Reporter on Postgres + Spoolman saw weight never decremented after
prints. Traced to _report_spool_usage_for_slots calling only
client.find_spool_by_tag() — which returns None when extra.tag is empty.
Non-RFID spools assigned via the Bambuddy UI intentionally leave
extra.tag empty (per #1457 — we don't want fallback tags polluting
Spoolman), so tag-less spools never got matched and weight tracking
silently no-op'd. The tracker never consulted the local
spoolman_slot_assignments table that has the binding.
Adds _resolve_spool_id_via_slot_assignment() as stage 2 of the
resolution chain. Stage 1 (existing tag-lookup) wins when present so
RFID auto-sync remains unchanged. (ams_id, tray_id) derived from
global_tray_id via the existing _global_tray_id_to_ams_slot helper,
so external slots and AMS-HT slots resolve correctly. Threaded
printer_id through the three callers (partial G-code, partial linear,
final-usage report). Resolution path is logged ("via tag" vs "via
slot-assignment") so support bundles confirm the fix is live.
extra.tag is deliberately NOT auto-populated — that would re-introduce
the exact pollution #1457 cleaned up. Slot-assignment table is the
source of truth for non-RFID; extra.tag is reserved for hardware RFID.
Reporter on a P1S with non-RFID spools saw an old, almost-empty spool in
the AMS hover card's "Spulen-ID" block while the "Zugewiesen" block
correctly showed the freshly assigned full spool. Two layers compounded:
(1) Non-RFID slots fall back to a deterministic per-slot tag
(hash(serial) + ams_id + tray_id). The Link / Assign routes wrote
that tag to Spoolman extra.tag but never cleared it from the
previous holder on re-binding.
(2) The frontend's hover-card resolver preferred the (stale) tag-link
over the user's explicit slot-assignment. Same precedence bug in
SpoolBuddy's fill-bar resolver and slot-action picker.
Frontend: swap precedence at 5 sites — slot-assignment outranks tag-link
everywhere. FilamentHoverCard's existing match-dedupe then collapses the
two "Open in Inventory" buttons back into one.
Backend: new _clear_stale_tag_links() in spoolman_inventory.py, called
from POST /spoolman/inventory/slot-assignments (with the slot's
deterministic fallback tag) and POST /spoolman/spools/{id}/link (with
the literal tag being bound — works for RFID and fallback). Best-effort:
Spoolman 5xx and per-spool patch failures log + continue, never wedge
the bind. get_fallback_spool_tag_for_slot promoted to a public helper
mirroring the frontend's signature exactly.
PrintLogEntry.failure_reason is captured once at print-completion time
(main.py:3641) by copying archive.failure_reason — which is NULL while
the user hasn't classified the failure yet. The PATCH /archives/{id}
route then writes only to print_archives via a generic setattr loop,
so the log entry stays NULL and failure_analysis.py keeps grouping the
print as "Unknown". Same desync hits status — flipping it in the modal
never reached the entry either.
Mirror failure_reason and status from the PATCH payload to the latest
PrintLogEntry for that archive (highest id). Latest-only because
archive.failure_reason / status already reflect the latest run's outcome
(each reprint clears the archive value at main.py:2195 and rewrites it
at completion), so the Edit Archive modal is implicitly editing the
latest run — reprints of an archive that succeeded on the second attempt
keep the earlier failed run's original classification intact.
Scoped to those two fields only. cost / print_name / printer_id stay
unmirrored because per-run values legitimately diverge from archive
ones (partial-print cost on a failed run vs source archive's full-print
cost — see _compute_run_filament_grams at main.py:596).
Reporter — the same person who originally requested the labels
feature in #809 — discovered that the ams_30x15 preset's 30x15 mm
dimension didn't actually fit any variant of the MakerWorld AMS
Filament Label Holder (model 752566) it advertised. Two new
presets replace it:
- ams_holder_74x33 (74 x 33 mm) matches the printable label STL
bundled in the MakerWorld project
- ams_holder_75x55 (75 x 55 mm) fits the cardstock-insert variant
the reporter validated on bench
Both cross the 20 mm height threshold so they land in the roomy
layout branch — swatch on the left, QR on the right, multi-line
text (brand, material, hex code, spool ID) in the middle. The
old 30x15 mm preset couldn't fit a QR code; the new ones do.
No DB migration: the preset name was never persisted. Callers
scripting the old ams_30x15 value get a clean 422 at the route's
Literal validator with the new valid values listed.
i18n: replaced inventory.labels.templates.ams.{label,hint} with
amsHolderSmall and amsHolderLarge across all 8 locales with real
translations; parity guard cleaned of the stale English-fallback
cognate entries. Parity holds at 4856 leaves per locale.
Tests: backend label renderer + integration tests cover both new
presets; LabelTemplatePickerModal test updated for the 6-button
grid and the new template value in the API-call assertion.
Cover endpoint had no negative cache: when every FTP path returned
550 for a print whose 3MF wasn't on the printer (typical SD-card
print), each frontend refresh re-ran the full 8-path fan-out. Add
_cover_404_cache keyed by (subtask_name, view_key) and short-circuit
to 404 on hit; clear alongside _cover_cache on print start. Only
populated on genuine 404 paths, not transient FTP errors, so flaky
network doesn't lock out future retries.
GitHub update-check had no backoff on 403 rate-limit. Add module-
level _github_rate_limit_until plus three helpers; check before
every api.github.com call in /updates/check and
_discover_target_release. Read X-RateLimit-Reset from the 403 with a
1-hour fallback when the header is absent and a 60-second floor to
guard against container/GitHub clock skew. Route surfaces
retry_after_seconds so the UI can display real wait time.
The "ffmpeg didn't terminate gracefully" line the reporter quoted
is the standard SIGTERM/SIGKILL pattern in camera.py and unrelated
to the FTP loop; it goes away on its own once the cover endpoint
stops hammering the printer.
Three independent root causes behind the divergences the reporter
flagged after the archived-spool fix shipped — fixed together.
(1) Filament Used vs Total Consumed.
_compute_run_filament_grams returned the slicer estimate for completed
prints even when inventory had measured the actual AMS weight delta.
That made Stats and Inventory two different sources of truth: Stats
showed slicer-estimate grams, Inventory showed AMS-tracked grams, and
the two never agreed. Reordered the helper so the tracked spool delta
(same source that drives weight_used behind Total Consumed) takes
priority for every status. Slicer estimate stays as the fallback when
no inventory was tracked; partial-progress scale stays as the fallback
for failed/cancelled with no tracker. The _run_cost block right next
to it was already tracker-first; only filament_used_grams was
inconsistent.
(2) Printer Stats By Time vs Quick Stats Print Time.
/archives/slim only set actual_time_seconds when status == "completed".
For failed/cancelled rows the frontend fell back to print_time_seconds
(the slicer's full-print estimate — wrong number for a print that
failed at 15%). Quick Stats already summed elapsed duration across
all statuses, so the two halves of the page disagreed by the
(estimate - actual-elapsed) gap on every non-completed event. Dropped
the completed-only gate; failed/cancelled now report measured elapsed.
(3) Success Rate %.
Was successful / (successful + failed), excluding cancelled / stopped
from the denominator. With "Total Prints: N" displayed right above
the gauge that produced confusing numbers — 4 successful, 0 failed,
48 cancelled showed 100% out of an apparent 52 prints. Switched to
successful / total_prints — matches the count the user reads from
the widget header.
Editing a Spoolman spool used to mint a brand-new filament every time
a match-key field (subtype/material/brand/color_hex) changed, orphan
the previous one, and re-link the spool. The reporter ended up with
dozens of duplicate "Amazon Basics / PLA Glow" filament rows.
PATCH /spoolman/inventory/spools/{id} now:
- Reuses the current filament_id when no filament-shaping field
changed (a note/weight_used edit never touches the catalogue).
- PATCHes the existing filament in place when it's a singleton
(only this spool points at it, archived spools included).
- Falls back to find_or_create_filament only when the filament is
genuinely shared with another spool.
Mirrors internal-inventory behaviour where editing a spool updates
the thing the spool points at instead of proliferating new entities.
Step 2 of the camera architecture overhaul agreed after #1395. When
the camera viewer hits its error state OR before a print at any
time, a Diagnose button runs a staged check against the printer and
renders the result inline: which stage failed, how long it took,
and a translated remediation hint. Cuts off the "user opens a
'camera broken' ticket → ask for support bundle → triage" loop at
the user's screen.
Backend
- New `backend/app/services/camera_diagnose.py` orchestrator with
CameraDiagnoseResult / CameraDiagnoseStage dataclasses.
- New POST /printers/{id}/camera/diagnose route in camera.py.
- Stages:
tcp_reachable — TCP socket open to 322 (RTSP) / 6000 (chamber)
with 3 s timeout. Distinguishes timeout, refused, and host-
unreachable into distinct summary codes so the frontend can
show a precise remediation (firewall vs LAN-only off vs
wrong IP).
first_frame — captures one JPEG end-to-end via the existing
capture_camera_frame_bytes pipeline. Auth + RTSP handshake +
first keyframe collapse into one stage; the user-facing
answer is the same regardless of which sub-layer failed.
- Live-stream shortcut: when a viewer is currently watching the
camera with a buffered frame < 10 s old, the diagnostic skips
the real test and returns live_stream_active_healthy. Opening a
fresh socket would kick the live viewer off on single-camera-
connection firmwares (the #1348 reconnect-storm trigger), so we
trust the real-world evidence instead.
- Response surfaces protocol, port, and profile name for support
triage — lets us ask "what does your modal say?" instead of
"send the support bundle".
Frontend
- New CameraDiagnoseModal renders one row per stage with green-
check / red-X / grey-skipped icons, the per-stage duration in
ms, a remediation banner styled by overall status, and a Run
again button.
- Two entry points:
1. The viewer's error overlay grows a Diagnose button next to
Retry. Retry stays the primary action; Diagnose is the
escape hatch for users who can't see what's wrong.
2. A stethoscope icon in the viewer's always-visible control
bar, between Refresh and Fullscreen. Pre-flight testing
("did my firmware update break the camera?", "is the
camera up before I send a print?") doesn't require waiting
for the stream to fail first.
- Also lifted the previously-hard-coded "Camera unavailable" /
"Retry" strings into camera.unavailable / camera.retry so the
error UI is fully translated alongside the new keys.
Reporter sliced in OrcaSlicer with timelapse on, sent the job to a VP
queue, started from the queue, and got no timelapse video. Their
dispatch chain itself was correct (queue item -> scheduler -> MQTT
command honors `timelapse`); the gap was at queue-add time.
The VP's `_add_to_print_queue` reads `default_timelapse` (and the four
other print-option settings) from the workflow settings card. That was
introduced in #1235 to stop column-level defaults from winning. But it
also discarded the slicer's actual choice carried on the MQTT
`project_file` command, which all the slicers (Studio / Handy / Orca)
ship as `timelapse: true|1`. Result: a user with the new-install value
`default_timelapse=false` had to either flip the global setting or
edit every queue item by hand, even though their slicer's "Print
options" UI clearly said "record timelapse".
Investigation went wider than #1403 because Martin's hypothesis was
"the print options modal isn't respected either." Cross-checking
86 captured P1S `project_file` commands across the support packages
shows 46 from the queue scheduler and 33 from background_dispatch
emitting `"timelapse": true` correctly to real printers - the modal +
re-print path is intact end-to-end. The slicer-side gap was the only
real bug. Two unrelated dead-code issues turned up in the same dig and
are folded in below.
Fix (VP queue inheritance)
- `on_print_command` in the VP manager now stashes the slicer's
project_file dict keyed by filename, then signals an asyncio.Event.
- `_add_to_print_queue` checks the dict first; if empty, creates the
event and waits up to 2 s for it before reading the settings
fallback. Each option flows through per-field - slicer value wins
if present, else the existing settings default (so users who
explicitly set `default_timelapse=true` in their VP workflow card
still get that on slicers that don't send a print command).
- MQTT field naming preserved exactly: `bed_leveling` (single L) on
the wire stays mapped to `bed_levelling` (double L) on the Bambuddy
column. Integer 0/1 from H-family slicers and bool true/false from
P1/X1 slicers both coerce via `bool()`.
- Capture is gated on `mode == "print_queue"` so immediate / review /
proxy modes keep their pre-fix no-op `on_print_command` and don't
accumulate stashed entries over the VP's uptime.
- Wait is also skipped when there's no MQTT server attached
(`self._mqtt is None`), so unit tests that invoke
`_add_to_print_queue` directly don't pay the 2 s tax.
- Capture is consumed on use so the dict stays bounded.
- `printer_manager.get_status(...).get(...)` against a `PrinterState`
dataclass that has no `.get()` method.
- Every print option discarded (timelapse, bed_levelling, AMS mapping).
The route 500'd before ever reaching the printer. Rewritten to mirror
`POST /print-queue/{item_id}/start`: clear `manual_start=False` on the
next pending queue item and let the scheduler dispatch with the
queue's stored options intact. Response shape preserved.
Side-bug b: vibration_cali default drift in background_dispatch
- `ReprintRequest.vibration_cali` and `FilePrintRequest.vibration_cali`
both default to `True` (matches Bambu Studio behavior for X1/P1).
- Both `_process_job` call sites read
`job.options.get("vibration_cali", False)`.
Cosmetic today because the frontend always sends the field, but a
latent landmine for any future caller that bypasses the schema. Both
sites flipped to `True`.
Reporter saw a 544 g spool jump to 1000 g after pressing the eraser.
"Spools and remaining weights are not changed" - the dialog promised
this; the implementation did the opposite. Root cause was an
architectural conflation: `weight_used` did double duty as the
resettable "consumed since tracking started" counter AND as the basis
for the displayed remaining (`label_weight - weight_used`), so zeroing
it correctly cleared the stat but unavoidably reset remaining to full.
Spoolman has separate `used_weight` and `remaining_weight` fields, so
the API call there was correct - but Bambuddy's frontend was also
computing remaining as `label_weight - weight_used` for Spoolman
spools (ignoring Spoolman's real `remaining_weight` field), so the
same visual bug bit there too. Inventory-mode parity required fixing
both halves in one drop.
Internal mode
- New `weight_used_baseline` column (Float DEFAULT 0) on `spool`.
- Reset stamps `baseline = weight_used` and leaves `weight_used` alone.
- Displayed consumed = `weight_used - baseline`; remaining =
`label_weight - weight_used` (unchanged).
- Subsequent prints continue to grow `weight_used`, so the resettable
counter naturally tracks post-reset delta and remaining keeps
decrementing across the reset.
Spoolman mode
- `_map_spoolman_spool` now reads Spoolman's `remaining_weight` field
and returns a synthetic `weight_used = label - remaining` so the
frontend's remaining calc matches Spoolman's real stored value;
`weight_used_baseline = synthetic - real_used_weight` so the consumed
counter (`weight_used - baseline`) matches Spoolman's `used_weight`.
- Fallback path (no `remaining_weight` set) preserves the old behavior.
- Related fix: `update_spool` (Spoolman PATCH) was deriving the default
`weight_used` from `used_weight`, so editing unrelated fields AFTER
a reset would patch Spoolman with `remaining_weight = label - 0 =
label`, trampling the real value. Now derives from
`remaining_weight` so non-weight edits preserve physical state.
Frontend
- `InventoryPage` `totalConsumed` aggregate switched to
`Math.max(0, weight_used - (weight_used_baseline ?? 0))`.
- `ForecastPanel` `computeDeltaRate`, `totalUsedG`, and the per-spool
"consumed" table cell got the same treatment so forecast and
inventory aggregates stay coherent across a reset.
- `?? 0` keeps pre-migration installs rendering correctly until
`init_db()` runs the idempotent ALTER TABLE.
Migration
- `ALTER TABLE spool ADD COLUMN weight_used_baseline REAL DEFAULT 0`
via `_safe_execute` - SQLite and Postgres both accept it; verified
end-to-end on Postgres 16.
Several support reports traced back to one root cause: a mistyped access
code in Add Printer left an empty card on the dashboard. POST /printers/
was persisting the row first, then firing connect_printer() fire-and-forget.
Now we test_connection() BEFORE the insert; failure returns HTTP 400 and
the row is never written.
Structured error response -- detail={"code", "message"} -- so the toast
shows the localized message instead of the English fallback. New
ApiError.code field on the frontend; printers.toast.connectionFailedNotAdded
While auditing real-world Bambuddy backup repos on GitHub I found
several left public. That's a serious leak: the settings backup only
filters bambu_cloud_token and auth_secret_key, so mqtt_username,
mqtt_password, ha_token, prometheus_token, bambu_cloud_email,
external_url, and the printer access codes (via K-profiles) were going
to whatever visibility the user picked.
Hard guard at every save and re-checked on every push:
- POST /github-backup/config and PATCH /github-backup/config (when URL,
token, or provider changes) run a connection test internally and
return 400 unless is_private comes back True.
- run_backup() re-checks before each scheduled or manual push, so a
repository that flipped from private to public gets a clear
"Backup aborted: the target repository is no longer private" failure.
Each provider's test_connection now returns is_private (GitHub /
Gitea / Forgejo read data.private, GitLab reads visibility=="private";
"internal" is treated as non-private). None means "couldn't determine"
and is also rejected -- safer to fail closed.
Frontend renders visibility inline on Test Connection: green check when
private, red warning panel listing every credential at risk when public,
yellow when unknown.
---
ui(github-backup): show save-failure messages inline on the card
The new "repository is not private" rejection message is ~250 characters
listing every credential the backup carries (MQTT password, HA token,
Prometheus token, Bambu Cloud email, printer access codes), which clips
badly in a toast.
Both the initial-setup save and the debounced autosave now stash the
backend's error message into a saveError state and render it as a red
inline banner above the test-result block, with whitespace-pre-wrap so
the full message stays readable. The banner clears on success, on the
next save attempt, and when the user starts editing URL / token / provider
-- the three fields whose changes invalidate the privacy check -- so it
doesn't linger after the user has already addressed the cause.
Short success toasts (Settings saved, Token updated, Backup enabled) are
unchanged.
First cut of this action only wired the built-in inventory path, so the
eraser buttons vanished when the user switched to Spoolman mode. Mirror
the endpoints on the Spoolman router:
- POST /spoolman/inventory/spools/{id}/reset-usage
- POST /spoolman/inventory/spools/reset-usage-bulk
Both route to a new SpoolmanClient.reset_spool_usage() helper that PATCHes
/spool/{id} with used_weight=0. The bulk variant keeps the same typo-wipe
guard (rejects empty/missing spool_ids), and individual Spoolman failures
are logged + counted out without aborting the batch.
InventoryPage mutations now switch on spoolmanMode to pick the right
client method, and the three "spoolmanMode ? undefined : ..." gates on
the eraser buttons are gone.
After #1378 moved Quick Stats to print_log_entries, six widgets and
Failure Analysis still iterated the archive list. That made reprints
multiply event-based widgets while leaving archive-based ones unchanged,
and made hard-deleted archives drop from archive-based widgets while
their orphan events kept feeding Quick Stats.
Swap the data source in two places:
- GET /archives/slim now reads PrintLogEntry, LEFT JOINs the archive for
the sliced print_time_seconds estimate, prefers PrintLogEntry's own
duration_seconds as the measured-time field. StatsPage is the only
caller -- every widget realigns in one step.
- FailureAnalysisService swapped from PrintArchive to PrintLogEntry for
every aggregation. project_id filter still resolves through archives
but counts matching events.
Conftest archive_factory now syncs the synthesized event's created_at
with the archive's so backdated test data survives the change.
Reporter @iitazz uploaded slicer output to Bambuddy, clicked Print,
and the printer rejected every job with "Printing stopped because
the printer was unable to parse the 3mf file". Support bundle showed
the stored library file ended in .gcode (not .gcode.3mf), and
background_dispatch.py appends ".3mf" to filenames that don't
already end in .gcode.3mf/.3mf — so raw gcode shipped to the printer
named .gcode.3mf and the firmware's 3MF parser choked. Same shape
also surfaced as "File is not a zip file" on Bambuddy's own plate
parser.
New validate_print_file_upload() helper in library.py runs at upload
time:
- Reject filenames ending in .gcode (but not .gcode.3mf) with a
clear message — Bambu printers need .gcode.3mf zip containers,
not raw gcode.
- For .3mf / .gcode.3mf uploads, verify body starts with PK\x03\x04
(ZIP magic); reject otherwise pointing at the slicer's "Export
Plate Sliced File" action.
Applied to every relevant upload route: POST /library/files (covers
File Manager + printer-card drag-drop), POST /archives/upload,
POST /archives/upload-bulk (rejects per-row so one bad file doesn't
abort the batch), POST /archives/{id}/source, POST /archives/upload-source.
Runs after _resolve_upload_destination so folder-permission errors
(403 readonly, 400 missing-path, 409 collision) still take precedence.
STL / image / other non-print uploads bypass the validator.
FileUploadModal frontend fix: the modal auto-closed after every
batch regardless of per-file results, so a 400 rejection was captured
but invisible. Now:
- Errors render inline as red text under the file row instead of
as a hover-only title tooltip.
- Modal stays open if any file ended with status='error', so the
user can read the backend's remediation message before closing.
- Successful-only batches still auto-close as before.
UploadModal (bulk archive) was already showing inline errors and
not auto-closing — no change needed there.
Reporter IndividualGhost1905 followed up after the #1378 / #1343
backfill landed and pointed at the next inconsistency: the per-
archive delete dialog has had a "Also remove this print from Quick
Stats" checkbox since #1343, but the "Purge Old" button and the
scheduled daily auto-purge sweeper both ignored that choice and
hard-deleted unconditionally. From the user side this looked like
"automatically deleted from statistics without any warning" — half-
true, and the inconsistency was real either way.
The actual current shape (before this fix):
- POST /archives/purge -> archive_purge_service.purge_older_than
-> ArchiveService.delete_archive (hard). Archive row dropped.
Linked PrintLogEntry rows have ON DELETE SET NULL so they
survive as orphans with archive_id=NULL. Quick Stats keeps the
filament / cost / energy contribution because the log rows are
still there, but the archive-list-iterating widgets (Filament
Trends, By Material, Color Distribution, Printer Stats) lose
the row, and Time Accuracy loses its join target. Visibly
inconsistent.
- Scheduled _maybe_run_auto_purge -> same code path, same effect.
- Single-archive DELETE /archives/{id} -> already takes
purge_stats=true|false (default false=soft) and routes either
soft_delete_archive (keeps everything, flips deleted_at) or
deletes PrintLogEntry rows first + hard-deletes archive.
The fix threads the same purge_stats flag through every bulk surface
with soft as the default, matching the single-archive default:
Backend:
- archive_purge_service.purge_older_than(..., purge_stats=False)
-> per-row soft_delete_archive when False, per-row
PrintLogEntry deletion + delete_archive when True. Each runs in
its own session (same pattern the sweeper already used).
- preview_purge gains the same kwarg so the eligible-count
matches what an actual run would touch: soft mode excludes
already-soft-deleted rows, hard mode counts them as eligible
for promotion.
- get_settings / set_settings now persist archive_auto_purge_stats
(default False). _maybe_run_auto_purge reads it on every tick.
- ArchivePurgeRequest / ArchivePurgeResponse / ArchivePurgeSettings
schemas extended.
- Route /archives/purge accepts the body flag, /purge/preview
accepts the query param, /purge/settings GET + PUT echo the
setting.
Frontend:
- "Purge old archives" modal: new "Also remove from statistics"
checkbox under the preview, unchecked by default. Plumbed into
the preview query key + the execute mutation.
- Settings -> Archives auto-purge card: matching toggle next to
the days slider, disabled when auto-purge itself is off.
- api.previewArchivePurge / api.executeArchivePurge accept the
flag; ArchivePurgeSettings type gains purge_stats.
- All 8 locales (en, de, fr, it, ja, pt-BR, zh-CN, zh-TW) get
new purgeStatsLabel / purgeStatsHint / purgeStatsDescription
keys, plus rewritten effect / warning copy in archivePurge and
archiveAutoPurge to reflect that the default no longer
"permanently removes from the database" but instead hides the
row + removes files while keeping Quick Stats intact. i18n
parity check clean: 4814 keys across all 8 locales, no fallback.
Behaviour change for existing users on auto-purge: the sweeper used
to hard-delete by default and now soft-deletes by default. After
the upgrade those installs start *preserving* more data in Quick
Stats rather than losing it — safer direction of the two, but worth
the explicit call-out. Anyone who wants the old behaviour ticks the
new toggle once and it persists.
Reporter pgladel edited a spool's Color Name in Spoolman mode, hit
Save, and saw the value snap back to the subtype on the next read.
The earlier #1319 fix correctly handled the read/form-prefill half
(the color_name_is_synthesized flag, blank-on-synth form init), but
the write half assumed Spoolman has a `color_name` field on Filament.
It doesn't. Verified against the live FilamentUpdateParameters schema
on Spoolman 0.23.1 — the accepted fields are name, vendor_id, material,
price, density, diameter, weight, spool_weight, article_number,
comment, settings_extruder_temp, settings_bed_temp, color_hex,
multi_color_hexes, multi_color_direction, external_id, extra. No
color_name. Spoolman's PATCH happily returns 200 for
{"color_name": "Red"} and silently discards the unknown key, so
find_or_create_filament was either patching a void or creating
filament after filament with the same field-that-doesn't-stick (which
is what produced the "BB also created a bunch of new filaments"
duplicate trail on each save attempt).
The fix follows the same pattern as the existing BambuStudio slicer-
preset storage: persist color_name on spool.extra.bambu_color_name as
a JSON-encoded string, register the extra field via
ensure_extra_field before write (Spoolman 400s on unknown extra keys),
and read it back in _map_spoolman_spool with priority
extra > filament.color_name (forward-compat for any future Spoolman
release that adds the field) > subtype synth.
Dropped the now-dead color_name passing through
find_or_create_filament and create_filament — Spoolman would discard
it anyway and keeping the dead pipe risked the same confusion the
next time someone reads this code. The previous "match by name then
patch color_name" loop is gone; what survives is the name-match
resilience that lets an AMS-sync-created filament named "Glow" still
match the user-driven edit's composed "PLA Glow", which prevents
re-introducing the duplicate-filament trail.
The frontend form's color_name_is_synthesized handling is unchanged
— that part already worked.
In-app updater was failing on installs whose local clone had stale
tags (e.g. v0.2.1 re-pointed upstream after a post-release re-tag).
git fetch --prune --tags returns a non-zero exit when even one tag
would be clobbered, even though origin/main and the target release
tag itself fetched cleanly:
! [rejected] v0.2.1 -> v0.2.1 (would clobber existing tag)
...
ERROR Git fetch failed: From https://github.com/maziggy/bambuddy
The updater surfaced this as "Failed to fetch updates" and aborted,
leaving the user stuck on the previous release.
Adding --force lets the moved tag overwrite the local stale copy.
That matches the in-app updater's contract ("sync me to the remote")
and is what the native update.sh sidesteps entirely by not using
--tags at all. The in-app path can't drop --tags because release-tag
refs (v0.2.4b1, v0.2.4.1, etc.) need to be resolvable locally for the
subsequent git reset --hard.
Regression test asserts --force is in the fetch args alongside the
existing --tags assertion.
Opening the Print Queue page fired 404s on /archives/{id}/thumbnail,
/archives/{id}/plates, and /archives/{id}/plate-thumbnail/{n} for any
row pointing at a soft-deleted archive. Two underlying problems wearing
one mask:
1. Cosmetic: the queue API was copying item.archive.thumbnail_path into
archive_thumbnail without checking deleted_at. Soft-delete leaves the
row (so the relationship resolves) but removes the file from disk, so
the cached path was always stale.
2. Functional: a queue item whose 3MF was removed can never dispatch.
Without an explicit cancel, the item sits in 'pending' forever with
no indication to the user about why nothing is printing.
Fix in three parts:
- New _cancel_pending_queue_items() helper, called from soft_delete_archive
alongside the existing print-log thumbnail cleanup. Sets status='cancelled'
+ waiting_reason='Source archive deleted' on every pending queue item
linked to the archive. Only 'pending' is touched - completed/failed/
cancelled rows are historical and untouched. Hard-delete is already
covered by ON DELETE CASCADE on print_queue.archive_id.
- Queue API serializer now checks item.archive.deleted_at before
populating any archive-derived field. New archive_deleted: bool field
on PrintQueueItemResponse signals the soft-deleted state.
- Frontend's getArchivePlates query in QueuePage was gated on archive_id
only - archive_id is the real FK and stays exposed for dispatch/audit,
so added an explicit && !item.archive_deleted clause to respect the
new flag. Thumbnail render and CompactHistoryRow/QueueTimelineView
already gate on archive_thumbnail so the backend suppression alone
covers them.
Regression tests pin cancel-only-pending behavior, soft-deleted
suppression + archive_deleted=True flag, and the sanity guard that
live-archive fields keep flowing through unchanged.
Archives → Print Log was 404-storming the thumbnail endpoint on every
render: PrintLogEntry.thumbnail_path is copied by value from the archive
at write-time, but the FK on archive_id is ON DELETE SET NULL (#1378) so
log entries survive archive deletion to preserve stats history - and the
cached path keeps pointing at a file that was removed when the archive's
directory was deleted. Same shape for failed prints whose extractor never
wrote the thumbnail.
Two-part fix:
1. Route self-heals: get_print_log_thumbnail NULLs thumbnail_path on the
entry and commits before returning 404 when the file is missing on
disk. The frontend's <img> tag is gated on entry.thumbnail_path being
truthy, so the next fetch of the log list skips the request entirely.
2. Eager clear on archive delete: new _null_print_log_thumbnail_paths()
helper called from both soft_delete_archive and delete_archive before
the on-disk files are removed. Avoids the one-time storm for future
deletes; covers both the manual delete route and the auto-purge
sweeper at archive_purge.py.
Regression tests cover soft delete, hard delete via ArchiveService, and
the route's lazy-NULL for failed-print orphans where the file was never
written.
Discord's "Copy Webhook URL" button emits discordapp.com URLs; both
hostnames serve the same webhooks. The validation now accepts either
prefix while keeping the check itself in place to catch the
paste-the-wrong-thing error.
archives stop reporting near-zero cost (#1344)
Reporter @nicktags hit $0.01 on a 110.3g multi-color print with the
global default filament cost set to $10/kg. archive.py initial cost
calc was correct (~$1.10), then usage_tracker.on_print_complete
overwrote archive.cost with sum(r.cost for r in results) -- where
results only includes AMS trays mapped to a spool in Bambuddy's
inventory. On a multi-color print where 3 of 4 used trays had no
inventory spool, only the one tracked slot's tiny share (~1g) survived
and the archive recorded $0.01.
The overwrite logic dates to #505 (Feb 2026) and is correct for
fully-tracked single-color prints, but the multi-color slicer feature
in 0.2.4 (988c0055) made the partial-inventory state common -- users
slice + print multi-color from Bambuddy without first setting up an
inventory entry for every tray.
Cover the gap: any filament weight not represented in results gets
charged at the global default rate. For a fully-tracked print,
untracked grams = 0 and the top-up adds nothing, so the single-color
behavior is preserved. For a partial print, the missing slots are
priced at the user's documented default rate so the archive cost
reflects the whole print.
Three call sites updated to share the same logic:
- usage_tracker.py: live cost-update on print complete
- archives.py rescan_archive: per-archive manual recalc
- archives.py recalculate_all_costs: bulk recalc button
Reporter @Andlar94 hit a permanent "Build plate not empty" on every print
start on an A1 with an external RTSP camera. The runtime auto-check at
main.py:1819 called check_plate_empty with use_external=external_camera_enabled,
but the manual UI routes (camera.py) declared use_external: bool = False
and the frontend client always sent use_external=false. So calibration
captured a built-in frame and stamped it as the reference; the runtime
check captured an external frame and diffed it against that reference --
a permanent mismatch.
Centralise the default on the backend: both routes now take bool | None,
deriving the default from the printer's external_camera_enabled +
external_camera_url + external_camera_type. The frontend client stops
sending the flag unless the caller explicitly sets it, so the existing
UI call sites immediately benefit and any future caller gets the right
camera automatically. Explicit overrides still win.
Adds 4 regression tests pinning the new default for both the
external-enabled and external-disabled cases, plus the explicit override
path so a future "always built-in" caller stays supported.
Reporter @maziggy followed the Energy Tracking wiki literally - "create a
key with Write Settings permission, PATCH /api/v1/settings with
{energy_cost_per_kwh: ...}" - and hit:
{"detail":"API keys cannot be used for administrative operations"}.
Triage showed three independent drifts:
1. Wiki listed nine fictional API-key permissions (Read Printers / Write
Settings / Admin / ...) but the UI only ever exposed four toggles
(Read Status, Manage Queue, Control Printer, Allow Cloud Access).
There was no Write Settings toggle to tick.
2. Even if it had existed, the backend hard-denies SETTINGS_UPDATE for
every API key via _APIKEY_DENIED_PERMISSIONS - intentional protection
because PATCH /settings can rewrite SMTP/LDAP/MQTT credentials and the
HA access token. Wider surface than any documented use case needs.
3. So the wiki had been promising a workflow that was never deliverable.
Fix: introduce a narrowly-scoped door rather than relax the deny list.
- New column can_update_energy_cost (default FALSE - existing keys
never silently gain settings-write capability on upgrade).
- New route POST /api/v1/settings/electricity-price accepting
{"energy_cost_per_kwh": <float >= 0>}. Field name matches what the
wiki already documented so the HA rest_command example needs only a
URL+method change, not a payload change.
- Custom dependency require_energy_cost_update() bypasses
_APIKEY_DENIED_PERMISSIONS for this one route for API keys with the
flag set. JWT users still go through standard SETTINGS_UPDATE.
- General PATCH /settings remains denied for API keys - flipping the
narrow flag does NOT widen general settings-write access. Pinned by
test_patch_settings_still_denied_with_energy_flag.
Frontend: fifth "Update electricity price" toggle on the create-API-key
card + amber "Energy" badge on existing keys with the flag set. Three
new i18n keys across all 8 locales (German translated, English fallbacks
elsewhere).
Add `# pragma: allowlist secret` markers to the three lines GitGuardian
flagged (ldap_server_url, ldap_bind_dn, admin_password) and pull the
duplicated AdminPass1! literal into a single test_password variable so
the marker only needs to live in one place.
All values are test fixtures (test directory + admin password used only
by the LDAP provisioning integration suite), not real credentials.
Reporter @Fuechslein flagged that disabling LDAP auto-provision left admins
with no UI path to onboard new users — the create-user form had zero LDAP
awareness and the only workaround was hand-editing the database.
Add a Local / LDAP tab toggle to the create-user modal (hidden when LDAP is
disabled). The LDAP tab is a debounced directory search (≥2 chars, 300ms)
that returns up to 25 matches via the service-account bind, annotated with
already_provisioned so existing usernames render disabled. Clicking
"Provision user" re-resolves via the service bind and creates the user
through the same _provision_ldap_user helper the auto-provision login path
uses, so group mapping, default-group fallback, and email sync are identical
regardless of which path created the user.
The picker component is shared across all four create-user modal paths
(UsersPage basic + advanced, SettingsPage basic + advanced).
Two ldap3 schema-check workarounds were needed for OpenLDAP installs:
- Open the search connection with check_names=False so ldap3 doesn't reject
the cross-schema OR filter (sAMAccountName/displayName are AD-only)
- Request attributes=["*"] because ldap3's build_attribute_selection
validates each named attribute against the server schema regardless of
check_names, and only the * wildcard is in its hard-coded exclusion list
Login/lookup paths keep check_names=True so typos in user_filter still fail
loudly.
Backend
- New routes: GET /auth/ldap/search, POST /auth/ldap/provision (both gated
by USERS_CREATE; 503 details include ldap3 exception class + message)
- Extract _open_service_connection + _extract_user_info helpers so
authenticate_ldap_user, lookup_ldap_user, and search_ldap_users share the
bind and attribute-extraction logic
Frontend
- New LdapUserPicker component (debounced search, result list, provision
mutation, already-provisioned guard, error surface)
- Tab toggle wired into UsersPage and SettingsPage modals, plus
CreateUserAdvancedAuthModal props
- 14 i18n keys added to en.ts (other locales fall back to English)
The original #1322 fix widened empty-slot detection to (state == 11 OR
tray_type != ""), which closed the configured-slot reconfig case but
didn't help the "Reset Slot on printer screen with spool still inserted"
flow. On these firmwares the AMS reports state=3, tray_type="" after a
Reset Slot regardless of whether a spool is physically present, so the
empty-detection still decided "empty", skipped MQTT, marked pending —
and on_ams_change replay never re-fired because the AMS never reported
any state change either.
RosdasHH traced the path: tray_state=3 falls into the else: branch,
slot_is_empty = not (fingerprint_type and fingerprint_type.strip()),
fingerprint_type is "", so slot_is_empty=True, MQTT is skipped, and the
slot stays unconfigured forever. He verified empirically that removing
the gate makes the firmware accept the push when a spool is physically
present.
Drop the tray_type fallback entirely. Only state in {9, 10} (firmware's
explicit "no spool" / "spool present but no feed") short-circuits the
MQTT publish. Every other state — including 3 (default-idle, ambiguous)
and missing-state (older firmwares) — attempts the publish. Bambu's
"firmware silently drops on empty slots" behavior makes the worst case
a no-op for a truly-empty slot, and on_ams_change replay still serves
as the safety net for state=9/10 slots whose spools get inserted later.
pending_config is now (slot_is_definitely_empty OR not configured) so a
printer-offline / no-client publish failure correctly flags the
assignment for replay instead of falsely showing "configured".
Reported by @IndividualGhost1905: printing the same model ten times and
then deleting nine archive entries (to keep the file list tidy) silently
rewound the totals on the Statistics page — total prints, filament,
cost, and per-print energy all dropped back to whatever the surviving
row contributed, as if the other nine prints had never happened.
Root cause: every metric in get_archive_stats is recomputed live from
PrintArchive rows via COUNT / SUM, so removing a row removes its
contribution. Energy in the default "Total" mode already survived
deletion because it reads the smart-plug lifetime counters — that's
the architectural shape we now generalise to the rest.
Fix: soft delete with opt-in hard purge.
Backend:
- New nullable, indexed deleted_at column on print_archives, dialect-
conditional migration (DATETIME on SQLite, TIMESTAMP on PostgreSQL).
- ArchiveService.soft_delete_archive flips deleted_at and removes the
files from disk (still reclaims storage); the path-safety checks were
extracted into _resolve_archive_dir_for_delete so soft and hard delete
share the rules.
- DELETE /archives/{id} accepts ?purge_stats=true; default is soft.
- Listings filter deleted_at IS NULL: list_archives, search FTS + LIKE
fallback, GET /{id} (404 on soft-deleted), tag listing, duplicate
detection (so a 1-live + 9-soft-deleted group no longer marks the
survivor as a duplicate), and ArchiveComparisonService's "similar"
suggestions. GET /stats and GET /slim deliberately do NOT filter so
Quick Stats and the dashboard widgets keep counting deleted prints.
Frontend:
- ConfirmModal gained an optional children slot.
- ArchivesPage (both card and detail views) own a per-instance
deletePurgeStats boolean and render an opt-in checkbox in the delete
dialog; resets to off on every close so the destructive option is
never sticky.
- api.deleteArchive(id, purgeStats?) appends ?purge_stats=true only
when the box is ticked.
- One new i18n key archives.modal.deletePurgeStats added across all 8
locales (full German, English fallbacks elsewhere).
* feat(auth): proxy OIDC provider icons server-side (#1333)
Strict img-src CSP blocked external OIDC icon hosts on the login page.
Loosening CSP was rejected via the MakerWorld precedent, so icons are
proxied: admin sets icon_url, backend fetches and caches the bytes in a
deferred BLOB column, the SPA renders from a same-origin
/api/v1/auth/oidc/providers/{id}/icon endpoint.
The assign flow was sending slicer-invalid values for tray_info_idx and an
empty setting_id, which the slicer rejected — slot detail modal showed
empty fields. With a stored k-profile the realignment path masked the
issue; without one, garbage hit MQTT.
Backend (apply_spool_to_slot_via_mqtt):
- Discard tray_info_idx values that aren't real preset IDs: literal
material names ("PLA", "PETG-CF") AND PFUS-prefix cloud setting_ids
(valid as setting_id but rejected as tray_info_idx). Same check applied
to current_tray_info_idx so stale slot values don't get reused as
garbage.
- Local-preset path now reads the printer-recognized filament_id from
the preset's setting JSON (e.g. P4d64437) instead of falling through
to a generic material ID.
- Derive setting_id from filament_id_to_setting_id when empty so
ams_filament_setting always carries a matched pair.
- No stored k-profile: always send cali_idx=-1 (Default K), regardless
of the live cali_idx on the slot. The live value belongs to whatever
filament was there before, so reusing it would apply the wrong K to
the new spool.
Frontend (spool-form/utils.ts):
- Local preset options use String(preset.id) as the unique code instead
of preset.filament_type — every PLA local preset was collapsing onto
the same "PLA" code, so picking any of them saved slicer_filament=
"PLA" and lost the specific preset identity.
Spoolman counterpart in spoolman_inventory.py mirrors the cali_idx=-1
reset.
Slicing an STL via the integrated slicer always defaulted to whatever
curr_bed_type lived in the chosen process preset (typically "Cool
Plate"), which the slicer CLI rejected for high-temp filaments with
"Plate 1: Cool Plate does not support filament 1". The user had no
way to switch plates without cloning the preset in BambuStudio.
The Slice modal now exposes a Build plate dropdown with the six
canonical BambuStudio / OrcaSlicer plates (Cool Plate, Cool Plate
SuperTack, Engineering Plate, High Temp Plate, Textured PEI Plate,
Smooth PEI Plate) plus an "Auto (use process preset)" option that
preserves the previous behavior. Positioned between Process profile
and Filament rows so a long filament list never pushes it off the
modal's scrolled viewport, and always enabled regardless of whether
the user picked a Printer Preset Bundle.
A new bed_type field on SliceRequest flows through both dispatch
paths:
- Resolved-preset path: _patch_process_bed_type overwrites
curr_bed_type on the process JSON before forwarding to the sidecar.
Works end-to-end today, no sidecar change needed.
- Bundle dispatch path: slice_with_bundle adds a bedType form field
to the sidecar multipart. The sidecar (maziggy/orca-slicer-api
fork) needs a matching change to honor it as --curr_bed_type on
the CLI invocation; until then the field is silently ignored and
the slice runs with the bundle's default plate.
A1 Mini BMCU (01.07.02.00) and P1S Standard AMS (00.00.06.75) always
report tray.state=3, even for loaded configured slots. The empty-slot
detection preferred state==11 with tray_type as a fallback only when
state was absent, so every assign was classified as empty and MQTT
was skipped — both for "assign to unconfigured slot" and the secondary
"PETG over a PLA-configured slot won't reconfigure" symptom.
Empty-slot detection in the assign route and the on_ams_change replay
now treats the slot as loaded when EITHER state==11 OR tray_type is
non-empty. Reset-slot case (state=11 + tray_type="") still works
through the first clause; configured slots on these firmwares now
work through the second.
Truly empty unconfigured slots (state!=11 + tray_type="") still hit
the pending-config path, and the deferred publish now fires when the
user later configures the slot in Bambu Studio (tray_type goes
non-empty), since the replay uses the same disjunction.
Editing a spool's color name on Spoolman-backed inventory appeared to
accept the new value but the inventory list column and the next edit
showed it back to the subtype. Three layers stacked to produce this:
1. find_or_create_filament matches by material/name/color_hex/vendor —
color_name is intentionally not part of the match key, but on a
match it returned the existing filament's id unchanged, silently
dropping the new value.
2. The read helper falls back to subtype when filament.color_name is
empty (kept on purpose: without it Spoolman installs that don't
fill the field render every spool as "Unknown color").
3. The edit form prefilled color_name from spool.color_name — which
on those installs was the synth value. Changing subtype but not
color_name silently round-tripped the OLD subtype back to Spoolman
as if it were a real user-set color_name.
Fixes:
- find_or_create_filament now patches the matched filament's
color_name via the existing patch_filament wrapper when the request
differs. Parameter convention: None = don't touch, "" = explicit
clear, any other string = set/update. A patch failure is logged but
does not block the match.
- The PATCH route uses model_fields_set to distinguish "field omitted"
from "field explicitly set to null" (mirrors the existing
storage_location pattern at the same site).
- The map helper returns color_name_is_synthesized: bool. The edit
form leaves the input blank when true, so the user sees the real
stored state and can't accidentally round-trip the synth value back.
fix(auth): cleanup orphan OIDC/MFA rows on user delete (#1285)
Three User-FK tables (user_oidc_links, user_totp, user_otp_codes)
declare ON DELETE CASCADE in their models, but SQLite ships with
PRAGMA foreign_keys=OFF (the project's existing pattern, mirrored
for APIKey in PR #1182). Without explicit DELETEs, deleting a user
on SQLite leaves orphan rows behind:
The Clear Plate button (and 4 other features on the Printers page) read
their state from /settings, which requires SETTINGS_READ. Granting that
permission also adds the Settings nav item and leaks SMTP/LDAP/MQTT
credentials — exactly what users were trying to avoid by giving an
operator only printers:clear_plate.
New /settings/ui-preferences endpoint returns a curated, opt-in subset
of non-sensitive fields. Matches the existing /default-sidebar-order
precedent. PrintersPage switched to the new endpoint; admin pages still
use /settings for full access.
_sync_ldap_user used to replace user.groups entirely on every login,
wiping manual admin assignments to groups outside the LDAP mapping.
Now partitions on LDAP-managed group names (mapping values + default
group) and only rebuilds that slice from LDAP truth. Manual assignments
to non-managed groups are preserved; revocation in LDAP still
propagates for managed groups.
H2C / H2D AMS-HT units report ams_id 128+ (one ams_id per unit, single
tray), but spoolman_slot_assignments.ck_ams_id_range only admitted 0-7
and 255. Every attempt to link a Spoolman spool to an AMS-HT slot died
with `CHECK constraint failed: ck_ams_id_range`. The internal
spool_assignment table has no such constraint and works fine.
Widen the formula to (0-7) OR (128-191) OR 255 in the model, the
CREATE TABLE DDL, and an idempotent in-place migration for existing
installs (Postgres: DROP/ADD CONSTRAINT; SQLite: detect stale formula
in sqlite_master, rebuild via _v2 rename pattern).
The MJPEG fan-out broadcaster from #1089 only solved viewer-side
concurrency. Obico polling (every 5s) and the manual /camera/snapshot
endpoint kept opening their own fresh RTSP sockets, which X1/H2/P2
firmwares tolerated but X2D firmware 01.01.00.00 enforces strict
single-connection on — every poll kicked the live stream.
Add try_get_active_buffered_frame(printer_id): returns the broadcaster's
last buffered frame when a viewer is connected, None otherwise. Obico
and /camera/snapshot consult it before opening a fresh socket. When no
viewer is active they fall through to the existing fresh-capture path.
plate_detection and layer_timelapse intentionally not converted.
Severity: Warning ×4
Issue: "Probable insecure usage of temp file/directory" — /tmp/<filename> literals used as synthetic DB field values in two integration tests
Status: Fixed
────────────────────────────────────────
Tool: CodeQL Python / JS
Severity: Pending
Issue: Still running on the head SHA
Status: —
────────────────────────────────────────
Tool: Trivy container scan
Severity: Pending
Issue: Still running
Status: —
────────────────────────────────────────
Tool: Bandit (Python Security Analysis)
Severity: Pass
Issue: The separate Bandit run on the changes already passes
Status: ✓
feat(#1239): first cut at Gitea backups silently failing after 1st run
feat(#1239): Added Token Scope for Forgejo edge case. Also included: test coverage for fixes
The kiosk's Settings -> Update Daemon button returned "API keys cannot
be used for administrative operations" because POST /spoolbuddy/devices/
{id}/update was gated on Permission.SETTINGS_UPDATE, and SETTINGS_UPDATE
is in the _APIKEY_DENIED_PERMISSIONS deny-list introduced by PR #1241.
Every kiosk-side request tripped the deny-list before the API key's
scope set (Read / Print Queue / Control / Legacy) was even consulted.
Same root cause as the four QuickMenu System buttons fixed in 0.2.4b3
(Restart Daemon / Restart Browser / Reboot / Shutdown). Missed /update
in that audit on the reasoning "replaces the daemon binary, different
threat surface" — but that's wrong: restart_daemon already replaces
the running daemon process, so daemon-replacement is not a step up in
blast radius. The SSH update is also strictly scoped to the one device
the operator physically controls (git fetch + pip install + systemctl
restart on that host) — same threat profile as the system commands
already running on INVENTORY_UPDATE.
Lower /spoolbuddy/devices/{id}/update from SETTINGS_UPDATE to
INVENTORY_UPDATE so it aligns with the rest of the kiosk-scoped routes
(calibration/tare, display, cancel-write, system/command,
system/command-result, update-status). The main Bambuddy in-app updater
at POST /api/v1/updates/apply keeps SETTINGS_UPDATE — that one runs on
the Bambuddy host and is correctly fenced behind the deny-list.
Reported by @1000Delta. The printer file download (and three sibling
endpoints) raised UnicodeEncodeError: 'latin-1' codec can't encode
characters... on any filename outside U+0000..U+00FF (Chinese,
Japanese, Arabic, accented Latin), because the route pushed `filename`
straight into Content-Disposition: attachment; filename="...".
Starlette/uvicorn encodes response headers as latin-1, so the assignment
crashed at write-time.
New backend/app/utils/http.py::build_content_disposition emits both an
ASCII-stripped legacy filename="..." fallback and an RFC 5987
filename*=UTF-8''<percent-encoded> parameter. Every modern browser
prefers the *= form, so the original Unicode filename round-trips
through Save-As intact.
Same shape was latent in three siblings and fixed in the same PR
(no deferred follow-ups): archive QR endpoint (archive.print_name
from 3MF metadata), project ZIP export (project.name — the existing
isalnum() sanitiser passes non-ASCII through), and the PDF label
streamer (latent today, callers ASCII-only but the helper hardens it).
Seven intertwined SpoolBuddy + Spoolman bugs from feature/spoolman-inventory-ui
testing, fixed as one batch since they all live on the same path:
1. /spoolbuddy/nfc/tag-scanned always tried local DB first and only
consulted Spoolman as a fallback on local-DB miss. A stale local
row silently won over the authoritative Spoolman record. Now gates
on _get_spoolman_client_or_none() so the route uses Spoolman
exclusively when enabled, local exclusively otherwise.
2. Dashboard "Assign to AMS" button was a no-op when the matched
spool wasn't yet in the cached spools query (newly created in
Spoolman, or unarchived after page load). The card rendered via
`displayedSpool ?? sbState.matchedSpool` fallback but the modal's
stricter guard silently failed to mount. New effectiveModalSpool
synthesises an InventorySpool-shaped object from the WebSocket-
delivered MatchedSpool (9-field subset, sufficient for the modal
since it only needs `id` to route the assign API).
3. AMS-page slot picker explicitly returned null for the
assign/unassign branch when a slot had a SpoolmanSlotAssignment
but no tag-linked spool — only Configure stayed visible. Now
resolves the assignment via spoolmanSlotAssignmentsAll +
spoolmanInventorySpoolsCache, renders a "Assigned spool" info
card, and exposes an Unassign button wired to a new
unassignSpoolmanSlotMutation (DELETE
/spoolman/inventory/slot-assignments/<id>).
4. LinkSpoolModal showed "Unknown color" for every Spoolman spool
because Spoolman doesn't standardise color_name — most installs
only populate color_hex and filament.name (which often carries
the colour, e.g. "PLA Basic Red"). _map_spoolman_spool now falls
back to the filament's subtype (filament name minus material
prefix) when color_name is empty, so spools are visually
distinguishable. The NFC write-tag warning specifically checks
the raw filament.color_name (not the mapped value) so the
"tag encodes empty color name" warning still fires on installs
that genuinely lack the field.
5. Writing a tag for spool B didn't clear the same tag from spool A,
so a single NFC UID could map to two spools at once and
find_spool_by_tag returned whichever came first in the cached
list. nfc_write_result now searches Spoolman for any other spool
currently bound to the target UID and clears its extra.tag
(best-effort: cleanup failure logs a warning but doesn't block
the write, since the chip is already written).
6. The kiosk display held stale spoolmanSlotAssignments cache
permanently because a long-running browser window has no
focus/remount triggers to fire a refetch. Adds
refetchInterval: 3_000 so the kiosk picks up changes from another
client (Bambuddy main UI, direct Spoolman edit) within seconds.
7. Kiosk QuickMenu System buttons (Restart Daemon / Restart Browser /
Reboot / Shutdown) all 403'd silently. /system/command was gated
on Permission.SETTINGS_UPDATE (T-Gap 2 from a prior security
audit) but every other kiosk-scoped device route uses
INVENTORY_UPDATE; the kiosk operator's session has the latter,
not the former. Lowered to INVENTORY_UPDATE so operators can
recover the kiosk from the kiosk. Risk is bounded — only the 4
named commands are accepted (no RCE), reboot/shutdown require
physical-access recovery anyway, the same operator already
controls printers + weighs spools. /update keeps SETTINGS_UPDATE
because it can replace the daemon binary.
feat(spoolman-inventory): squashed feature work for rebase onto dev
Squashed all commits from feature/spoolman-inventory-ui onto a single commit
to enable a clean rebase onto dev. Original per-commit history preserved at
backup tag backup/spoolman-inventory-ui-prerebase-20260507-105721.
chore(i18n): extend parity gate to all locales with strict/info tiers
Previously the script only inspected en/zh-CN/zh-TW, leaving de/fr/it/ja/pt-BR
drift invisible. Now locales are auto-discovered from src/i18n/locales/, and a
STRICT list (de, zh-CN, zh-TW — currently in parity) gates CI while the rest
report informationally until their drift is caught up. ja notably has 27 real
placeholder bugs worth fixing before promotion to strict.
feat(spoolman-inventory): squashed feature work for rebase onto dev
Squashed all commits from feature/spoolman-inventory-ui onto a single commit
to enable a clean rebase onto dev. Original per-commit history preserved at
backup tag backup/spoolman-inventory-ui-prerebase-20260507-105721.
The embedded GCode viewer's static assets (gcode_viewer/) were never
copied into the production Docker image, so /gcode-viewer/ returned a
bare FastAPI 404 ({"detail":"Not Found"}) and 3D Preview broke for every
Docker user since the viewer landed in 0.2.4b1. The Vite production
build doesn't stage the directory either — the dev server serves it via
a configureServer middleware that's dev-only.
Dockerfile now copies gcode_viewer/ alongside the React build output.
Defence in depth: main.py logs an ERROR at startup when
_gcode_viewer_dir/index.html is missing so future packaging gaps surface
in docker logs and the support bundle instead of as silent runtime 404s.
The existing integration test accepted 404 unconditionally
(assert response.status_code in (200, 404)) so CI never caught the
missing files. Add test_gcode_viewer_index_served_when_assets_present
which skips when the directory is intentionally absent (unit-test envs)
but asserts 200 + non-empty HTML body when the assets do exist on disk —
so a broken COPY fails CI loudly rather than shipping a broken image.
maziggy