verify-binary.md

How To Verify Binary Blobs

Contents

• Examples
  • Approving images
  • Approving NumPy arrays

Using the verify_binary function, you can also check binary data blobs. For instance, you could check that your program produces correct image frames. Another example would be approving and checking numpy arrays in scientific applications.

Examples

Approving images

In this example, you can see how verify_binary is used to approve rendered images.

name = "icon.png"
filename = get_adjacent_file(name)
with open(filename, mode="rb") as f:
    verify_binary(f.read(), ".png")

^{snippet source | anchor}

Which will produce

Approving NumPy arrays

This more detailed example demonstrates how you could use verify_binary to approve NumPy arrays.

It uses serializers instead of dumping the binary blob directly to disk. In this case, the serialization provided by NumPy is used.

def test_simulator_produces_correct_output():
    np_array = np.full(shape=(32, 16), fill_value=42, dtype=np.int64)
    verify_binary(
        serialize_ndarray(np_array),
        ".npy",
        options=Options().with_reporter(NDArrayDiffReporter()),
    )

^{snippet source | anchor}

It also shows the use of a custom reporter, which uses the NumPy testing features to produce a well-formatted error report highlighting the differences between arrays.

def load_ndarray(path):
    with open(path, mode="rb") as f:
        return np.load(f)

class NDArrayDiffReporter(Reporter):
    def report(self, received_path: str, approved_path: str) -> bool:
        if not Path(approved_path).is_file():
            self._create_empty_array(approved_path)
        received = load_ndarray(received_path)
        approved = load_ndarray(approved_path)
        to_approve_msg = (
            f"To approve run:\n {get_command_text(received_path,approved_path)}"
        )
        print(np.testing.build_err_msg([received, approved], err_msg=to_approve_msg))
        return True

^{snippet source | anchor}