feat: Switch Python generateContent to call Unary API endpoint

matthew29tang · copybara-github · commit 9a19545e864c · 2024-02-05T10:57:21.000-08:00
PiperOrigin-RevId: 604369375
diff --git a/tests/unit/vertexai/test_generative_models.py b/tests/unit/vertexai/test_generative_models.py
@@ -163,6 +163,56 @@ def mock_stream_generate_content(
     yield response
 
 
+def mock_generate_content(
+    self,
+    request: gapic_prediction_service_types.GenerateContentRequest,
+    *,
+    model: Optional[str] = None,
+    contents: Optional[MutableSequence[gapic_content_types.Content]] = None,
+) -> Iterable[gapic_prediction_service_types.GenerateContentResponse]:
+    is_continued_chat = len(request.contents) > 1
+    has_tools = bool(request.tools)
+
+    if has_tools:
+        has_function_response = any(
+            "function_response" in content.parts[0] for content in request.contents
+        )
+        needs_function_call = not has_function_response
+        if needs_function_call:
+            response_part_struct = _RESPONSE_FUNCTION_CALL_PART_STRUCT
+        else:
+            response_part_struct = _RESPONSE_AFTER_FUNCTION_CALL_PART_STRUCT
+    elif is_continued_chat:
+        response_part_struct = {"text": "Other planets may have different sky color."}
+    else:
+        response_part_struct = _RESPONSE_TEXT_PART_STRUCT
+
+    return gapic_prediction_service_types.GenerateContentResponse(
+        candidates=[
+            gapic_content_types.Candidate(
+                index=0,
+                content=gapic_content_types.Content(
+                    # Model currently does not identify itself
+                    # role="model",
+                    parts=[
+                        gapic_content_types.Part(response_part_struct),
+                    ],
+                ),
+                finish_reason=gapic_content_types.Candidate.FinishReason.STOP,
+                safety_ratings=[
+                    gapic_content_types.SafetyRating(rating)
+                    for rating in _RESPONSE_SAFETY_RATINGS_STRUCT
+                ],
+                citation_metadata=gapic_content_types.CitationMetadata(
+                    citations=[
+                        gapic_content_types.Citation(_RESPONSE_CITATION_STRUCT),
+                    ]
+                ),
+            ),
+        ],
+    )
+
+
 @pytest.mark.usefixtures("google_auth_mock")
 class TestGenerativeModels:
     """Unit tests for the generative models."""
@@ -178,8 +228,8 @@ def teardown_method(self):
 
     @mock.patch.object(
         target=prediction_service.PredictionServiceClient,
-        attribute="stream_generate_content",
-        new=mock_stream_generate_content,
+        attribute="generate_content",
+        new=mock_generate_content,
     )
     def test_generate_content(self):
         model = generative_models.GenerativeModel("gemini-pro")
@@ -212,8 +262,8 @@ def test_generate_content_streaming(self):
 
     @mock.patch.object(
         target=prediction_service.PredictionServiceClient,
-        attribute="stream_generate_content",
-        new=mock_stream_generate_content,
+        attribute="generate_content",
+        new=mock_generate_content,
     )
     def test_chat_send_message(self):
         model = generative_models.GenerativeModel("gemini-pro")
@@ -225,8 +275,8 @@ def test_chat_send_message(self):
 
     @mock.patch.object(
         target=prediction_service.PredictionServiceClient,
-        attribute="stream_generate_content",
-        new=mock_stream_generate_content,
+        attribute="generate_content",
+        new=mock_generate_content,
     )
     def test_chat_function_calling(self):
         get_current_weather_func = generative_models.FunctionDeclaration(
diff --git a/vertexai/generative_models/_generative_models.py b/vertexai/generative_models/_generative_models.py
@@ -431,15 +431,7 @@ def _generate_content(
             safety_settings=safety_settings,
             tools=tools,
         )
-        # generate_content is not available
-        # gapic_response = self._prediction_client.generate_content(request=request)
-        gapic_response = None
-        stream = self._prediction_client.stream_generate_content(request=request)
-        for gapic_chunk in stream:
-            if gapic_response:
-                _append_gapic_response(gapic_response, gapic_chunk)
-            else:
-                gapic_response = gapic_chunk
+        gapic_response = self._prediction_client.generate_content(request=request)
         return self._parse_response(gapic_response)
 
     async def _generate_content_async(
@@ -473,17 +465,9 @@ async def _generate_content_async(
             safety_settings=safety_settings,
             tools=tools,
         )
-        # generate_content is not available
-        # gapic_response = await self._prediction_async_client.generate_content(request=request)
-        gapic_response = None
-        stream = await self._prediction_async_client.stream_generate_content(
+        gapic_response = await self._prediction_async_client.generate_content(
             request=request
         )
-        async for gapic_chunk in stream:
-            if gapic_response:
-                _append_gapic_response(gapic_response, gapic_chunk)
-            else:
-                gapic_response = gapic_chunk
         return self._parse_response(gapic_response)
 
     def _generate_content_streaming(
