fix(camera): recover automatically after printer reboot rotates stream token (Issue #99)

The printer rotates its FLV/RTSP stream token on reboot. CameraCache.set_url()
was a bare assignment, so the three running ffmpeg loops never noticed - they
only re-read self._url at the top of their outer loop, which they never reach
while permanently blocked in a stdout read on the now-silent, stale-token
connection (TCP stays ESTABLISHED with no data and no FIN, so a passive
reader can't tell "peer is quiet" from "peer is gone").

Three-part fix, all per the excellent root-cause analysis and reproduction
in the issue (thanks @fmontagna):

1. set_url() now detects a URL change and calls reset() to tear down the
   stale ffmpeg loops, so the next ensure_running() respawns them against
   the new URL.
2. ffmpeg's -timeout option (10s, applies to both RTSP and HTTP-FLW since
   both share _input_args) as a second line of defense for a source that
   goes silent while the URL stays the same (network loss, camera hang).
3. handle_camera_stream now waits up to 5s for the first frame BEFORE
   calling resp.prepare() and returns 503 on timeout. Previously it had no
   first-frame timeout at all, and prepare() commits the response status
   to 200 - so a stalled source could never surface as an error to the
   client, only as an infinite hang.

NIGHTLY_CHANGELOG.md

@@ -4,3 +4,4 @@
 - Feat: the GCode browser now supports multi-select and bulk delete — click any file's checkbox to enter select mode, use "Select All" to grab everything currently visible (respecting your active search/filter), then delete the selection with one confirmation instead of one-by-one (Issue #94, thanks @Blaim)
 - Fix: **printing via OrcaSlicer "Upload and print" failed (or fed the wrong spool) when a slot below the used filament was empty** — e.g. printing with Filament 4 while slot 3 was empty. The generated AMS slot mapping inserted a placeholder pointing at the physically empty tray, which the printer rejects. Placeholders now point at a loaded tray instead. Printing with all slots full already worked and is unchanged.
 - Fix: manually assigning a filament profile to an ACE slot could crash the MQTT callback (`'list' object has no attribute 'get'`) when the printer rejected the assignment, e.g. for custom-RFID/third-party filament. The bridge now handles the printer's failure response cleanly instead of crashing, and the log now correlates a rejected assignment with the request that triggered it (slot/type/color) so a rejection reason can be diagnosed from bridge logs alone (Issue #100, thanks @Blaim)
+- 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)

kobrax_moonraker_bridge.py

@@ -631,7 +631,16 @@ class CameraCache:
         self._fail_count_mjpeg: int = 0
 
     def set_url(self, url: str):
+        # A changed URL means the printer rotated its stream token (typically
+        # after a reboot). Running ffmpeg processes still hold the stale URL
+        # and will never pick it up on their own - they only re-read self._url
+        # at the top of their outer loop, which they never reach while blocked
+        # in a stdout read on the old, now-silent connection. Tear them down;
+        # the next ensure_running() respawns them against the new URL.
+        changed = bool(url and self._url and url != self._url)
         self._url = url
+        if changed:
+            self.reset()
 
     def reset(self):
         """Reset backoff counters and forcefully tear down any running
@@ -679,7 +688,12 @@ class CameraCache:
             self._task_mjpeg = asyncio.create_task(self._run_mjpeg_loop())
 
     def _input_args(self, url: str) -> list[str]:
-        args = ["-fflags", "nobuffer", "-flags", "low_delay"]
+        args = ["-fflags", "nobuffer", "-flags", "low_delay",
+                 # Bail out if the source goes silent. A printer reboot or
+                 # network loss leaves the TCP connection ESTABLISHED with no
+                 # data and no FIN, so a passive stdout read blocks forever
+                 # without this (Issue #99). Value is microseconds.
+                 "-timeout", "10000000"]
         if url.lower().startswith("rtsp://"):
             args += ["-probesize", "32", "-analyzeduration", "0", "-rtsp_transport", "tcp"]
         else:
@@ -4368,6 +4382,16 @@ class KobraXBridge:
         q: asyncio.Queue[bytes] = asyncio.Queue(maxsize=8)
         self.camera_cache.mjpeg_subscribers.add(q)
 
+        # Wait for the first frame BEFORE resp.prepare() - once prepare() sends
+        # the response headers the status is committed to 200, so a stalled
+        # source (Issue #99) must be caught here to actually return a 503
+        # instead of hanging the client forever with no frame ever arriving.
+        try:
+            first_frame = await asyncio.wait_for(q.get(), timeout=5.0)
+        except asyncio.TimeoutError:
+            self.camera_cache.mjpeg_subscribers.discard(q)
+            return web.Response(status=503, text="No frame in cache yet")
+
         boundary = "kobraxframe"
         resp = web.StreamResponse(headers={
             "Content-Type": f"multipart/x-mixed-replace;boundary={boundary}",
@@ -4376,8 +4400,8 @@ class KobraXBridge:
         })
         await resp.prepare(request)
         try:
+            frame = first_frame
             while True:
-                frame = await q.get()
                 header = (
                     f"--{boundary}\r\n"
                     f"Content-Type: image/jpeg\r\n"
@@ -4389,6 +4413,7 @@ class KobraXBridge:
                     break
                 except Exception:
                     break
+                frame = await q.get()
         except Exception as e:
             log.warning(f"Camera stream interrupted: {e}")
         finally:

tests/test_camera_cache_url_rotation.py

@@ -0,0 +1,51 @@
+"""Camera stream hangs forever after printer reboot (Issue #99).
+
+The printer rotates its stream token on reboot, changing the camera URL.
+CameraCache.set_url() used to be a bare assignment - the running ffmpeg loops
+never noticed since they only re-read self._url at the top of their outer
+loop, which they never reach while permanently blocked reading stdout from
+the now-silent, stale-token connection. set_url() must detect the change and
+tear the loops down so the next ensure_running() respawns them against the
+new URL.
+"""
+from kobrax_moonraker_bridge import CameraCache
+
+
+def test_set_url_first_time_does_not_reset():
+    """No prior URL - nothing stale to tear down."""
+    c = CameraCache()
+    calls = []
+    c.reset = lambda: calls.append(True)
+    c.set_url("http://printer/live/tokenA")
+    assert c._url == "http://printer/live/tokenA"
+    assert not calls
+
+
+def test_set_url_same_value_does_not_reset():
+    c = CameraCache()
+    calls = []
+    c.set_url("http://printer/live/tokenA")
+    c.reset = lambda: calls.append(True)
+    c.set_url("http://printer/live/tokenA")
+    assert not calls
+
+
+def test_set_url_changed_triggers_reset():
+    """The actual bug scenario: token rotates after a printer reboot."""
+    c = CameraCache()
+    calls = []
+    c.set_url("http://printer/live/tokenA")
+    c.reset = lambda: calls.append(True)
+    c.set_url("http://printer/live/tokenB")
+    assert c._url == "http://printer/live/tokenB"
+    assert calls == [True]
+
+
+def test_set_url_empty_to_value_does_not_reset():
+    """Startup case: no URL known yet, first status push sets it - nothing
+    running to tear down."""
+    c = CameraCache()
+    calls = []
+    c.reset = lambda: calls.append(True)
+    c.set_url("http://printer/live/tokenA")
+    assert not calls