Skip to main content

crashtest.run

Functions to run tests, collect results, and produce run reports.

compare_opcodes​

def compare_opcodes(original_code: Bytecode, recompiled_code: Bytecode, class_name: str) -> OpcodeComparison

Compare per-method opcode sequences between original and recompiled bytecode.

file_to_name​

def file_to_name(file: str) -> str

gen_id​

def gen_id() -> str

Generate a unique ID for a test run.

gen_status​

def gen_status(results: List[TestCase]) -> Tuple[str, str]

Generate a status message and color based on test results. Returns a tuple of (status_message, color_hex).

Colors:

  • Green (#22C55E): All tests passed
  • Yellow (#EAB308): < 10% failures
  • Orange (#F97316): 10-20% failures
  • Red-Orange (#EF4444): 20-50% failures
  • Red (#DC2626): > 50% failures
  • Dark Red (#991B1B): All tests failed

get_repo_info​

def get_repo_info() -> GitInfo

Get the git branch and commit hash, if available.

op_similarity​

def op_similarity(a: List[str], b: List[str]) -> float

Sequence similarity of two opcode name lists, 0.0-1.0.

op_source_lines​

def op_source_lines(func: Function) -> List[int]

Per-op source line numbers (1-indexed), parallel to to_asm_resolved's output line-for-line. Each op's debug entry points back into whatever source file was compiled to produce this bytecode (the original .hx for the original function, or the decompiled pseudocode .hx for the recompiled one) — this is what lets the UI match a diffed opcode line up with the source line that produced it. -1 where no debug info is available.

recompile_pseudo​

def recompile_pseudo(pseudo_content: str, class_name: str, original_path: Path) -> Tuple[Optional[Bytecode], Optional[str], BehavioralComparison]

Compile pseudo, then compare behavior unless this shipped fixture is exempt.

run​

def run() -> bool

Run all tests.

run_case​

def run_case(case: str, id: int) -> TestCase

Runs a single test case by decompiling the main class in the file. It generates both class-level pseudocode and a combined IR view of all methods, then recompiles and executes both versions. Opcode similarity is diagnostic only.

run_case_isolated​

def run_case_isolated(case: str, id: int) -> TestCase

Runs run_case in a subprocess, enforcing TIME_LIMIT_SECONDS and MEMORY_LIMIT_MB. Exceeding either is reported as a failed case rather than hanging or OOMing the host.

run_single_case​

def run_single_case(args: Any) -> bool

Run a single test case by name and print results to the terminal.

sweep​

def sweep(bytecode_path: str, count: int, out_path: str) -> None

Decompile the first count functions of a bytecode image to a file.

to_asm_resolved​

def to_asm_resolved(func: Function, code: Bytecode) -> str

One opcode per line, with constant-pool indices resolved to their actual values.