Function unsafe_execute

evaluation/execution.py:58–361

(tmp_dir)

    """

    def unsafe_execute(tmp_dir):
        random_id = random.uniform(1, 1000)
        if "python" in language_type.lower():
            with create_tempdir():

                # These system calls are needed when cleaning up tempdir.
                import os
                import shutil
                rmtree = shutil.rmtree
                rmdir = os.rmdir
                chdir = os.chdir

                # Disable functionalities that can make destructive changes to the test.
                reliability_guard()

                try:
                    exec_globals = {}
                    with swallow_io():
                        with time_limit(timeout):
                            # WARNING
                            # This program exists to execute untrusted model-generated code. Although
                            # it is highly unlikely that model-generated code will do something overtly
                            # malicious in response to this test suite, model-generated code may act
                            # destructively due to a lack of model capability or alignment.
                            # Users are strongly encouraged to sandbox this evaluation suite so that it
                            # does not perform destructive actions on their host or network.
                            # Once you have read this disclaimer and taken appropriate precautions,
                            # uncomment the following line and proceed at your own risk:
                            exec(sample["test_code"], exec_globals)
                        result.append("passed")
                except TimeoutException:
                    result.append("timed out")
                except AssertionError as e:
                    result.append(f"failed: AssertionError")
                except BaseException as e:
                    result.append(f"failed: {e}")

                # Needed for cleaning up.
                shutil.rmtree = rmtree
                os.rmdir = rmdir
                os.chdir = chdir

        elif "go" in language_type.lower():
            assert tmp_dir is not None, "Go should be evaluated in a dir where necessary module files installed."

            import os
            import shutil

            if "tmp" not in tmp_dir:
                tmp_dir = os.path.join(tmp_dir, "tmp")
            tmp_dir = os.path.join(tmp_dir, f"{task_id.replace('/', '-')}-{random_id}")
            if not os.path.exists(tmp_dir):
                os.makedirs(tmp_dir)

            os.chdir(tmp_dir)
            open(f"main_test.go", 'w').write(sample["test_code"])
            try:
                exec_result = None