Fix afni contrast bool

[Shotgunosine]

closes #265

[pep8speaks]

Hello @Shotgunosine, Thank you for updating!

Cheers! There are no style issues detected in this Pull Request. 🍻 To test for issues locally, pip install flake8 and then run flake8 fitlins.

Comment last updated at 2020-11-20 23:28:02 UTC

[effigies]

No obvious problems with the logic, but here are some simplifications.

[effigies]

Tests passing. Is this branch working for you? Do you have a test model snippet that can exercise this?

[Shotgunosine]

Double checking the changes now.

here's a slightly tweaked ds003_model001 that would provoke the bug if it was still present:

{
  "Name": "ds003_model001",
  "Description": "",
  "Input": {
    "task": "rhymejudgment"
  },
  "Steps": [
    {
      "Level": "run",
      "Transformations": [
        {
          "Name": "Factor",
          "Input": [
            "trial_type"
          ]
        },
        {
          "Name": "Convolve",
          "Input": ["trial_type.word", "trial_type.pseudoword"],
          "Model": "spm"
        }
      ],
      "Model": {
        "X": [
          "trial_type.word",
          "trial_type.pseudoword",
          "framewise_displacement",
          "trans_x", "trans_y", "trans_z", "rot_x", "rot_y", "rot_z",
          "a_comp_cor_00", "a_comp_cor_01", "a_comp_cor_02",
          "a_comp_cor_03", "a_comp_cor_04", "a_comp_cor_05"
        ]
      },
      "DummyContrasts": {
        "Conditions": ["trial_type.word", "trial_type.pseudoword"],
        "Type": "t"
        },
      "Contrasts": [
        {
          "Name": "word_gt_pseudo",
          "ConditionList": [
            "trial_type.word",
            "trial_type.pseudoword"
          ],
          "Weights": [1, -1],
          "Type": "t"
        },
        {
          "Name": "word_gt_pseudo_rev",
          "ConditionList": [
            "trial_type.word",
            "trial_type.pseudoword"
          ],
          "Weights": [-1, 1],
          "Type": "t"
        },
        {
          "Name": "task_vs_baseline",
          "ConditionList": [
            "trial_type.word",
            "trial_type.pseudoword"
          ],
          "Weights": [0.5, 0.5],
          "Type": "t"
        }
      ]
    },
    {
      "Level": "dataset",
      "DummyContrasts": {
        "Conditions": ["word_gt_pseudo", "task_vs_baseline"],
        "Type": "t"
        },
      "Contrasts":[
        {
          "Name": "any_words",
          "ConditionList": [
            "trial_type.word",
            "trial_type.pseudoword"
          ],
          "Weights": [
            [1, 0],
            [0, 1]
          ],
          "Type": "F"
        }
      ]
    }
  ]
}

As for that line that's not getting tested, I don't think that case should normally come up, but I didn't want the counts to get out of wack if it did come up for some reason.

[Shotgunosine]

It's working locally for me, I'd be ok if you wanted to go ahead and merge it.

[effigies]

Sounds good. Looks like we're covering the cases. Would be good to set up more rigorous tests when we have time.