Base entity ids on English for languages not using Latin script (#91357)

homeassistant/generated/languages.py

@@ -3,6 +3,8 @@
 To update, run python3 -m script.languages [frontend_tag]
 """
 
+DEFAULT_LANGUAGE = "en"
+
 LANGUAGES = {
     "af",
     "ar",
@@ -66,3 +68,46 @@ LANGUAGES = {
     "zh-Hans",
     "zh-Hant",
 }
+
+NATIVE_ENTITY_IDS = {
+    "af",
+    "bs",
+    "ca",
+    "cs",
+    "cy",
+    "da",
+    "de",
+    "en",
+    "en-GB",
+    "eo",
+    "es",
+    "es-419",
+    "et",
+    "eu",
+    "fi",
+    "fr",
+    "fy",
+    "gl",
+    "gsw",
+    "hr",
+    "hu",
+    "id",
+    "is",
+    "it",
+    "ka",
+    "lb",
+    "lt",
+    "lv",
+    "nb",
+    "nl",
+    "nn",
+    "pl",
+    "pt",
+    "pt-BR",
+    "ro",
+    "sk",
+    "sl",
+    "sr-Latn",
+    "sv",
+    "tr",
+}

homeassistant/helpers/entity.py

@@ -381,17 +381,31 @@ class Entity(ABC):
             return self.entity_description.has_entity_name
         return False
 
-    @cached_property
-    def _device_class_name(self) -> str | None:
+    def _device_class_name_helper(
+        self,
+        component_translations: dict[str, Any],
+    ) -> str | None:
         """Return a translated name of the entity based on its device class."""
         if not self.has_entity_name:
             return None
         device_class_key = self.device_class or "_"
         platform = self.platform
         name_translation_key = (
-            f"component.{platform.domain}.entity_component." f"{device_class_key}.name"
+            f"component.{platform.domain}.entity_component.{device_class_key}.name"
         )
-        return platform.component_translations.get(name_translation_key)
+        return component_translations.get(name_translation_key)
+
+    @cached_property
+    def _object_id_device_class_name(self) -> str | None:
+        """Return a translated name of the entity based on its device class."""
+        return self._device_class_name_helper(
+            self.platform.object_id_component_translations
+        )
+
+    @cached_property
+    def _device_class_name(self) -> str | None:
+        """Return a translated name of the entity based on its device class."""
+        return self._device_class_name_helper(self.platform.component_translations)
 
     def _default_to_device_class_name(self) -> bool:
         """Return True if an unnamed entity should be named by its device class."""
@@ -408,15 +422,18 @@ class Entity(ABC):
             f".{self.translation_key}.name"
         )
 
-    @property
-    def name(self) -> str | UndefinedType | None:
+    def _name_internal(
+        self,
+        device_class_name: str | None,
+        platform_translations: dict[str, Any],
+    ) -> str | UndefinedType | None:
         """Return the name of the entity."""
         if hasattr(self, "_attr_name"):
             return self._attr_name
         if (
             self.has_entity_name
             and (name_translation_key := self._name_translation_key)
-            and (name := self.platform.platform_translations.get(name_translation_key))
+            and (name := platform_translations.get(name_translation_key))
         ):
             if TYPE_CHECKING:
                 assert isinstance(name, str)
@@ -424,15 +441,42 @@ class Entity(ABC):
         if hasattr(self, "entity_description"):
             description_name = self.entity_description.name
             if description_name is UNDEFINED and self._default_to_device_class_name():
-                return self._device_class_name
+                return device_class_name
             return description_name
 
         # The entity has no name set by _attr_name, translation_key or entity_description
         # Check if the entity should be named by its device class
         if self._default_to_device_class_name():
-            return self._device_class_name
+            return device_class_name
         return UNDEFINED
 
+    @property
+    def suggested_object_id(self) -> str | None:
+        """Return input for object id."""
+        # The check for self.platform guards against integrations not using an
+        # EntityComponent and can be removed in HA Core 2024.1
+        # mypy doesn't know about fget: https://github.com/python/mypy/issues/6185
+        if self.__class__.name.fget is Entity.name.fget and self.platform:  # type: ignore[attr-defined]
+            name = self._name_internal(
+                self._object_id_device_class_name,
+                self.platform.object_id_platform_translations,
+            )
+        else:
+            name = self.name
+        return None if name is UNDEFINED else name
+
+    @property
+    def name(self) -> str | UndefinedType | None:
+        """Return the name of the entity."""
+        # The check for self.platform guards against integrations not using an
+        # EntityComponent and can be removed in HA Core 2024.1
+        if not self.platform:
+            return self._name_internal(None, {})
+        return self._name_internal(
+            self._device_class_name,
+            self.platform.platform_translations,
+        )
+
     @property
     def state(self) -> StateType:
         """Return the state of the entity."""

homeassistant/helpers/entity_platform.py

@@ -31,6 +31,7 @@ from homeassistant.exceptions import (
     PlatformNotReady,
     RequiredParameterMissing,
 )
+from homeassistant.generated import languages
 from homeassistant.setup import async_start_setup
 from homeassistant.util.async_ import run_callback_threadsafe
 
@@ -128,6 +129,8 @@ class EntityPlatform:
         self.entities: dict[str, Entity] = {}
         self.component_translations: dict[str, Any] = {}
         self.platform_translations: dict[str, Any] = {}
+        self.object_id_component_translations: dict[str, Any] = {}
+        self.object_id_platform_translations: dict[str, Any] = {}
         self._tasks: list[asyncio.Task[None]] = []
         # Stop tracking tasks after setup is completed
         self._setup_complete = False
@@ -294,22 +297,43 @@ class EntityPlatform:
         logger = self.logger
         hass = self.hass
         full_name = f"{self.domain}.{self.platform_name}"
+        object_id_language = (
+            hass.config.language
+            if hass.config.language in languages.NATIVE_ENTITY_IDS
+            else languages.DEFAULT_LANGUAGE
+        )
 
-        try:
-            self.component_translations = await translation.async_get_translations(
-                hass, hass.config.language, "entity_component", {self.domain}
+        async def get_translations(
+            language: str, category: str, integration: str
+        ) -> dict[str, Any]:
+            """Get entity translations."""
+            try:
+                return await translation.async_get_translations(
+                    hass, language, category, {integration}
+                )
+            except Exception as err:  # pylint: disable=broad-exception-caught
+                _LOGGER.debug(
+                    "Could not load translations for %s",
+                    integration,
+                    exc_info=err,
+                )
+            return {}
+
+        self.component_translations = await get_translations(
+            hass.config.language, "entity_component", self.domain
+        )
+        self.platform_translations = await get_translations(
+            hass.config.language, "entity", self.platform_name
+        )
+        if object_id_language == hass.config.language:
+            self.object_id_component_translations = self.component_translations
+            self.object_id_platform_translations = self.platform_translations
+        else:
+            self.object_id_component_translations = await get_translations(
+                object_id_language, "entity_component", self.domain
             )
-        except Exception as err:  # pylint: disable=broad-exception-caught
-            _LOGGER.debug(
-                "Could not load translations for %s", self.domain, exc_info=err
-            )
-        try:
-            self.platform_translations = await translation.async_get_translations(
-                hass, hass.config.language, "entity", {self.platform_name}
-            )
-        except Exception as err:  # pylint: disable=broad-exception-caught
-            _LOGGER.debug(
-                "Could not load translations for %s", self.platform_name, exc_info=err
+            self.object_id_platform_translations = await get_translations(
+                object_id_language, "entity", self.platform_name
             )
 
         logger.info("Setting up %s", full_name)
@@ -652,9 +676,11 @@ class EntityPlatform:
                     if entity.use_device_name:
                         suggested_object_id = device_name
                     else:
-                        suggested_object_id = f"{device_name} {entity_name}"
+                        suggested_object_id = (
+                            f"{device_name} {entity.suggested_object_id}"
+                        )
                 if not suggested_object_id:
-                    suggested_object_id = entity_name
+                    suggested_object_id = entity.suggested_object_id
 
             if self.entity_namespace is not None:
                 suggested_object_id = f"{self.entity_namespace} {suggested_object_id}"
@@ -709,7 +735,7 @@ class EntityPlatform:
         # Generate entity ID
         if entity.entity_id is None or generate_new_entity_id:
             suggested_object_id = (
-                suggested_object_id or entity_name or DEVICE_DEFAULT_NAME
+                suggested_object_id or entity.suggested_object_id or DEVICE_DEFAULT_NAME
             )
 
             if self.entity_namespace is not None:

script/languages.py

@@ -15,10 +15,61 @@ req = requests.get(
 data = json.loads(req.content)
 languages = set(data.keys())
 
+# Languages which can be used for entity IDs.
+# Languages in the set are those which use a writing system based on the Latin
+# script. Languages not in this set will instead base the entity ID on English.
+
+# Note: Although vietnamese writing is based on the Latin script, it's too ambiguous
+# after accents and diacritics have been removed by slugify
+NATIVE_ENTITY_IDS = {
+    "af",  # Afrikaans
+    "bs",  # Bosanski
+    "ca",  # Català
+    "cs",  # Čeština
+    "cy",  # Cymraeg
+    "da",  # Dansk
+    "de",  # Deutsch
+    "en",  # English
+    "en-GB",  # English (GB)
+    "eo",  # Esperanto
+    "es",  # Español
+    "es-419",  # Español (Latin America)
+    "et",  # Eesti
+    "eu",  # Euskara
+    "fi",  # Suomi
+    "fr",  # Français
+    "fy",  # Frysk
+    "gl",  # Galego
+    "gsw",  # Schwiizerdütsch
+    "hr",  # Hrvatski
+    "hu",  # Magyar
+    "id",  # Indonesia
+    "is",  # Íslenska
+    "it",  # Italiano
+    "ka",  # Kartuli
+    "lb",  # Lëtzebuergesch
+    "lt",  # Lietuvių
+    "lv",  # Latviešu
+    "nb",  # Nederlands
+    "nl",  # Norsk Bokmål
+    "nn",  # Norsk Nynorsk"
+    "pl",  # Polski
+    "pt",  # Português
+    "pt-BR",  # Português (BR)
+    "ro",  # Română
+    "sk",  # Slovenčina
+    "sl",  # Slovenščina
+    "sr-Latn",  # Srpski
+    "sv",  # Svenska
+    "tr",  # Türkçe
+}
+
 Path("homeassistant/generated/languages.py").write_text(
     format_python_namespace(
         {
+            "DEFAULT_LANGUAGE": "en",
             "LANGUAGES": languages,
+            "NATIVE_ENTITY_IDS": NATIVE_ENTITY_IDS,
         },
         generator="script.languages [frontend_tag]",
     )

tests/helpers/test_entity_platform.py

@@ -1,5 +1,6 @@
 """Tests for the EntityPlatform helper."""
 import asyncio
+from collections.abc import Iterable
 from datetime import timedelta
 import logging
 from typing import Any
@@ -18,6 +19,7 @@ from homeassistant.helpers import (
 )
 from homeassistant.helpers.entity import (
     DeviceInfo,
+    Entity,
     EntityCategory,
     async_generate_entity_id,
 )
@@ -1669,3 +1671,159 @@ async def test_entity_name_influences_entity_id(
 
     assert len(hass.states.async_entity_ids()) == 1
     assert registry.async_get(expected_entity_id) is not None
+
+
+@pytest.mark.parametrize(
+    ("language", "has_entity_name", "expected_entity_id"),
+    (
+        ("en", False, "test_domain.test_qwer"),  # Set to <platform>_<unique_id>
+        ("en", True, "test_domain.device_bla_english_name"),
+        ("sv", True, "test_domain.device_bla_swedish_name"),
+        # Chinese uses english for entity_id
+        ("cn", True, "test_domain.device_bla_english_name"),
+    ),
+)
+async def test_translated_entity_name_influences_entity_id(
+    hass: HomeAssistant,
+    language: str,
+    has_entity_name: bool,
+    expected_entity_id: str,
+) -> None:
+    """Test entity_id is influenced by translated entity name."""
+
+    class TranslatedEntity(Entity):
+        _attr_unique_id = "qwer"
+        _attr_device_info = {
+            "identifiers": {("hue", "1234")},
+            "connections": {(dr.CONNECTION_NETWORK_MAC, "abcd")},
+            "name": "Device Bla",
+        }
+
+        _attr_translation_key = "test"
+
+        def __init__(self, has_entity_name: bool) -> None:
+            """Initialize."""
+            self._attr_has_entity_name = has_entity_name
+
+    registry = er.async_get(hass)
+
+    translations = {
+        "en": {"component.test.entity.test_domain.test.name": "English name"},
+        "sv": {"component.test.entity.test_domain.test.name": "Swedish name"},
+        "cn": {"component.test.entity.test_domain.test.name": "Chinese name"},
+    }
+    hass.config.language = language
+
+    async def async_get_translations(
+        hass: HomeAssistant,
+        language: str,
+        category: str,
+        integrations: Iterable[str] | None = None,
+        config_flow: bool | None = None,
+    ) -> dict[str, Any]:
+        """Return all backend translations."""
+        return translations[language]
+
+    async def async_setup_entry(hass, config_entry, async_add_entities):
+        """Mock setup entry method."""
+        async_add_entities([TranslatedEntity(has_entity_name)])
+        return True
+
+    platform = MockPlatform(async_setup_entry=async_setup_entry)
+    config_entry = MockConfigEntry(entry_id="super-mock-id")
+    entity_platform = MockEntityPlatform(
+        hass, platform_name=config_entry.domain, platform=platform
+    )
+
+    with patch(
+        "homeassistant.helpers.entity_platform.translation.async_get_translations",
+        side_effect=async_get_translations,
+    ):
+        assert await entity_platform.async_setup_entry(config_entry)
+        await hass.async_block_till_done()
+
+    assert len(hass.states.async_entity_ids()) == 1
+    assert registry.async_get(expected_entity_id) is not None
+
+
+@pytest.mark.parametrize(
+    ("language", "has_entity_name", "device_class", "expected_entity_id"),
+    (
+        ("en", False, None, "test_domain.test_qwer"),  # Set to <platform>_<unique_id>
+        (
+            "en",
+            False,
+            "test_class",
+            "test_domain.test_qwer",
+        ),  # Set to <platform>_<unique_id>
+        ("en", True, "test_class", "test_domain.device_bla_english_cls"),
+        ("sv", True, "test_class", "test_domain.device_bla_swedish_cls"),
+        # Chinese uses english for entity_id
+        ("cn", True, "test_class", "test_domain.device_bla_english_cls"),
+    ),
+)
+async def test_translated_device_class_name_influences_entity_id(
+    hass: HomeAssistant,
+    language: str,
+    has_entity_name: bool,
+    device_class: str | None,
+    expected_entity_id: str,
+) -> None:
+    """Test entity_id is influenced by translated entity name."""
+
+    class TranslatedDeviceClassEntity(Entity):
+        _attr_unique_id = "qwer"
+        _attr_device_info = {
+            "identifiers": {("hue", "1234")},
+            "connections": {(dr.CONNECTION_NETWORK_MAC, "abcd")},
+            "name": "Device Bla",
+        }
+
+        def __init__(self, device_class: str | None, has_entity_name: bool) -> None:
+            """Initialize."""
+            self._attr_device_class = device_class
+            self._attr_has_entity_name = has_entity_name
+
+        def _default_to_device_class_name(self) -> bool:
+            """Return True if an unnamed entity should be named by its device class."""
+            return self.device_class is not None
+
+    registry = er.async_get(hass)
+
+    translations = {
+        "en": {"component.test_domain.entity_component.test_class.name": "English cls"},
+        "sv": {"component.test_domain.entity_component.test_class.name": "Swedish cls"},
+        "cn": {"component.test_domain.entity_component.test_class.name": "Chinese cls"},
+    }
+    hass.config.language = language
+
+    async def async_get_translations(
+        hass: HomeAssistant,
+        language: str,
+        category: str,
+        integrations: Iterable[str] | None = None,
+        config_flow: bool | None = None,
+    ) -> dict[str, Any]:
+        """Return all backend translations."""
+        return translations[language]
+
+    async def async_setup_entry(hass, config_entry, async_add_entities):
+        """Mock setup entry method."""
+        async_add_entities([TranslatedDeviceClassEntity(device_class, has_entity_name)])
+        return True
+
+    platform = MockPlatform(async_setup_entry=async_setup_entry)
+    config_entry = MockConfigEntry(entry_id="super-mock-id")
+    entity_platform = MockEntityPlatform(
+        hass, platform_name=config_entry.domain, platform=platform
+    )
+
+    with patch(
+        "homeassistant.helpers.entity_platform.translation.async_get_translations",
+        side_effect=async_get_translations,
+    ):
+        assert await entity_platform.async_setup_entry(config_entry)
+        await hass.async_block_till_done()
+
+    assert len(hass.states.async_entity_ids()) == 1
+    assert registry.async_get(expected_entity_id) is not None