Add JSON rubric editor and modernize grading UI with additive scoring

.github/workflows/matlab-tests.yml

@@ -0,0 +1,34 @@
+name: MATLAB tests
+
+on:
+  push:
+    branches: [master, main, 'claude/**']
+  pull_request:
+  workflow_dispatch:
+
+jobs:
+  test:
+    name: Run function-based tests
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check out repository
+        uses: actions/checkout@v4
+
+      - name: Set up MATLAB
+        uses: matlab-actions/setup-matlab@v2
+        with:
+          release: latest
+
+      - name: Run tests
+        uses: matlab-actions/run-tests@v2
+        with:
+          source-folder: .
+          select-by-folder: tests
+          test-results-junit: test-results/junit.xml
+
+      - name: Upload test results
+        if: always()
+        uses: actions/upload-artifact@v4
+        with:
+          name: test-results-junit
+          path: test-results/junit.xml

tests/fixtures/tiny_rubric.json

@@ -0,0 +1,66 @@
+{
+  "assignment_name": "Tiny Test Rubric",
+  "assignment_url": "http://example.invalid/tiny",
+  "points_model": "rubric_additive",
+  "shared_comment_bank": [
+    { "label": "Nice",         "text": "Nice work.",                   "points_delta": 0  },
+    { "label": "Missing",      "text": "Missing / incomplete.",        "points_delta": -1 }
+  ],
+  "items": [
+    {
+      "item_name": "Q1 response text",
+      "subfolder": "question1",
+      "item_filename": "q1.mat",
+      "points_possible": 5,
+      "description": "Answer Q1 in response.md.",
+      "skills": ["writing"],
+      "code": "",
+      "code_files": [],
+      "autograder": { "type": "response_name", "response_name": "(\\s*)#(\\s*)Q(\\s*)1" },
+      "rubric": [
+        { "name": "Attempted", "points": 3 },
+        { "name": "Reasoned",  "points": 2 }
+      ],
+      "comment_bank": [
+        { "label": "Too short", "text": "Response is too brief.", "points_delta": -1 }
+      ]
+    },
+    {
+      "item_name": "Q2 anyvartest",
+      "subfolder": "question2",
+      "item_filename": "q2.mat",
+      "points_possible": 10,
+      "description": "Set variables to expected values.",
+      "skills": ["code"],
+      "code": "close all; clear; q2;",
+      "code_files": ["q2.m"],
+      "autograder": {
+        "type": "anyvartest",
+        "checks": [
+          { "criterion": "answer_a", "value": 3.5,  "tolerance": 0.001, "points": 6 },
+          { "criterion": "answer_b", "value": 17,   "tolerance": 0.001, "points": 4 }
+        ]
+      },
+      "rubric": [
+        { "name": "answer_a correct", "points": 6 },
+        { "name": "answer_b correct", "points": 4 }
+      ],
+      "comment_bank": []
+    },
+    {
+      "item_name": "Q3 manual only",
+      "subfolder": "question3",
+      "item_filename": "q3.mat",
+      "points_possible": 4,
+      "description": "Manual only.",
+      "skills": [],
+      "code": "",
+      "code_files": [],
+      "autograder": { "type": "manual" },
+      "rubric": [
+        { "name": "Whole thing", "points": 4 }
+      ],
+      "comment_bank": []
+    }
+  ]
+}

tests/runalltests.m

@@ -0,0 +1,18 @@
+function results = runalltests()
+% RUNALLTESTS - run every function-based test in this folder.
+%
+% RESULTS = RUNALLTESTS()
+%
+% Adds the parent (toolbox root) to the path, then runs every test_*.m
+% file in this directory via matlab.unittest. Returns the
+% TestResult array so callers can inspect failures.
+
+here = fileparts(mfilename('fullpath'));
+addpath(fileparts(here));
+
+import matlab.unittest.TestSuite
+suite = TestSuite.fromFolder(here);
+results = run(suite);
+
+disp(table(results));
+end

tests/test_vhgradeloadrubric.m

@@ -0,0 +1,90 @@
+function tests = test_vhgradeloadrubric
+% TEST_VHGRADELOADRUBRIC - function-based tests for vhgradeloadrubric.
+%
+% Run with:  results = runtests('test_vhgradeloadrubric')
+tests = functiontests(localfunctions);
+end
+
+function setupOnce(tc)
+tc.TestData.here    = fileparts(mfilename('fullpath'));
+tc.TestData.fixture = fullfile(tc.TestData.here, 'fixtures', 'tiny_rubric.json');
+addpath(fileparts(tc.TestData.here));
+end
+
+function testMetadata(tc)
+[~, meta] = vhgradeloadrubric(tc.TestData.fixture);
+verifyEqual(tc, meta.assignment_name, 'Tiny Test Rubric');
+verifyEqual(tc, meta.assignment_url,  'http://example.invalid/tiny');
+verifyEqual(tc, meta.points_model,    'rubric_additive');
+verifyEqual(tc, numel(meta.shared_comment_bank), 2);
+verifyEqual(tc, meta.shared_comment_bank(1).label, 'Nice');
+verifyEqual(tc, meta.shared_comment_bank(2).points_delta, -1);
+end
+
+function testItemCountAndTopFields(tc)
+items = vhgradeloadrubric(tc.TestData.fixture);
+verifyEqual(tc, numel(items), 3);
+verifyEqual(tc, items(1).Item_name, 'Q1 response text');
+verifyEqual(tc, items(2).Points_possible, 10);
+verifyEqual(tc, items(3).Item_filename, 'q3.mat');
+verifyEqual(tc, items(1).Skills, {'writing'});
+verifyEqual(tc, items(2).CodeFiles, {'q2.m'});
+end
+
+function testResponseNameAutograder(tc)
+items = vhgradeloadrubric(tc.TestData.fixture);
+p = items(1).Parameters;
+verifyEqual(tc, lower(char(p(1).type)), 'response_name');
+verifyEqual(tc, p(1).response_name, '(\s*)#(\s*)Q(\s*)1');
+end
+
+function testAnyVartestExpandedToChecks(tc)
+items = vhgradeloadrubric(tc.TestData.fixture);
+p = items(2).Parameters;
+verifyEqual(tc, numel(p), 2);
+verifyEqual(tc, lower(char(p(1).type)), 'anyvartest');
+verifyEqual(tc, p(1).criterion, 'answer_a');
+verifyEqual(tc, p(1).value, 3.5);
+verifyEqual(tc, p(1).tolerance, 0.001);
+verifyEqual(tc, p(1).points, 6);
+verifyEqual(tc, p(2).points, 4);
+end
+
+function testManualAutograder(tc)
+items = vhgradeloadrubric(tc.TestData.fixture);
+verifyEqual(tc, lower(char(items(3).Parameters(1).type)), 'manual');
+end
+
+function testRubricParsed(tc)
+items = vhgradeloadrubric(tc.TestData.fixture);
+r = items(1).Rubric;
+verifyEqual(tc, numel(r), 2);
+verifyEqual(tc, r(1).name, 'Attempted');
+verifyEqual(tc, r(1).points, 3);
+verifyEqual(tc, r(2).points, 2);
+end
+
+function testCommentBankAndShared(tc)
+items = vhgradeloadrubric(tc.TestData.fixture);
+verifyEqual(tc, numel(items(1).Comment_bank), 1);
+verifyEqual(tc, items(1).Comment_bank(1).points_delta, -1);
+
+% Shared bank propagates to every item.
+for i = 1:numel(items)
+    verifyEqual(tc, numel(items(i).Shared_comment_bank), 2, ...
+        sprintf('shared bank size on item %d', i));
+    verifyEqual(tc, items(i).Shared_comment_bank(1).label, 'Nice');
+end
+end
+
+function testRubricSumsMatchPointsPossible(tc)
+% Every item in the fixture is designed so rubric points sum to
+% points_possible; this test catches accidental fixture drift.
+items = vhgradeloadrubric(tc.TestData.fixture);
+for i = 1:numel(items)
+    if isempty(items(i).Rubric), continue; end
+    s = sum([items(i).Rubric.points]);
+    verifyEqual(tc, s, items(i).Points_possible, ...
+        sprintf('item %d rubric sum', i));
+end
+end

tests/test_vhgradequestion_auto.m

@@ -0,0 +1,87 @@
+function tests = test_vhgradequestion_auto
+% TEST_VHGRADEQUESTION_AUTO - end-to-end autograder happy-path tests.
+%
+% Only exercises the branch where every autograder check passes, since
+% that branch saves the grade and returns without opening the GUI.
+% Failing / partial-credit paths open vhgraderesponsegui and would block
+% a headless test run.
+tests = functiontests(localfunctions);
+end
+
+function setupOnce(tc)
+tc.TestData.here    = fileparts(mfilename('fullpath'));
+tc.TestData.fixture = fullfile(tc.TestData.here, 'fixtures', 'tiny_rubric.json');
+addpath(fileparts(tc.TestData.here));
+end
+
+function setup(tc)
+tc.TestData.parentDir = tempname;
+mkdir(tc.TestData.parentDir);
+end
+
+function teardown(tc)
+if isfield(tc.TestData,'parentDir') && isfolder(tc.TestData.parentDir)
+    rmdir(tc.TestData.parentDir, 's');
+end
+end
+
+function testAllChecksPassAwardsFullCredit(tc)
+items = vhgradeloadrubric(tc.TestData.fixture);
+item = items(2);   % anyvartest, 10 points
+
+qdir = fullfile(tc.TestData.parentDir, item.Subfolder);
+mkdir(qdir);
+writeQ2(qdir, 3.5, 17);
+
+grade = vhgradequestion(tc.TestData.parentDir, item);
+
+verifyEqual(tc, grade.Points_earned, item.Points_possible, ...
+    'all checks pass → full credit');
+verifyEqual(tc, grade.CodeError, 0, 'no code error expected');
+verifyTrue(tc, isfile(fullfile(tc.TestData.parentDir,'GRADING',item.Item_filename)), ...
+    'grade .mat should be written');
+verifyTrue(tc, all(grade.Autograder_score.checks_passed), ...
+    'every check should pass');
+end
+
+function testCachedGradeIsReturnedUnchanged(tc)
+% Second call without forceRegrade must NOT re-run the sandbox; it should
+% just reload the saved grade.
+items = vhgradeloadrubric(tc.TestData.fixture);
+item = items(2);
+qdir = fullfile(tc.TestData.parentDir, item.Subfolder);
+mkdir(qdir);
+writeQ2(qdir, 3.5, 17);
+
+g1 = vhgradequestion(tc.TestData.parentDir, item);
+
+% Overwrite q2.m so the code would fail if it re-ran:
+writeQ2(qdir, 0, 0);
+g2 = vhgradequestion(tc.TestData.parentDir, item);
+
+verifyEqual(tc, g2.Points_earned, g1.Points_earned, ...
+    'cached grade must be reused when forceRegrade is 0');
+end
+
+function testDoesNotPolluteBaseWorkspace(tc)
+% Belt-and-braces: make sure the autograder path leaves no trace in base.
+items = vhgradeloadrubric(tc.TestData.fixture);
+item = items(2);
+qdir = fullfile(tc.TestData.parentDir, item.Subfolder);
+mkdir(qdir);
+writeQ2(qdir, 3.5, 17);
+
+evalin('base', 'clear answer_a answer_b');
+vhgradequestion(tc.TestData.parentDir, item);
+baseNames = evalin('base', 'who');
+verifyFalse(tc, any(strcmp(baseNames,'answer_a')), 'answer_a leaked to base');
+verifyFalse(tc, any(strcmp(baseNames,'answer_b')), 'answer_b leaked to base');
+end
+
+
+% ----- helpers -----
+function writeQ2(qdir, a, b)
+fid = fopen(fullfile(qdir,'q2.m'),'w');
+c = onCleanup(@() fclose(fid));
+fprintf(fid, 'answer_a = %.6f;\nanswer_b = %.6f;\n', a, b);
+end