Test failing against historical example

[jpgoldberg]

There is a fair chance that I am using it wrong, but my attempts to test against Message from Dönitz have failed.

My (pytest) test looks like

class TestHistorical:
    """Tests taken from https://gist.github.com/Jither/d8dbc4d38998c18686bb646b49b9a8a6"""

    def test_p1030681(self) -> None:
        wheels: list[str] = ["Beta", "V", "VI", "VIII"]
        rotor_positions = "CDSZ"
        ring_settings = "EPEL"
        reflector = "Reflector C Thin"
        plug_board = "AE BF CM DQ HU JN LX PR SZ VW"
        rotors = list(zip(wheels, list(rotor_positions)))

        ctext = "LANOTCTOUARBBFPMHPHGCZXTDYGAHGUFXGEWKBLKGJWLQXXTGPJJAVTOCKZFSLPPQIHZFXOEBWIIEKFZLCLOAQJULJOYHSSMBBGWHZANVOIIPYRBRTDJQDJJOQKCXWDNBBTYVXLYTAPGVEATXSONPNYNQFUDBBHHVWEPYEYDOHNLXKZDNWRHDUWUJUMWWVIIWZXIVIUQDRHYMNCYEFUAPNHOTKHKGDNPSAKNUAGHJZSMJBMHVTREQEDGXHLZWIFUSKDQVELNMIMITHBHDBWVHDFYHJOQIHORTDJDBWXEMEAYXGYQXOHFDMYUXXNOJAZRSGHPLWMLRECWWUTLRTTVLBHYOORGLGOWUXNXHMHYFAACQEKTHSJW"

        ptext = "KRKRALLEXXFOLGENDESISTSOFORTBEKANNTZUGEBENXXICHHABEFOLGELNBEBEFEHLERHALTENXXJANSTERLEDESBISHERIGXNREICHSMARSCHALLSJGOERINGJSETZTDERFUEHRERSIEYHVRRGRZSSADMIRALYALSSEINENNACHFOLGEREINXSCHRIFTLSCHEVOLLMACHTUNTERWEGSXABSOFORTSOLLENSIESAEMTLICHEMASSNAHMENVERFUEGENYDIESICHAUSDERGEGENWAERTIGENLAGEERGEBENXGEZXREICHSLEITEIKKTULPEKKJBORMANNJXXOBXDXMMMDURNHFKSTXKOMXADMXUUUBOOIEXKP"

        machine = enigma.Enigma(
            catalog="default",
            stecker=plug_board,
            stator="military",
            rotors=rotors,
            reflector=reflector,
            operator=False,
        )
        machine.set_wheels(ring_settings)

        decrypted = machine.parse(ctext)
        assert decrypted == ptext

The test fails with

>       assert decrypted == ptext
E       AssertionError: assert 'YSTCDYZBWPSM...UIHTWERVHZEPR' == 'KRKRALLEXXFO...MXUUUBOOIEXKP'
...

I have tried swapping my ring settings with the wheel_positions, as I might have been confused which is to be used where, but I still get a failure.

[ZAdamMac]

This actually came up once before, with a different historical example message. I'm going to have to hunt down and go through my lab book from back then to see what I changed. The good news there is that I'm staring down the barrel of a pretty sizeable chunk of PTO and can put some real time into this script for a while.

I am curious though; is this happening on the version of the code that you've got set up with the PRs you've submitted, or with the version currently available through pip//latest master?

[jpgoldberg]

My test file also tests against your Hello World example (which passes), but I will create a branch that is identical to your master branch with only the test file difference.

[jpgoldberg]

I've created a PR #16 that only includes my test file. This way we can both be confident that my "should be entirely harmless" tidying didn't accidentally change behavior.

[jpgoldberg]

The good news there is that I'm staring down the barrel of a pretty sizeable chunk of PTO

I should add that I don't have a real need for this to be fixed. I was just very impressed with the research you did in getting the historical details. As you know, many Enigma simulators out there use simplified variants (leaving out the Stator, notches, variation in reflector, etc). But my actual needs are satisfied by anything that is self-consistent with encryption and decryption and maintains the property that no input symbol ever encrypts to itself.

So enjoy your PTO as you see fit. I'm happy to work with you on this if that is what you want to do, but it is perfectly fine if you don't.

[jpgoldberg]

I believe I have identified the bug (but haven't specially tested). Because the Greek rotors (Beta in this case) are static, the subsequent wheels never advance.

Update: That is not the problem

[jpgoldberg]

the subsequent wheels never advance.

I have added a test, and I see that wheels are advancing.