Automl edge predict test (GoogleCloudPlatform#2195)

* update test

* use python-docker image

* update test to get test image from path

* use updated python-docker image

* update common.cfg to use the udpated python image

Author: dizcology

.kokoro/common.cfg

@@ -9,7 +9,7 @@ build_file: "python-docs-samples/.kokoro/trampoline.sh"
 # Use the Python worker docker image.
 env_vars: {
     key: "TRAMPOLINE_IMAGE"
-    value: "gcr.io/cloud-devrel-kokoro-resources/python@sha256:e11a459d01e5dcd3613fda35c7c94edfecfe911ed79c078580ff59de300b1938"
+    value: "gcr.io/cloud-devrel-kokoro-resources/python@sha256:9f83a3711eef244a4d45ac25028e200c87c4d4060d445438683a1ed4d3f093b6"
 }
 
 # Specify project ID

vision/automl/edge_container_predict/automl_vision_edge_container_predict_test.py

@@ -37,14 +37,14 @@
 
 # The absolute path of the current file. This will locate the model_path when
 # run docker containers.
-ROOT_DIR = os.path.abspath(os.path.dirname(__file__))
+ROOT_DIR = os.environ.get('KOKORO_ROOT', os.path.abspath(os.path.dirname(__file__)))
 MODEL_PATH = os.path.join(ROOT_DIR, 'model_path')
+
+IMAGE_FILE_PATH = os.path.join(os.path.dirname(__file__), 'test.jpg')
 # The cpu docker gcs path is from 'Edge container tutorial'.
-DOCKER_GCS_DIR = 'gcr.io/automl-vision-ondevice/'
-CPU_DOCKER_GCS_PATH = DOCKER_GCS_DIR + 'gcloud-container-1.12.0:latest'
+CPU_DOCKER_GCS_PATH = 'gcr.io/automl-vision-ondevice/gcloud-container-1.12.0:latest'
 # The path of a sample saved model.
-MODEL_GCS_DIR = 'gs://cloud-samples-data/vision/edge_container_predict/'
-SAMPLE_SAVED_MODEL = MODEL_GCS_DIR + 'saved_model.pb'
+SAMPLE_SAVED_MODEL = 'gs://cloud-samples-data/vision/edge_container_predict/saved_model.pb'
 # Container Name.
 NAME = 'AutomlVisionEdgeContainerPredictTest'
 # Port Number.
@@ -56,8 +56,8 @@ def edge_container_predict_server_port():
     # set up
     # Pull the CPU docker.
     subprocess.check_output(['docker', 'pull', CPU_DOCKER_GCS_PATH])
-    # Get the sample saved model.
 
+    # Get the sample saved model.
     if not os.path.exists(MODEL_PATH):
         os.mkdir(MODEL_PATH)
     subprocess.check_output(
@@ -79,18 +79,16 @@ def edge_container_predict_server_port():
     # Remove the docker image.
     subprocess.check_output(['docker', 'rmi', CPU_DOCKER_GCS_PATH])
 
-# TODO(dizcology): Enable tests in future.
-@pytest.mark.skip(reason='skipping to avoid running docker in docker')
+
 def test_edge_container_predict(capsys, edge_container_predict_server_port):
-    image_file_path = 'test.jpg'
     # If you send requests with one image each time, the key value does not
     # matter. If you send requests with multiple images, please used different
     # keys to indicated different images, which can make sure that the
     # responses corresponding to the given image.
     image_key = '1'
     # Send a request.
     response = predict.container_predict(
-            image_file_path, image_key, PORT_NUMBER)
+            IMAGE_FILE_PATH, image_key, PORT_NUMBER)
     # Verify the response.
     assert 'predictions' in response
     assert 'key' in response['predictions'][0]