feat: add support for accepting an Artifact Registry URL in pipeline_job (#1405)

chongyouquan · parthea · web-flow · commit e138cfd8b8f8 · 2022-06-17T14:51:13.000-07:00
* Add support for Artifact Registry in template_path * fix typo * update tests * fix AR path * remove unused project * add code for refreshing credentials * add import for google.auth.transport * fix AR path * fix AR path * fix runtime_config * test removing v1beta1 * try using v1 directly instead * update to use v1beta1 * use select_version * add back template_uri * try adding back v1beta1 * use select_version * differentiate when to use select_version * test removing v1beta1 for pipeline_complete_states * add tests for creating pipelines using v1beta1 * fix merge * fix typo * fix lint using blacken * fix regex * update to use v1 instead of v1beta1 * add test for invalid url * update error type * implement failure_policy * use urllib.request instead of requests * Revert "implement failure_policy" This reverts commit 72cdd9e. * fix lint Co-authored-by: Anthonios Partheniou <partheniou@google.com>
diff --git a/google/cloud/aiplatform/pipeline_jobs.py b/google/cloud/aiplatform/pipeline_jobs.py
@@ -56,6 +56,9 @@
 # Pattern for valid names used as a Vertex resource name.
 _VALID_NAME_PATTERN = re.compile("^[a-z][-a-z0-9]{0,127}$")
 
+# Pattern for an Artifact Registry URL.
+_VALID_AR_URL = re.compile(r"^https:\/\/([\w-]+)-kfp\.pkg\.dev\/.*")
+
 
 def _get_current_time() -> datetime.datetime:
     """Gets the current timestamp."""
@@ -125,8 +128,9 @@ def __init__(
                 Required. The user-defined name of this Pipeline.
             template_path (str):
                 Required. The path of PipelineJob or PipelineSpec JSON or YAML file. It
-                can be a local path or a Google Cloud Storage URI.
-                Example: "gs://project.name"
+                can be a local path, a Google Cloud Storage URI (e.g. "gs://project.name"),
+                or an Artifact Registry URI (e.g.
+                "https://us-central1-kfp.pkg.dev/proj/repo/pack/latest").
             job_id (str):
                 Optional. The unique ID of the job run.
                 If not specified, pipeline name + timestamp will be used.
@@ -237,15 +241,20 @@ def __init__(
         if enable_caching is not None:
             _set_enable_caching_value(pipeline_job["pipelineSpec"], enable_caching)
 
-        self._gca_resource = gca_pipeline_job.PipelineJob(
-            display_name=display_name,
-            pipeline_spec=pipeline_job["pipelineSpec"],
-            labels=labels,
-            runtime_config=runtime_config,
-            encryption_spec=initializer.global_config.get_encryption_spec(
+        pipeline_job_args = {
+            "display_name": display_name,
+            "pipeline_spec": pipeline_job["pipelineSpec"],
+            "labels": labels,
+            "runtime_config": runtime_config,
+            "encryption_spec": initializer.global_config.get_encryption_spec(
                 encryption_spec_key_name=encryption_spec_key_name
             ),
-        )
+        }
+
+        if _VALID_AR_URL.match(template_path):
+            pipeline_job_args["template_uri"] = template_path
+
+        self._gca_resource = gca_pipeline_job.PipelineJob(**pipeline_job_args)
 
     @base.optional_sync()
     def run(
diff --git a/google/cloud/aiplatform/utils/yaml_utils.py b/google/cloud/aiplatform/utils/yaml_utils.py
@@ -15,11 +15,17 @@
 # limitations under the License.
 #
 
+import re
 from typing import Any, Dict, Optional
+from urllib import request
 
 from google.auth import credentials as auth_credentials
+from google.auth import transport
 from google.cloud import storage
 
+# Pattern for an Artifact Registry URL.
+_VALID_AR_URL = re.compile(r"^https:\/\/([\w-]+)-kfp\.pkg\.dev\/.*")
+
 
 def load_yaml(
     path: str,
@@ -42,6 +48,8 @@ def load_yaml(
     """
     if path.startswith("gs://"):
         return _load_yaml_from_gs_uri(path, project, credentials)
+    elif _VALID_AR_URL.match(path):
+        return _load_yaml_from_ar_uri(path, credentials)
     else:
         return _load_yaml_from_local_file(path)
 
@@ -95,3 +103,37 @@ def _load_yaml_from_local_file(file_path: str) -> Dict[str, Any]:
         )
     with open(file_path) as f:
         return yaml.safe_load(f)
+
+
+def _load_yaml_from_ar_uri(
+    uri: str,
+    credentials: Optional[auth_credentials.Credentials] = None,
+) -> Dict[str, Any]:
+    """Loads data from a YAML document referenced by a Artifact Registry URI.
+
+    Args:
+      path (str):
+          Required. Artifact Registry URI for YAML document.
+      credentials (auth_credentials.Credentials):
+          Optional. Credentials to use with Artifact Registry.
+
+    Returns:
+      A Dict object representing the YAML document.
+    """
+    try:
+        import yaml
+    except ImportError:
+        raise ImportError(
+            "pyyaml is not installed and is required to parse PipelineJob or PipelineSpec files. "
+            'Please install the SDK using "pip install google-cloud-aiplatform[pipelines]"'
+        )
+    req = request.Request(uri)
+
+    if credentials:
+        if not credentials.valid:
+            credentials.refresh(transport.requests.Request())
+        if credentials.token:
+            req.add_header("Authorization", "Bearer " + credentials.token)
+    response = request.urlopen(req)
+
+    return yaml.safe_load(response.read().decode("utf-8"))
diff --git a/tests/unit/aiplatform/test_pipeline_jobs.py b/tests/unit/aiplatform/test_pipeline_jobs.py
@@ -22,6 +22,7 @@
 from unittest import mock
 from importlib import reload
 from unittest.mock import patch
+from urllib import request
 from datetime import datetime
 
 from google.auth import credentials as auth_credentials
@@ -50,6 +51,7 @@
 _TEST_SERVICE_ACCOUNT = "abcde@my-project.iam.gserviceaccount.com"
 
 _TEST_TEMPLATE_PATH = f"gs://{_TEST_GCS_BUCKET_NAME}/job_spec.json"
+_TEST_AR_TEMPLATE_PATH = "https://us-central1-kfp.pkg.dev/proj/repo/pack/latest"
 _TEST_PARENT = f"projects/{_TEST_PROJECT}/locations/{_TEST_LOCATION}"
 _TEST_NETWORK = f"projects/{_TEST_PROJECT}/global/networks/{_TEST_PIPELINE_JOB_ID}"
 
@@ -289,6 +291,17 @@ def mock_load_yaml_and_json(job_spec):
         yield mock_load_yaml_and_json
 
 
+@pytest.fixture
+def mock_request_urlopen(job_spec):
+    with patch.object(request, "urlopen") as mock_urlopen:
+        mock_read_response = mock.MagicMock()
+        mock_decode_response = mock.MagicMock()
+        mock_decode_response.return_value = job_spec.encode()
+        mock_read_response.return_value.decode = mock_decode_response
+        mock_urlopen.return_value.read = mock_read_response
+        yield mock_urlopen
+
+
 @pytest.mark.usefixtures("google_auth_mock")
 class TestPipelineJob:
     def setup_method(self):
@@ -376,6 +389,85 @@ def test_run_call_pipeline_service_create(
             gca_pipeline_state.PipelineState.PIPELINE_STATE_SUCCEEDED
         )
 
+    @pytest.mark.parametrize(
+        "job_spec",
+        [_TEST_PIPELINE_SPEC_JSON, _TEST_PIPELINE_SPEC_YAML, _TEST_PIPELINE_JOB],
+    )
+    @pytest.mark.parametrize("sync", [True, False])
+    def test_run_call_pipeline_service_create_artifact_registry(
+        self,
+        mock_pipeline_service_create,
+        mock_pipeline_service_get,
+        mock_request_urlopen,
+        job_spec,
+        mock_load_yaml_and_json,
+        sync,
+    ):
+        aiplatform.init(
+            project=_TEST_PROJECT,
+            staging_bucket=_TEST_GCS_BUCKET_NAME,
+            location=_TEST_LOCATION,
+            credentials=_TEST_CREDENTIALS,
+        )
+
+        job = pipeline_jobs.PipelineJob(
+            display_name=_TEST_PIPELINE_JOB_DISPLAY_NAME,
+            template_path=_TEST_AR_TEMPLATE_PATH,
+            job_id=_TEST_PIPELINE_JOB_ID,
+            parameter_values=_TEST_PIPELINE_PARAMETER_VALUES,
+            enable_caching=True,
+        )
+
+        job.run(
+            service_account=_TEST_SERVICE_ACCOUNT,
+            network=_TEST_NETWORK,
+            sync=sync,
+            create_request_timeout=None,
+        )
+
+        if not sync:
+            job.wait()
+
+        expected_runtime_config_dict = {
+            "gcsOutputDirectory": _TEST_GCS_BUCKET_NAME,
+            "parameterValues": _TEST_PIPELINE_PARAMETER_VALUES,
+        }
+        runtime_config = gca_pipeline_job.PipelineJob.RuntimeConfig()._pb
+        json_format.ParseDict(expected_runtime_config_dict, runtime_config)
+
+        job_spec = yaml.safe_load(job_spec)
+        pipeline_spec = job_spec.get("pipelineSpec") or job_spec
+
+        # Construct expected request
+        expected_gapic_pipeline_job = gca_pipeline_job.PipelineJob(
+            display_name=_TEST_PIPELINE_JOB_DISPLAY_NAME,
+            pipeline_spec={
+                "components": {},
+                "pipelineInfo": pipeline_spec["pipelineInfo"],
+                "root": pipeline_spec["root"],
+                "schemaVersion": "2.1.0",
+            },
+            runtime_config=runtime_config,
+            service_account=_TEST_SERVICE_ACCOUNT,
+            network=_TEST_NETWORK,
+            template_uri=_TEST_AR_TEMPLATE_PATH,
+        )
+
+        mock_pipeline_service_create.assert_called_once_with(
+            parent=_TEST_PARENT,
+            pipeline_job=expected_gapic_pipeline_job,
+            pipeline_job_id=_TEST_PIPELINE_JOB_ID,
+            timeout=None,
+        )
+
+        mock_pipeline_service_get.assert_called_with(
+            name=_TEST_PIPELINE_JOB_NAME, retry=base._DEFAULT_RETRY
+        )
+
+        assert job._gca_resource == make_pipeline_job(
+            gca_pipeline_state.PipelineState.PIPELINE_STATE_SUCCEEDED
+        )
+
     @pytest.mark.parametrize(
         "job_spec",
         [
diff --git a/tests/unit/aiplatform/test_utils.py b/tests/unit/aiplatform/test_utils.py
@@ -20,6 +20,8 @@
 import json
 import os
 from typing import Callable, Dict, Optional
+from unittest import mock
+from urllib import request
 
 import pytest
 import yaml
@@ -564,13 +566,34 @@ def json_file(tmp_path):
     yield json_file_path
 
 
+@pytest.fixture(scope="function")
+def mock_request_urlopen():
+    data = {"key": "val", "list": ["1", 2, 3.0]}
+    with mock.patch.object(request, "urlopen") as mock_urlopen:
+        mock_read_response = mock.MagicMock()
+        mock_decode_response = mock.MagicMock()
+        mock_decode_response.return_value = json.dumps(data)
+        mock_read_response.return_value.decode = mock_decode_response
+        mock_urlopen.return_value.read = mock_read_response
+        yield "https://us-central1-kfp.pkg.dev/proj/repo/pack/latest"
+
+
 class TestYamlUtils:
-    def test_load_yaml_from_local_file__with_json(self, yaml_file):
+    def test_load_yaml_from_local_file__with_yaml(self, yaml_file):
         actual = yaml_utils.load_yaml(yaml_file)
         expected = {"key": "val", "list": ["1", 2, 3.0]}
         assert actual == expected
 
-    def test_load_yaml_from_local_file__with_yaml(self, json_file):
+    def test_load_yaml_from_local_file__with_json(self, json_file):
         actual = yaml_utils.load_yaml(json_file)
         expected = {"key": "val", "list": ["1", 2, 3.0]}
         assert actual == expected
+
+    def test_load_yaml_from_ar_uri(self, mock_request_urlopen):
+        actual = yaml_utils.load_yaml(mock_request_urlopen)
+        expected = {"key": "val", "list": ["1", 2, 3.0]}
+        assert actual == expected
+
+    def test_load_yaml_from_invalid_uri(self):
+        with pytest.raises(FileNotFoundError):
+            yaml_utils.load_yaml("https://us-docker.pkg.dev/v2/proj/repo/img/tags/list")
