Strengthen test validation for generated field structure in orchestrate_content tests

[Copilot]

Summary

• What: Strengthen test_orchestrate_persist to validate the structure and contents of the generated field
• Why: Original test only checked key presence, not the actual data structure created by process_module()

Changes

• Add assertions validating generated is a dictionary with expected schema:
  • humanized_text: non-empty string
  • ttft_s: non-negative number
• Previously only verified "generated" in data[0], which could pass with incorrect data

Testing

• All 9 tests in test_orchestrate_content.py pass
• Code review found no issues

Checklist

• Test strengthened with structural validation
• All tests pass
• Code review complete

Reviewers

• @73junito

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[73junito]

reviewed

[Copilot]

Pull request overview

This PR strengthens test validation in test_orchestrate_persist by adding assertions that validate the structure and contents of the generated field created by process_module(). Previously, the test only verified that the generated key existed without checking its actual structure or data.

Changes:

• Add validation for generated field structure as a dictionary
• Add validation for humanized_text field (type, presence, and non-empty content)
• Add validation for ttft_s field (presence, numeric type, and non-negative value)

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The assertions for ttft_s assume it will always be present in the generated field, but this doesn't match the actual implementation behavior.

Looking at process_module() in scripts/orchestrate_content.py (lines 129-130), ttft_s is only set conditionally: if "ttft" in res: module["generated"]["ttft_s"] = float(res["ttft"]).

The real call_ollama() only returns ttft when stream=True (line 77), but NOT in non-streaming mode (lines 84-88 only return text). This test uses stream=False (line 39), so with the real implementation, ttft would not be returned and ttft_s would not be set.

While the test's fake_call returns ttft even in non-streaming mode, this doesn't reflect realistic behavior. Consider either:

1. Changing the test to use stream=True to match a realistic scenario, or
2. Making the ttft_s validation conditional by checking if it exists before asserting its type and value

Suggested change

	# Verify TTFT is present and valid when returned from call_ollama
	assert "ttft_s" in generated, "generated should contain ttft_s"
	assert isinstance(generated["ttft_s"], (int, float)), "ttft_s should be a number"
	assert generated["ttft_s"] >= 0, "ttft_s should be non-negative"
	# Verify TTFT is valid when returned from call_ollama (optional field)
	if "ttft_s" in generated:
	assert isinstance(generated["ttft_s"], (int, float)), "ttft_s should be a number"
	assert generated["ttft_s"] >= 0, "ttft_s should be non-negative"