Python: fix: prevent inner_exception from being lost in AgentFrameworkException

[Bahtya]

Problem

AgentFrameworkException.__init__() unconditionally calls super().__init__(message, *args) after the conditional super().__init__(message, inner_exception, *args). When inner_exception is provided, the second call overwrites the exception args, losing the inner exception reference.

Root Cause

Missing else branch — both super().__init__() calls execute when inner_exception is truthy.

Fix

Add else: branch so super().__init__() is called exactly once with the correct arguments.

Before:

if inner_exception:
    super().__init__(message, inner_exception, *args)
super().__init__(message, *args)  # always runs, overwrites args

After:

if inner_exception:
    super().__init__(message, inner_exception, *args)
else:
    super().__init__(message, *args)

Fixes #5155

[Bahtya]

Hi team, just wanted to follow up on this PR. It has been approved by @eavanvalkenburg. There appear to be some CI failures — happy to investigate if they're related to this change. Would appreciate any feedback!

[Bahtya]

CI Failure Analysis

The Python 3.11 test failures are not caused by this PR. They are pre-existing issues in the durabletask package:

TypeError: argument of type 'Mock' is not iterable
durabletask/entities/entity_instance_id.py:83

This error occurs in EntityInstanceId.__init__ when a Mock object is passed as entity_key — the if not entity_key check fails because Mock is not iterable on Python 3.11.

I have added explicit unit tests for the AgentFrameworkException inner_exception handling in python/packages/core/tests/core/test_exceptions.py to ensure both code branches are covered.

The coverage and merge-gatekeeper failures are downstream of the Python 3.11 test failure.

[Bahtya]

@microsoft-github-policy-service agree

[moonbox3]

@Bahtya please fix the CI/CD failures

[Bahtya]

@moonbox3 The CI workflows (Python - Tests, Code Quality, etc.) are stuck at action_required status — they require maintainer approval to trigger for fork PRs. The previous test failure from April 13 was a pre-existing issue in the durabletask package (TypeError: argument of type 'Mock' is not iterable on Python 3.11), not related to this PR. Could a maintainer approve the CI workflow runs so we can verify the latest commit?

[moonbox3]

@moonbox3 The CI workflows (Python - Tests, Code Quality, etc.) are stuck at action_required status — they require maintainer approval to trigger for fork PRs. The previous test failure from April 13 was a pre-existing issue in the durabletask package (TypeError: argument of type 'Mock' is not iterable on Python 3.11), not related to this PR. Could a maintainer approve the CI workflow runs so we can verify the latest commit?

Yes, CI/CD checks require approval. But all checks can also be run locally. If they pass locally, they will pass when pushed to origin/upstream.

[moonbox3]

Python Test Coverage Report •

File	Stmts	Miss	Cover	Missing
packages/core/agent_framework
exceptions.py	66	0	100%
TOTAL	29865	3479	88%

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
6026	30 💤	0 ❌	0 🔥	1m 39s ⏱️