Python: Improve Json parsing for declarative workflow

[Peter Ibekwe (peibekwe)]

Motivation & Context

Improve Json parsing in declarative workflow

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

Warning

Copilot couldn't run its full agentic review because it didn't start before the timeout. Make sure your repository has a runner available, or add a copilot-code-review.yml file specifying one with the runs-on attribute. See the docs for more details.

Improves JSON extraction/parsing from agent responses in the declarative workflow executor, with more robust handling of fenced code blocks and malformed/streamed JSON segments.

Changes:

• Replaced regex-based fenced code parsing with a deterministic fenced-block iterator and “qualified fence first” precedence.
• Added bounded JSON “recovery” scanning to safely extract the last valid JSON object/array from mixed text.
• Expanded unit tests to cover multiple fences, invalid/partial candidates, scalars, CRLF, and escape/bracket edge cases.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
python/packages/declarative/agent_framework_declarative/_workflows/_executors_agents.py	Implements new fenced-block iteration and bounded recovery scanning for JSON extraction.
python/packages/declarative/tests/test_graph_executors.py	Adds extensive unit tests covering new JSON extraction behaviors and edge cases.

[github-actions]

Python Test Coverage Report •

File	Stmts	Miss	Cover	Missing
packages/declarative/agent_framework_declarative/_workflows
_executors_agents.py	523	78	85%	192, 215, 307, 310–311, 319–320, 330, 341–346, 612, 650–652, 655, 671, 687–691, 715, 720, 810, 816–821, 828, 852–853, 862–864, 913–914, 922, 941, 1048–1051, 1058–1059, 1068–1069, 1081–1086, 1090–1091, 1093, 1098, 1100, 1104–1107, 1110, 1112–1113, 1116, 1124–1126, 1128, 1131–1132, 1134
TOTAL	45790	4241	90%

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
9329	36 💤	0 ❌	0 🔥	2m 27s ⏱️

[github-actions]

Agent Framework Review — Iteration 1

Completed passes: 4 | Result: Partial

Scope: full PR (1 commit(s)): d5e999abf77b

Review passes

• Correctness (gpt-5.6-sol) — The three-pass recovery cap can miss valid JSON after three malformed openings, regressing prior recovery behavior.
• Security Reliability (claude-opus-4.8) — No issues found in this pass.
• Test Coverage (gpt-5.6-sol) — No issues found in this pass.
• Design Approach (claude-opus-4.8) — No issues found in this pass.

[Evan Mattson (moonbox3)]

/review

[github-actions]

Agent Framework Review — Iteration 1

Completed passes: 4 | Result: Partial

Scope: full PR (1 commit(s)): d5e999abf77b

Review passes

• Security Reliability (claude-opus-4.8) — No issues found in this pass.
• Test Coverage (gpt-5.6-sol) — No issues found in this pass.
• Failure Modes (claude-opus-4.8) — No issues found in this pass.
• Design Approach (claude-opus-4.8) — No issues found in this pass.