Fix OpenAI Token Extraction Issues

[fenilfaldu]

📥 Pull Request

Issue:Token usage wasn't being extracted from ChatCompletion, Responses API response.

Response object wasn't being properly converted to dictionary due to incorrect condition checking and added proper handling for Pydantic models and improved fallback logic.

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 26 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
agentops/instrumentation/openai/wrappers/chat.py	0.00%	22 Missing ⚠️
...tops/instrumentation/openai/wrappers/embeddings.py	0.00%	4 Missing ⚠️

📢 Thoughts on this report? Let us know!

[Copilot]

Pull Request Overview

This pull request fixes the OpenAI token extraction issues by adjusting the response handling logic. Key changes include adding support for Pydantic models via model_dump in both embeddings and chat wrappers, and adding explicit checks to ensure tool_calls and function_call values are handled only when present.

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
agentops/instrumentation/openai/wrappers/embeddings.py	Adds fallback conditions to handle Pydantic models and dict responses
agentops/instrumentation/openai/wrappers/chat.py	Introduces explicit checks for tool_calls and function_call, ensuring proper extraction and logging

Comments suppressed due to low confidence (3)

agentops/instrumentation/openai/wrappers/embeddings.py:65

• The new branch for handling Pydantic models using model_dump is correctly placed before the dict check, which is good. Consider adding unit tests to confirm that the precedence between model_dump and dict conversions behaves as intended.

elif hasattr(return_value, "model_dump"):

agentops/instrumentation/openai/wrappers/chat.py:86

• The explicit check for tool_calls helps avoid errors when the value is None; however, please verify that an empty list is handled appropriately as it will evaluate to false.

if tool_calls:  # Check if tool_calls is not None

agentops/instrumentation/openai/wrappers/chat.py:186

• The guard for function_call adds a useful safeguard; ensure that an empty dictionary is handled as intended in downstream processing.

if function_call:  # Check if function_call is not None