arkindex-base-worker 0.5.1b1__py3-none-any.whl → 0.5.1b4__py3-none-any.whl

{arkindex_base_worker-0.5.1b1.dist-info → arkindex_base_worker-0.5.1b4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: arkindex-base-worker
-Version: 0.5.1b1
+Version: 0.5.1b4
 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.dist-info → arkindex_base_worker-0.5.1b4.dist-info}/RECORD

@@ -1,15 +1,15 @@
-arkindex_base_worker-0.5.1b1.dist-info/licenses/LICENSE,sha256=NVshRi1efwVezMfW7xXYLrdDr2Li1AfwfGOd5WuH1kQ,1063
+arkindex_base_worker-0.5.1b4.dist-info/licenses/LICENSE,sha256=NVshRi1efwVezMfW7xXYLrdDr2Li1AfwfGOd5WuH1kQ,1063
 arkindex_worker/__init__.py,sha256=Sdt5KXn8EgURb2MurYVrUWaHbH3iFA1XLRo0Lc5AJ44,250
 arkindex_worker/cache.py,sha256=x1d1oVF297ItLoZnPkZQoEefa39ZigrwRoHC_6az94k,10731
 arkindex_worker/image.py,sha256=GvIpW7LNSalVw3Obt9nySDWnW7-NbC0__SWREEQqVCk,20696
 arkindex_worker/models.py,sha256=bPQzGZNs5a6z6DEcygsa8T33VOqPlMUbwKzHqlKzwbw,9923
 arkindex_worker/utils.py,sha256=MbbJT8oh8DMHHR-vidFeXdUH0TSXGWm7ZDGWzrRXoEY,9933
-arkindex_worker/worker/__init__.py,sha256=Ho4rzGDtpFwd13cl5tL45yAVZJuyGCOIOroLPGZsvmk,18399
-arkindex_worker/worker/base.py,sha256=03WLu7R8vca54LI00g-S0EqIbFiNaIsgZjN6zmfHisw,20126
+arkindex_worker/worker/__init__.py,sha256=SzD0s1_m6gMV02EUF-NeciqZdVPA4dpXI84tSj-g494,17869
+arkindex_worker/worker/base.py,sha256=fbRJ5vDON3DfSQfwxFqto85HY8Dw2_YgOmnm5cxbQ2g,21725
 arkindex_worker/worker/classification.py,sha256=qvykymkgd4nGywHCxL8obo4egstoGsmWNS4Ztc1qNWQ,11024
 arkindex_worker/worker/corpus.py,sha256=MeIMod7jkWyX0frtD0a37rhumnMV3p9ZOC1xwAoXrAA,2291
 arkindex_worker/worker/dataset.py,sha256=tVaPx43vaH-KTtx4w5V06e26ha8XPfiJTRzBXlu928Y,5273
-arkindex_worker/worker/element.py,sha256=YUYKPhWrt2J5bfCA58slhI9CVIBmgIdcgC6-hhR9Zwg,46942
+arkindex_worker/worker/element.py,sha256=sLfnf09AfJ5tSCKQ7cAkl7WsGhjsfq14swsT30MDnYk,47385
 arkindex_worker/worker/entity.py,sha256=Aj6EOfzHEm7qQV-Egm0YKLZgCrLS_3ggOKTY81M2JbI,12323
 arkindex_worker/worker/image.py,sha256=L6Ikuf0Z0RxJk7JarY5PggJGrYSHLaPK0vn0dy0CIaQ,623
 arkindex_worker/worker/metadata.py,sha256=rBjU057xngwrf32vAo-2cpgYfmrdEj3lfDg_kv4-zr0,6810
@@ -21,15 +21,16 @@ examples/standalone/python/worker.py,sha256=Zr4s4pHvgexEjlkixLFYZp1UuwMLeoTxjyNG
 examples/tooled/python/worker.py,sha256=kIYlHLsO5UpwX4XtERRq4tf2qTsvqKK30C-w8t0yyhA,1821
 hooks/pre_gen_project.py,sha256=xQJERv3vv9VzIqcBHI281eeWLWREXUF4mMw7PvJHHXM,269
 tests/__init__.py,sha256=DG--S6IpGl399rzSAjDdHL76CkOIeZIjajCcyUSDhOQ,241
-tests/conftest.py,sha256=NwYfoWO5ZTRfK5REYLGExt64CS1OPTcTpJ4LfbvuouE,22048
-tests/test_base_worker.py,sha256=dA00oxauTSCwnFX3ZFBl-RI71HN6GmK48FBBW_oYN-k,30627
+tests/conftest.py,sha256=OaajGB0FS0Jh3rFhL8a3HdaN84XzIxNbRORvzI5W2ns,24928
+tests/test_base_worker.py,sha256=3YjhjxSWVjEWFYS8m8pYYoaVAhHFkJLNTs0QPQIkBDM,32651
 tests/test_cache.py,sha256=nnEFfAAqtYHk2ymOwN0spXJd8nrRiwp3voj0tOmIbQ8,10407
-tests/test_dataset_worker.py,sha256=z8ydliUlwW2j-irgLAotJMacgJXkVvF5TgsWLyCn1Jo,22087
+tests/test_dataset_worker.py,sha256=iDJM2C4PfQNH0r4_QqSWoPt8BcM0geUUdODtWY0Z9PA,22412
 tests/test_element.py,sha256=2G9M15TLxQRmvrWM9Kw2ucnElh4kSv_oF_5FYwwAxTY,13181
 tests/test_image.py,sha256=NEIp5evr6QoTWgJ-_fze19IEFm_hG6YEcuW1kxnxS_I,28013
 tests/test_merge.py,sha256=REpZ13jkq_qm_4L5URQgFy5lxvPZtXxQEiWfYLMdmF0,7956
+tests/test_modern_config.py,sha256=Bm-a4LYQXgLZWQX7AmVyfJW0LNoLy1wj2d2GjzDkcBk,2683
 tests/test_utils.py,sha256=nYL1s2ViZoLoMiNpLGDaWwxf8dJ1D8aT522AO-PVaEQ,3607
-tests/test_elements_worker/__init__.py,sha256=Fh4nkbbyJSMv_VtjQxnWrOqTnxXaaWI8S9WU0VrzCHs,179
+tests/test_elements_worker/__init__.py,sha256=2t3NciCIOun_N-Wv63FWGsTm5W9N3mbwAWVuFORlMg8,308
 tests/test_elements_worker/test_classification.py,sha256=nya7veSPR_O9G41Enodp2-o6AifMBcaSTWJP2vXSSJ4,30133
 tests/test_elements_worker/test_cli.py,sha256=a23i1pUDbXi23MUtbWwGEcLLrmc_YlrbDgOG3h66wLM,2620
 tests/test_elements_worker/test_corpus.py,sha256=kscJyM8k1njYJJFGuvliVzn89lWh41mEyDCCawnp3W8,5483
@@ -48,13 +49,13 @@ tests/test_elements_worker/test_training.py,sha256=qgK7BLucddRzc8ePbQtY75x17QvGD
 tests/test_elements_worker/test_transcription_create.py,sha256=yznO9B_BVsOR0Z_VY5ZL8gJp0ZPCz_4sPUs5dXtixAg,29281
 tests/test_elements_worker/test_transcription_create_with_elements.py,sha256=tmcyglgssEqMnt1Mdy_u6X1m2wgLWTo_HdWst3GrK2k,33056
 tests/test_elements_worker/test_transcription_list.py,sha256=ikz7HYPCoQWTdTRCd382SB-y-T2BbigPLlIcx5Eow-I,15324
-tests/test_elements_worker/test_worker.py,sha256=tFxLZhX83Z-g2pZyZ2vRzTh_yBhHjpBnZ9kdYxlOt4g,28272
+tests/test_elements_worker/test_worker.py,sha256=ypAQS_DJj9qGlRJCs9g5qUXe7IgqaKXWDcxqwlhAqSg,28598
 worker-demo/tests/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 worker-demo/tests/conftest.py,sha256=XzNMNeg6pmABUAH8jN6eZTlZSFGLYjS3-DTXjiRN6Yc,1002
 worker-demo/tests/test_worker.py,sha256=3DLd4NRK4bfyatG5P_PK4k9P9tJHx9XQq5_ryFEEFVg,304
 worker-demo/worker_demo/__init__.py,sha256=2BPomV8ZMNf3YXJgloatKeHQCE6QOkwmsHGkO6MkQuM,125
 worker-demo/worker_demo/worker.py,sha256=Rt-DjWa5iBP08k58NDZMfeyPuFbtNcbX6nc5jFX7GNo,440
-arkindex_base_worker-0.5.1b1.dist-info/METADATA,sha256=LZx_PDuHE_v64WNubUcSPXN2FCWBwnHxEIxogiprVtU,3137
-arkindex_base_worker-0.5.1b1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-arkindex_base_worker-0.5.1b1.dist-info/top_level.txt,sha256=-vNjP2VfROx0j83mdi9aIqRZ88eoJjxeWz-R_gPgyXU,49
-arkindex_base_worker-0.5.1b1.dist-info/RECORD,,
+arkindex_base_worker-0.5.1b4.dist-info/METADATA,sha256=hLR30IODhVUqKVq_4qXv_AcNswb0afCA-2Nce13UPno,3137
+arkindex_base_worker-0.5.1b4.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+arkindex_base_worker-0.5.1b4.dist-info/top_level.txt,sha256=-vNjP2VfROx0j83mdi9aIqRZ88eoJjxeWz-R_gPgyXU,49
+arkindex_base_worker-0.5.1b4.dist-info/RECORD,,

arkindex_worker/worker/__init__.py

@@ -33,13 +33,10 @@ from arkindex_worker.worker.transcription import TranscriptionMixin
 
 
 class WorkerActivityIterator:
-    def __init__(self, api_client, types):
+    def __init__(self, api_client):
         # Use same api client as main class
         self.api_client = api_client
 
-        # Index element types by ID
-        self.types = {t["id"]: t["slug"] for t in types}
-
         logger.info(
             "Using StartWorkerActivity instead of reading init_elements JSON file"
         )
@@ -53,7 +50,7 @@ class WorkerActivityIterator:
 
     def __next__(self):
         """
-        Provide a new element from a worker activity upon each iteration
+        Provide a new element ID from a worker activity upon each iteration
         """
         try:
             data = self.api_client.request("StartWorkerActivity")
@@ -67,12 +64,7 @@ class WorkerActivityIterator:
             )
             raise e
 
-        # Find the type slug using API provided type_id
-        type_id = data["type_id"]
-        if type_id not in self.types:
-            raise Exception(f"Unknown type {type_id}")
-
-        return Element(type=self.types[type_id], **data)
+        return data["id"]
 
 
 class ElementsWorker(
@@ -154,14 +146,9 @@ class ElementsWorker(
         elif self.process_mode == ProcessMode.Export:
             # For export mode processes, use list_process_elements and return element IDs
             return {item["id"] for item in self.list_process_elements()}
-        elif self.args.consume_worker_activities:
-            # We need to list corpus types as the StartWorkerActivity endpoint only provide type_id
-            self.list_corpus_types()
-
+        elif self.consume_worker_activities:
             # Consume worker activitives one by one
-            return WorkerActivityIterator(
-                self.api_client, types=self.corpus_types.values()
-            )
+            return WorkerActivityIterator(self.api_client)
 
         invalid_element_ids = list(filter(invalid_element_id, out))
         assert not invalid_element_ids, (
@@ -195,7 +182,7 @@ class ElementsWorker(
          - when running with init_elements, we have a known list
          - when running with StartWorkerActivity, we have a queue of unknown size
         """
-        return self.args.consume_worker_activities
+        return self.consume_worker_activities
 
     def run(self):
         """
@@ -241,7 +228,7 @@ class ElementsWorker(
 
                 # Process the element and report its progress if activities are enabled
                 # We do not update the worker activity to "Started" state when consuming them
-                if self.args.consume_worker_activities or self.update_activity(
+                if self.consume_worker_activities or self.update_activity(
                     element.id, ActivityState.Started
                 ):
                     self.process_element(element)

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_worker/worker/element.py

@@ -55,6 +55,17 @@ class ElementMixin:
         )
         super().add_arguments()
 
+    @property
+    def consume_worker_activities(self) -> bool:
+        """
+        Helper to detect if the worker rely on an elements.json or consume directly worker activities
+        Uses the process information when available, fallback to CLI args
+        """
+        if self.process_information is not None:
+            return self.process_information.get("skip_elements_json") is True
+
+        return self.args.consume_worker_activities
+
     def list_corpus_types(self):
         """
         Loads available element types in corpus.

tests/conftest.py

@@ -153,6 +153,7 @@ def _mock_worker_run_api(responses):
             "train_folder_id": None,
             "validation_folder_id": None,
             "test_folder_id": None,
+            "skip_elements_json": False,
         },
         "summary": "Worker Fake worker @ 123412",
     }
@@ -165,6 +166,13 @@ 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 +241,56 @@ 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):
     """
@@ -289,29 +347,57 @@ def mock_elements_worker_consume_wa(monkeypatch, responses, mock_elements_worker
     instead of reading a JSON file
     """
 
-    # Enable consume worker activities mode from CLI args
-    mock_elements_worker.args.consume_worker_activities = True
-
-    # Worker requires element types from corpus details as they are not provided by StartWorkerActivity
-    responses.add(
+    # Enable consume worker activities through the process configuration
+    responses.replace(
         responses.GET,
-        "http://testserver/api/v1/corpus/11111111-1111-1111-1111-111111111111/",
+        "http://testserver/api/v1/process/workers/56785678-5678-5678-5678-567856785678/",
         status=200,
         json={
-            "id": "11111111-1111-1111-1111-111111111111",
-            "name": "Test corpus",
-            "types": [
-                {
-                    "id": "page-aaaa-aaaa-aaaa-aaaaaaaaaaaa",
-                    "slug": "page",
-                    "display_name": "Page",
-                    "folder": False,
-                    "color": "28b62c",
-                }
-            ],
+            "id": "56785678-5678-5678-5678-567856785678",
+            "parents": [],
+            "worker": {
+                "id": "deadbeef-1234-5678-1234-worker",
+                "name": "Fake worker",
+                "slug": "fake_worker",
+                "type": "classifier",
+            },
+            "worker_version": {
+                "id": "12341234-1234-1234-1234-123412341234",
+                "configuration": {
+                    "docker": {"image": "python:3"},
+                    "configuration": {"someKey": "someValue"},
+                    "secrets": [],
+                },
+                "worker": {
+                    "id": "deadbeef-1234-5678-1234-worker",
+                    "name": "Fake worker",
+                    "slug": "fake_worker",
+                    "type": "classifier",
+                },
+            },
+            "configuration": None,
+            "model_version": None,
+            "process": {
+                "name": None,
+                "id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff",
+                "state": "running",
+                "mode": "workers",
+                "corpus": CORPUS_ID,
+                "use_cache": False,
+                "activity_state": "ready",
+                "model_id": None,
+                "train_folder_id": None,
+                "validation_folder_id": None,
+                "test_folder_id": None,
+                "skip_elements_json": True,
+            },
+            "summary": "Worker Fake worker @ 123412",
         },
     )
 
+    # Call configure again to use updated process infos
+    mock_elements_worker.configure()
+
     return mock_elements_worker
 
 

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()

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)",

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/",
+    ),
 ]

tests/test_elements_worker/test_worker.py

@@ -698,7 +698,8 @@ def test_run_consuming_worker_activities(
     assert mock_elements_worker_consume_wa.is_read_only is False
 
     # Provide 2 worker activities to run and the corresponding update call
-    for elt_id in ("page_1", "page_2"):
+    # and 2 element details response
+    for i, elt_id in enumerate(("page_1", "page_2"), 1):
         responses.add(
             responses.POST,
             "http://testserver/api/v1/process/start-activity/",
@@ -706,7 +707,7 @@ def test_run_consuming_worker_activities(
             json={
                 "id": elt_id,
                 "type_id": "page-aaaa-aaaa-aaaa-aaaaaaaaaaaa",  # Element type provided by mock corpus
-                "name": "Page XXX",
+                "name": f"Page n°{i}",
             },
         )
         responses.add(
@@ -714,6 +715,16 @@ def test_run_consuming_worker_activities(
             "http://testserver/api/v1/workers/versions/56785678-5678-5678-5678-567856785678/activity/",
             status=200,
         )
+        responses.add(
+            responses.GET,
+            f"http://testserver/api/v1/element/{elt_id}/",
+            status=200,
+            json={
+                "id": elt_id,
+                "type": "page",
+                "name": f"Page n°{i}",
+            },
+        )
 
     # Then a 404 to stop iterating
     responses.add(
@@ -725,18 +736,19 @@ def test_run_consuming_worker_activities(
     # Simply run the process
     mock_elements_worker_consume_wa.run()
 
-    assert len(responses.calls) == len(BASE_API_CALLS) + 6
+    # We call twice configure in the conftest
+    assert len(responses.calls) == len(BASE_API_CALLS) * 2 + 7
     assert [
         (call.request.method, call.request.url) for call in responses.calls
-    ] == BASE_API_CALLS + [
-        (
-            "GET",
-            "http://testserver/api/v1/corpus/11111111-1111-1111-1111-111111111111/",
-        ),
+    ] == BASE_API_CALLS * 2 + [
         (
             "POST",
             "http://testserver/api/v1/process/start-activity/",
         ),
+        (
+            "GET",
+            "http://testserver/api/v1/element/page_1/",
+        ),
         (
             "PUT",
             "http://testserver/api/v1/workers/versions/56785678-5678-5678-5678-567856785678/activity/",
@@ -745,6 +757,10 @@ def test_run_consuming_worker_activities(
             "POST",
             "http://testserver/api/v1/process/start-activity/",
         ),
+        (
+            "GET",
+            "http://testserver/api/v1/element/page_2/",
+        ),
         (
             "PUT",
             "http://testserver/api/v1/workers/versions/56785678-5678-5678-5678-567856785678/activity/",
@@ -756,21 +772,17 @@ def test_run_consuming_worker_activities(
     ]
 
     assert [(record.levelno, record.message) for record in caplog.records] == [
-        (
-            logging.INFO,
-            "Loaded 1 element type in corpus (11111111-1111-1111-1111-111111111111).",
-        ),
         (
             logging.INFO,
             "Using StartWorkerActivity instead of reading init_elements JSON file",
         ),
         (
             logging.INFO,
-            "Processing page Page XXX (page_1) (n°1)",
+            "Processing page Page n°1 (page_1) (n°1)",
         ),
         (
             logging.INFO,
-            "Processing page Page XXX (page_2) (n°2)",
+            "Processing page Page n°2 (page_2) (n°2)",
         ),
         (
             logging.INFO,

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"
+    }