arkindex-base-worker 0.5.1b1__tar.gz → 0.5.1b2__tar.gz

{arkindex_base_worker-0.5.1b1 → arkindex_base_worker-0.5.1b2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: arkindex-base-worker
-Version: 0.5.1b1
+Version: 0.5.1b2
 Summary: Base Worker to easily build Arkindex ML workflows
 Author-email: Teklia <contact@teklia.com>
 Maintainer-email: Teklia <contact@teklia.com>

{arkindex_base_worker-0.5.1b1 → arkindex_base_worker-0.5.1b2}/arkindex_base_worker.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: arkindex-base-worker
-Version: 0.5.1b1
+Version: 0.5.1b2
 Summary: Base Worker to easily build Arkindex ML workflows
 Author-email: Teklia <contact@teklia.com>
 Maintainer-email: Teklia <contact@teklia.com>

{arkindex_base_worker-0.5.1b1 → arkindex_base_worker-0.5.1b2}/arkindex_base_worker.egg-info/SOURCES.txt

@@ -35,6 +35,7 @@ tests/test_dataset_worker.py
 tests/test_element.py
 tests/test_image.py
 tests/test_merge.py
+tests/test_modern_config.py
 tests/test_utils.py
 tests/test_elements_worker/__init__.py
 tests/test_elements_worker/test_classification.py

{arkindex_base_worker-0.5.1b1 → arkindex_base_worker-0.5.1b2}/arkindex_worker/worker/base.py

@@ -9,12 +9,13 @@ import os
 import shutil
 from pathlib import Path
 from tempfile import mkdtemp
+from typing import Any
 
 import gnupg
 import yaml
 
 from arkindex import options_from_env
-from arkindex.exceptions import ErrorResponse
+from arkindex.exceptions import ClientError, ErrorResponse
 from arkindex_worker import logger
 from arkindex_worker.cache import (
     check_version,
@@ -260,6 +261,48 @@ class BaseWorker:
 
         logger.info(f"Loaded {worker_run['summary']} from API")
 
+        def _process_config_item(item: dict) -> tuple[str, Any]:
+            if not item["secret"]:
+                return (item["key"], item["value"])
+
+            # Load secret, only available in Arkindex EE
+            try:
+                secret = self.load_secret(Path(item["value"]))
+            except ClientError as e:
+                logger.error(
+                    f"Failed to retrieve the secret {item['value']}, probably an Arkindex Community Edition: {e}"
+                )
+                return (item["key"], None)
+
+            return (item["key"], secret)
+
+        # Load worker run information
+        try:
+            config = self.api_client.request(
+                "RetrieveWorkerRunConfiguration", id=self.worker_run_id
+            )
+
+            # Provide the same configuration through all previous attributes
+            self.config = self.user_configuration = dict(
+                map(_process_config_item, config["configuration"])
+            )
+
+            # Provide secret values through the previous attribute
+            self.secrets = {
+                item["key"]: self.config[item["key"]]
+                for item in config["configuration"]
+                if item["secret"]
+            }
+            logger.info("Using modern configuration")
+
+            return  # Stop here once we have modern configuration
+
+        except ErrorResponse as e:
+            if e.status_code != 400:
+                raise
+            logger.info("Modern configuration is not available")
+
+        # Use old-style configuration with local merge
         # Load model version configuration when available
         model_version = worker_run.get("model_version")
         if model_version:

{arkindex_base_worker-0.5.1b1 → arkindex_base_worker-0.5.1b2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "arkindex-base-worker"
-version = "0.5.1b1"
+version = "0.5.1b2"
 description = "Base Worker to easily build Arkindex ML workflows"
 license = { file = "LICENSE" }
 dependencies = [

{arkindex_base_worker-0.5.1b1 → arkindex_base_worker-0.5.1b2}/tests/conftest.py

@@ -165,6 +165,13 @@ def _mock_worker_run_api(responses):
         content_type="application/json",
     )
 
+    # By default, stick to classic configuration
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+        status=400,
+    )
+
 
 @pytest.fixture
 def _mock_worker_run_no_revision_api(responses):
@@ -233,6 +240,56 @@ def _mock_worker_run_no_revision_api(responses):
     )
 
 
+@pytest.fixture
+def mock_base_worker_modern_conf(mocker, responses):
+    """
+    Provide a base worker to test modern configuration with (not provided in the fixture)
+    """
+    worker = BaseWorker()
+    mocker.patch.object(sys, "argv")
+    worker.args = worker.parser.parse_args()
+
+    payload = {
+        "id": "56785678-5678-5678-5678-567856785678",
+        "parents": [],
+        "worker_version": {
+            "id": "12341234-1234-1234-1234-123412341234",
+            "worker": {
+                "id": "deadbeef-1234-5678-1234-worker",
+                "name": "Fake worker",
+                "slug": "fake_worker",
+                "type": "classifier",
+            },
+            "revision": {"hash": "deadbeef1234"},
+            "configuration": {
+                "configuration": {"extra_key1": "not showing up"},
+                "user_configuration": {"extra_key2": "not showing up"},
+            },
+        },
+        "configuration": {
+            "id": "af0daaf4-983e-4703-a7ed-a10f146d6684",
+            "name": "my-userconfig",
+            "configuration": {
+                "extra_key3": "not showing up",
+            },
+        },
+        "model_version": None,
+        "process": {
+            "id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff",
+            "corpus": CORPUS_ID,
+        },
+        "summary": "Worker Fake worker @ 123412",
+    }
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/process/workers/56785678-5678-5678-5678-567856785678/",
+        status=200,
+        json=payload,
+    )
+
+    return worker
+
+
 @pytest.fixture
 def _mock_activity_calls(responses):
     """

{arkindex_base_worker-0.5.1b1 → arkindex_base_worker-0.5.1b2}/tests/test_base_worker.py

@@ -190,6 +190,14 @@ def test_configure_worker_run(mocker, responses, caplog):
         body=json.dumps(payload),
         content_type="application/json",
     )
+
+    # By default, stick to classic configuration
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+        status=400,
+    )
+
     worker.args = worker.parser.parse_args()
     assert worker.is_read_only is False
     assert worker.worker_run_id == "56785678-5678-5678-5678-567856785678"
@@ -205,6 +213,11 @@ def test_configure_worker_run(mocker, responses, caplog):
             logging.INFO,
             "Loaded Worker Fake worker @ 123412 from API",
         ),
+        (
+            "arkindex_worker",
+            logging.INFO,
+            "Modern configuration is not available",
+        ),
         ("arkindex_worker", logging.INFO, "Loaded user configuration from WorkerRun"),
         ("arkindex_worker", logging.INFO, "User configuration retrieved"),
     ]
@@ -213,9 +226,16 @@ def test_configure_worker_run(mocker, responses, caplog):
 
 
 @pytest.mark.usefixtures("_mock_worker_run_no_revision_api")
-def test_configure_worker_run_no_revision(mocker, caplog):
+def test_configure_worker_run_no_revision(mocker, caplog, responses):
     worker = BaseWorker()
 
+    # By default, stick to classic configuration
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+        status=400,
+    )
+
     mocker.patch.object(sys, "argv", ["worker"])
     worker.args = worker.parser.parse_args()
     assert worker.is_read_only is False
@@ -227,7 +247,12 @@ def test_configure_worker_run_no_revision(mocker, caplog):
         worker.configure()
 
     assert caplog.record_tuples == [
-        ("arkindex_worker", logging.INFO, "Loaded Worker Fake worker @ 1 from API")
+        ("arkindex_worker", logging.INFO, "Loaded Worker Fake worker @ 1 from API"),
+        (
+            "arkindex_worker",
+            logging.INFO,
+            "Modern configuration is not available",
+        ),
     ]
 
 
@@ -283,6 +308,13 @@ def test_configure_user_configuration_defaults(mocker, responses):
         content_type="application/json",
     )
 
+    # By default, stick to classic configuration
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+        status=400,
+    )
+
     worker.configure()
 
     assert worker.user_configuration == {
@@ -340,6 +372,13 @@ def test_configure_user_config_debug(mocker, responses, debug):
         body=json.dumps(payload),
         content_type="application/json",
     )
+
+    # By default, stick to classic configuration
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+        status=400,
+    )
     worker.args = worker.parser.parse_args()
     worker.configure()
 
@@ -388,6 +427,12 @@ def test_configure_worker_run_missing_conf(mocker, responses):
         body=json.dumps(payload),
         content_type="application/json",
     )
+    # By default, stick to classic configuration
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+        status=400,
+    )
     worker.args = worker.parser.parse_args()
     worker.configure()
 
@@ -430,6 +475,12 @@ def test_configure_worker_run_no_worker_run_conf(mocker, responses):
         body=json.dumps(payload),
         content_type="application/json",
     )
+    # By default, stick to classic configuration
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+        status=400,
+    )
     worker.args = worker.parser.parse_args()
     worker.configure()
 
@@ -480,6 +531,12 @@ def test_configure_load_model_configuration(mocker, responses):
         body=json.dumps(payload),
         content_type="application/json",
     )
+    # By default, stick to classic configuration
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+        status=400,
+    )
     worker.args = worker.parser.parse_args()
     assert worker.is_read_only is False
     assert worker.worker_run_id == "56785678-5678-5678-5678-567856785678"
@@ -947,6 +1004,13 @@ def test_worker_config_multiple_source(
         content_type="application/json",
     )
 
+    # By default, stick to classic configuration
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+        status=400,
+    )
+
     # Create and configure a worker
     monkeypatch.setattr(sys, "argv", ["worker"])
     worker = BaseWorker()

{arkindex_base_worker-0.5.1b1 → arkindex_base_worker-0.5.1b2}/tests/test_dataset_worker.py

@@ -430,6 +430,7 @@ def test_run_no_sets(mocker, caplog, mock_dataset_worker):
 
     assert [(level, message) for _, level, message in caplog.record_tuples] == [
         (logging.INFO, "Loaded Worker Fake worker @ 123412 from API"),
+        (logging.INFO, "Modern configuration is not available"),
         (logging.WARNING, "No sets to process, stopping."),
     ]
 
@@ -453,6 +454,7 @@ def test_run_initial_dataset_state_error(
 
     assert [(level, message) for _, level, message in caplog.record_tuples] == [
         (logging.INFO, "Loaded Worker Fake worker @ 123412 from API"),
+        (logging.INFO, "Modern configuration is not available"),
         (
             logging.WARNING,
             "Failed running worker on Set (train) from Dataset (dataset_id): AssertionError('When processing a set, its dataset state should be Complete.')",
@@ -497,6 +499,7 @@ def test_run_download_dataset_artifact_api_error(
 
     assert [(level, message) for _, level, message in caplog.record_tuples] == [
         (logging.INFO, "Loaded Worker Fake worker @ 123412 from API"),
+        (logging.INFO, "Modern configuration is not available"),
         (
             logging.INFO,
             "Retrieving data for Set (train) from Dataset (dataset_id) (1/1)",
@@ -550,6 +553,7 @@ def test_run_no_downloaded_dataset_artifact_error(
 
     assert [(level, message) for _, level, message in caplog.record_tuples] == [
         (logging.INFO, "Loaded Worker Fake worker @ 123412 from API"),
+        (logging.INFO, "Modern configuration is not available"),
         (
             logging.INFO,
             "Retrieving data for Set (train) from Dataset (dataset_id) (1/1)",
@@ -626,6 +630,7 @@ def test_run(
 
     assert [(level, message) for _, level, message in caplog.record_tuples] == [
         (logging.INFO, "Loaded Worker Fake worker @ 123412 from API"),
+        (logging.INFO, "Modern configuration is not available"),
         (
             logging.INFO,
             "Retrieving data for Set (train) from Dataset (dataset_id) (1/1)",

{arkindex_base_worker-0.5.1b1 → arkindex_base_worker-0.5.1b2}/tests/test_elements_worker/__init__.py

@@ -4,4 +4,8 @@ BASE_API_CALLS = [
         "GET",
         "http://testserver/api/v1/process/workers/56785678-5678-5678-5678-567856785678/",
     ),
+    (
+        "GET",
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+    ),
 ]

arkindex_base_worker-0.5.1b2/tests/test_modern_config.py

@@ -0,0 +1,81 @@
+def test_simple_configuration(mock_base_worker_modern_conf, responses):
+    # Provide the full configuration directly from the worker run
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+        status=200,
+        json={"configuration": [{"key": "some_key", "value": "test", "secret": False}]},
+    )
+
+    mock_base_worker_modern_conf.configure()
+
+    assert mock_base_worker_modern_conf.config == {"some_key": "test"}
+    assert (
+        mock_base_worker_modern_conf.user_configuration
+        == mock_base_worker_modern_conf.config
+    )
+    assert mock_base_worker_modern_conf.secrets == {}
+
+
+def test_empty(mock_base_worker_modern_conf, responses):
+    # Provide the full configuration directly from the worker run
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+        status=200,
+        json={"configuration": []},
+    )
+
+    mock_base_worker_modern_conf.configure()
+
+    assert mock_base_worker_modern_conf.config == {}
+    assert (
+        mock_base_worker_modern_conf.user_configuration
+        == mock_base_worker_modern_conf.config
+    )
+    assert mock_base_worker_modern_conf.secrets == {}
+
+
+def test_with_secrets(mock_base_worker_modern_conf, responses):
+    # Provide the full configuration directly from the worker run
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/workers/runs/56785678-5678-5678-5678-567856785678/configuration/",
+        status=200,
+        json={
+            "configuration": [
+                {"key": "some_key", "value": "test", "secret": False},
+                {
+                    "key": "a_secret",
+                    "value": "471b9e64-29af-48dc-8bda-1a64a2da0c12",
+                    "secret": True,
+                },
+            ]
+        },
+    )
+
+    # Provide a secret value
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/secret/471b9e64-29af-48dc-8bda-1a64a2da0c12",
+        status=200,
+        json={
+            "id": "471b9e64-29af-48dc-8bda-1a64a2da0c12",
+            "name": "a_secret",
+            "content": "My super duper secret value",
+        },
+    )
+
+    mock_base_worker_modern_conf.configure()
+
+    assert mock_base_worker_modern_conf.config == {
+        "a_secret": "My super duper secret value",
+        "some_key": "test",
+    }
+    assert (
+        mock_base_worker_modern_conf.user_configuration
+        == mock_base_worker_modern_conf.config
+    )
+    assert mock_base_worker_modern_conf.secrets == {
+        "a_secret": "My super duper secret value"
+    }