Python: Sanitize author_name for the Chat Completions message name field

[itjuba]

Motivation & Context

OpenAI validates the Chat Completions request-message name against ^[^\s<|\\/>]+$, and _prepare_message_for_openai forwards message.author_name into that field verbatim. Since the framework stamps the agent's display name onto author_name, any agent whose name contains a space (e.g. Agent(name="My Agent")) fails every request on the Chat Completions path with:

Error code: 400 - Invalid 'messages[1].name': string does not match pattern.
Expected a string that matches the pattern '^[^\s<|\\/>]+$'.

The .NET client hit the identical problem (#1195) and solved it with SanitizeAuthorName (dotnet/extensions#6827); the Python implementation of the name forwarding (#2951) omitted that step.

Description & Review Guide

• What are the major changes? Adds a _sanitize_author_name helper to _chat_completion_client.py, mirroring the .NET SanitizeAuthorName semantics: characters outside [a-zA-Z0-9_] are removed, the name is omitted entirely when nothing remains, and the result is truncated to 64 characters. Applied at all three name assignment sites (system/developer path, the general per-content path, and the reasoning-only fallback).
• What is the impact of these changes? Agents with human-readable display names work on the Chat Completions path instead of 400-ing. Names that are already valid pass through unchanged. The agent's display name itself is untouched — only the wire-level participant id is sanitized. The Responses client is unaffected (it does not send author_name).
• What do you want reviewers to focus on? The sanitization semantics deliberately mirror the .NET client for cross-SDK consistency (remove invalid characters rather than replace, drop empty results, 64-char cap) — flag if Python should diverge from that.

Tests: test_prepare_message_sanitizes_author_name (space, invalid-only, truncation, and system-path cases) and test_prepare_message_keeps_valid_author_name in tests/openai/test_openai_chat_completion_client.py. Full test_openai_chat_completion_client* files pass locally; ruff check and ruff format --check clean.

Related Issue

Fixes #7126

Contribution Checklist

• The code builds clean without any errors or warnings
• All unit tests pass, and I have added new tests where possible

[Copilot]

Pull request overview

This PR fixes a Python OpenAI Chat Completions compatibility bug where Message.author_name was forwarded verbatim into the request-message name field, causing 400s for common agent display names (e.g., names containing spaces). It introduces a small sanitization helper aligned with the .NET SDK behavior so agent display names remain human-friendly while the wire name becomes API-compliant.

Changes:

• Added _sanitize_author_name (strip to [a-zA-Z0-9_], omit if empty, truncate to 64 chars) in the Chat Completions client module.
• Applied sanitization at each of the name assignment sites in _prepare_message_for_openai.
• Added unit tests covering space removal, invalid-only omission, truncation, system-path handling, and pass-through for already-valid names.

Reviewed changes

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

File	Description
python/packages/openai/agent_framework_openai/_chat_completion_client.py	Introduces and applies author-name sanitization before setting Chat Completions message name.
python/packages/openai/tests/openai/test_openai_chat_completion_client.py	Adds regression and behavior tests for author-name sanitization and valid-name pass-through.

[itjuba]

@microsoft-github-policy-service agree

[eavanvalkenburg]

LGTM

[eavanvalkenburg]

@itjuba there is a merge conflict, please have a look!

[github-actions]

Python Test Coverage Report •

File	Stmts	Miss	Cover	Missing
packages/openai/agent_framework_openai
_chat_completion_client.py	396	17	95%	486, 582–583, 587, 833, 835, 840, 843, 946, 963, 984, 992, 1016, 1029, 1062, 1085, 1406
TOTAL	45590	4478	90%

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
9409	34 💤	0 ❌	0 🔥	2m 29s ⏱️

[Copilot]

Pull request overview

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

[itjuba]

@eavanvalkenburg conflict resolved in 7aa0433 — branch is now up to date with main (0 behind), no code changes. Workflows are pending approval to run. Could you kick them off and take another look? Thanks!