fix(azure-ai-inference): guard against IndexError when tool message items list is empty

[frankgoldfish]

Summary

In _format_tool_message within the Azure AI Inference utilities, when a ChatMessageContent with role TOOL has an empty items list, the existing guard only checks len(message.items) != 1 and logs a warning — but then immediately proceeds to access message.items[0], raising an unhandled IndexError.

Bug

def _format_tool_message(message: ChatMessageContent) -> ToolMessage:
    if len(message.items) != 1:
        # This logs a warning for items==0, but does NOT stop execution
        logger.warning("Unsupported number of items ...")

    # IndexError: list index out of range when message.items is empty
    if not isinstance(message.items[0], FunctionResultContent):
        raise ValueError("No FunctionResultContent found in the message items")

When message.items is an empty list, message.items[0] raises IndexError: list index out of range, which is an opaque error that obscures the real problem.

Fix

Add an explicit guard for the empty-items case before the length-mismatch warning, raising a descriptive ValueError:

def _format_tool_message(message: ChatMessageContent) -> ToolMessage:
    if len(message.items) == 0:
        raise ValueError("Tool message has no items; expected exactly one FunctionResultContent item.")

    if len(message.items) != 1:
        logger.warning("Unsupported number of items ...")

    if not isinstance(message.items[0], FunctionResultContent):
        raise ValueError("No FunctionResultContent found in the message items")

Files Changed

• python/semantic_kernel/connectors/ai/azure_ai_inference/services/utils.py

Test plan

• Pass a ChatMessageContent(role=AuthorRole.TOOL, items=[]) to _format_tool_message and confirm it raises ValueError with a clear message instead of IndexError
• Run existing Azure AI Inference unit tests to verify no regression

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR prevents an unhandled IndexError when formatting Azure AI Inference tool messages whose ChatMessageContent.items list is empty, by adding an explicit early validation and raising a clearer ValueError.

Changes:

• Add an explicit empty list guard in _format_tool_message before indexing message.items[0].
• Raise a descriptive ValueError when a TOOL message has zero items.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Add a unit test covering the new empty-items guard (TOOL ChatMessageContent with items=[]), asserting a ValueError is raised and the message is descriptive. This prevents regressions back to an IndexError and locks in the intended error behavior.