From 27d6becfd0f2b3071fc650eee0ae8e15a7ae8115 Mon Sep 17 00:00:00 2001
From: Ashwin Bharambe <ashwin.bharambe@gmail.com>
Date: Mon, 18 Aug 2025 12:20:50 -0700
Subject: [PATCH] fix(misc): pin openai dependency to < 1.100.0 (#3192)

This OpenAI client release
https://github.com/openai/openai-python/commit/0843a1116498bc3312db9904adf71a4fb0a0a77e
ends up breaking litellm
https://github.com/BerriAI/litellm/blob/169a17400f1f5e36815c7d89128754975cd0584d/litellm/types/llms/openai.py#L40

Update the dependency pin. Also make the imports a bit more defensive
anyhow if something else during `llama stack build` ends up moving
openai to a previous version.

## Test Plan

Run pre-release script integration tests.
---
 .../utils/inference/openai_compat.py          | 12 +++++++++---
 pyproject.toml                                |  2 +-
 scripts/integration-tests.sh                  | 19 ++++++++++++++++---
 uv.lock                                       |  2 +-
 4 files changed, 27 insertions(+), 8 deletions(-)

diff --git a/llama_stack/providers/utils/inference/openai_compat.py b/llama_stack/providers/utils/inference/openai_compat.py
index 6297cc2ed..5e6c26884 100644
--- a/llama_stack/providers/utils/inference/openai_compat.py
+++ b/llama_stack/providers/utils/inference/openai_compat.py
@@ -31,9 +31,15 @@ from openai.types.chat import (
 from openai.types.chat import (
     ChatCompletionContentPartTextParam as OpenAIChatCompletionContentPartTextParam,
 )
-from openai.types.chat import (
-    ChatCompletionMessageFunctionToolCall as OpenAIChatCompletionMessageFunctionToolCall,
-)
+
+try:
+    from openai.types.chat import (
+        ChatCompletionMessageFunctionToolCall as OpenAIChatCompletionMessageFunctionToolCall,
+    )
+except ImportError:
+    from openai.types.chat.chat_completion_message_tool_call import (
+        ChatCompletionMessageToolCall as OpenAIChatCompletionMessageFunctionToolCall,
+    )
 from openai.types.chat import (
     ChatCompletionMessageParam as OpenAIChatCompletionMessage,
 )
diff --git a/pyproject.toml b/pyproject.toml
index db0ad1f00..f02c02c41 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -33,7 +33,7 @@ dependencies = [
     "jsonschema",
     "llama-stack-client>=0.2.17",
     "llama-api-client>=0.1.2",
-    "openai>=1.99.6",
+    "openai>=1.99.6,<1.100.0",
     "prompt-toolkit",
     "python-dotenv",
     "python-jose[cryptography]",
diff --git a/scripts/integration-tests.sh b/scripts/integration-tests.sh
index e9a5283e1..66e6d8e57 100755
--- a/scripts/integration-tests.sh
+++ b/scripts/integration-tests.sh
@@ -144,6 +144,19 @@ else
     export LLAMA_STACK_TEST_RECORDING_DIR="tests/integration/recordings"
 fi
 
+# check if "llama" and "pytest" are available. this script does not use `uv run` given
+# it can be used in a pre-release environment where we have not been able to tell
+# uv about pre-release dependencies properly (yet).
+if ! command -v llama &> /dev/null; then
+    echo "llama could not be found, ensure llama-stack is installed"
+    exit 1
+fi
+
+if ! command -v pytest &> /dev/null; then
+    echo "pytest could not be found, ensure pytest is installed"
+    exit 1
+fi
+
 # Start Llama Stack Server if needed
 if [[ "$STACK_CONFIG" == *"server:"* ]]; then
     # check if server is already running
@@ -151,7 +164,7 @@ if [[ "$STACK_CONFIG" == *"server:"* ]]; then
         echo "Llama Stack Server is already running, skipping start"
     else
         echo "=== Starting Llama Stack Server ==="
-        nohup uv run llama stack run ci-tests --image-type venv > server.log 2>&1 &
+        nohup llama stack run ci-tests --image-type venv > server.log 2>&1 &
 
         echo "Waiting for Llama Stack Server to start..."
         for i in {1..30}; do
@@ -189,7 +202,7 @@ fi
 if [[ "$RUN_VISION_TESTS" == "true" ]]; then
     echo "Running vision tests..."
     set +e
-    uv run pytest -s -v tests/integration/inference/test_vision_inference.py \
+    pytest -s -v tests/integration/inference/test_vision_inference.py \
         --stack-config="$STACK_CONFIG" \
         -k "$PYTEST_PATTERN" \
         --vision-model=ollama/llama3.2-vision:11b \
@@ -257,7 +270,7 @@ echo "=== Running all collected tests in a single pytest command ==="
 echo "Total test files: $(echo $TEST_FILES | wc -w)"
 
 set +e
-uv run pytest -s -v $TEST_FILES \
+pytest -s -v $TEST_FILES \
     --stack-config="$STACK_CONFIG" \
     -k "$PYTEST_PATTERN" \
     --text-model="$TEXT_MODEL" \
diff --git a/uv.lock b/uv.lock
index a09406770..3e3bf7e24 100644
--- a/uv.lock
+++ b/uv.lock
@@ -1856,7 +1856,7 @@ requires-dist = [
     { name = "llama-api-client", specifier = ">=0.1.2" },
     { name = "llama-stack-client", specifier = ">=0.2.17" },
     { name = "llama-stack-client", marker = "extra == 'ui'", specifier = ">=0.2.17" },
-    { name = "openai", specifier = ">=1.99.6" },
+    { name = "openai", specifier = ">=1.99.6,<1.100.0" },
     { name = "opentelemetry-exporter-otlp-proto-http", specifier = ">=1.30.0" },
     { name = "opentelemetry-sdk", specifier = ">=1.30.0" },
     { name = "pandas", marker = "extra == 'ui'" },