li3: correct status gate and current scaling; harden BLE/MQTT recovery

- status gate and decode_status now both key off one parsed int
  (0 == OK in any repr; unparseable = unreliable), so they can't diverge
  on "0"/"00"/"0000"
- main-pack current parsed at its real 0.1 A scale (was raw)
- run `systemctl restart bluetooth` + settle delay off the event loop
  (asyncio.to_thread, timeout=30; errors logged, not fatal); call site awaits
- bounded MQTT startup: exit (let systemd restart) if broker never connects
  within MQTT_STARTUP_TIMEOUT_S
- re-assert HA discovery on every (re)connect so entities survive a
  mosquitto restart that drops retained config
- expire_after *4 -> *8 so routine scan misses don't flap the battery
  to unavailable

li3/van-li3-battery

@@ -39,6 +39,7 @@ from bleak import BleakClient, BleakScanner
 BLE_RETRY_DELAY_S = 8
 STUCK_DISCOVERY_THRESHOLD = 10      # consecutive scan failures before we intervene
 BLUETOOTH_RESTART_COOLDOWN_S = 300  # don't restart bluetooth.service more than this often
+MQTT_STARTUP_TIMEOUT_S = 30         # exit (let systemd restart) if the broker never connects
 
 config_path = sys.argv[1] if len(sys.argv) > 1 else "/etc/van-li3/config.json"
 with open(config_path) as f:
@@ -122,10 +123,19 @@ STATUS_FLAGS = [
 ]
 
 
-def decode_status(hex_code):
+def status_value(status):
+    """Parse the raw hex status field to an int; None if unparseable.
+    Single source of truth for both the "is this OK" gate and the flag
+    decode, so the two can't disagree on e.g. "0" vs "000000"."""
     try:
-        value = int(hex_code, 16)
+        return int(status, 16)
     except (ValueError, TypeError):
+        return None
+
+
+def decode_status(hex_code):
+    value = status_value(hex_code)
+    if value is None:
         return hex_code
     active = [
         label for i, label in enumerate(STATUS_FLAGS)
@@ -152,7 +162,7 @@ def parse_line(line):
             "cell4": round(int(parts[4]) * 0.01, 2),
             "bms_temp_f": int(parts[5]),
             "batt_temp_f": int(parts[6]),
-            "current_a": int(parts[7]),
+            "current_a": round(int(parts[7]) * 0.1, 1),  # reported in 0.1A units
             "soc_pct": int(parts[8]),
             "status": parts[9] if len(parts) > 9 else "?",
         }
@@ -231,7 +241,7 @@ def publish_discovery(mqc):
         topic = f"{discovery_prefix}/sensor/{device_id}/{key}/config"
         payload = _sensor_config(
             key, name, unit, device_class, precision, state_topic, device_info,
-            expire_after=publish_interval_s * 4,
+            expire_after=publish_interval_s * 8,
         )
         mqc.publish(topic, json.dumps(payload), retain=True)
 
@@ -242,7 +252,7 @@ def publish_discovery(mqc):
         "state_topic": state_topic,
         "value_template": "{{ value_json.status_text }}",
         "device": device_info,
-        "expire_after": publish_interval_s * 4,
+        "expire_after": publish_interval_s * 8,
     }
     mqc.publish(status_topic, json.dumps(status_payload), retain=True)
 
@@ -267,7 +277,11 @@ def build_state_payload(cs_reading, trace_fields):
     # ("&" line) fields come from a separate message and are published
     # regardless.
     status = cs_reading["status"]
-    if status != "000000":
+    # "OK" == integer zero in ANY representation. Gate on the parsed int (not
+    # the raw string) so "0"/"0000"/"000000" all agree with decode_status, and
+    # an unparseable status (e.g. "?") is treated as unreliable and suppressed.
+    value = status_value(status)
+    if value is None or value != 0:
         payload = {"status": status}
     else:
         payload = {
@@ -287,10 +301,20 @@ def build_state_payload(cs_reading, trace_fields):
     return payload
 
 
-def restart_bluetooth_service():
+async def restart_bluetooth_service():
     print("scan failed; restarting bluetooth.service")
-    subprocess.run(["systemctl", "restart", "bluetooth.service"], check=False)
+    # Run systemctl + the settle delay off the event loop (worker thread) so
-    time.sleep(3)
+    # they don't block the loop; paho drives its own socket thread regardless.
+    try:
+        await asyncio.to_thread(
+            subprocess.run,
+            ["systemctl", "restart", "bluetooth.service"],
+            check=False,
+            timeout=30,
+        )
+    except (subprocess.SubprocessError, OSError) as e:
+        print("bluetooth restart failed:", e)
+    await asyncio.sleep(3)
 
 
 async def find_device():
@@ -308,7 +332,7 @@ async def find_device():
         if consecutive_failures >= STUCK_DISCOVERY_THRESHOLD:
             now = time.time()
             if (now - last_bluetooth_restart) >= BLUETOOTH_RESTART_COOLDOWN_S:
-                restart_bluetooth_service()
+                await restart_bluetooth_service()
                 last_bluetooth_restart = now
             consecutive_failures = 0
         await asyncio.sleep(BLE_RETRY_DELAY_S)
@@ -366,8 +390,33 @@ async def stream_from_device(dev, mqc):
 async def main():
     mqc = mqtt.Client(mqtt.CallbackAPIVersion.VERSION2, client_id=client_id)
     mqc.username_pw_set(mq_user, mq_pw)
+
+    # Reassert HA discovery on every (re)connect. If mosquitto restarts and
+    # loses its retained messages, the whole device disappears from HA until
+    # discovery is re-sent; doing it from here (paho thread) keeps the entities
+    # self-restoring. Idempotent on the very first connect.
+    def _reassert_discovery(client, *args):
+        publish_discovery(client)
+
+    mqc.on_connect_callback_add(_reassert_discovery)
+
     mqc.connect(broker, mqtt_port, keepalive=60)
     mqc.loop_start()
+
+    # The one failure that makes this service pointless is "no broker at all":
+    # streaming BLE to a dead broker just wastes power and hides the fault. So
+    # give the connection a bounded startup window and exit (letting systemd
+    # restart us) if it never establishes. A good start's transient drops are
+    # left to paho's built-in auto-reconnect above.
+    deadline = time.monotonic() + MQTT_STARTUP_TIMEOUT_S
+    while not mqc.is_connected() and time.monotonic() < deadline:
+        await asyncio.sleep(1)
+    if not mqc.is_connected():
+        print("ERROR: MQTT broker %s:%s not reachable; exiting" % (broker, mqtt_port))
+        mqc.loop_stop()
+        mqc.disconnect()
+        sys.exit(1)
+
     publish_discovery(mqc)
 
     while True: