v1.2.0

This version introduces the ability to refactor test cases as part of the Analyze command! The only refactoring strategy available in this version is to add t.Run() subtests to table-driven tests that don't have it already, but a framework has been established to allow additional strategies in the future!

What's Changed

✨ New Features

• Implement "subtest" refactoring strategy in Analyze command
  • Execute all tests before and after refactoring to verify success without modifying behavior
  • Output clearly differentiates between errors in the refactor generation and execution phases using various new struct types
  • Save refactoring changes on disk for manual analysis using keep-refactored-files option, or revert files to their original state by default
  • Avoids compilation errors by modifying the scenario runner if needed:
    • Replaces some continue statements with return to indicate that the test should pass without performing more logic
    • Modifies the scenario runner's "loop key" variable if scenario names are stored as a map key, instead of trying to access a scenario struct field
    • Detects the *testing.T parameter name dynamically as opposed to hardcoding it
  • Example output: ollama_api_TestMessage_UnmarshalJSON.json
• Clone helper functions internally during refactoring to avoid interference with multiple test cases
• Completely expand package-local helper function calls within test cases
  • This opens the door for more detailed analysis when helper functions are used to extract test logic
• Minor scenario field detection improvements
  • Prioritize detecting scenario name field from the first t.Run() argument
  • Enforce that scenario name field must be a string
  • Update scenario runner field to store entire loop statement instead of just the body
  • Require scenario runner to be detected directly in the test function to reduce table-driven test false positives
• Remove timer option, leaving it enabled permanently

🐛 Bug Fixes

• Fix nil pointer dereference for empty or non-analyzed test cases
• Scenario fields of user-defined type with underlying string type are no longer considered a valid scenario name
• Table-driven tests using a map like map[string]bool now properly detect the use of t.Run()

🛠️ Internal Improvements

• Completely restructure testcase package structure (see b9fa7c1)
• Create internal/asttools package to hold AST-related util functions

Full Changelog: v1.1.1...v1.2.0