Enhance info_from_service function in zeroconf integration (#31059)

* enhance zeroconf service info decoding to include raw bytes

* Update homeassistant/components/zeroconf/__init__.py

Co-Authored-By: Paulus Schoutsen <paulus@home-assistant.io>

* fix test based on last commit

* fix test based on last commit

* remove .keys() when asserting processed and raw service info properties

Co-authored-by: Paulus Schoutsen <paulus@home-assistant.io>

homeassistant/components/zeroconf/__init__.py

@@ -148,15 +148,20 @@ def handle_homekit(hass, info) -> bool:
 
 def info_from_service(service):
     """Return prepared info from mDNS entries."""
-    properties = {}
+    properties = {"_raw": {}}
 
     for key, value in service.properties.items():
+        # See https://ietf.org/rfc/rfc6763.html#section-6.4 and
+        # https://ietf.org/rfc/rfc6763.html#section-6.5 for expected encodings
+        # for property keys and values
+        key = key.decode("ascii")
+        properties["_raw"][key] = value
+
         try:
             if isinstance(value, bytes):
-                value = value.decode("utf-8")
-            properties[key.decode("utf-8")] = value
+                properties[key] = value.decode("utf-8")
         except UnicodeDecodeError:
-            _LOGGER.warning("Unicode decode error on %s: %s", key, value)
+            pass
 
     address = service.addresses[0]
 

tests/components/zeroconf/test_init.py

@@ -31,7 +31,7 @@ def get_service_info_mock(service_type, name):
         weight=0,
         priority=0,
         server="name.local.",
-        properties={b"macaddress": b"ABCDEF012345"},
+        properties={b"macaddress": b"ABCDEF012345", b"non-utf8-value": b"ABCDEF\x8a"},
     )
 
 
@@ -93,3 +93,16 @@ async def test_homekit_match_full(hass, mock_zeroconf):
     assert len(mock_service_browser.mock_calls) == 1
     assert len(mock_config_flow.mock_calls) == 2
     assert mock_config_flow.mock_calls[0][1][0] == "hue"
+
+
+async def test_info_from_service_non_utf8(hass):
+    """Test info_from_service handles non UTF-8 property values correctly."""
+    service_type = "_test._tcp.local."
+    info = zeroconf.info_from_service(
+        get_service_info_mock(service_type, f"test.{service_type}")
+    )
+    raw_info = info["properties"].pop("_raw", False)
+    assert raw_info
+    assert len(info["properties"]) <= len(raw_info)
+    assert "non-utf8-value" not in info["properties"]
+    assert raw_info["non-utf8-value"] is not None