mirror of
https://github.com/maziggy/bambuddy.git
synced 2026-09-30 03:01:21 +02:00
Reshapes the embedded PrettyGCode viewer (landed in #963) into a focused archive-preview tool, matching Bambuddy's data model instead of the OctoPrint-style "connected-printer + library file picker" flow it shipped with. Reached only from the Archives page 3D-preview button; URL /gcode-viewer?archive=<id>[&plate=<N>]. Backend: - /archives/{id}/gcode accepts ?plate=N and resolves the filename by parsing the suffix as int, so zero-padded names like plate_01.gcode are found when the plates endpoint reports index 1. - /archives/{id}/plates gains top-level has_gcode: bool. Source-only 3MFs (PNG/JSON fallback path) surface the flag so the frontend can skip the picker instead of sending the user into a dead viewer. - printer_state_to_dict injects name + model into every WS snapshot so consumers render proper labels on the initial tick without racing a separate /printers fetch. - /gcode-viewer (no trailing slash) dropped from the backend so reloads fall through to the SPA catch-all and keep the layout shell; only /gcode-viewer/ (trailing slash) and /gcode-viewer/<path> remain for the iframe + static assets. Frontend: - PlatePickerModal shown only for multi-plate archives with sliced gcode, grid layout with thumbnails matching the Re-print modal. - Source-only archives show a noGcode toast instead of the empty viewer. - ArchivesPage navigate path swapped to /gcode-viewer?archive=<id> with no trailing slash; GCodeViewerPage iframe forwards window.location.search so the archive reference survives both the initial navigate and a full-page reload. - Viewer iframe's auth path: fetch intercept injects Bearer; a 401 redirects to / so the SPA handles login. Viewer adapter: - Stripped the printer selector, WebSocket subscription, library file picker, tryAutoLoadPrintingFile, BAMBU_BED_SIZES, and updatePrinter- Selector. The viewer no longer observes live printer state. - Bed size derived from /archives/{id}/capabilities.build_volume (extracted from the 3MF's printable_area/printable_height), so H2D, H-family, and any future printer render on the correct bed without a hardcoded map. - loadArchiveById accepts a plate param; fetch intercept rewrites __bambuddy_archive_<id>[_plate<N>] to /archives/<id>/gcode[?plate=N]. Nav + locale cleanup: - Sidebar "GCode Viewer" nav entry removed (viewer is archive-scoped now, not a destination page). - 32 orphaned gcodeViewer locale keys deleted across all 8 locales. - platePicker.{title, hint, plateLabel, objectCount, noGcode} keys added in all 8 locales. ArchivesPage: the now-unreachable ModelViewerModal render paths + its showViewer state removed. ModelViewerModal itself stays — File Manager still uses it for library file previews (plate picker + .3mf 3D model). pre-commit: - gcode_viewer/ excluded from trailing-whitespace + end-of-file-fixer so vendored third-party JS libs don't drift away from upstream. Incidental sweeps picked up by pre-commit and kept (unrelated but benign): - NotificationsPage.tsx: single trailing-whitespace line removed. - spoolbuddy/scripts/pn5180_diag.py: dead `import gpiod` dropped — the pn5180 driver module imported at line 27 does its own `import gpiod` and `gpiod.Chip()` calls, so the diag script's top-level import was never referenced. Tests: - 6 new cases in test_gcode_viewer.py for the backend plate / has_gcode behaviour (plate=N resolution, zero-padded filenames, missing-plate 404, no-plate fallback, plate=0 rejection, has_gcode true/false). - 3 new cases in test_printer_manager.py for name/model WS injection. - PlatePickerModal.test.tsx — 6 frontend cases covering render, plate-name composition, onSelect payload, backdrop close, and thumbnail fallback.