Smart Test errors

Common error signatures

Most smart-test failures fall into two categories: coverage file problems and malformed source code that the AST analyzer cannot parse.

Exception	Message pattern	Likely cause
`FileNotFoundError`	`Coverage file not found: {path}`	`parse_coverage_json()` was called before `pytest --cov` produced a `coverage.json`, or the path passed to it is wrong
`ValueError`	`Invalid coverage JSON at {path}: {detail}`	The `coverage.json` file exists but is not valid JSON — typically a truncated or partially written file
`ValueError`	`Unexpected coverage JSON structure in {path}: 'files' key missing or not a dict`	The JSON is valid but doesn't match the expected pytest-cov schema — usually caused by a coverage.py version mismatch

Where errors originate

Failures in the coverage-parsing stage happen in parse_coverage_json() before any test generation begins. Failures in the generation stage come from ASTFunctionAnalyzer, generate_test_for_function(), or generate_test_for_class() when the source module has syntax the analyzer doesn't handle.

parse_coverage_json(json_path) — reads and validates coverage.json; raises FileNotFoundError or ValueError (see above)
prioritize_modules(modules, min_threshold) — filters by coverage threshold; a completely empty result here means every module is above min_threshold (default 50.0), not that parsing failed
group_into_batches(modules, max_batches) — groups by package path; an empty batch list means prioritize_modules() returned nothing
generate_test_for_function(module, func) — depends on ASTFunctionAnalyzer.analyze() having run first; if func is missing expected keys the output will be an empty or broken test stub
TestAuditWorkflow.execute() and TestGenerationWorkflow.execute() — orchestrate the subagents; errors from the functions above bubble up through these

How to diagnose

Identify which stage failed. Coverage-parsing errors (FileNotFoundError, ValueError) happen before any test generation. If you see one of the message patterns in the table above, the problem is in your coverage.json, not in your source code.
Confirm coverage.json was generated. Run pytest --cov=<your_src_dir> --cov-report=json and verify the file exists at the path you're passing to smart-test. If the file is missing, pytest-cov either wasn't installed or the run failed before producing output.
Check for a 'files' key mismatch. If you see the 'files' key missing error, your coverage.json was generated by a coverage.py version that uses a different schema. Upgrade or pin coverage to a version compatible with pytest-cov 4.x and regenerate the file.
Isolate an empty prioritization result. If smart-test produces no modules to test, check the min_threshold in prioritize_modules() (default 50.0). All your modules may already be above 50% coverage. Lower the threshold or target a specific module path directly.
Debug AST analysis failures. If test generation produces empty or malformed stubs, run ASTFunctionAnalyzer().analyze(source_code, file_path) directly on the failing module to see what FunctionSignature and ClassSignature objects it produces. Missing raises, empty params, or complexity=1 on a clearly complex function indicate the analyzer couldn't parse that construct.

Source files

src/attune/workflows/test_audit/coverage_parser.py
src/attune/workflows/test_gen/
src/attune/workflows/test_gen_parallel.py

Tags: tests, coverage, generation