tests/test_base_worker.py

@@ -86,7 +86,6 @@ def test_cli_default(mocker):
     assert logger.level == logging.NOTSET
     assert worker.api_client
     assert worker.config == {"someKey": "someValue"}  # from API
-    assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
 
     logger.setLevel(logging.NOTSET)
 
@@ -106,7 +105,6 @@ def test_cli_arg_verbose_given(mocker):
     assert logger.level == logging.DEBUG
     assert worker.api_client
     assert worker.config == {"someKey": "someValue"}  # from API
-    assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
 
     logger.setLevel(logging.NOTSET)
 
@@ -126,7 +124,6 @@ def test_cli_envvar_debug_given(mocker, monkeypatch):
     assert logger.level == logging.DEBUG
     assert worker.api_client
     assert worker.config == {"someKey": "someValue"}  # from API
-    assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
 
     logger.setLevel(logging.NOTSET)
 
@@ -142,7 +139,6 @@ def test_configure_dev_mode(mocker):
 
     assert worker.args.dev is True
     assert worker.process_information is None
-    assert worker.worker_run_id == "56785678-5678-5678-5678-567856785678"
     assert worker.is_read_only is True
     assert worker.user_configuration == {}
 
@@ -212,7 +208,6 @@ def test_configure_worker_run(mocker, responses, caplog):
         ("arkindex_worker", logging.INFO, "Loaded user configuration from WorkerRun"),
     ]
 
-    assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
     assert worker.user_configuration == {"a": "b"}
 
 
@@ -482,7 +477,6 @@ def test_configure_load_model_configuration(mocker, responses):
 
     worker.configure()
 
-    assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
     assert worker.model_configuration == {
         "param1": "value1",
         "param2": 2,

tests/test_dataset_worker.py

@@ -430,7 +430,7 @@ def test_run_no_datasets(mocker, caplog, mock_dataset_worker, generator):
 @pytest.mark.parametrize(
     ("generator", "error"),
     [
-        (True, "When generating a new dataset, its state should be Open."),
+        (True, "When generating a new dataset, its state should be Open or Error."),
         (False, "When processing an existing dataset, its state should be Complete."),
     ],
 )
@@ -470,10 +470,17 @@ def test_run_initial_dataset_state_error(
             logging.WARNING,
             f"Failed running worker on dataset dataset_id: AssertionError('{error}')",
         ),
-        (
-            logging.ERROR,
-            "Ran on 1 datasets: 0 completed, 1 failed",
-        ),
+    ] + (
+        [
+            (
+                logging.WARNING,
+                "This API helper `update_dataset_state` did not update the cache database",
+            )
+        ]
+        if generator
+        else []
+    ) + [
+        (logging.ERROR, "Ran on 1 datasets: 0 completed, 1 failed"),
     ]
 
 
@@ -697,7 +704,15 @@ def test_run(
         ] * 2
         extra_logs += [
             (logging.INFO, "Building Dataset (dataset_id) (1/1)"),
+            (
+                logging.WARNING,
+                "This API helper `update_dataset_state` did not update the cache database",
+            ),
             (logging.INFO, "Completed Dataset (dataset_id) (1/1)"),
+            (
+                logging.WARNING,
+                "This API helper `update_dataset_state` did not update the cache database",
+            ),
         ]
     else:
         archive_path = (

tests/test_elements_worker/test_cli.py

@@ -24,7 +24,7 @@ def test_cli_default(monkeypatch):
         )
     )
 
-    monkeypatch.setenv("TASK_ELEMENTS", path)
+    monkeypatch.setenv("TASK_ELEMENTS", str(path))
     monkeypatch.setattr(sys, "argv", ["worker"])
     worker = ElementsWorker()
     worker.configure()

tests/test_elements_worker/test_elements.py

@@ -1770,13 +1770,11 @@ def test_list_element_children_wrong_with_metadata(mock_elements_worker):
 @pytest.mark.parametrize(
     ("param", "value"),
     [
-        ("worker_version", 1234),
         ("worker_run", 1234),
-        ("transcription_worker_version", 1234),
         ("transcription_worker_run", 1234),
     ],
 )
-def test_list_element_children_wrong_worker_version(mock_elements_worker, param, value):
+def test_list_element_children_wrong_worker_run(mock_elements_worker, param, value):
     elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
 
     with pytest.raises(AssertionError, match=f"{param} should be of type str or bool"):
@@ -1786,16 +1784,36 @@ def test_list_element_children_wrong_worker_version(mock_elements_worker, param,
         )
 
 
+@pytest.mark.parametrize(
+    ("param", "alternative", "value"),
+    [
+        ("worker_version", "worker_run", 1234),
+        ("transcription_worker_version", "transcription_worker_run", 1234),
+    ],
+)
+def test_list_element_children_wrong_worker_version(
+    mock_elements_worker, param, alternative, value
+):
+    elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+    # WARNING: pytest.deprecated_call must be placed BEFORE pytest.raises, otherwise `match` argument won't be checked
+    with pytest.deprecated_call(
+        match=f"`{param}` usage is deprecated. Consider using `{alternative}` instead."
+    ), pytest.raises(AssertionError, match=f"{param} should be of type str or bool"):
+        mock_elements_worker.list_element_children(
+            element=elt,
+            **{param: value},
+        )
+
+
 @pytest.mark.parametrize(
     "param",
     [
-        "worker_version",
         "worker_run",
-        "transcription_worker_version",
         "transcription_worker_run",
     ],
 )
-def test_list_element_children_wrong_bool_worker_version(mock_elements_worker, param):
+def test_list_element_children_wrong_bool_worker_run(mock_elements_worker, param):
     elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
 
     with pytest.raises(
@@ -1807,6 +1825,30 @@ def test_list_element_children_wrong_bool_worker_version(mock_elements_worker, p
         )
 
 
+@pytest.mark.parametrize(
+    ("param", "alternative"),
+    [
+        ("worker_version", "worker_run"),
+        ("transcription_worker_version", "transcription_worker_run"),
+    ],
+)
+def test_list_element_children_wrong_bool_worker_version(
+    mock_elements_worker, param, alternative
+):
+    elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+    # WARNING: pytest.deprecated_call must be placed BEFORE pytest.raises, otherwise `match` argument won't be checked
+    with pytest.deprecated_call(
+        match=f"`{param}` usage is deprecated. Consider using `{alternative}` instead."
+    ), pytest.raises(
+        AssertionError, match=f"if of type bool, {param} can only be set to False"
+    ):
+        mock_elements_worker.list_element_children(
+            element=elt,
+            **{param: True},
+        )
+
+
 def test_list_element_children_api_error(responses, mock_elements_worker):
     elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
     responses.add(
@@ -1942,10 +1984,15 @@ def test_list_element_children_manual_worker_version(responses, mock_elements_wo
         },
     )
 
-    for idx, child in enumerate(
-        mock_elements_worker.list_element_children(element=elt, worker_version=False)
+    with pytest.deprecated_call(
+        match="`worker_version` usage is deprecated. Consider using `worker_run` instead."
     ):
-        assert child == expected_children[idx]
+        for idx, child in enumerate(
+            mock_elements_worker.list_element_children(
+                element=elt, worker_version=False
+            )
+        ):
+            assert child == expected_children[idx]
 
     assert len(responses.calls) == len(BASE_API_CALLS) + 1
     assert [
@@ -2038,56 +2085,84 @@ def test_list_element_children_with_cache_unhandled_param(
             },
             ("22222222-2222-2222-2222-222222222222",),
         ),
-        # Filter on element and worker version should give first two elements
+        # Filter on element and worker run should give second
         (
             {
                 "element": CachedElement(id="12341234-1234-1234-1234-123412341234"),
-                "worker_version": "56785678-5678-5678-5678-567856785678",
+                "worker_run": "56785678-5678-5678-5678-567856785678",
             },
-            (
-                "11111111-1111-1111-1111-111111111111",
-                "22222222-2222-2222-2222-222222222222",
-            ),
+            ("22222222-2222-2222-2222-222222222222",),
         ),
-        # Filter on element, type something and worker version should give first
+        # Filter on element, manual worker run should give first and third
         (
             {
                 "element": CachedElement(id="12341234-1234-1234-1234-123412341234"),
-                "type": "something",
-                "worker_version": "56785678-5678-5678-5678-567856785678",
+                "worker_run": False,
             },
-            ("11111111-1111-1111-1111-111111111111",),
+            (
+                "11111111-1111-1111-1111-111111111111",
+                "33333333-3333-3333-3333-333333333333",
+            ),
         ),
-        # Filter on element, manual worker version should give third
+    ],
+)
+def test_list_element_children_with_cache(
+    responses,
+    mock_elements_worker_with_cache,
+    filters,
+    expected_ids,
+):
+    # Check we have 5 elements already present in database
+    assert CachedElement.select().count() == 5
+
+    # Query database through cache
+    elements = mock_elements_worker_with_cache.list_element_children(**filters)
+    assert elements.count() == len(expected_ids)
+    for child, expected_id in zip(elements.order_by("id"), expected_ids, strict=True):
+        assert child.id == UUID(expected_id)
+
+    # Check the worker never hits the API for elements
+    assert len(responses.calls) == len(BASE_API_CALLS)
+    assert [
+        (call.request.method, call.request.url) for call in responses.calls
+    ] == BASE_API_CALLS
+
+
+@pytest.mark.usefixtures("_mock_cached_elements")
+@pytest.mark.parametrize(
+    ("filters", "expected_ids"),
+    [
+        # Filter on element and worker version
         (
             {
                 "element": CachedElement(id="12341234-1234-1234-1234-123412341234"),
-                "worker_version": False,
+                "worker_version": "56785678-5678-5678-5678-567856785678",
             },
-            ("33333333-3333-3333-3333-333333333333",),
+            (
+                "11111111-1111-1111-1111-111111111111",
+                "22222222-2222-2222-2222-222222222222",
+            ),
         ),
-        # Filter on element and worker run should give second
+        # Filter on element, type double_page and worker version
         (
             {
                 "element": CachedElement(id="12341234-1234-1234-1234-123412341234"),
-                "worker_run": "56785678-5678-5678-5678-567856785678",
+                "type": "page",
+                "worker_version": "56785678-5678-5678-5678-567856785678",
             },
             ("22222222-2222-2222-2222-222222222222",),
         ),
-        # Filter on element, manual worker run should give first and third
+        # Filter on element, manual worker version
         (
             {
                 "element": CachedElement(id="12341234-1234-1234-1234-123412341234"),
-                "worker_run": False,
+                "worker_version": False,
             },
-            (
-                "11111111-1111-1111-1111-111111111111",
-                "33333333-3333-3333-3333-333333333333",
-            ),
+            ("33333333-3333-3333-3333-333333333333",),
         ),
     ],
 )
-def test_list_element_children_with_cache(
+def test_list_element_children_with_cache_deprecation(
     responses,
     mock_elements_worker_with_cache,
     filters,
@@ -2096,8 +2171,11 @@ def test_list_element_children_with_cache(
     # Check we have 5 elements already present in database
     assert CachedElement.select().count() == 5
 
-    # Query database through cache
-    elements = mock_elements_worker_with_cache.list_element_children(**filters)
+    with pytest.deprecated_call(
+        match="`worker_version` usage is deprecated. Consider using `worker_run` instead."
+    ):
+        # Query database through cache
+        elements = mock_elements_worker_with_cache.list_element_children(**filters)
     assert elements.count() == len(expected_ids)
     for child, expected_id in zip(elements.order_by("id"), expected_ids, strict=True):
         assert child.id == UUID(expected_id)
@@ -2218,13 +2296,11 @@ def test_list_element_parents_wrong_with_metadata(mock_elements_worker):
 @pytest.mark.parametrize(
     ("param", "value"),
     [
-        ("worker_version", 1234),
         ("worker_run", 1234),
-        ("transcription_worker_version", 1234),
         ("transcription_worker_run", 1234),
     ],
 )
-def test_list_element_parents_wrong_worker_version(mock_elements_worker, param, value):
+def test_list_element_parents_wrong_worker_run(mock_elements_worker, param, value):
     elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
 
     with pytest.raises(AssertionError, match=f"{param} should be of type str or bool"):
@@ -2234,16 +2310,36 @@ def test_list_element_parents_wrong_worker_version(mock_elements_worker, param,
         )
 
 
+@pytest.mark.parametrize(
+    ("param", "alternative", "value"),
+    [
+        ("worker_version", "worker_run", 1234),
+        ("transcription_worker_version", "transcription_worker_run", 1234),
+    ],
+)
+def test_list_element_parents_wrong_worker_version(
+    mock_elements_worker, param, alternative, value
+):
+    elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+    # WARNING: pytest.deprecated_call must be placed BEFORE pytest.raises, otherwise `match` argument won't be checked
+    with pytest.deprecated_call(
+        match=f"`{param}` usage is deprecated. Consider using `{alternative}` instead."
+    ), pytest.raises(AssertionError, match=f"{param} should be of type str or bool"):
+        mock_elements_worker.list_element_parents(
+            element=elt,
+            **{param: value},
+        )
+
+
 @pytest.mark.parametrize(
     "param",
     [
-        "worker_version",
         "worker_run",
-        "transcription_worker_version",
         "transcription_worker_run",
     ],
 )
-def test_list_element_parents_wrong_bool_worker_version(mock_elements_worker, param):
+def test_list_element_parents_wrong_bool_worker_run(mock_elements_worker, param):
     elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
 
     with pytest.raises(
@@ -2255,6 +2351,30 @@ def test_list_element_parents_wrong_bool_worker_version(mock_elements_worker, pa
         )
 
 
+@pytest.mark.parametrize(
+    ("param", "alternative"),
+    [
+        ("worker_version", "worker_run"),
+        ("transcription_worker_version", "transcription_worker_run"),
+    ],
+)
+def test_list_element_parents_wrong_bool_worker_version(
+    mock_elements_worker, param, alternative
+):
+    elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+    # WARNING: pytest.deprecated_call must be placed BEFORE pytest.raises, otherwise `match` argument won't be checked
+    with pytest.deprecated_call(
+        match=f"`{param}` usage is deprecated. Consider using `{alternative}` instead."
+    ), pytest.raises(
+        AssertionError, match=f"if of type bool, {param} can only be set to False"
+    ):
+        mock_elements_worker.list_element_parents(
+            element=elt,
+            **{param: True},
+        )
+
+
 def test_list_element_parents_api_error(responses, mock_elements_worker):
     elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
     responses.add(
@@ -2390,10 +2510,13 @@ def test_list_element_parents_manual_worker_version(responses, mock_elements_wor
         },
     )
 
-    for idx, parent in enumerate(
-        mock_elements_worker.list_element_parents(element=elt, worker_version=False)
+    with pytest.deprecated_call(
+        match="`worker_version` usage is deprecated. Consider using `worker_run` instead."
     ):
-        assert parent == expected_parents[idx]
+        for idx, parent in enumerate(
+            mock_elements_worker.list_element_parents(element=elt, worker_version=False)
+        ):
+            assert parent == expected_parents[idx]
 
     assert len(responses.calls) == len(BASE_API_CALLS) + 1
     assert [
@@ -2482,6 +2605,50 @@ def test_list_element_parents_with_cache_unhandled_param(
             },
             "12341234-1234-1234-1234-123412341234",
         ),
+        # Filter on element and worker run
+        (
+            {
+                "element": CachedElement(id="22222222-2222-2222-2222-222222222222"),
+                "worker_run": "56785678-5678-5678-5678-567856785678",
+            },
+            "12341234-1234-1234-1234-123412341234",
+        ),
+        # Filter on element, manual worker run
+        (
+            {
+                "element": CachedElement(id="12341234-1234-1234-1234-123412341234"),
+                "worker_run": False,
+            },
+            "99999999-9999-9999-9999-999999999999",
+        ),
+    ],
+)
+def test_list_element_parents_with_cache(
+    responses,
+    mock_elements_worker_with_cache,
+    filters,
+    expected_id,
+):
+    # Check we have 5 elements already present in database
+    assert CachedElement.select().count() == 5
+
+    # Query database through cache
+    elements = mock_elements_worker_with_cache.list_element_parents(**filters)
+    assert elements.count() == 1
+    for parent in elements.order_by("id"):
+        assert parent.id == UUID(expected_id)
+
+    # Check the worker never hits the API for elements
+    assert len(responses.calls) == len(BASE_API_CALLS)
+    assert [
+        (call.request.method, call.request.url) for call in responses.calls
+    ] == BASE_API_CALLS
+
+
+@pytest.mark.usefixtures("_mock_cached_elements")
+@pytest.mark.parametrize(
+    ("filters", "expected_id"),
+    [
         # Filter on element and worker version
         (
             {
@@ -2507,25 +2674,9 @@ def test_list_element_parents_with_cache_unhandled_param(
             },
             "99999999-9999-9999-9999-999999999999",
         ),
-        # Filter on element and worker run
-        (
-            {
-                "element": CachedElement(id="22222222-2222-2222-2222-222222222222"),
-                "worker_run": "56785678-5678-5678-5678-567856785678",
-            },
-            "12341234-1234-1234-1234-123412341234",
-        ),
-        # Filter on element, manual worker run
-        (
-            {
-                "element": CachedElement(id="12341234-1234-1234-1234-123412341234"),
-                "worker_run": False,
-            },
-            "99999999-9999-9999-9999-999999999999",
-        ),
     ],
 )
-def test_list_element_parents_with_cache(
+def test_list_element_parents_with_cache_deprecation(
     responses,
     mock_elements_worker_with_cache,
     filters,
@@ -2534,8 +2685,11 @@ def test_list_element_parents_with_cache(
     # Check we have 5 elements already present in database
     assert CachedElement.select().count() == 5
 
-    # Query database through cache
-    elements = mock_elements_worker_with_cache.list_element_parents(**filters)
+    with pytest.deprecated_call(
+        match="`worker_version` usage is deprecated. Consider using `worker_run` instead."
+    ):
+        # Query database through cache
+        elements = mock_elements_worker_with_cache.list_element_parents(**filters)
     assert elements.count() == 1
     for parent in elements.order_by("id"):
         assert parent.id == UUID(expected_id)

tests/test_elements_worker/test_entities.py

@@ -708,7 +708,7 @@ def test_create_transcription_entity_with_confidence_with_cache(
     ]
 
 
-def test_list_transcription_entities(fake_dummy_worker):
+def test_list_transcription_entities_deprecation(fake_dummy_worker):
     transcription = Transcription({"id": "fake_transcription_id"})
     worker_version = "worker_version_id"
     fake_dummy_worker.api_client.add_response(
@@ -717,8 +717,28 @@ def test_list_transcription_entities(fake_dummy_worker):
         worker_version=worker_version,
         response={"id": "entity_id"},
     )
+    with pytest.deprecated_call(
+        match="`worker_version` usage is deprecated. Consider using `worker_run` instead."
+    ):
+        assert fake_dummy_worker.list_transcription_entities(
+            transcription, worker_version=worker_version
+        ) == {"id": "entity_id"}
+
+    assert len(fake_dummy_worker.api_client.history) == 1
+    assert len(fake_dummy_worker.api_client.responses) == 0
+
+
+def test_list_transcription_entities(fake_dummy_worker):
+    transcription = Transcription({"id": "fake_transcription_id"})
+    worker_run = "worker_run_id"
+    fake_dummy_worker.api_client.add_response(
+        "ListTranscriptionEntities",
+        id=transcription.id,
+        worker_run=worker_run,
+        response={"id": "entity_id"},
+    )
     assert fake_dummy_worker.list_transcription_entities(
-        transcription, worker_version
+        transcription, worker_run=worker_run
     ) == {"id": "entity_id"}
 
     assert len(fake_dummy_worker.api_client.history) == 1

tests/test_elements_worker/test_metadata.py

@@ -508,14 +508,23 @@ def test_create_metadatas_api_error(responses, mock_elements_worker):
     ]
 
 
+def test_list_element_metadata_wrong_load_parents(fake_dummy_worker):
+    element = Element({"id": "element_id"})
+    with pytest.raises(AssertionError, match="load_parents should be of type bool"):
+        fake_dummy_worker.list_element_metadata(
+            element=element,
+            load_parents="not bool",
+        )
+
+
 def test_list_element_metadata(fake_dummy_worker):
     element = Element({"id": "element_id"})
     fake_dummy_worker.api_client.add_response(
         "ListElementMetaData",
         id=element.id,
-        response={"id": "metadata_id"},
+        response=[{"id": "metadata_id"}],
     )
-    assert fake_dummy_worker.list_element_metadata(element) == {"id": "metadata_id"}
+    assert fake_dummy_worker.list_element_metadata(element) == [{"id": "metadata_id"}]
 
     assert len(fake_dummy_worker.api_client.history) == 1
     assert len(fake_dummy_worker.api_client.responses) == 0
@@ -527,11 +536,28 @@ def test_list_element_metadata_cached_element(mock_elements_worker_with_cache):
     mock_elements_worker_with_cache.api_client.add_response(
         "ListElementMetaData",
         id="element_id",
-        response={"id": "metadata_id"},
+        response=[{"id": "metadata_id"}],
     )
-    assert mock_elements_worker_with_cache.list_element_metadata(element) == {
-        "id": "metadata_id"
-    }
+    assert mock_elements_worker_with_cache.list_element_metadata(element) == [
+        {"id": "metadata_id"}
+    ]
 
     assert len(mock_elements_worker_with_cache.api_client.history) == 1
     assert len(mock_elements_worker_with_cache.api_client.responses) == 0
+
+
+def test_list_element_metadata_with_load_parents(fake_dummy_worker):
+    element = Element({"id": "element_id"})
+    fake_dummy_worker.api_client.add_response(
+        "ListElementMetaData",
+        id=element.id,
+        load_parents=True,
+        response=[{"id": "metadata_id"}, {"id": "parent_metadata_id"}],
+    )
+    assert fake_dummy_worker.list_element_metadata(element, load_parents=True) == [
+        {"id": "metadata_id"},
+        {"id": "parent_metadata_id"},
+    ]
+
+    assert len(fake_dummy_worker.api_client.history) == 1
+    assert len(fake_dummy_worker.api_client.responses) == 0

tests/test_elements_worker/test_transcriptions.py

@@ -1711,10 +1711,25 @@ def test_list_transcriptions_wrong_recursive(mock_elements_worker):
         )
 
 
-def test_list_transcriptions_wrong_worker_version(mock_elements_worker):
+def test_list_transcriptions_wrong_worker_run(mock_elements_worker):
     elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
 
     with pytest.raises(
+        AssertionError, match="worker_run should be of type str or bool"
+    ):
+        mock_elements_worker.list_transcriptions(
+            element=elt,
+            worker_run=1234,
+        )
+
+
+def test_list_transcriptions_wrong_worker_version(mock_elements_worker):
+    elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+    # WARNING: pytest.deprecated_call must be placed BEFORE pytest.raises, otherwise `match` argument won't be checked
+    with pytest.deprecated_call(
+        match="`worker_version` usage is deprecated. Consider using `worker_run` instead."
+    ), pytest.raises(
         AssertionError, match="worker_version should be of type str or bool"
     ):
         mock_elements_worker.list_transcriptions(
@@ -1723,10 +1738,25 @@ def test_list_transcriptions_wrong_worker_version(mock_elements_worker):
         )
 
 
-def test_list_transcriptions_wrong_bool_worker_version(mock_elements_worker):
+def test_list_transcriptions_wrong_bool_worker_run(mock_elements_worker):
     elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
 
     with pytest.raises(
+        AssertionError, match="if of type bool, worker_run can only be set to False"
+    ):
+        mock_elements_worker.list_transcriptions(
+            element=elt,
+            worker_run=True,
+        )
+
+
+def test_list_transcriptions_wrong_bool_worker_version(mock_elements_worker):
+    elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+    # WARNING: pytest.deprecated_call must be placed BEFORE pytest.raises, otherwise `match` argument won't be checked
+    with pytest.deprecated_call(
+        match="`worker_version` usage is deprecated. Consider using `worker_run` instead."
+    ), pytest.raises(
         AssertionError, match="if of type bool, worker_version can only be set to False"
     ):
         mock_elements_worker.list_transcriptions(
@@ -1784,6 +1814,7 @@ def test_list_transcriptions(responses, mock_elements_worker):
             "text": "hey",
             "confidence": 0.42,
             "worker_version_id": "56785678-5678-5678-5678-567856785678",
+            "worker_run_id": "56785678-5678-5678-5678-567856785678",
             "element": None,
         },
         {
@@ -1791,6 +1822,7 @@ def test_list_transcriptions(responses, mock_elements_worker):
             "text": "it's",
             "confidence": 0.42,
             "worker_version_id": "56785678-5678-5678-5678-567856785678",
+            "worker_run_id": "56785678-5678-5678-5678-567856785678",
             "element": None,
         },
         {
@@ -1798,6 +1830,7 @@ def test_list_transcriptions(responses, mock_elements_worker):
             "text": "me",
             "confidence": 0.42,
             "worker_version_id": "56785678-5678-5678-5678-567856785678",
+            "worker_run_id": "56785678-5678-5678-5678-567856785678",
             "element": None,
         },
     ]
@@ -1836,6 +1869,7 @@ def test_list_transcriptions_manual_worker_version(responses, mock_elements_work
             "text": "hey",
             "confidence": 0.42,
             "worker_version_id": None,
+            "worker_run_id": None,
             "element": None,
         }
     ]
@@ -1850,8 +1884,50 @@ def test_list_transcriptions_manual_worker_version(responses, mock_elements_work
         },
     )
 
+    with pytest.deprecated_call(
+        match="`worker_version` usage is deprecated. Consider using `worker_run` instead."
+    ):
+        for idx, transcription in enumerate(
+            mock_elements_worker.list_transcriptions(element=elt, worker_version=False)
+        ):
+            assert transcription == trans[idx]
+
+    assert len(responses.calls) == len(BASE_API_CALLS) + 1
+    assert [
+        (call.request.method, call.request.url) for call in responses.calls
+    ] == BASE_API_CALLS + [
+        (
+            "GET",
+            "http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/transcriptions/?worker_version=False",
+        ),
+    ]
+
+
+def test_list_transcriptions_manual_worker_run(responses, mock_elements_worker):
+    elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+    trans = [
+        {
+            "id": "0000",
+            "text": "hey",
+            "confidence": 0.42,
+            "worker_version_id": None,
+            "worker_run_id": None,
+            "element": None,
+        }
+    ]
+    responses.add(
+        responses.GET,
+        "http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/transcriptions/?worker_run=False",
+        status=200,
+        json={
+            "count": 1,
+            "next": None,
+            "results": trans,
+        },
+    )
+
     for idx, transcription in enumerate(
-        mock_elements_worker.list_transcriptions(element=elt, worker_version=False)
+        mock_elements_worker.list_transcriptions(element=elt, worker_run=False)
     ):
         assert transcription == trans[idx]
 
@@ -1861,7 +1937,7 @@ def test_list_transcriptions_manual_worker_version(responses, mock_elements_work
     ] == BASE_API_CALLS + [
         (
             "GET",
-            "http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/transcriptions/?worker_version=False",
+            "http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/transcriptions/?worker_run=False",
         ),
     ]
 
@@ -1895,16 +1971,26 @@ def test_list_transcriptions_manual_worker_version(responses, mock_elements_work
                 "66666666-6666-6666-6666-666666666666",
             ),
         ),
-        # Filter on element and worker_version should give first transcription
+        # Filter on element and worker run should give the first transcription
         (
             {
                 "element": CachedElement(
                     id="11111111-1111-1111-1111-111111111111", type="page"
                 ),
-                "worker_version": "56785678-5678-5678-5678-567856785678",
+                "worker_run": "56785678-5678-5678-5678-567856785678",
             },
             ("11111111-1111-1111-1111-111111111111",),
         ),
+        # Filter on element, manual worker run should give the sixth transcription
+        (
+            {
+                "element": CachedElement(
+                    id="11111111-1111-1111-1111-111111111111", type="page"
+                ),
+                "worker_run": False,
+            },
+            ("66666666-6666-6666-6666-666666666666",),
+        ),
         # Filter recursively on element should give all transcriptions inserted
         (
             {
@@ -1922,33 +2008,70 @@ def test_list_transcriptions_manual_worker_version(responses, mock_elements_work
                 "66666666-6666-6666-6666-666666666666",
             ),
         ),
-        # Filter recursively on element and worker_version should give four transcriptions
+        # Filter recursively on element and element_type should give three transcriptions
         (
             {
                 "element": CachedElement(
                     id="11111111-1111-1111-1111-111111111111", type="page"
                 ),
-                "worker_version": "90129012-9012-9012-9012-901290129012",
+                "element_type": "something_else",
                 "recursive": True,
             },
             (
                 "22222222-2222-2222-2222-222222222222",
-                "33333333-3333-3333-3333-333333333333",
                 "44444444-4444-4444-4444-444444444444",
                 "55555555-5555-5555-5555-555555555555",
             ),
         ),
-        # Filter recursively on element and element_type should give three transcriptions
+    ],
+)
+def test_list_transcriptions_with_cache(
+    responses, mock_elements_worker_with_cache, filters, expected_ids
+):
+    # Check we have 5 elements already present in database
+    assert CachedTranscription.select().count() == 6
+
+    # Query database through cache
+    transcriptions = mock_elements_worker_with_cache.list_transcriptions(**filters)
+    assert transcriptions.count() == len(expected_ids)
+    for transcription, expected_id in zip(
+        transcriptions.order_by(CachedTranscription.id), expected_ids, strict=True
+    ):
+        assert transcription.id == UUID(expected_id)
+
+    # Check the worker never hits the API for elements
+    assert len(responses.calls) == len(BASE_API_CALLS)
+    assert [
+        (call.request.method, call.request.url) for call in responses.calls
+    ] == BASE_API_CALLS
+
+
+@pytest.mark.usefixtures("_mock_cached_transcriptions")
+@pytest.mark.parametrize(
+    ("filters", "expected_ids"),
+    [
+        # Filter on element and worker_version should give first transcription
         (
             {
                 "element": CachedElement(
                     id="11111111-1111-1111-1111-111111111111", type="page"
                 ),
-                "element_type": "something_else",
+                "worker_version": "56785678-5678-5678-5678-567856785678",
+            },
+            ("11111111-1111-1111-1111-111111111111",),
+        ),
+        # Filter recursively on element and worker_version should give four transcriptions
+        (
+            {
+                "element": CachedElement(
+                    id="11111111-1111-1111-1111-111111111111", type="page"
+                ),
+                "worker_version": "90129012-9012-9012-9012-901290129012",
                 "recursive": True,
             },
             (
                 "22222222-2222-2222-2222-222222222222",
+                "33333333-3333-3333-3333-333333333333",
                 "44444444-4444-4444-4444-444444444444",
                 "55555555-5555-5555-5555-555555555555",
             ),
@@ -1965,14 +2088,17 @@ def test_list_transcriptions_manual_worker_version(responses, mock_elements_work
         ),
     ],
 )
-def test_list_transcriptions_with_cache(
+def test_list_transcriptions_with_cache_deprecation(
     responses, mock_elements_worker_with_cache, filters, expected_ids
 ):
     # Check we have 5 elements already present in database
     assert CachedTranscription.select().count() == 6
 
-    # Query database through cache
-    transcriptions = mock_elements_worker_with_cache.list_transcriptions(**filters)
+    with pytest.deprecated_call(
+        match="`worker_version` usage is deprecated. Consider using `worker_run` instead."
+    ):
+        # Query database through cache
+        transcriptions = mock_elements_worker_with_cache.list_transcriptions(**filters)
     assert transcriptions.count() == len(expected_ids)
     for transcription, expected_id in zip(
         transcriptions.order_by(CachedTranscription.id), expected_ids, strict=True

tests/test_elements_worker/test_worker.py

@@ -20,7 +20,8 @@ def test_get_worker_version(fake_dummy_worker):
 
     api_client.add_response("RetrieveWorkerVersion", response, id=TEST_VERSION_ID)
 
-    res = fake_dummy_worker.get_worker_version(TEST_VERSION_ID)
+    with pytest.deprecated_call(match="WorkerVersion usage is deprecated."):
+        res = fake_dummy_worker.get_worker_version(TEST_VERSION_ID)
 
     assert res == response
     assert fake_dummy_worker._worker_version_cache[TEST_VERSION_ID] == response
@@ -33,8 +34,11 @@ def test_get_worker_version__uses_cache(fake_dummy_worker):
 
     api_client.add_response("RetrieveWorkerVersion", response, id=TEST_VERSION_ID)
 
-    response_1 = fake_dummy_worker.get_worker_version(TEST_VERSION_ID)
-    response_2 = fake_dummy_worker.get_worker_version(TEST_VERSION_ID)
+    with pytest.deprecated_call(match="WorkerVersion usage is deprecated."):
+        response_1 = fake_dummy_worker.get_worker_version(TEST_VERSION_ID)
+
+    with pytest.deprecated_call(match="WorkerVersion usage is deprecated."):
+        response_2 = fake_dummy_worker.get_worker_version(TEST_VERSION_ID)
 
     assert response_1 == response
     assert response_1 == response_2
@@ -51,12 +55,16 @@ def test_get_worker_version_slug(mocker, fake_dummy_worker):
         "worker": {"slug": "mock_slug"},
     }
 
-    slug = fake_dummy_worker.get_worker_version_slug(TEST_VERSION_ID)
+    with pytest.deprecated_call(match="WorkerVersion usage is deprecated."):
+        slug = fake_dummy_worker.get_worker_version_slug(TEST_VERSION_ID)
     assert slug == "mock_slug"
 
 
 def test_get_worker_version_slug_none(fake_dummy_worker):
-    with pytest.raises(ValueError, match="No worker version ID"):
+    # WARNING: pytest.deprecated_call must be placed BEFORE pytest.raises, otherwise `match` argument won't be checked
+    with pytest.deprecated_call(
+        match="WorkerVersion usage is deprecated."
+    ), pytest.raises(ValueError, match="No worker version ID"):
         fake_dummy_worker.get_worker_version_slug(None)
 
 

worker-{{cookiecutter.slug}}/.arkindex.yml

 ---
 version: 2
 
-type: worker
-
 workers:
   - slug: {{ cookiecutter.slug }}
     name: {{ cookiecutter.name }}

worker-{{cookiecutter.slug}}/.gitlab-ci.yml

@@ -10,7 +10,7 @@ include:
   - template: 'Workflows/Branch-Pipelines.gitlab-ci.yml'
 
 variables:
-  VERSION: $CI_COMMIT_SHA
+  VERSION: commit-$CI_COMMIT_SHORT_SHA
   DEBIAN_FRONTEND: non-interactive
 
 test:

worker-{{cookiecutter.slug}}/requirements.txt

-arkindex-base-worker==0.3.5
+arkindex-base-worker==0.3.6