Input dataset MD5s are now tested when a run is launched, as part of #…

…751.
cfe-lab · Feb 22, 2019 · 5aaad54 · 5aaad54
1 parent 8b446c1
commit 5aaad54
Show file tree

Hide file tree

Showing 2 changed files with 36 additions and 2 deletions.
diff --git a/kive/container/management/commands/runcontainer.py b/kive/container/management/commands/runcontainer.py
@@ -93,6 +93,18 @@ def run_container(self, run):
         stdout_path = os.path.join(logs_path, 'stdout.txt')
         stderr_path = os.path.join(logs_path, 'stderr.txt')
 
+        for input_cd in run.datasets.filter(argument__type=ContainerArgument.INPUT):
+            input_dataset = input_cd.dataset
+            current_md5 = input_dataset.compute_md5()
+            if current_md5 != input_dataset.MD5_checksum:
+                raise ValueError(
+                    "Dataset with pk={} has an inconsistent checksum (original {}; current {})".format(
+                        input_dataset.pk,
+                        input_dataset.MD5_checksum,
+                        current_md5
+                    )
+                )
+
         container_to_run = run.app.container
         container_to_run.validate_md5()
         if not container_to_run.is_singularity():

diff --git a/kive/container/tests.py b/kive/container/tests.py
@@ -1483,6 +1483,23 @@ def test_run(self):
         dataset_groups = list(output_dataset.groups_allowed.values_list('name'))
         self.assertEqual(expected_groups, dataset_groups)
 
+    def test_run_input_bad_md5(self):
+        run = ContainerRun.objects.get(name='fixture run')
+        everyone = Group.objects.get(name='Everyone')
+        run.groups_allowed.clear()
+        run.groups_allowed.add(everyone)
+
+        # Tamper with the file.
+        input_containerdataset = run.datasets.get(
+            argument__type=ContainerArgument.INPUT,
+            argument__position=1
+        )
+        input_dataset = input_containerdataset.dataset
+        input_dataset.dataset_file.save("tampered", ContentFile(b"foo"), save=True)
+
+        with self.assertRaises(ValueError):
+            call_command('runcontainer', str(run.id))
+
     def test_run_bad_md5(self):
         run = ContainerRun.objects.get(name='fixture run')
         everyone = Group.objects.get(name='Everyone')
@@ -1699,8 +1716,13 @@ def test_run_multistep_archive(self):
         run.app = container.apps.create(memory=200, threads=1)
         run.app.write_inputs('pairs_csv')
         run.app.write_outputs('summary_csv')
-        pairs_dataset = Dataset.objects.create(user=run.user, name='pairs.csv')
-        pairs_dataset.dataset_file.save('pairs.csv', ContentFile(pairs_text))
+        pairs_dataset = Dataset.create_dataset(
+            file_path=None,
+            user=run.user,
+            file_handle=ContentFile(pairs_text, name="pairs.csv")
+        )
+        # pairs_dataset = Dataset.objects.create(user=run.user, name='pairs.csv')
+        # pairs_dataset.dataset_file.save('pairs.csv', ContentFile(pairs_text))
         run_input = run.datasets.get()
         run_input.dataset = pairs_dataset
         run_input.argument = run.app.arguments.get(type=ContainerArgument.INPUT)