get tests to 100% coverage (#131)

* restructure test_tmjob to hit all lines

* fix missing line hits test_weight

* test that the other tests would not test that binscript if plotting is not available

* exclude coverage for uncoverable line

* try different syntax

* try this syntax instead?

* use documented way instead

* use non-relative import instead?

* revert changes to test_broken_imports.py

* actually run test without the plotting code enabled

* skip test if plotting is not available

* use correct pyroject.toml flag for parellel unittest

* catch all errors instead

* update coverage to 79 (currently at 80)

.github/workflows/coverage.yml

@@ -44,7 +44,7 @@ jobs:
           path: coverage.xml
           repo_token: ${{ secrets.GITHUB_TOKEN }}
           pull_request_number: ${{ steps.get-pr.outputs.PR }}
-          minimum_coverage: 78
+          minimum_coverage: 79
           show_missing: True
           fail_below_threshold: True
           link_missing_lines: True

.github/workflows/unit-tests.yml

@@ -27,11 +27,18 @@ jobs:
         run: |
           conda install -y -c conda-forge python=3 cupy cuda-version=11.8 
           python -m pip install coverage
-      - name: Install code and list dependencies 
+      - name: Install base code and list dependencies 
+        run: |
+          python -m pip install -e .
+          conda list
+      - name: Run tests
+        run: |
+          coverage run -m unittest discover tests/
+      - name: Install plotting code and list dependencies
         run: |
           python -m pip install -e .[plotting]
           conda list
-      - name: Run tests and generate coverage
+      - name: Run complete tests and generate coverage report
         run: |
           coverage run -m unittest discover tests/
           coverage combine

pyproject.toml

@@ -61,3 +61,4 @@ package-data = {"pytom_tm.angle_lists" = ["*.txt"]}
 
 [tool.coverage.run]
 concurrency = ["multiprocessing", "thread"]
+parallel = true

tests/test00_parallel.py

@@ -77,7 +77,7 @@ def test_parallel_breaking(self):
             self.assertEqual(len(multiprocessing.active_children()), 0,
                              msg='a process was still lingering after a parallel job with partially invalid resources '
                                  'was started')
-        else:
+        else: # pragma: no cover
             self.fail('This should have given a RuntimeError')
 
     def test_parallel_manager(self):

tests/test_broken_imports.py

@@ -4,6 +4,13 @@
 # Mock out installed dependencies
 orig_import = __import__
 
+# skip tests if optional stuff is not installed
+SKIP_PLOT = False
+try:
+    import pytom_tm.plotting
+except:
+    SKIP_PLOT = True
+
 def module_not_found_mock(missing_name):
     def import_mock(name, *args):
         if name == missing_name:
@@ -42,6 +49,7 @@ def test_broken_cupy(self):
             self.assertEqual(len(cm.output), 1)
             self.assertIn("cupy installation not found or not functional", cm.output[0])
 
+    @unittest.skipIf(SKIP_PLOT, "plotting module not installed")
     def test_missing_matplotlib(self):
         # assert working import
         import pytom_tm
@@ -63,6 +71,7 @@ def test_missing_matplotlib(self):
             with self.assertRaisesRegex(RuntimeError, "matplotlib and seaborn"):
                 reload(pytom_tm.plotting)
 
+    @unittest.skipIf(SKIP_PLOT, "plotting module not installed")
     def test_missing_seaborn(self):
         # assert working import
         import pytom_tm

tests/test_tmjob.py

@@ -141,6 +141,7 @@ def test_tm_job_errors(self):
             with self.assertRaises(ValueError, msg='Invalid start index in search should raise ValueError'):
                 TMJob('0', 10, TEST_TOMOGRAM, TEST_TEMPLATE, TEST_MASK, TEST_DATA_DIR, voxel_size=1.,
                       **{param: [-10, 100]})
+            with self.assertRaises(ValueError, msg='Invalid start index in search should raise ValueError'):
                 TMJob('0', 10, TEST_TOMOGRAM, TEST_TEMPLATE, TEST_MASK, TEST_DATA_DIR, voxel_size=1.,
                       **{param: [110, 130]})
             with self.assertRaises(ValueError, msg='Invalid end index in search should raise ValueError'):

tests/test_weights.py

@@ -114,6 +114,9 @@ def test_radial_reduced_grid(self):
         with self.assertRaises(ValueError, msg='Radial reduced grid should raise ValueError if the shape is '
                                                'not 2- or 3-dimensional.'):
             radial_reduced_grid((5, ))
+        with self.assertRaises(ValueError, msg='Radial reduced grid should raise ValueError if the shape is '
+                                               'not 2- or 3-dimensional.'):
+
             radial_reduced_grid((5, ) * 4)
 
         self.assertEqual(radial_reduced_grid(self.volume_shape_even).shape, self.reduced_even_shape_3d,
@@ -291,6 +294,9 @@ def test_power_spectrum_profile(self):
         with self.assertRaises(ValueError, msg='Power spectrum profile should raise ValueError if input image is '
                                                'not 2- or 3-dimensional.'):
             power_spectrum_profile(np.zeros(5))
+        with self.assertRaises(ValueError, msg='Power spectrum profile should raise ValueError if input image is '
+                                               'not 2- or 3-dimensional.'):
+
             power_spectrum_profile(np.zeros((5, ) * 4))
         profile = power_spectrum_profile(np.zeros(self.volume_shape_irregular))
         self.assertEqual(profile.shape, (max(self.volume_shape_irregular) // 2 + 1, ),
@@ -305,6 +311,8 @@ def test_profile_to_weighting(self):
         with self.assertRaises(ValueError, msg='Profile to weighting should raise a ValueError if the output shape '
                                                'for the weighting is not 2- or 3-dimensional.'):
             profile_to_weighting(np.zeros(5), (5, ))
+        with self.assertRaises(ValueError, msg='Profile to weighting should raise a ValueError if the output shape '
+                                               'for the weighting is not 2- or 3-dimensional.'):
             profile_to_weighting(np.zeros(5), (5,) * 4)
 
         profile = power_spectrum_profile(np.zeros(self.volume_shape_irregular))