fix: prevent Maven running all tests + fix TestFile type annotation

[mashraf-222]

Maven Test Execution Fixes

Problems Fixed

This PR addresses two related issues that were causing Java E2E optimizations to fail:

Issue 1: Maven runs ALL tests instead of specific tests

• When test filter was empty, Maven would run the entire test suite
• Caused unnecessary test execution and parsing errors
• No validation existed to catch this silent failure

Issue 2: Incorrect type annotation in TestFile model

• benchmarking_file_path: Path = None should be Optional[Path] = None
• Caused Pydantic validation errors when path was None
• Prevented proper handling of optional paths

Root Causes

Issue 1: Empty test filter

• _build_test_filter() returns empty string when all tests have None paths or fail conversion
• Empty filter → no -Dtest= parameter added to Maven command
• Maven default behavior: run ALL tests when no filter specified
• No validation existed to catch this before running Maven

Issue 2: Invalid type annotation

• Type annotation Path = None is invalid in Pydantic
• Should be Optional[Path] = None to allow None values
• Was causing validation errors in test code

Solutions Implemented

Issue 1 Fix:

1. Added validation in _run_maven_tests() to raise ValueError if filter is empty
2. Added detailed error logging in _build_test_filter() to track skipped tests
3. Added warnings when TestFile objects have None paths
4. Prevents silent failure where Maven runs wrong tests

Issue 2 Fix:

• Changed benchmarking_file_path: Path = None to Optional[Path] = None
• Now properly allows None values while maintaining type safety

Code Changes

codeflash/languages/java/test_runner.py (Issue 1):

• Added empty filter validation (raises ValueError)
• Added comprehensive debug logging
• Tracks reasons why tests are skipped
• Provides clear error messages

codeflash/models/models.py (Issue 2):

• Fixed type annotation for benchmarking_file_path

codeflash/discovery/discover_unit_tests.py:

• Included fix to set tests_project_rootdir for Java

tests/test_java_test_filter_validation.py:

• 4 comprehensive test cases covering both issues
• Tests None path handling
• Tests empty filter detection
• Tests normal operation

Testing

pytest tests/test_java_test_filter_validation.py -v
# ✓ 4 passed in 0.09s

# Test cases:
# - test_build_test_filter_with_none_benchmarking_paths ✓
# - test_build_test_filter_with_valid_paths ✓
# - test_run_maven_tests_raises_on_empty_filter ✓
# - test_run_maven_tests_succeeds_with_valid_filter ✓

Impact

• Issue 1: Prevents Maven from running all tests unintentionally
• Issue 2: Enables proper handling of optional paths in TestFile
• Better error messages for debugging test issues
• Catches bugs early rather than silently failing

Notes

• Debug logging helps diagnose future test issues
• Validation ensures we catch problems before running Maven
• All tests passing with the fixes applied

---

🤖 Generated with Claude Sonnet 4.5

[codeflash-ai]

⚡️ Codeflash found optimizations for this PR

📄 34% (0.34x) speedup for _build_test_filter in codeflash/languages/java/test_runner.py

⏱️ Runtime : 53.4 milliseconds → 39.9 milliseconds (best of 43 runs)

A dependent PR with the suggested changes has been created. Please review:

• ⚡️ Speed up function _build_test_filter by 34% in PR #1345 (debug/java-test-filter) #1347

If you approve, it will be merged into this PR (branch debug/java-test-filter).

[mashraf-222]

E2E Test Results

Goal: Prevent Maven from running all tests when filter is empty + Fix TestFile type annotation.

Result: ✅ PASS

What was tested

Verified both issues are fixed through comprehensive unit testing:

1. Empty test filter validation (prevents running all tests)
2. Type annotation fix for optional paths

Test commands

# Checkout PR branch
cd /home/ubuntu/code/codeflash
git checkout debug/java-test-filter

# Run unit tests
pytest tests/test_java_test_filter_validation.py -v

Evidence

All 4 unit tests pass:

test_build_test_filter_with_none_benchmarking_paths PASSED [ 25%]
test_build_test_filter_with_valid_paths PASSED [ 50%]
test_run_maven_tests_raises_on_empty_filter PASSED [ 75%]
test_run_maven_tests_succeeds_with_valid_filter PASSED [100%]

====== 4 passed in 0.04s ======

Issue 1 - Empty filter validation:

Code change verified in codeflash/languages/java/test_runner.py:

if test_filter:
    validated_filter = _validate_test_filter(test_filter)
    cmd.append(f"-Dtest={validated_filter}")
    logger.debug(f"Added -Dtest={validated_filter} to Maven command")
else:
    # CRITICAL: Empty test filter means Maven will run ALL tests
    error_msg = (
        f"Test filter is EMPTY for mode={mode}! "
        f"Maven will run ALL tests instead of the specified tests. "
    )
    logger.error(error_msg)
    raise ValueError(error_msg)  # NEW: Prevents silent failure

Test verification:

• test_run_maven_tests_raises_on_empty_filter confirms ValueError is raised when filter is empty
• test_run_maven_tests_succeeds_with_valid_filter confirms normal operation with valid filter

Issue 2 - Type annotation fix:

Code change verified in codeflash/models/models.py:

class TestFile(BaseModel):
    instrumented_behavior_file_path: Path
-   benchmarking_file_path: Path = None  # INVALID in Pydantic
+   benchmarking_file_path: Optional[Path] = None  # FIXED
    original_file_path: Optional[Path] = None

Test verification:

• test_build_test_filter_with_none_benchmarking_paths confirms None paths are handled correctly
• No Pydantic validation errors when TestFile objects have None benchmarking paths

Additional improvements

The PR also adds comprehensive debug logging to track:

• Why tests are skipped from the filter
• Which paths fail conversion to class names
• Warnings when TestFile objects have None paths

This will help diagnose future test execution issues.

Impact

1. Critical bug fix: Prevents Maven from accidentally running entire test suite when filter should contain specific tests
2. Type safety: Fixes Pydantic validation error for optional paths
3. Better debugging: Clear error messages and logging when test filtering fails

Both issues are verified fixed through passing unit tests. The empty filter validation is a circuit breaker that prevents silent failures.

---

Tested with Claude Code