Speed up openai tests (#9943)

Saves ~8-10 seconds from total unit tests times

---------

Co-authored-by: Bagatur <baskaryan@gmail.com>

libs/langchain/tests/unit_tests/llms/test_openai.py

@@ -4,7 +4,9 @@ from typing import Any
 from unittest.mock import MagicMock, patch
 
 import pytest
+from tenacity import wait_none
 
+from langchain.llms import base
 from langchain.llms.openai import OpenAI
 from tests.unit_tests.callbacks.fake_callback_handler import (
     FakeAsyncCallbackHandler,
@@ -55,6 +57,16 @@ def mock_completion() -> dict:
     }
 
 
+def _patched_retry(*args: Any, **kwargs: Any) -> Any:
+    """Patched retry for unit tests that does not wait."""
+    from tenacity import retry
+
+    assert "wait" in kwargs
+    kwargs["wait"] = wait_none()
+    r = retry(*args, **kwargs)
+    return r
+
+
 @pytest.mark.requires("openai")
 def test_openai_retries(mock_completion: dict) -> None:
     llm = OpenAI()
@@ -73,13 +85,16 @@ def test_openai_retries(mock_completion: dict) -> None:
 
     mock_client.create = raise_once
     callback_handler = FakeCallbackHandler()
-    with patch.object(
-        llm,
-        "client",
-        mock_client,
-    ):
-        res = llm.predict("bar", callbacks=[callback_handler])
-        assert res == "Bar Baz"
+
+    # Patch the retry to avoid waiting during a unit test
+    with patch.object(base, "retry", _patched_retry):
+        with patch.object(
+            llm,
+            "client",
+            mock_client,
+        ):
+            res = llm.predict("bar", callbacks=[callback_handler])
+            assert res == "Bar Baz"
     assert completed
     assert raised
     assert callback_handler.retries == 1
@@ -105,13 +120,15 @@ async def test_openai_async_retries(mock_completion: dict) -> None:
 
     mock_client.acreate = araise_once
     callback_handler = FakeAsyncCallbackHandler()
-    with patch.object(
-        llm,
-        "client",
-        mock_client,
-    ):
-        res = await llm.apredict("bar", callbacks=[callback_handler])
-        assert res == "Bar Baz"
+    # Patch the retry to avoid waiting during a unit test
+    with patch.object(base, "retry", _patched_retry):
+        with patch.object(
+            llm,
+            "client",
+            mock_client,
+        ):
+            res = await llm.apredict("bar", callbacks=[callback_handler])
+            assert res == "Bar Baz"
     assert completed
     assert raised
     assert callback_handler.retries == 1