add initial realtime tests for aoai (Azure#38939)

* add initial realtime tests * retrigger CI
kingernupur · Jan 3, 2025 · eb751a5 · eb751a5
1 parent e4ee11e
commit eb751a5
Show file tree

Hide file tree

Showing 5 changed files with 232 additions and 2 deletions.
diff --git a/sdk/openai/azure-openai/dev_requirements.txt b/sdk/openai/azure-openai/dev_requirements.txt
@@ -1,6 +1,6 @@
 -e ../../../tools/azure-sdk-tools
 ../../identity/azure-identity
 aiohttp
-openai
+openai[realtime]
 pillow
 pydantic
diff --git a/sdk/openai/azure-openai/tests/conftest.py b/sdk/openai/azure-openai/tests/conftest.py
@@ -21,7 +21,7 @@
 
 # for pytest.parametrize
 GA = "2024-10-21"
-PREVIEW = "2024-11-01-preview"
+PREVIEW = "2024-10-01-preview"
 LATEST = PREVIEW
 
 AZURE = "azure"
@@ -40,6 +40,7 @@
 ENV_AZURE_OPENAI_KEY = "AZURE_OPENAI_KEY"
 ENV_AZURE_OPENAI_NORTHCENTRALUS_ENDPOINT = "AZURE_OPENAI_NORTHCENTRALUS_ENDPOINT"
 ENV_AZURE_OPENAI_SWEDENCENTRAL_ENDPOINT = "AZURE_OPENAI_SWEDENCENTRAL_ENDPOINT"
+ENV_AZURE_OPENAI_SWEDENCENTRAL_KEY = "AZURE_OPENAI_SWEDENCENTRAL_KEY"
 ENV_AZURE_OPENAI_SEARCH_ENDPOINT = "AZURE_OPENAI_SEARCH_ENDPOINT"
 ENV_AZURE_OPENAI_SEARCH_INDEX = "AZURE_OPENAI_SEARCH_INDEX"
 
@@ -178,6 +179,11 @@ def build_kwargs(args, api_type):
             return {"model": ENV_AZURE_OPENAI_CHAT_COMPLETIONS_GPT4_NAME}
         elif api_type == "gpt_4_openai":
             return {"model": ENV_OPENAI_CHAT_COMPLETIONS_GPT4_MODEL}
+    if test_feature.startswith("test_realtime"):
+        if api_type in ["gpt_4_azure"]:
+            return {"model": "gpt-4o-realtime-preview-1001"}
+        elif api_type == "gpt_4_openai":
+            return {"model": "gpt-4o-realtime-preview-2024-10-01"}
     if test_feature.startswith(("test_module_client", "test_cli")):
         return {}
     raise ValueError(f"Test feature: {test_feature} needs to have its kwargs configured.")

diff --git a/sdk/openai/azure-openai/tests/test_realtime.py b/sdk/openai/azure-openai/tests/test_realtime.py
@@ -0,0 +1,109 @@
+# ------------------------------------
+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.
+# ------------------------------------
+
+import os
+import pytest
+import openai
+from devtools_testutils import AzureRecordedTestCase, get_credential
+from conftest import (
+    ENV_AZURE_OPENAI_SWEDENCENTRAL_ENDPOINT,
+    ENV_AZURE_OPENAI_SWEDENCENTRAL_KEY,
+    GPT_4_AZURE,
+    GPT_4_OPENAI,
+    configure,
+    PREVIEW,
+)
+
+
+@pytest.mark.live_test_only
+class TestRealtime(AzureRecordedTestCase):
+
+    @configure
+    @pytest.mark.parametrize(
+        "api_type, api_version",
+        [(GPT_4_AZURE, PREVIEW), (GPT_4_OPENAI, "v1")],
+    )
+    def test_realtime_text(self, client, api_type, api_version, **kwargs):
+        with client.beta.realtime.connect(
+            **kwargs,
+        ) as connection:
+            connection.session.update(session={"modalities": ["text"]})
+            connection.conversation.item.create(
+                item={
+                    "type": "message",
+                    "role": "user",
+                    "content": [{"type": "input_text", "text": "Say hello!"}],
+                }
+            )
+            connection.response.create()
+            for event in connection:
+                if event.type == "response.text.delta":
+                    assert event.delta
+                elif event.type == "response.text.done":
+                    assert event.text
+                elif event.type == "response.done":
+                    break
+
+    @configure
+    @pytest.mark.parametrize(
+        "api_type, api_version",
+        [(GPT_4_AZURE, PREVIEW)],
+    )
+    def test_realtime_text_api_key(self, client, api_type, api_version, **kwargs):
+        client = openai.AzureOpenAI(
+            azure_endpoint=os.environ[ENV_AZURE_OPENAI_SWEDENCENTRAL_ENDPOINT],
+            api_key=os.environ[ENV_AZURE_OPENAI_SWEDENCENTRAL_KEY],
+            api_version=api_version,
+        )
+        with client.beta.realtime.connect(
+            **kwargs
+        ) as connection:
+            connection.session.update(session={"modalities": ["text"]})
+            connection.conversation.item.create(
+                item={
+                    "type": "message",
+                    "role": "user",
+                    "content": [{"type": "input_text", "text": "Say hello!"}],
+                }
+            )
+            connection.response.create()
+            for event in connection:
+                if event.type == "response.text.delta":
+                    assert event.delta
+                elif event.type == "response.text.done":
+                    assert event.text
+                elif event.type == "response.done":
+                    break
+
+    @configure
+    @pytest.mark.parametrize(
+        "api_type, api_version",
+        [(GPT_4_AZURE, PREVIEW)],
+    )
+    def test_realtime_text_ad_token(self, client, api_type, api_version, **kwargs):
+        client = openai.AzureOpenAI(
+            azure_endpoint=os.environ[ENV_AZURE_OPENAI_SWEDENCENTRAL_ENDPOINT],
+            azure_ad_token=get_credential().get_token("https://cognitiveservices.azure.com/.default").token,
+            api_version=api_version,
+        )
+        with client.beta.realtime.connect(
+            **kwargs
+        ) as connection:
+            connection.session.update(session={"modalities": ["text"]})
+            connection.conversation.item.create(
+                item={
+                    "type": "message",
+                    "role": "user",
+                    "content": [{"type": "input_text", "text": "Say hello!"}],
+                }
+            )
+            connection.response.create()
+            for event in connection:
+                if event.type == "response.text.delta":
+                    assert event.delta
+                elif event.type == "response.text.done":
+                    assert event.text
+                elif event.type == "response.done":
+                    break
diff --git a/sdk/openai/azure-openai/tests/test_realtime_async.py b/sdk/openai/azure-openai/tests/test_realtime_async.py
@@ -0,0 +1,114 @@
+# ------------------------------------
+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.
+# ------------------------------------
+
+import os
+import pytest
+import openai
+from devtools_testutils import AzureRecordedTestCase, get_credential
+from conftest import (
+    ENV_AZURE_OPENAI_SWEDENCENTRAL_ENDPOINT,
+    ENV_AZURE_OPENAI_SWEDENCENTRAL_KEY,
+    GPT_4_AZURE,
+    GPT_4_OPENAI,
+    configure_async,
+    PREVIEW,
+)
+
+
+@pytest.mark.live_test_only
+class TestRealtimeAsync(AzureRecordedTestCase):
+
+    @configure_async
+    @pytest.mark.asyncio
+    @pytest.mark.parametrize(
+        "api_type, api_version",
+        [(GPT_4_AZURE, PREVIEW), (GPT_4_OPENAI, "v1")],
+    )
+    async def test_realtime_text(self, client_async, api_type, api_version, **kwargs):
+        async with client_async.beta.realtime.connect(
+            **kwargs,
+        ) as connection:
+            await connection.session.update(session={"modalities": ["text"]})
+            await connection.conversation.item.create(
+                item={
+                    "type": "message",
+                    "role": "user",
+                    "content": [{"type": "input_text", "text": "Say hello!"}],
+                }
+            )
+            await connection.response.create()
+            async for event in connection:
+                if event.type == "response.text.delta":
+                    assert event.delta
+                elif event.type == "response.text.done":
+                    assert event.text
+                elif event.type == "response.done":
+                    break
+
+    @configure_async
+    @pytest.mark.asyncio
+    @pytest.mark.parametrize(
+        "api_type, api_version",
+        [(GPT_4_AZURE, PREVIEW)],
+    )
+    async def test_realtime_text_api_key(self, client_async, api_type, api_version, **kwargs):
+        client_async = openai.AsyncAzureOpenAI(
+            azure_endpoint=os.environ[ENV_AZURE_OPENAI_SWEDENCENTRAL_ENDPOINT],
+            api_key=os.environ[ENV_AZURE_OPENAI_SWEDENCENTRAL_KEY],
+            api_version=api_version,
+        )
+        async with client_async.beta.realtime.connect(
+            **kwargs
+        ) as connection:
+            await connection.session.update(session={"modalities": ["text"]})
+            await connection.conversation.item.create(
+                item={
+                    "type": "message",
+                    "role": "user",
+                    "content": [{"type": "input_text", "text": "Say hello!"}],
+                }
+            )
+            await connection.response.create()
+            async for event in connection:
+                if event.type == "response.text.delta":
+                    assert event.delta
+                elif event.type == "response.text.done":
+                    assert event.text
+                elif event.type == "response.done":
+                    break
+
+    @configure_async
+    @pytest.mark.asyncio
+    @pytest.mark.parametrize(
+        "api_type, api_version",
+        [(GPT_4_AZURE, PREVIEW)],
+    )
+    async def test_realtime_text_ad_token(self, client_async, api_type, api_version, **kwargs):
+        credential = get_credential(is_async=True)
+        access_token = await credential.get_token("https://cognitiveservices.azure.com/.default")
+        client_async = openai.AsyncAzureOpenAI(
+            azure_endpoint=os.environ[ENV_AZURE_OPENAI_SWEDENCENTRAL_ENDPOINT],
+            azure_ad_token=access_token.token,
+            api_version=api_version,
+        )
+        async with client_async.beta.realtime.connect(
+            **kwargs
+        ) as connection:
+            await connection.session.update(session={"modalities": ["text"]})
+            await connection.conversation.item.create(
+                item={
+                    "type": "message",
+                    "role": "user",
+                    "content": [{"type": "input_text", "text": "Say hello!"}],
+                }
+            )
+            await connection.response.create()
+            async for event in connection:
+                if event.type == "response.text.delta":
+                    assert event.delta
+                elif event.type == "response.text.done":
+                    assert event.text
+                elif event.type == "response.done":
+                    break
diff --git a/sdk/openai/tests.yml b/sdk/openai/tests.yml
@@ -17,6 +17,7 @@ extends:
         AZURE_OPENAI_COMPLETIONS_DEPLOYMENT: $(AOAI-COMPLETIONS-MODEL-DEPLOYMENT)
         AZURE_OPENAI_NORTHCENTRALUS_ENDPOINT: $(AOAI-ENDPOINT-NORTHCENTRALUS)
         AZURE_OPENAI_SWEDENCENTRAL_ENDPOINT: $(AOAI-ENDPOINT-SWEDENCENTRAL)
+        AZURE_OPENAI_SWEDENCENTRAL_KEY: $(AOAI-API-KEY-SWEDENCENTRAL)
         OPENAI_KEY: $(python-nonazure-openai-key)
         AZURE_OPENAI_SEARCH_ENDPOINT: $(COGNITIVE-SEARCH-API-ENDPOINT)
         AZURE_OPENAI_SEARCH_INDEX: $(COGNITIVE-SEARCH-API-INDEX)