P2S firmware sends local slot IDs (0-3) in tray_now, not global tray
IDs, when multiple AMS units are connected. The UI highlighted the
wrong AMS unit (e.g., AMS-A instead of AMS-B). Usage tracking was
unaffected because it uses the MQTT mapping field.
Added disambiguation in _handle_ams_data for single-nozzle printers
with ams_exist_bits indicating >1 AMS: cross-references tray_now with
the MQTT mapping field (snow-encoded) to resolve the correct global
tray ID. Falls back to raw value when no mapping is available or
ambiguous. No behavior change for single-AMS printers or dual-nozzle
(H2D) printers.
Add Quick Add mode to the spool form for simplified entry (material +
color + weight only), and a quantity field (1-100) for creating multiple
identical spools at once. Stock spools (no slicer profile) are computed
rather than stored — any spool without slicer_filament shows an amber
"Stock" badge. New filter chips (All/Stock/Configured) on the inventory
page. Backend bulk endpoint creates N spools in one transaction.
Backend:
- SpoolBulkCreate schema with quantity validation (1-100)
- POST /inventory/spools/bulk endpoint
Frontend:
- Quick Add toggle in SpoolFormModal (create mode only)
- Quantity field in FilamentSection (both modes)
- bulkCreateSpools API method and bulkCreateMutation
- Stock column (hidden by default) and stock filter chips
- Inline error rendering moved into FilamentSection
- CellCtx extended with t() for translatable cell content
Tests:
- 13 backend tests (schema validation + endpoint logic)
- 10 frontend tests (validateForm quickAdd + UI behavior)
i18n: 6 locales (en, de, fr, it, ja, pt-BR)
Docs: CHANGELOG, README, website features, wiki inventory
Three bugs allowed uploaded gcode files to survive on the printer's SD
card, causing firmware restarts to auto-start old prints hours later.
1. The post-print cleanup retry loop always broke after attempt 1
regardless of delete success — delete_file_async returns False on
failure instead of raising, so the except-based retry never fired.
2. When start_print() failed after uploading, the file was never cleaned
up since on_print_complete never fires for prints that never started.
3. Cleanup failures were logged at DEBUG level, invisible in production.
When a spool runs empty mid-print and the AMS auto-switches to a backup,
the entire print weight was attributed to the original spool. Track tray
changes during the print via tray_change_log on PrinterState, then split
the 3MF weight across trays using per-layer gcode segment boundaries at
print completion. Falls back to linear layer-ratio splitting when gcode
data is unavailable.
When a print was started from BambuStudio (not Bambuddy), the
auto-archive has an empty file_path. The photo save code uses
base_dir / Path("").parent which resolves correctly, but the
serve/upload/delete endpoints used (base_dir / "").parent which
goes one directory level too high — returning 404 despite the
photo existing on disk.
ntfy servers with attachments disabled reject thumbnail uploads with
"attachments not allowed" (HTTP 400 / code 40014), causing the entire
notification to fail. Now automatically retries without the image so
the notification still gets delivered.
Infrastructure/auth HMS error codes like 0500_0007 (MQTT command
verification failed) triggered false-alarm printer error notifications
even though they don't indicate actual print problems.
- Add suppress list for non-actionable error codes: 0500_0007 (MQTT
auth failure), 0500_4001 (cloud connection failure), 0500_400E
(print cancelled by user)
- Skip suppressed codes in the notification loop so they are still
tracked in HMS state and shown in the UI, but don't fire notifications
The log sanitizer only used regex patterns, missing arbitrary user-chosen
strings (printer names, usernames). Tasmota smart plug credentials were
logged verbatim in URLs by httpx.
- Make _sanitize_log_content() database-aware: query Printer names/serials,
User usernames, and Bambu Cloud email for exact-string replacement
(longest-first, skip <3 chars to prevent over-redaction)
- Fix serial regex leaking first 3 chars (remove capture group partial
redaction), add case-insensitive flag
- Move Tasmota credentials from URL-embedded (http://user:pass@host) to
httpx auth= parameter so they never appear in logs
- Add URL credentials regex as defense-in-depth for user:pass@ in logs
- Add 'username' and 'path' to settings sensitive_keys filter (catches
smtp_username, slicer_binary_path in support-info.json)
When sending notifications with image attachments (progress/error
events), the message body was placed in an HTTP Message header. Messages
containing newlines (e.g. "Printer: file.stl\nRemaining: 4h 52m") were
rejected by httpx since HTTP headers cannot contain newline characters.
Escape newlines to literal \n which ntfy renders as line breaks.
Two bugs in the inventory page:
1. Dates always displayed as DD/MM/YY regardless of the user's date
format setting. The page had a local formatDate() hardcoding en-GB
locale. Replaced with the shared formatDateOnly() from utils/date.ts.
2. AMS-HT slot locations showed accented characters (e.g., "Á1") instead
of proper labels (e.g., "HT-A"). The code used String.fromCharCode(65
+ ams_id), which produces garbled chars for AMS-HT IDs (128+). Now
uses the shared formatSlotLabel() from utils/amsHelpers.ts.
The post-print SD card cleanup used printer_manager.get_printer(),
which returns a PrinterInfo object with only name and serial_number.
The code then accessed .ip_address, .access_code, and .model, which
don't exist on PrinterInfo — raising AttributeError on every print
completion. The outer except Exception handler silently swallowed it
at debug level.
Result: uploaded gcode files were never deleted from the printer's
SD card root. On some printers (e.g. P1S), files left in root
auto-start on power cycle, causing phantom prints.
Replaced with a DB query for the Printer model, matching the pattern
used everywhere else in on_print_complete().
MQTT raw data returns AMS unit and tray IDs as strings (e.g. "0"),
but _build_loaded_filaments() compared them to integers without
casting. After the scheduler assigned an idle printer to an "Any
Model" job, it crashed computing AMS mapping with:
'>=' not supported between instances of 'str' and 'int'
The exception prevented the assignment from committing, so the
scheduler retried every 30 seconds in an infinite loop. The job
stayed in "Waiting" and the "Clear Plate" button kept reappearing.
Cast ams_id and tray_id to int(), matching the pattern already used
for external spool IDs on line 550.
When a spool ran empty during a print and the AMS auto-switched to a
backup, on_ams_change eagerly deleted the SpoolAssignment (fingerprint
mismatch). on_print_complete then queried the now-deleted assignment
and silently dropped usage.
Snapshot all spool assignments into PrintSession at print start. At
completion, prefer the snapshot over live DB queries in both the 3MF
primary path and the AMS remain% delta fallback. Fall back to live
query for backward compatibility with in-flight sessions started
before upgrade.
The MQTT callback thread checked self._pending_kprofile_response before
calling .set(), but the asyncio thread could clear it to None between
the check and the use (TOCTOU race after get_kprofiles() timeout).
Capture the event reference in a local variable to eliminate the race.
Add frontend-typecheck (tsc --noEmit) and frontend-lint (eslint .)
hooks to .pre-commit-config.yaml. Both hooks only trigger when
frontend/src/**/*.{ts,tsx} files are staged.
This allows users sending particularly large prints to slow printers such as the P1-series to start a print and then move onto another task in Bambuddy immediately (such as starting more prints on more printers). It also gives the user visibility into what's happening instead of a loading indicator appearing for an indefinite period of time.
The new toast-based UI uses websockets to update in real time. It will also appear for other users / instances of Bambuddy, not just the user who started the prints, allowing more transparency and handling cases where the user closes the page and then comes back wanting to know the status of the dispatching.
fix: review fixes for background dispatch PR #408
- Restore missing imports in main.py (inventory, print_log, virtual_printers, mqtt_smart_plug_service)
- Guard voidresp() for A1 printers to prevent hang after upload
- Don't fail upload on voidresp() error since data transfer already completed
- Add ams_mapping to register_expected_print calls for Spoolman usage tracking
- Fix cancel_job TOCTOU race by using single lock acquisition
- Fix batch counter reset TOCTOU by re-checking condition inside second lock
- Add backgroundDispatch translations to fr.ts and pt-BR.ts
- Remove dead upload_progress_callback definitions
- Skip redundant "Print queued" toast in reprint mode (dispatch toast handles it)
- 5 backend tests: cancel_job single-lock TOCTOU, batch reset re-check,
job lifecycle
- 2 FTP regression tests: voidresp error handling (upload-loop fix),
A1 model voidresp skip
- 1 frontend test: reprint toast suppression
- CHANGELOG: background dispatch feature + test coverage entries
- README: add background dispatch to Scheduling & Automation
- Website: add feature item to Automation section
- Wiki: add Background Print Dispatch section to print-queue.md
The previous fix passed target_model from the frontend, but this relied
on printer.model being set on the client. When a printer was added
without selecting a model (the field is optional), the frontend sent no
target_model parameter, causing the OR logic to be bypassed entirely —
the widget only queried printer_id=X and missed unassigned model-based
items.
The backend now looks up the printer's model from the database when
target_model isn't provided by the client, ensuring "Clear Plate &
Start Next" appears for model-based queue jobs regardless of how the
printer was configured.
Users on the Docker `latest` tag were seeing update notifications for beta
releases (e.g. v0.2.1b) they couldn't install. The update checker now fetches
/releases instead of /releases/latest and filters by parse_version() prerelease
detection. A new toggle in Settings (default: off) lets users opt in to beta
notifications.
When a print was queued with "Any [Model]", the queue widget only
queried items with printer_id=X after dispatch, missing the next
pending model-based item (printer_id IS NULL, target_model="P1S").
The "Clear Plate & Start Next" button never appeared, leaving the
scheduler stuck reporting "Busy".
Add optional target_model query parameter to GET /queue/. When
combined with printer_id, uses OR logic to also return unassigned
items whose target_model matches. The frontend now passes the
printer's model through PrinterQueueWidget to this query.
BambuStudio actively resets AMS slots configured with unrecognized PFUS*
(user-local) preset IDs. Replace PFUS* with generic Bambu filament IDs
(e.g. GFL99 for PLA) in both the slot configure and inventory assignment
endpoints. When the slot already has a recognized cloud-synced preset for
the same material, reuse it to preserve K-profile calibration.
Also fix fill level bar not showing for brand new spools (weight_used=0)
by changing the condition from weight_used > 0 to weight_used != null.
When the snow field hadn't arrived yet, the ams_extruder_map fallback
computed ams_id * 4 + slot for AMS-HT units (128-135), producing
invalid tray IDs like 512+. AMS-HT has a single slot; global ID = unit
ID.
- Single-AMS fallback: return ams_id directly for AMS-HT (128-135)
- Multi-AMS matching: decompose AMS-HT trays correctly (slot=0)
- Multi-AMS narrowing: slot > 0 eliminates AMS-HT candidates
- last_loaded_tray: tighten to valid IDs only (0-15, 128-135, 254)
- Add 10 tests for AMS-HT fallback and last_loaded_tray validation
Multiple Virtual Printers:
- Each VP gets a dedicated bind IP with independent FTP, MQTT, SSDP, and Bind services
- New VirtualPrinter DB model, CRUD API (/api/virtual-printers), React UI
- VirtualPrinterList, VirtualPrinterCard, VirtualPrinterAddDialog components
- Per-instance TLS certificates (shared CA), 11 printer models, all 4 modes
- Auto-incremented serial suffixes, network interface override per VP
Dual Bind/Detect Ports (#445):
- Listen on both ports 3000 and 3002 for slicer bind/detect handshake
- Different BambuStudio/OrcaSlicer versions use different ports
- Applies to BindServer (server mode) and SlicerProxyManager (proxy mode)
- Updated Dockerfile, docker-compose.yml, firewall rules in wiki
Also:
- Rewrote VP test suite for new multi-instance architecture (75 tests)
- Rewritten "How it works" section with 3-step workflow explanation
- Updated all 5 locales (en, de, ja, fr, it)
- Updated wiki and website for multi-VP + dual ports
- New multi-VP screenshot
The printer only sends PAUSE via MQTT gcode_state, never PAUSED.
Removed all unreachable PAUSED comparisons from frontend (4 files)
and backend (2 files).
Add a new `printers:clear_plate` permission so admins can grant
plate-clearing ability without full `printers:control` access.
Existing groups with `printers:control` automatically receive the
new permission on startup via migration.
Replace the cramped permission modal in Settings with a dedicated
full-page group editor (`/groups/new`, `/groups/:id/edit`) featuring
search filtering, select all/clear all, category-level toggles,
and a responsive 2-column permission grid. Remove dead GroupsPage
code that was never routed.
- Backend: new Permission enum, category, defaults, endpoint guard, migration
- Frontend: GroupEditPage, updated routes, removed modal from SettingsPage
- Tests: 10 backend + 10 frontend tests for new permission and editor
- Docs: updated wiki (auth, printer-control, API), website, changelog
- i18n: added group editor keys to all 6 locale files (en/de/ja/fr/pt-BR/it)
high) by changing audit-level from "high" to "critical" in .npmrc. Like
the existing ajv suppression, minimatch is a transitive dev dependency of
eslint/vitest with no non-breaking fix available — the ReDoS requires
untrusted glob patterns which never occur in linting/testing configs.
Library files are stored on disk with UUID-hex filenames for uniqueness,
but archive_print() used the disk path as the display name. Pass the
original LibraryFile.filename through from both the print scheduler and
direct-print-from-library flow so the archive's filename, print_name,
and directory name all use the human-readable name.
Add color-matching fallback for slot-to-tray mapping on printers that
don't provide the MQTT mapping field (snow encoding) or reject request
topic subscription. Matches 3MF filament slot colors against AMS tray
colors to resolve the correct physical tray. Returns no match when
colors are ambiguous (duplicate tray colors) to safely fall through
to existing fallbacks.
Mapping priority chain: print_cmd → MQTT snow → queue → color match →
tray_now → slot_id-1.
Slicer protocol handlers (bambustudio://, orcaslicer://) launch the
slicer app which fetches files via HTTP but cannot send auth headers.
The global auth middleware returned 401, causing "importing failed."
Add short-lived, single-use download tokens: the frontend fetches a
token via authenticated POST, then builds a /dl/{token}/{filename} URL
the slicer can access without auth headers. Tokens are validated
server-side (5-min expiry, single-use). Applies to archive files,
source 3MFs, and library files.
Also fix platform-specific URL formats to match slicer source code:
- macOS: bambustudioopen:// with encodeURIComponent
- Windows/Linux: bambustudio://open?file= (was wrongly using macOS scheme on Linux)
- OrcaSlicer: orcaslicer://open?file=
The spool add/edit modal allowed saving without brand or subtype,
causing incomplete tray_sub_brands when assigned to AMS slots.
BambuStudio then failed to recognize the filament profile.
Brand and Subtype are now mandatory with validation errors on submit.
The /filaments/ endpoint manages material type definitions (cost, temps,
density) but shares its name with the UI's "Filament" page which shows
the spool inventory (/inventory/spools/). This caused users to expect
the API to return their spool inventory.
Rename to /filament-catalog/ to make the distinction clear. No frontend
behavior change — these API methods are defined but unused in the UI.
On dual-nozzle printers (H2C, H2D Pro) with multiple AMS units, usage
tracking attributed filament consumption to the wrong spools. The MQTT
mapping field — a snow-encoded array mapping slicer slots to physical
AMS trays — was preserved in state but never parsed.
Add _decode_mqtt_mapping() to convert snow encoding (ams_hw_id * 256 +
slot) to bambuddy global tray IDs. Insert as priority 2 in the mapping
resolution chain, after print_cmd ams_mapping but before queue/tray_now
fallbacks. Works for all printer models and print sources.
AMS-HT trays (IDs 128-135) are single-slot units, but the snow_slot
computation used `snow_tray % 4 if snow_tray < 128 else -1`, yielding
-1 for AMS HT. This caused a misleading "slot mismatch" debug log on
every MQTT update despite producing the correct tray_now value. Now
returns slot 0 for AMS-HT IDs (128-135), matching the firmware's
tray_now=0 for these single-slot units.
P2S (and X1C) with multiple AMS units incorrectly entered the H2D
dual-nozzle tray_now disambiguation code path, causing warning log
spam. The check `tray_now <= 3` matched any printer loading from
AMS 0. Now uses a persistent `_is_dual_nozzle` flag detected from
device.extruder.info (>= 2 entries), which only dual-nozzle printers
(H2D, H2D Pro) report. Flag is set early in message processing so
it's available before AMS data is parsed.
maziggy