Fix some tests

homeassistant/components/assist_pipeline/pipeline.py

@@ -20,9 +20,6 @@ import hass_nabucasa
 import voluptuous as vol
 
 from homeassistant.components import conversation, stt, tts, wake_word, websocket_api
-from homeassistant.components.tts import (
-    generate_media_source_id as tts_generate_media_source_id,
-)
 from homeassistant.const import ATTR_SUPPORTED_FEATURES, MATCH_ALL
 from homeassistant.core import Context, HomeAssistant, callback
 from homeassistant.exceptions import HomeAssistantError
@@ -1275,26 +1272,10 @@ class PipelineRun:
             )
         )
 
-        try:
-            # Synthesize audio and get URL
-            tts_media_id = tts_generate_media_source_id(
-                self.hass,
-                tts_input,
-                engine=self.tts_stream.engine,
-                language=self.tts_stream.language,
-                options=self.tts_stream.options,
-            )
-        except Exception as src_error:
-            _LOGGER.exception("Unexpected error during text-to-speech")
-            raise TextToSpeechError(
-                code="tts-failed",
-                message="Unexpected error during text-to-speech",
-            ) from src_error
-
         self.tts_stream.async_set_message(tts_input)
 
         tts_output = {
-            "media_id": tts_media_id,
+            "media_id": self.tts_stream.media_source_id,
             "url": self.tts_stream.url,
             "mime_type": self.tts_stream.content_type,
         }

homeassistant/components/assist_satellite/entity.py

@@ -23,9 +23,6 @@ from homeassistant.components.assist_pipeline import (
     vad,
 )
 from homeassistant.components.media_player import async_process_play_media_url
-from homeassistant.components.tts import (
-    generate_media_source_id as tts_generate_media_source_id,
-)
 from homeassistant.core import Context, callback
 from homeassistant.exceptions import HomeAssistantError
 from homeassistant.helpers import chat_session, entity
@@ -484,6 +481,9 @@ class AssistSatelliteEntity(entity.Entity):
             pipeline_id = self._resolve_pipeline()
             pipeline = async_get_pipeline(self.hass, pipeline_id)
 
+            if pipeline.tts_engine is None:
+                raise HomeAssistantError("Pipeline has no TTS engine configured")
+
             tts_options: dict[str, Any] = {}
             if pipeline.tts_voice is not None:
                 tts_options[tts.ATTR_VOICE] = pipeline.tts_voice
@@ -491,14 +491,15 @@ class AssistSatelliteEntity(entity.Entity):
             if self.tts_options is not None:
                 tts_options.update(self.tts_options)
 
-            media_id = tts_generate_media_source_id(
+            stream = tts.async_create_stream(
                 self.hass,
-                message,
                 engine=pipeline.tts_engine,
                 language=pipeline.tts_language,
                 options=tts_options,
             )
-            original_media_id = media_id
+            stream.async_set_message(message)
+            media_id = stream.url
+            original_media_id = stream.media_source_id
 
         if media_source.is_media_source_id(media_id):
             if not media_id_source:

homeassistant/components/tts/__init__.py

@@ -27,6 +27,10 @@ import voluptuous as vol
 
 from homeassistant.components import ffmpeg, websocket_api
 from homeassistant.components.http import HomeAssistantView
+from homeassistant.components.media_source import (
+    Unresolvable,
+    generate_media_source_id as ms_generate_media_source_id,
+)
 from homeassistant.config_entries import ConfigEntry
 from homeassistant.const import EVENT_HOMEASSISTANT_STOP, PLATFORM_FORMAT
 from homeassistant.core import (
@@ -188,10 +192,19 @@ async def async_get_media_source_audio(
 ) -> tuple[str, bytes]:
     """Get TTS audio as extension, data."""
     manager = hass.data[DATA_TTS_MANAGER]
-    cache_key = manager.async_cache_message_in_memory(
+
-        **media_source_id_to_kwargs(media_source_id)
+    if not media_source_id.startswith("media-source://tts/temporary/"):
-    )
+        cache_key = manager.async_cache_message_in_memory(
-    return await manager.async_get_tts_audio(cache_key)
+            **media_source_id_to_kwargs(media_source_id)
+        )
+        return await manager.async_get_tts_audio(cache_key)
+
+    token = media_source_id.split("/")[-1]
+    if (stream := manager.token_to_stream.get(token)) is None:
+        raise Unresolvable("Token from media source not found")
+
+    data = b"".join([chunk async for chunk in stream.async_stream_result()])
+    return stream.extension, data
 
 
 @callback
@@ -394,6 +407,11 @@ class ResultStream:
         """Get the URL to stream the result."""
         return f"/api/tts_proxy/{self.token}"
 
+    @cached_property
+    def media_source_id(self) -> str:
+        """Get the media source ID for the result."""
+        return ms_generate_media_source_id(DOMAIN, f"temporary/{self.token}")
+
     @cached_property
     def _result_cache_key(self) -> asyncio.Future[str]:
         """Get the future that returns the cache key."""

homeassistant/components/tts/media_source.py

@@ -3,7 +3,7 @@
 from __future__ import annotations
 
 import json
-from typing import TypedDict
+from typing import TypedDict, TYPE_CHECKING
 
 from yarl import URL
 
@@ -22,12 +22,15 @@ from homeassistant.exceptions import HomeAssistantError
 from .const import DATA_COMPONENT, DATA_TTS_MANAGER, DOMAIN
 from .helper import get_engine_instance
 
+if TYPE_CHECKING:
+    from . import SpeechManager
+
 URL_QUERY_TTS_OPTIONS = "tts_options"
 
 
 async def async_get_media_source(hass: HomeAssistant) -> TTSMediaSource:
     """Set up tts media source."""
-    return TTSMediaSource(hass)
+    return TTSMediaSource(hass, hass.data[DATA_TTS_MANAGER])
 
 
 @callback
@@ -109,22 +112,31 @@ class TTSMediaSource(MediaSource):
     """Provide text-to-speech providers as media sources."""
 
     name: str = "Text-to-speech"
+    manager: SpeechManager
 
-    def __init__(self, hass: HomeAssistant) -> None:
+    def __init__(self, hass: HomeAssistant, manager: SpeechManager) -> None:
         """Initialize TTSMediaSource."""
         super().__init__(DOMAIN)
         self.hass = hass
+        self.manager = manager
 
     async def async_resolve_media(self, item: MediaSourceItem) -> PlayMedia:
         """Resolve media to a url."""
-        try:
+        if item.identifier.startswith("temporary/"):
-            stream = self.hass.data[DATA_TTS_MANAGER].async_create_result_stream(
+            token = item.identifier.partition("/")[2]
-                **media_source_id_to_kwargs(item.identifier)
+            stream = self.manager.token_to_stream.get(token)
-            )
+            if stream is None:
-        except Unresolvable:
+                raise Unresolvable("Temporary media not found")
-            raise
+
-        except HomeAssistantError as err:
+        else:
-            raise Unresolvable(str(err)) from err
+            try:
+                stream = self.manager.async_create_result_stream(
+                    **media_source_id_to_kwargs(item.identifier)
+                )
+            except Unresolvable:
+                raise
+            except HomeAssistantError as err:
+                raise Unresolvable(str(err)) from err
 
         return PlayMedia(stream.url, stream.content_type)
 
@@ -134,6 +146,9 @@ class TTSMediaSource(MediaSource):
     ) -> BrowseMediaSource:
         """Return media."""
         if item.identifier:
+            if item.identifier.startswith("temporary/"):
+                raise BrowseError("Temporary media cannot be browsed")
+
             engine, _, params = item.identifier.partition("?")
             return self._engine_item(engine, params)
 

tests/components/assist_pipeline/snapshots/test_init.ambr

@@ -83,7 +83,7 @@
     dict({
       'data': dict({
         'tts_output': dict({
-          'media_id': "media-source://tts/test?message=Sorry,+I+couldn't+understand+that&language=en-US&tts_options=%7B%22voice%22:%22james_earl_jones%22%7D",
+          'media_id': 'media-source://tts/temporary/test_token.mp3',
           'mime_type': 'audio/mpeg',
           'url': '/api/tts_proxy/test_token.mp3',
         }),
@@ -180,7 +180,7 @@
     dict({
       'data': dict({
         'tts_output': dict({
-          'media_id': "media-source://tts/test?message=Sorry,+I+couldn't+understand+that&language=en-US&tts_options=%7B%22voice%22:%22Arnold+Schwarzenegger%22%7D",
+          'media_id': 'media-source://tts/temporary/test_token.mp3',
           'mime_type': 'audio/mpeg',
           'url': '/api/tts_proxy/test_token.mp3',
         }),
@@ -277,7 +277,7 @@
     dict({
       'data': dict({
         'tts_output': dict({
-          'media_id': "media-source://tts/test?message=Sorry,+I+couldn't+understand+that&language=en-US&tts_options=%7B%22voice%22:%22Arnold+Schwarzenegger%22%7D",
+          'media_id': 'media-source://tts/temporary/test_token.mp3',
           'mime_type': 'audio/mpeg',
           'url': '/api/tts_proxy/test_token.mp3',
         }),
@@ -398,7 +398,7 @@
     dict({
       'data': dict({
         'tts_output': dict({
-          'media_id': "media-source://tts/test?message=Sorry,+I+couldn't+understand+that&language=en-US&tts_options=%7B%22voice%22:%22james_earl_jones%22%7D",
+          'media_id': 'media-source://tts/temporary/test_token.mp3',
           'mime_type': 'audio/mpeg',
           'url': '/api/tts_proxy/test_token.mp3',
         }),

tests/components/assist_pipeline/snapshots/test_websocket.ambr

@@ -79,7 +79,7 @@
 # name: test_audio_pipeline.6
   dict({
     'tts_output': dict({
-      'media_id': "media-source://tts/test?message=Sorry,+I+couldn't+understand+that&language=en-US&tts_options=%7B%22voice%22:%22james_earl_jones%22%7D",
+      'media_id': 'media-source://tts/temporary/test_token.mp3',
       'mime_type': 'audio/mpeg',
       'url': '/api/tts_proxy/test_token.mp3',
     }),
@@ -168,7 +168,7 @@
 # name: test_audio_pipeline_debug.6
   dict({
     'tts_output': dict({
-      'media_id': "media-source://tts/test?message=Sorry,+I+couldn't+understand+that&language=en-US&tts_options=%7B%22voice%22:%22james_earl_jones%22%7D",
+      'media_id': 'media-source://tts/temporary/test_token.mp3',
       'mime_type': 'audio/mpeg',
       'url': '/api/tts_proxy/test_token.mp3',
     }),
@@ -269,7 +269,7 @@
 # name: test_audio_pipeline_with_enhancements.6
   dict({
     'tts_output': dict({
-      'media_id': "media-source://tts/test?message=Sorry,+I+couldn't+understand+that&language=en-US&tts_options=%7B%22voice%22:%22james_earl_jones%22%7D",
+      'media_id': 'media-source://tts/temporary/test_token.mp3',
       'mime_type': 'audio/mpeg',
       'url': '/api/tts_proxy/test_token.mp3',
     }),
@@ -380,7 +380,7 @@
 # name: test_audio_pipeline_with_wake_word_no_timeout.8
   dict({
     'tts_output': dict({
-      'media_id': "media-source://tts/test?message=Sorry,+I+couldn't+understand+that&language=en-US&tts_options=%7B%22voice%22:%22james_earl_jones%22%7D",
+      'media_id': 'media-source://tts/temporary/test_token.mp3',
       'mime_type': 'audio/mpeg',
       'url': '/api/tts_proxy/test_token.mp3',
     }),

tests/components/tts/test_media_source.py

@@ -2,11 +2,11 @@
 
 from http import HTTPStatus
 import re
-from unittest.mock import MagicMock
+from unittest.mock import MagicMock, patch
 
 import pytest
 
-from homeassistant.components import media_source
+from homeassistant.components import media_source, tts
 from homeassistant.components.media_player import BrowseError
 from homeassistant.components.tts.media_source import (
     MediaSourceOptions,
@@ -302,3 +302,33 @@ async def test_generate_media_source_id_and_media_source_id_to_kwargs(
         "options": {"age": {"k1": [5, 6], "k2": "v2"}},
         "use_file_cache": True,
     }
+
+
+async def test_stream_media_sources(hass: HomeAssistant, setup_media_source) -> None:
+    """Test ResultStream as media sources."""
+    assert await async_setup_component(hass, "tts", {})
+    stream = tts.ResultStream(
+        token="mock-token.flac",
+        extension="flac",
+        content_type="audio/flac",
+        engine="test",
+        use_file_cache=True,
+        language="en",
+        options={},
+        _manager=None,
+    )
+    hass.data[tts.DATA_TTS_MANAGER].token_to_stream[stream.token] = stream
+    assert stream.media_source_id == "media-source://tts/temporary/mock-token.flac"
+
+    assert await media_source.async_resolve_media(
+        hass, stream.media_source_id, None
+    ) == media_source.PlayMedia(url=stream.url, mime_type=stream.content_type)
+
+    async def async_stream_result():
+        yield b"test"
+
+    with patch.object(stream, "async_stream_result", async_stream_result):
+        assert await tts.async_get_media_source_audio(hass, stream.media_source_id) == (
+            stream.extension,
+            b"test",
+        )