Python: Align OpenAI parser conversions

[Eduard van Valkenburg (eavanvalkenburg)]

Motivation & Context

The Python OpenAI clients still had a few conversion paths that conflated valid falsey values with missing values, emitted Agent Framework's internal content representation into Chat Completions requests, or assumed embedding responses arrived in input order. These mismatches could lose provider data, produce invalid request content, or associate an embedding with the wrong input.

Description & Review Guide

• What are the major changes? Preserve SDK-supported falsey Responses function outputs, omit unsupported Agent Framework content from Chat Completions payloads, retain explicitly reported zero-valued optional usage details, and order embedding response items by their indexes.
• What is the impact of these changes? OpenAI request and response conversion is more faithful while remaining compatible with the OpenAI 2.x wire types. Unsupported content is omitted instead of being sent as a schema-invalid dictionary.
• What do you want reviewers to focus on? Please focus on the narrow conversion behavior changes and their regression coverage, especially the unsupported-content omission semantics.

Related Issue

N/A — follow-up from the Responses conversion consistency sweep

Contribution Checklist

• The code builds clean without any errors or warnings
• All unit tests pass, and I have added new tests where possible
• The PR follows the Contribution Guidelines
• This PR is linked to an issue and there is no other open PR for this issue (see Related Issue above).
• This is not a breaking change. If it is a breaking change, add the breaking change label (or add "[BREAKING]" to the title prefix, before or after any language prefix) — a workflow keeps the label and title prefix in sync automatically.

[Copilot]

Pull request overview

Aligns Python OpenAI conversions with provider wire formats and response metadata.

Changes:

• Preserves supported falsey outputs and zero-valued usage details.
• Omits unsupported Chat Completions content.
• Orders embeddings by provider indexes and adds regression tests.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
_chat_client.py	Updates function-result output handling.
_chat_completion_client.py	Filters unsupported content and preserves zero usage values.
_embedding_client.py	Sorts embeddings by response index.
test_openai_chat_client.py	Tests falsey function outputs.
test_openai_chat_completion_client.py	Tests omission and usage conversions.
test_openai_embedding_client.py	Tests indexed embedding ordering.

---

💡 Add a code-review agent skill for context-aware, tailored reviews. Learn more in the docs.

[github-actions]

Python Test Coverage Report •

File	Stmts	Miss	Cover	Missing
packages/openai/agent_framework_openai
_chat_client.py	1433	117	91%	361, 374, 728–734, 743–746, 752–756, 764, 808–813, 817–820, 822–824, 831–833, 836, 914, 922, 945, 1162, 1221, 1223, 1225, 1227, 1293, 1307, 1387, 1397, 1402, 1445, 1557–1558, 1573, 1852, 1964, 1969–1970, 2020, 2024–2025, 2064, 2074, 2101, 2107, 2117, 2123, 2128, 2134, 2139–2140, 2220, 2264, 2267–2270, 2284, 2294–2295, 2307, 2414, 2431, 2434, 2461–2463, 2502, 2519, 2522, 2584, 2591, 2628–2629, 2670, 2708–2709, 2727–2728, 2900–2901, 2919, 2953, 3026–3029, 3213, 3228, 3317–3319, 3329–3330, 3336, 3351, 3484–3485
_chat_completion_client.py	454	16	96%	543, 637, 644–645, 649, 674, 849, 1007, 1075, 1092, 1114, 1122, 1148, 1161, 1194, 1560
_embedding_client.py	81	2	97%	240, 288
TOTAL	48418	4500	90%

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
9865	36 💤	0 ❌	0 🔥	2m 45s ⏱️

[github-actions]

MAF Automated Review — Iteration 1

Result: Findings reported
Scope: full PR (1 commit(s)): be809084dc76
Model: gpt-5.6-sol

Overview

The PR improves conversion fidelity by preserving SDK-supported falsey outputs, retaining explicit zero usage details, omitting unsupported Chat Completions content, and ordering embeddings by provider index, with focused regression coverage for each path. The remaining compatibility risk is that the broader falsey-value check also passes arbitrary deserialized values through to an OpenAI field that accepts only strings or rich-content lists.

Reviewed the supplied pull-request change set across correctness, security/reliability, architecture, and failure behavior.
1 verified finding remained after source verification (1 medium) across 1 file. Details are attached to the affected lines below.

Affected areas: python/packages/openai/agent_framework_openai/_chat_client.py

[github-actions]

MAF Automated Review — Iteration 2

Result: No findings
Scope: 1 net-new commit(s): dd765cd2e6c6
Model: gpt-5.6-sol

Overview

This incremental update addresses the prior function-output finding by preserving supported string and list values while serializing other non-None values with the same json.dumps(default=str) pattern used by the framework's canonical function-result constructor. The added tests cover the previously problematic falsey values False, 0, and {} as well as preserving "" and []. No Critical, High, or Medium issue remains in the supplied incremental diff.

Reviewed the supplied incremental change set across correctness, security/reliability, architecture, and failure behavior.
No publishable findings remained after source verification for this scope.