fix(moonraker): metadata state-leak + terminal-state reset gaps (Issue #102)

Three related gaps found via careful moonraker-obico observation:

1. _build_file_metadata() read layer_height/total_layers/estimated_time
   from live self._state before falling back to the queried file's own
   GCodeStore row - so querying metadata for any file other than the
   currently/last tracked job leaked that job's values into the response.
   Live state is now only used when the query targets that same tracked
   file; any other filename relies solely on its own stored row.

2. curr_layer/total_layers were never reset at print end in either
   _on_print or _on_info - only explicitly overwritten if a later
   payload happened to carry those keys, otherwise stuck indefinitely.
   Additionally, a successful "finished" print only ever cleared
   file_ready (Issue #29's fix), while stoped/canceled reset every
   other per-job field (progress, filename, duration, ...) - finished
   now gets the same full reset. Found and fixed a knock-on bug this
   surfaced: the unconditional `self._state["filename"] = d.get(...)`
   right after the reset block would immediately undo the filename
   reset, since the printer's own finished/stoped/canceled payload
   still carries the just-ended job's filename - now only applied
   outside terminal states.

3. The printer's own "progress" during pre-print phases (auto_leveling/
   preheating/checking/updated/init) was forwarded as-is to
   virtual_sdcard.progress/display_status.progress, causing a
   non-monotonic jump-then-reset once real printing began. These phases
   no longer update the tracked progress value.

Reported by @fmontagna, who correctly identified all three as genuine
gaps rather than intentional behavior.
This commit is contained in:
2026-07-26 23:29:32 +02:00
parent 6d6df59ff2
commit 0ca7618c85
3 changed files with 190 additions and 13 deletions

View File

@@ -7,3 +7,6 @@
- Fix: **the camera stream could hang forever after a printer reboot** — the printer rotates its stream token on reboot, but the running ffmpeg processes kept reading from the stale, now-silent connection and never noticed. The bridge now detects the URL change and automatically restarts the camera pipeline, ffmpeg itself now times out on a stalled read as a second line of defense, and `/api/camera/stream` now returns a proper 503 instead of hanging if no frame arrives within 5s (Issue #99, thanks @fmontagna for the excellent root-cause analysis and reproduction)
- Fix: shrinking the Progress dashboard tile below its default height clipped the lower content (time grid, filename, buttons) out of view instead of scaling with it; dashboard tiles in general could show a scrollbar even when there was nothing to scroll. Both fixed (Issue #97, thanks @Blaim)
- Feat: custom ACE-RFID filament tags (e.g. written via the "ACE RFID" app for third-party spools) are now matched automatically against your imported OrcaSlicer profiles. Anycubic's ACE RFID system concatenates vendor + material + a truncated serial into one type string (e.g. "GEEETECH PLA Bas") for custom tags — the bridge now recognizes the vendor prefix and resolves it to your already-imported profile instead of falling back to a generic default, so no manual per-slot reassignment is needed anymore (Issue #101, thanks @Blaim)
- Fix: `server/files/metadata` could leak the currently/last tracked print job's layer count and estimated time into a query for a *different* file, since those values were read from live state before falling back to the queried file's own stored metadata. Now only the currently-tracked file's own query uses live state — any other filename relies solely on its own record (Issue #102, thanks @fmontagna)
- Fix: `curr_layer`/`total_layers` (and, after a successful print, every other per-job field — progress, filename, duration, etc.) were never reset when a print finished normally, only on stop/cancel — they stayed at the last job's values until the next print happened to overwrite them (Issue #102, thanks @fmontagna)
- Fix: the printer's own progress value during pre-print phases (auto-leveling/preheating/bed-checking) was passed straight through to `virtual_sdcard.progress`/`display_status.progress`, causing it to jump non-monotonically once real printing started and progress reset. Pre-print progress is no longer forwarded (Issue #102, thanks @fmontagna)

View File

@@ -1394,12 +1394,12 @@ class KobraXBridge:
self._spoolman_notify_end()
self._current_job_id = ""
# Hide the upload banner after the print finishes (Issue #29): the
# printer reports "finished" after a successful print - file_ready used to be
# cleared only on stoped/canceled, which brought the banner back.
if kobra_state == "finished":
self._state["file_ready"] = ""
if kobra_state in ("stoped", "canceled"):
# Terminal states (successful finish AND stop/cancel) must leave the
# same clean end state - a "finished" print used to only clear
# file_ready (Issue #29), leaving progress/filename/duration/layer
# fields stuck at the last job's values until the *next* print
# happened to overwrite them (Issue #102).
if kobra_state in ("finished", "stoped", "canceled"):
self._state["progress"] = 0.0
self._state["filename"] = ""
self._state["file_ready"] = ""
@@ -1409,9 +1409,20 @@ class KobraXBridge:
self._state["layer_height"] = 0.0
self._state["first_layer_height"] = 0.0
self._state["supplies_usage"] = 0
self._state["curr_layer"] = 0
self._state["total_layers"] = 0
self._thumbnail_b64 = ""
self._state["filename"] = d.get("filename", self._state["filename"])
if "progress" in d:
else:
# Only adopt the payload's filename outside terminal states - the
# printer often still reports the just-finished job's filename in
# the same "finished"/"stoped"/"canceled" message that triggered
# the reset above, which would otherwise immediately undo it.
self._state["filename"] = d.get("filename", self._state["filename"])
# Pre-print phases (leveling/preheating/checking) report their own
# "progress" - passing it through would make display_status.progress/
# virtual_sdcard.progress jump non-monotonically once real printing
# starts and the value resets (Issue #102).
if "progress" in d and kobra_state not in ("preheating", "auto_leveling", "checking", "updated", "init"):
self._state["progress"] = float(d["progress"]) / 100.0
if "print_time" in d:
self._state["print_duration"] = int(d["print_time"]) * 60
@@ -1447,8 +1458,13 @@ class KobraXBridge:
self._state["kobra_state"] = kobra_state
# Hide the upload banner after the print ends (Issue #29) - the state also
# arrives via info/report (project.state) depending on the printer, not only print/report.
# Layer fields must reset here too (Issue #102) - info/report is the
# only source for curr_layer/total_layers on some printers, and they
# otherwise stay stuck at the last job's values indefinitely.
if kobra_state in ("finished", "stoped", "canceled"):
self._state["file_ready"] = ""
self._state["curr_layer"] = 0
self._state["total_layers"] = 0
# Camera auto-start here as well (OrcaSlicer often reports the start via info/report).
# The _camera_autostarted guard prevents a double start with _on_print.
if kobra_state == "printing":
@@ -1467,7 +1483,8 @@ class KobraXBridge:
if project:
if "filename" in project:
self._state["filename"] = project["filename"]
if "progress" in project:
# Same non-monotonic-progress guard as _on_print (Issue #102).
if "progress" in project and kobra_state not in ("preheating", "auto_leveling", "checking", "updated", "init"):
self._state["progress"] = float(project["progress"]) / 100.0
if "print_time" in project:
self._state["print_duration"] = int(project["print_time"]) * 60
@@ -3405,10 +3422,16 @@ class KobraXBridge:
`modified` are non-nullable in GCodeFile; `print_start_time` and the
Slicer-Felder optional."""
s = self._state
layer_h = float(s.get("layer_height") or 0.0)
first_h = float(s.get("first_layer_height") or 0.0)
total_layers = int(s.get("total_layers") or 0)
est_time = int(s.get("slicer_time") or 0)
# Live _state values are only relevant for the currently/last tracked
# job's own file - using them as a starting point for a DIFFERENT
# filename leaked the tracked job's layer count/time into unrelated
# metadata queries (Issue #102). For any other filename, rely solely
# on that file's own GCodeStore row.
is_tracked_file = bool(filename) and filename == s.get("filename")
layer_h = float(s.get("layer_height") or 0.0) if is_tracked_file else 0.0
first_h = float(s.get("first_layer_height") or 0.0) if is_tracked_file else 0.0
total_layers = int(s.get("total_layers") or 0) if is_tracked_file else 0
est_time = int(s.get("slicer_time") or 0) if is_tracked_file else 0
size_bytes = 0
try:
gf = self._store.get_file_by_name(filename) or {}

View File

@@ -0,0 +1,151 @@
"""server/files/metadata state-leak + terminal-state reset gaps (Issue #102).
Reported by @fmontagna via moonraker-obico:
1. Querying metadata for a file OTHER than the currently/last tracked job
leaked that job's live layer count / estimated time into the response,
because _build_file_metadata() read from self._state first and only fell
back to the file's own GCodeStore row when the state value was falsy.
2. curr_layer/total_layers (and, for a successful "finished" print, every
other per-job field) were never reset at print end - they stayed at the
last job's values until the next print happened to overwrite them.
3. The printer reports its own "progress" during pre-print phases
(preheating/auto_leveling/checking/...), which used to pass straight
through to display_status.progress/virtual_sdcard.progress and then jump
non-monotonically once real printing started and progress reset.
"""
import argparse
import tempfile
from unittest.mock import MagicMock
from kobrax_moonraker_bridge import KobraXBridge
def _bridge():
c = MagicMock(); c.callbacks = {}; c.connected = False
args = argparse.Namespace(
printer_ip="", mqtt_port=9883, username="", password="",
mode_id="20030", device_id="", host="127.0.0.1", port=7125,
data_dir=tempfile.mkdtemp(prefix="kxmeta-"),
)
return KobraXBridge(c, args=args)
def _insert_store_row(b, filename, layer_count=None, est_time=None, size_bytes=0):
with b._store._lock:
b._store._conn.execute(
"INSERT OR REPLACE INTO gcode_files (id, filename, path, size_bytes, uploaded_at, layer_count, est_print_time_sec) "
"VALUES (?,?,?,?,?,?,?)",
(filename, filename, "/tmp/" + filename, size_bytes, "2026-01-01T00:00:00Z", layer_count, est_time),
)
b._store._conn.commit()
# --- Fix 1: metadata state-leak -------------------------------------------
def test_metadata_for_untracked_file_does_not_leak_running_job_state():
b = _bridge()
# A job is "running": live state has its own layer/time values.
b._state["filename"] = "running.gcode"
b._state["total_layers"] = 999
b._state["slicer_time"] = 12345
b._state["layer_height"] = 0.3
# Querying a DIFFERENT, unrelated file must use ITS OWN store row, not
# the running job's live state.
_insert_store_row(b, "other.gcode", layer_count=42, est_time=600, size_bytes=1000)
meta = b._build_file_metadata("other.gcode")
assert meta["layer_count"] == 42
assert meta["estimated_time"] == 600
assert meta["size"] == 1000
def test_metadata_for_tracked_file_still_uses_live_state():
"""The currently-tracked file's OWN metadata query should still prefer
live state (fresher than what was known at upload time)."""
b = _bridge()
b._state["filename"] = "running.gcode"
b._state["total_layers"] = 55
b._state["slicer_time"] = 999
_insert_store_row(b, "running.gcode", layer_count=1, est_time=1)
meta = b._build_file_metadata("running.gcode")
assert meta["layer_count"] == 55
assert meta["estimated_time"] == 999
def test_metadata_for_nonexistent_file_falls_back_cleanly():
b = _bridge()
b._state["filename"] = "running.gcode"
b._state["total_layers"] = 999
meta = b._build_file_metadata("DOES_NOT_EXIST.gcode")
assert meta["layer_count"] is None
assert meta["size"] == 1 # documented fallback, unrelated to this fix
# --- Fix 2: terminal-state reset -------------------------------------------
def _print_payload(state, **extra):
"""print/report envelope: `state` is top-level, everything else is under
`data` (see _on_print: kobra_state = payload.get("state", "")). """
d = {"filename": "job.gcode"}
d.update(extra)
return {"state": state, "data": d}
def test_finished_resets_layer_fields_like_stoped_canceled():
b = _bridge()
b._state["curr_layer"] = 10
b._state["total_layers"] = 20
b._state["progress"] = 0.5
b._state["filename"] = "job.gcode"
b._on_print(_print_payload("finished"))
assert b._state["curr_layer"] == 0
assert b._state["total_layers"] == 0
assert b._state["progress"] == 0.0
assert b._state["filename"] == ""
def test_canceled_resets_layer_fields():
b = _bridge()
b._state["curr_layer"] = 7
b._state["total_layers"] = 20
b._on_print(_print_payload("canceled"))
assert b._state["curr_layer"] == 0
assert b._state["total_layers"] == 0
def test_on_info_resets_layer_fields_on_terminal_state():
b = _bridge()
b._state["curr_layer"] = 7
b._state["total_layers"] = 20
b._on_info({"data": {"project": {"state": "finished"}}})
assert b._state["curr_layer"] == 0
assert b._state["total_layers"] == 0
# --- Fix 3: progress clamping during pre-print phases ----------------------
def test_progress_not_updated_during_auto_leveling():
b = _bridge()
b._state["progress"] = 0.0
b._on_print(_print_payload("auto_leveling", progress=87))
assert b._state["progress"] == 0.0
def test_progress_not_updated_during_preheating():
b = _bridge()
b._state["progress"] = 0.0
b._on_print(_print_payload("preheating", progress=42))
assert b._state["progress"] == 0.0
def test_progress_updates_normally_once_printing():
b = _bridge()
b._on_print(_print_payload("printing", progress=33))
assert b._state["progress"] == 0.33
def test_on_info_progress_clamped_during_checking():
b = _bridge()
b._state["progress"] = 0.0
b._on_info({"data": {"project": {"state": "checking", "progress": 55}}})
assert b._state["progress"] == 0.0