[Bug]: tool_calls index starts from 1

[langpingxue]

What happened?

When I use Amazon Bedrock as LLM provider, claude-3-7-sonnet as LLM model.

I got an error: TypeError: Cannot read properties of undefined (reading 'id') when using litellm OpenAI-Compatible Endpoints, the root cause is tool_calls start from 1, it lead to an incorrect array [undefined, {tool_call_object}], then raise the error then (array[0].id)
tool_calls index should always starts from 0 rather than 1

Relevant log output

{
    "id": "chatcmpl-8cff77bc-ccea-4b88-a61a-e89d750fe236",
    "created": 1749546275,
    "model": "us.anthropic.claude-3-7-sonnet-20250219-v1:0",
    "object": "chat.completion.chunk",
    "choices": [
        {
            "index": 0,
            "delta": {
                "content": "",
                "role": "assistant",
                "tool_calls": [
                    {
                        "id": "tooluse_9116K8c5Q8ivMmdx3VdS7g",
                        "function": {
                            "arguments": "",
                            "name": "jira_get_issue"
                        },
                        "type": "function",
                        "index": 1
                    }
                ]
            }
        }
    ],
    "provider_specific_fields": {}
}

Are you a ML Ops Team?

No

What LiteLLM version are you on ?

v1.71.1-stable

Twitter / LinkedIn details

No response

[langpingxue]

seems related to litellm/litellm/litellm_core_utils/streaming_chunk_builder_utils.py need to enhance the index logic

[langpingxue]

for openAI models no such issue, specifically for bedrock models