fix : apricot test_generated comment bash error

[haddybhaiya]

This pull request improves the formatting and reliability of generated test code by introducing a normalization function to clean up LLM outputs, updating prompt instructions, and adding dedicated tests for the new normalization logic.

Test Output Normalization:

• Added a private function _normalize_generated_tests in test_generator.py to strip markdown code fences, language specifiers, and shell command lines (like pytest ...) from LLM-generated test code. This ensures only valid Python test code is returned.
• Updated the generate_tests function to use _normalize_generated_tests instead of manual string manipulation.

Prompt and Output Formatting:

• Enhanced the prompt in generate_tests to explicitly instruct the LLM to avoid markdown code fences and shell commands in its output.
• In main.py, changed the final output format to wrap suggested tests in a bash code block, improving readability.

Testing:

• Added new tests in test_test_generator_formatting.py to verify that _normalize_generated_tests correctly removes code fences and bash headers, and preserves valid Python code.

[github-actions]

Automated Code Review Comment:

Bugs

• None

Improvements

• The code quality has improved with the addition of the _normalize_generated_tests function, which removes markdown code fences and shell commands from the generated test content.
• The generate_tests function now uses the _normalize_generated_tests function to normalize the generated test content.

Suggestions

• Consider adding more test cases to cover different scenarios for the _normalize_generated_tests function.
• The generate_tests function could be improved by adding more error handling and logging to handle exceptions.
• It's a good practice to follow PEP 8 for Python code formatting, but the code looks well-formatted.
• The code looks good and follows best practices, but it would be beneficial to add more documentation to explain the purpose of each function and how they are used.

Code looks good ✅

---

Suggested Tests

def test_empty_diff():
    assert True

[haddybhaiya]

added bash for tests to be written in generated_test.py , working IMO

[haddybhaiya]

waiting for approvals , please check carefully!

[Copilot]

Pull request overview

This PR aims to make LLM-generated tests cleaner before they are written/commented by the agent, mainly by normalizing fenced output and adding focused tests for that normalization path.

Changes:

• Added _normalize_generated_tests and routed generate_tests() through it to strip code fences and leading pytest command lines.
• Added targeted tests for normalization behavior in a new test module.
• Changed the PR comment formatting for generated tests and replaced the checked-in tests/test_generated.py contents.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
tests/test_test_generator_formatting.py	Adds unit tests for normalization of generated test output.
tests/test_generated.py	Replaces prior contents with a trivial placeholder test.
agent/main.py	Changes how suggested tests are rendered in the final PR comment.
agent/llm/test_generator.py	Introduces output normalization and updates the LLM prompt rules.

---

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

[pleasingsunlight]

This PR aims at improving formatting for tests via introducing new functions to normalize tests and confirm the same via test_test_generated_formatting.py .

Upon proper review of the changes made, the intention is fulfilled as seen from the comment by 'github-actions-bot' ;

Tho there is still room for improvement here, especially when it comes to test coverage, preventing over-stripping of valid lines.

It is safe to merge the fixes to main.

• A few more tries would serve to better assess the improved formatting of the agent.

Hence for now :
We ball ! 🏐 😄 ~~~

[deoxyforge]

Everthing seems fine to me go ahead and merge this pr

[deoxyforge]

Go ahead everything looks good