[aceRage]

Description

Depends on #625 (merge #625 first).

#625 adoption: the shared map panel is used only for Happy Hare, AFC/BoxTurtle, and QIDI Box hosts — not Snapmaker/U1. Gates/lanes/Box slots group as 1A–1D with colour and type. Sent data is unchanged (Slot::index, QIDI SAVE_VARIABLE value_tN / filament_mapping()).

Klipper printers with Happy Hare or AFC/BoxTurtle can now populate the existing print-host send-dialog mapping table. Port of the parsing only from upstream OrcaSlicer src/slic3r/Utils/MoonrakerPrinterAgent.cpp into this fork's PrintHostDeviceStatus probe. The IPrinterAgent / printer_agent framework is not ported (research X1: keep EdgeSlicer's probe + send dialog; #610/#611 already extend that for QIDI).

This is S2 from the vendor-send-gaps research. Independent of the native Moonraker host PR — Sovol profiles stay on octoprint, which already probes as Moonraker. Based on main, not stacked.

Review follow-ups (Russ 1/3)

• F1 Add-on mmu / lane_data GETs use a 1 s timeout. probe() is called from the send dialog on the UI thread; the first query still uses the caller's timeout (3 s).
• F2 Extras are skipped when multi_color_controller.box_count > 0 (asked on the first query, same object QIDI Box: send the filament-to-slot mapping before starting a print #610 adds). Comment at the merge point for Status::qidi_box. Happy Hare/AFC indices cannot go out as QIDI SAVE_VARIABLE value_tN.
• F3 AFC lanes: index range [0, 32), de-duplication (first wins), cap 32.
• F4 num_gates > 32 is clamped, not dropped. CSS colour names map to hex (red → #FF0000).
• F5 Header no longer claims probe is off the GUI thread.
• Catch2: wrong-type fields, huge gate_status arrays, num_gates>32, QIDI Box vs HH, AFC dupes/out-of-range/40 lanes.

Re-review follow-up (Russ re-review 1/3)

• F6 Unparsable CSS names no longer go into Slot.color. That field stays #RRGGBB or empty so auto-match cannot treat chartreuse as a colour. Known names still map to hex; unknown ones are kept in Slot.color_name and appended to the dropdown label. print_host_slot_colour() is the GUI-free mapper used by Catch2.

Upstream source

• MoonrakerPrinterAgent::fetch_hh_filament_info — GET /printer/objects/query?mmu (num_gates, gate_status, gate_material, gate_color).
• MoonrakerPrinterAgent::fetch_moonraker_filament_data — GET /server/database/item?namespace=lane_data (per-lane lane / material / color).
• scripts/test_moonraker_lane_data.py for the AFC JSON shape.

What was adapted

• Parsers live next to slots_from_print_task_config / slots_from_extruders and fill the same Slot list the send dialog already draws.
• Priority: Snapmaker print_task_config first, then Happy Hare mmu (not if a QIDI Box is present), then AFC lane_data, then bare extruderN tools. Empty gates/lanes stay in the list so the unit is visible; only a loaded gate/lane with a type or colour (or a readable unmapped name) sets slots_have_filaments.
• Probe networking: first query is print_stats&print_task_config&multi_color_controller&extruder… (the QIDI object QIDI Box: send the filament-to-slot mapping before starting a print #610 also asks for). Happy Hare and AFC are extra 1 s GETs, only when that first answer named no filaments and is not a QIDI Box. A 404/timeout on the extras does not fail a probe that already got print_stats. X-Api-Key is sent when the device has an API key.
• Network-free entry points for Catch2: parse_happy_hare_mmu and parse_afc_lane_data (full HTTP body or the inner object). parse_moonraker_status also accepts mmu inside a status snapshot.
• Catch2 v2 tests appended to tests/slic3rutils/printhost_devices_tests.cpp. No CMake test edits. No G-code changes. No new enum values. No filament_is_mixed.

Mapping at start — not sent

Upstream MoonrakerPrinterAgent::start_print is a stub (return BAMBU_NETWORK_SUCCESS) and start_local_print only uploads. There is no Happy Hare T / MMU_CHANGE_TOOL or AFC lane mapping in the start path. This PR is read-only display + the dialog's existing auto-match.

Material-first auto-match (research X3) is the change in #610, not duplicated here, so the two PRs do not edit the same send-dialog loop. On main today the table still auto-matches by colour among loaded slots that have a colour; once #610 lands, material-first applies to these HH/AFC slots too.

Tests

Inline JSON fixtures:

• Happy Hare: four gates (loaded / loaded / empty / unknown), 0x colour prefix, named colours, unparsable names (chartreuse → empty colour + color_name), status-object form, empty mmu, Snapmaker print_task_config winning over mmu, clamp num_gates>32, wrong-type fields, huge arrays, QIDI Box suppressing HH. Tile grouping 1A–1D from index/4 and index%4; Slot::label() stays T1 PLA.
• AFC: lane_data value object sorted by tool number, empty lane with #None, lanes array form, garbage JSON, duplicate/out-of-range indices, 40-lane cap, unmapped colour name, 1A/1B grouping.
• QIDI Box: 1A–1D grouping; label() stays Box 1 slot 2 ASA.
• print_host_slot_colour("red") == "#FF0000", print_host_slot_colour("chartreuse").empty().

This VM has no EdgeSlicer deps prefix, so slic3rutils_tests was not compiled here. Wx-free ThirdPartyFilamentMap.cpp compiled.

Untested (no hardware)

Live probe against Happy Hare or AFC, and the send-dialog table on a real printer.

Smoke checklist

1. Stock Klipper, no MMU — still a tool list (or Snapmaker filaments) as today; extra mmu / lane_data GETs 404/empty and are ignored (1 s each).
2. Happy Hare: Fluidd/Mainsail mmu object with loaded gates → send dialog mapping table shows Bambu-style 1A–1D tiles with colour/type. Empty gates show as empty. Named colours (red) show as a swatch. An unknown name does not auto-match by colour. Auto-match picks a loaded gate with a #RRGGBB colour (material-first once QIDI Box: send the filament-to-slot mapping before starting a print #610 is in).
3. AFC / BoxTurtle: lane_data with lane / material / color → table shows those tools. Empty material → unloaded.
4. Snapmaker with print_task_config — still that list, not the shared panel.
5. QIDI Box (multi_color_controller, box_count>0) — HH/AFC extras are not asked; Box uses the shared panel.
6. API key set — extras send X-Api-Key.
7. Confirm Upload and Print does not send any MMU_* / AFC mapping commands (no upstream precedent). Switching devices in the send dialog should not stall ~9 s.

Screenshots/Recordings/Graphs

None — uses the existing mapping table; no new chrome.