[AIRFLOW-4363] Fix JSON encoding error (#7628)

From the docker-py code comments for APIClient pull, the decode parameter should be set to True, when the stream parameter is also set to True. This will allow decoding JSON data returned from the docker registry server into dicts Signed-off-by: Raymond Etornam <[email protected]>
apache · Mar 25, 2020 · 733d3d3 · 733d3d3
1 parent 4c067de
commit 733d3d3
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 2 deletions.
diff --git a/airflow/providers/docker/operators/docker.py b/airflow/providers/docker/operators/docker.py
@@ -275,7 +275,7 @@ def execute(self, context):
         # Pull the docker image if `force_pull` is set or image does not exist locally
         if self.force_pull or not self.cli.images(name=self.image):
             self.log.info('Pulling docker image %s', self.image)
-            for line in self.cli.pull(self.image, stream=True):
+            for line in self.cli.pull(self.image, stream=True, decode=True):
                 output = json.loads(line.decode('utf-8').strip())
                 if 'status' in output:
                     self.log.info("%s", output['status'])

diff --git a/tests/providers/docker/operators/test_docker.py b/tests/providers/docker/operators/test_docker.py
@@ -85,7 +85,8 @@ def test_execute(self, client_class_mock, tempdir_mock):
         client_mock.images.assert_called_once_with(name='ubuntu:latest')
         client_mock.attach.assert_called_once_with(container='some_id', stdout=True,
                                                    stderr=True, stream=True)
-        client_mock.pull.assert_called_once_with('ubuntu:latest', stream=True)
+        client_mock.pull.assert_called_once_with('ubuntu:latest', stream=True,
+                                                 decode=True)
         client_mock.wait.assert_called_once_with('some_id')
 
     @mock.patch('airflow.providers.docker.operators.docker.tls.TLSConfig')