Fix TTS languange characters (#39211)

homeassistant/components/tts/__init__.py

@@ -324,7 +324,9 @@ class SpeechManager:
         else:
             options_key = "-"
 
-        key = KEY_PATTERN.format(msg_hash, language, options_key, engine).lower()
+        key = KEY_PATTERN.format(
+            msg_hash, language.replace("_", "-"), options_key, engine
+        ).lower()
 
         # Is speech already in memory
         if key in self.mem_cache:
@@ -355,9 +357,14 @@ class SpeechManager:
         # Create file infos
         filename = f"{key}.{extension}".lower()
 
-        data = self.write_tags(filename, data, provider, message, language, options)
+        # Validate filename
+        if not _RE_VOICE_FILE.match(filename):
+            raise HomeAssistantError(
+                f"TTS filename '{filename}' from {engine} is invalid!"
+            )
 
         # Save to memory
+        data = self.write_tags(filename, data, provider, message, language, options)
         self._async_store_to_memcache(key, filename, data)
 
         if cache:

tests/components/tts/test_init.py

@@ -176,6 +176,41 @@ async def test_setup_component_and_test_service_with_config_language(
     ).is_file()
 
 
+async def test_setup_component_and_test_service_with_config_language_special(
+    hass, empty_cache_dir
+):
+    """Set up the demo platform and call service with extend language."""
+    import homeassistant.components.demo.tts as demo_tts
+
+    demo_tts.SUPPORT_LANGUAGES.append("en_US")
+    calls = async_mock_service(hass, DOMAIN_MP, SERVICE_PLAY_MEDIA)
+
+    config = {tts.DOMAIN: {"platform": "demo", "language": "en_US"}}
+
+    with assert_setup_component(1, tts.DOMAIN):
+        assert await async_setup_component(hass, tts.DOMAIN, config)
+
+    await hass.services.async_call(
+        tts.DOMAIN,
+        "demo_say",
+        {
+            "entity_id": "media_player.something",
+            tts.ATTR_MESSAGE: "There is someone at the door.",
+        },
+        blocking=True,
+    )
+    assert len(calls) == 1
+    assert calls[0].data[ATTR_MEDIA_CONTENT_TYPE] == MEDIA_TYPE_MUSIC
+    assert (
+        calls[0].data[ATTR_MEDIA_CONTENT_ID]
+        == "http://example.local:8123/api/tts_proxy/42f18378fd4393d18c8dd11d03fa9563c1e54491_en-us_-_demo.mp3"
+    )
+    await hass.async_block_till_done()
+    assert (
+        empty_cache_dir / "42f18378fd4393d18c8dd11d03fa9563c1e54491_en-us_-_demo.mp3"
+    ).is_file()
+
+
 async def test_setup_component_and_test_service_with_wrong_conf_language(hass):
     """Set up the demo platform and call service with wrong config."""
     config = {tts.DOMAIN: {"platform": "demo", "language": "ru"}}