add mock testing for vertex tts

litellm/llms/text_to_speech/vertex_ai.py

@@ -57,7 +57,7 @@ class VertexTextToSpeechAPI(VertexLLM):
         voice: Optional[dict] = None,
         _is_async: Optional[bool] = False,
         optional_params: Optional[dict] = None,
-        **kwargs,
+        kwargs: Optional[dict] = None,
     ):
         import base64
 
@@ -87,10 +87,11 @@ class VertexTextToSpeechAPI(VertexLLM):
         vertex_input = VertexInput(text=input)
         # required param
         optional_params = optional_params or {}
+        kwargs = kwargs or {}
         if voice is not None:
             vertex_voice = VertexVoice(**voice)
-        elif "voice" in optional_params:
-            vertex_voice = VertexVoice(**optional_params["voice"])
+        elif "voice" in kwargs:
+            vertex_voice = VertexVoice(**kwargs["voice"])
         else:
             # use defaults to not fail the request
             vertex_voice = VertexVoice(
@@ -98,8 +99,8 @@ class VertexTextToSpeechAPI(VertexLLM):
                 name="en-US-Studio-O",
             )
 
-        if "audioConfig" in optional_params:
-            vertex_audio_config = VertexAudioConfig(**optional_params["audioConfig"])
+        if "audioConfig" in kwargs:
+            vertex_audio_config = VertexAudioConfig(**kwargs["audioConfig"])
         else:
             # use defaults to not fail the request
             vertex_audio_config = VertexAudioConfig(

litellm/main.py

@@ -4867,6 +4867,7 @@ def speech(
             input=input,
             voice=voice,
             optional_params=optional_params,
+            kwargs=kwargs,
             logging_obj=logging_obj,
         )
 

litellm/tests/test_audio_speech.py

@@ -18,6 +18,7 @@ sys.path.insert(
     0, os.path.abspath("../..")
 )  # Adds the parent directory to the system path
 from pathlib import Path
+from unittest.mock import AsyncMock, MagicMock, patch
 
 import openai
 import pytest
@@ -123,6 +124,7 @@ async def test_audio_speech_router(mode):
     "sync_mode",
     [False, True],
 )
+@pytest.mark.skip(reason="local only test - we run testing using MockRequests below")
 @pytest.mark.asyncio
 async def test_audio_speech_litellm_vertex(sync_mode):
     litellm.set_verbose = True
@@ -147,3 +149,97 @@ async def test_audio_speech_litellm_vertex(sync_mode):
         from litellm.llms.openai import HttpxBinaryResponseContent
 
         response.stream_to_file(speech_file_path)
+
+
+@pytest.mark.asyncio
+async def test_speech_litellm_vertex_async():
+    # Mock the response
+    mock_response = AsyncMock()
+
+    def return_val():
+        return {
+            "audioContent": "dGVzdCByZXNwb25zZQ==",
+        }
+
+    mock_response.json = return_val
+    mock_response.status_code = 200
+
+    # Set up the mock for asynchronous calls
+    with patch(
+        "litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.post",
+        new_callable=AsyncMock,
+    ) as mock_async_post:
+        mock_async_post.return_value = mock_response
+        model = "vertex_ai/test"
+
+        response = await litellm.aspeech(
+            model=model,
+            input="async hello what llm guardrail do you have",
+        )
+
+        # Assert asynchronous call
+        mock_async_post.assert_called_once()
+        _, kwargs = mock_async_post.call_args
+        print("call args", kwargs)
+
+        assert kwargs["url"] == "https://texttospeech.googleapis.com/v1/text:synthesize"
+
+        assert "x-goog-user-project" in kwargs["headers"]
+        assert kwargs["headers"]["Authorization"] is not None
+
+        assert kwargs["json"] == {
+            "input": {"text": "async hello what llm guardrail do you have"},
+            "voice": {"languageCode": "en-US", "name": "en-US-Studio-O"},
+            "audioConfig": {"audioEncoding": "LINEAR16", "speakingRate": "1"},
+        }
+
+
+@pytest.mark.asyncio
+async def test_speech_litellm_vertex_async_with_voice():
+    # Mock the response
+    mock_response = AsyncMock()
+
+    def return_val():
+        return {
+            "audioContent": "dGVzdCByZXNwb25zZQ==",
+        }
+
+    mock_response.json = return_val
+    mock_response.status_code = 200
+
+    # Set up the mock for asynchronous calls
+    with patch(
+        "litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.post",
+        new_callable=AsyncMock,
+    ) as mock_async_post:
+        mock_async_post.return_value = mock_response
+        model = "vertex_ai/test"
+
+        response = await litellm.aspeech(
+            model=model,
+            input="async hello what llm guardrail do you have",
+            voice={
+                "languageCode": "en-UK",
+                "name": "en-UK-Studio-O",
+            },
+            audioConfig={
+                "audioEncoding": "LINEAR22",
+                "speakingRate": "10",
+            },
+        )
+
+        # Assert asynchronous call
+        mock_async_post.assert_called_once()
+        _, kwargs = mock_async_post.call_args
+        print("call args", kwargs)
+
+        assert kwargs["url"] == "https://texttospeech.googleapis.com/v1/text:synthesize"
+
+        assert "x-goog-user-project" in kwargs["headers"]
+        assert kwargs["headers"]["Authorization"] is not None
+
+        assert kwargs["json"] == {
+            "input": {"text": "async hello what llm guardrail do you have"},
+            "voice": {"languageCode": "en-UK", "name": "en-UK-Studio-O"},
+            "audioConfig": {"audioEncoding": "LINEAR22", "speakingRate": "10"},
+        }