fix: Look through all streaming chunks for tools calls

[vblagoje]

Why:

Enhances the processing logic to support various LLM platforms—not just OpenAI. Testing with the OpenAI Python package using the Mistral platform revealed that some platforms may place tool_calls in any streamed chunk (not only in the first). This change ensures that all chunks are scanned for the tool_calls payload.

What:

• Refined detection of tool calls across all streaming chunks.
• Changed the assumption that tool_calls appear only in the first chunk to a search that finds them anywhere.

How can it be used:

This update ensures robust tool call tracking in streaming messages, accurately constructing payloads regardless of where the tool_calls appear:

# Updated logic to search for tool_calls in all chunks
tool_call_index = next((i for i, chunk in enumerate(chunks) if chunk.meta.get("tool_calls")), None)
tools_len = len(chunks[tool_call_index].meta.get("tool_calls", []))

How did you test it:

• No tests were added, using old tests

Notes for the reviewer:

N/A

[coveralls]

Pull Request Test Coverage Report for Build 13202693684

Details

• 0 of 0 changed or added relevant lines in 0 files are covered.
• 2 unchanged lines in 1 file lost coverage.
• Overall coverage increased (+0.003%) to 92.712%

---

Files with Coverage Reduction	New Missed Lines	%
components/generators/chat/openai.py	2	96.4%

Totals
Change from base Build 13200853936:	0.003%
Covered Lines:	9057
Relevant Lines:	9769

---

💛 - Coveralls