chore: Add additional Home data to diagnostics (#723)

* chore: Add additional Home data to diagnostics

Add home API data including products devices, rooms, etc and
increase the number of redacted fields. Add a device manager
snapshot test to verify which fields are redacted.

* chore: Update device snapshots and lint errors

* chore: Improve redaction logic to support more complex paths

* chore: Fix schema redaction

* chore: Use built-in as_dict method for creating diagnostic data

* chore: fix diagnostic lint issues

Author: allenporter

roborock/devices/device.py

@@ -226,9 +226,11 @@ def diagnostic_data(self) -> dict[str, Any]:
         """Return diagnostics information about the device."""
         extra: dict[str, Any] = {}
         if self.v1_properties:
-            extra["traits"] = redact_device_data(self.v1_properties.as_dict())
-        return {
-            "device": redact_device_data(self.device_info.as_dict()),
-            "product": redact_device_data(self.product.as_dict()),
-            **extra,
-        }
+            extra["traits"] = self.v1_properties.as_dict()
+        return redact_device_data(
+            {
+                "device": self.device_info.as_dict(),
+                "product": self.product.as_dict(),
+                **extra,
+            }
+        )

roborock/devices/device_manager.py

@@ -16,7 +16,7 @@
     UserData,
 )
 from roborock.devices.device import DeviceReadyCallback, RoborockDevice
-from roborock.diagnostics import Diagnostics
+from roborock.diagnostics import Diagnostics, redact_device_data
 from roborock.exceptions import RoborockException
 from roborock.map.map_parser import MapParserConfig
 from roborock.mqtt.roborock_session import create_lazy_mqtt_session
@@ -76,6 +76,7 @@ def __init__(
         self._devices: dict[str, RoborockDevice] = {}
         self._mqtt_session = mqtt_session
         self._diagnostics = diagnostics
+        self._home_data: HomeData | None = None
 
     async def discover_devices(self, prefer_cache: bool = True) -> list[RoborockDevice]:
         """Discover all devices for the logged-in user."""
@@ -91,9 +92,9 @@ async def discover_devices(self, prefer_cache: bool = True) -> list[RoborockDevi
                     raise
                 _LOGGER.debug("Failed to fetch home data, using cached data: %s", ex)
             await self._cache.set(cache_data)
-        home_data = cache_data.home_data
+        self._home_data = cache_data.home_data
 
-        device_products = home_data.device_products
+        device_products = self._home_data.device_products
         _LOGGER.debug("Discovered %d devices", len(device_products))
 
         # These are connected serially to avoid overwhelming the MQTT broker
@@ -106,7 +107,7 @@ async def discover_devices(self, prefer_cache: bool = True) -> list[RoborockDevi
             if duid in self._devices:
                 continue
             try:
-                new_device = self._device_creator(home_data, device, product)
+                new_device = self._device_creator(self._home_data, device, product)
             except UnsupportedDeviceError:
                 _LOGGER.info("Skipping unsupported device %s %s", product.summary_info(), device.summary_info())
                 unsupported_devices_counter.increment(device.pv or "unknown")
@@ -136,7 +137,11 @@ async def close(self) -> None:
 
     def diagnostic_data(self) -> Mapping[str, Any]:
         """Return diagnostics information about the device manager."""
-        return self._diagnostics.as_dict()
+        return {
+            "home_data": redact_device_data(self._home_data.as_dict()) if self._home_data else None,
+            "devices": [device.diagnostic_data() for device in self._devices.values()],
+            "diagnostics": self._diagnostics.as_dict(),
+        }
 
 
 @dataclass

roborock/diagnostics.py

@@ -101,30 +101,50 @@ def reset(self) -> None:
     "imageContent",
     "mapData",
     "rawApiResponse",
+    # Home data
+    "id",  # We want to redact home_data.id but keep some other ids, see below
+    "name",
+    "productId",
+    "ipAddress",
+    "wifiName",
+    "lat",
+    "long",
+}
+KEEP_KEYS = {
+    # Product information not unique per user
+    "product.id",
+    "product.schema.id",
+    "product.schema.name",
+    # Room ids are likely unique per user, but don't seem too sensitive and are
+    # useful for debugging
+    "rooms.id",
 }
 DEVICE_UID = "duid"
 REDACTED = "**REDACTED**"
 
 
-def redact_device_data(data: T) -> T | dict[str, Any]:
+def redact_device_data(data: T, path: str = "") -> T | dict[str, Any]:
     """Redact sensitive data in a dict."""
     if not isinstance(data, (Mapping, list)):
         return data
 
     if isinstance(data, list):
-        return cast(T, [redact_device_data(item) for item in data])
+        return cast(T, [redact_device_data(item, path) for item in data])
 
     redacted = {**data}
 
     for key, value in redacted.items():
-        if key in REDACT_KEYS:
+        curr_path = f"{path}.{key}" if path else key
+        if key in KEEP_KEYS or curr_path in KEEP_KEYS:
+            continue
+        if key in REDACT_KEYS or curr_path in REDACT_KEYS:
             redacted[key] = REDACTED
         elif key == DEVICE_UID and isinstance(value, str):
             redacted[key] = redact_device_uid(value)
         elif isinstance(value, dict):
-            redacted[key] = redact_device_data(value)
+            redacted[key] = redact_device_data(value, curr_path)
         elif isinstance(value, list):
-            redacted[key] = [redact_device_data(item) for item in value]
+            redacted[key] = [redact_device_data(item, curr_path) for item in value]
 
     return redacted