Fix BLE RPC communication with devices that have frame length corruption bug (#991)

Author: bdraco

aioshelly/rpc_device/blerpc.py

@@ -319,6 +319,11 @@ async def _receive_response(self) -> bytes:
                 raise DeviceConnectionError(msg)
 
             frame_length = _UNPACK_UINT32_BE(length_data[:UINT32_BYTES])[0]
+            _LOGGER.debug(
+                "RX control raw bytes: %s, frame_length=%d",
+                length_data[:UINT32_BYTES].hex(),
+                frame_length,
+            )
             if frame_length == 0:
                 # Device hasn't prepared response yet, wait and retry
                 _LOGGER.debug("Frame length 0, polling again in %ss", RX_POLL_INTERVAL)
@@ -360,6 +365,22 @@ async def _receive_response(self) -> bytes:
             data_bytes.extend(chunk)
 
         if len(data_bytes) < frame_length:
+            # Workaround for firmware bug: some devices return corrupted frame length
+            # in RX control but provide complete valid JSON response in data
+            # characteristic.
+            # If we got valid complete JSON, use it despite the incorrect frame length.
+            try:
+                json_loads(data_bytes)
+                _LOGGER.debug(
+                    "Frame length mismatch (expected %d, got %d) but data is valid",
+                    frame_length,
+                    len(data_bytes),
+                )
+                return bytes(data_bytes)
+            except ValueError:
+                # Not valid JSON or incomplete - this is a real error
+                pass
+
             msg = (
                 f"Incomplete data received: expected {frame_length} bytes, "
                 f"got {len(data_bytes)}"

tests/rpc_device/test_blerpc.py

@@ -3,6 +3,7 @@
 from __future__ import annotations
 
 import asyncio
+import json
 from typing import Any
 from unittest.mock import AsyncMock, MagicMock, Mock, patch
 
@@ -323,6 +324,38 @@ async def test_blerpc_call_incomplete_data(
         await ble_rpc.call("Shelly.GetDeviceInfo")
 
 
+@pytest.mark.asyncio
+@pytest.mark.usefixtures("mock_establish_connection")
+async def test_blerpc_call_corrupted_frame_length_valid_json(
+    ble_device: BLEDevice, mock_ble_client: MagicMock
+) -> None:
+    """Test BLE RPC call with corrupted frame length but valid complete JSON.
+
+    Workaround for firmware bug where RX control returns wrong frame length
+    but the actual data is complete valid JSON.
+    """
+    ble_rpc = BleRPC(ble_device)
+
+    mock_ble_client.write_gatt_char = AsyncMock()
+    mock_ble_client.read_gatt_char = AsyncMock()
+
+    response = {"id": 1, "src": "test", "result": {"name": "Test Device"}}
+    response_bytes = json.dumps(response).encode()
+
+    # Mock RX control says 840106079 bytes (corrupted), but we get complete JSON
+    mock_ble_client.read_gatt_char.side_effect = [
+        (840106079).to_bytes(4, "big"),  # Corrupted frame length
+        response_bytes,  # Complete valid JSON response
+        b"",  # Empty chunk (no more data)
+    ]
+
+    await ble_rpc.connect()
+
+    # Should succeed despite corrupted frame length because JSON is valid
+    result = await ble_rpc.call("Shelly.GetDeviceInfo")
+    assert result == {"name": "Test Device"}
+
+
 @pytest.mark.asyncio
 @pytest.mark.usefixtures("mock_establish_connection")
 async def test_blerpc_call_invalid_json(