial HID backend (**aethersdr#3238**
Linux evdev, **aethersdr#3239** Windows hidapi + macOS IOKit). This plugin and
that backend do **not** collide — verified live 2026-05-28:

- The native Windows matcher keeps only HID devices whose hidapi
`product_string` contains the case-sensitive substring `"Ulanzi Dial"`
(`src/core/UlanziDialWindowsManager.cpp`, `kProductMatch`).
- `hid.enumerate()` on this machine reports: D100H → product `Dial_Lite`
(manuf `KEHWIN`, VID `0xFFF1` / PID `0x0082`, BLE HOGP); D200 → product
`ulanzi` (manuf `Zkswe`, VID `0x2207` / PID `0x0019`, USB). **Neither
contains `"Ulanzi Dial"`**, so the native backend claims neither device
— both stay free for this plugin via Ulanzi Studio.
- Tested against a clean build of this branch (rebased on `main`,
`-DHAVE_HIDAPI`, native backend compiled in) with both Ulanzi units
plugged in: no double-claim, no conflict.
- Trap for reviewers: Windows Device Manager may display the D100H's PnP
*FriendlyName* as "Ulanzi Dial", but that is a **different field** from
the hidapi `product_string` the matcher reads (which is `Dial_Lite`).

## Repository relationship

Independent public-facing releases / Ulanzi Marketplace path lives at
**[nigelfenton/aethersdr-ulanzi-plugin](https://github.com/nigelfenton/aethersdr-ulanzi-plugin)**.
Both repos stay alive — the standalone is canonical for end-user
downloads; this in-tree copy is bundled context for AetherSDR
developers. The README inside the plugin documents the relationship
explicitly and is the authoritative description of the plugin.

## Test plan

- [x] Plugin source loads in Ulanzi Studio without errors (D100H,
Windows)
- [x] TCI WebSocket connects to AetherSDR on `ws://127.0.0.1:40001`
- [x] MOX action flips radio state (`trx:0,<bool>;`) on D100H
- [x] AF / RF / mic gain seed from the radio's real values on connect —
no first-press jump (2026-05-28)
- [x] `mic_level:` verified to actually drive TX mic gain on AE
(2026-05-28)
- [x] Coexists with the native UlanziDial backend — neither Ulanzi unit
matches `kProductMatch` (2026-05-28)
- [x] D100H default profile imports cleanly into a fresh Studio
- [x] Icon generator (`scripts/Generate-Icons.ps1`) regenerates the PNGs
idempotently
- [ ] D200H full-keypad layout validated (in progress)
- [ ] D200X layout validated (waiting on hardware)
- [ ] macOS Studio compatibility re-confirmed against this in-repo path

## Out of scope / future work

- Per-app sub-profiles (TCI Monitor / ShackLog / IQ Capture /
aether-pad) using Studio's built-in `System.Hotkey` — will land here
once each app's hotkey table is known.
- Cross-platform icon generator (port `Generate-Icons.ps1` to
ImageMagick / SkiaSharp so macOS/Linux contributors can regenerate
without `pwsh`).
- Marketplace submission (independent of this PR).
- A properly-stateful slice-cycle action (the earlier local-index
version was removed in v0.1.7; a future one should be driven by AE's
slice-list event rather than a local counter).
- Broadening the native backend's device match so it also recognises
OEM-rebranded Ulanzi units (`Dial_Lite` / `ulanzi`) — raised separately
on aethersdr#3239.

---

<sub>_PR description refreshed to match v0.1.7 (rotary knobs,
slice-cycle removed, `Generate-Icons.ps1`, GPL-3.0-or-later). The
in-plugin README remains the authoritative source of truth._</sub>

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-authored-by: jensenpat <patjensen@gmail.com>