arkindex-base-worker 0.3.6rc5__py3-none-any.whl → 0.3.7.post1__py3-none-any.whl

arkindex_worker/worker/transcription.py

@@ -4,6 +4,7 @@ ElementsWorker methods for transcriptions.
 
 from collections.abc import Iterable
 from enum import Enum
+from warnings import warn
 
 from peewee import IntegrityError
 
@@ -366,14 +367,22 @@ class TranscriptionMixin:
         element_type: str | None = None,
         recursive: bool | None = None,
         worker_version: str | bool | None = None,
+        worker_run: str | bool | None = None,
     ) -> Iterable[dict] | Iterable[CachedTranscription]:
         """
         List transcriptions on an element.
 
+        Warns:
+        ----
+        The following parameters are **deprecated**:
+
+        - `worker_version` in favor of `worker_run`
+
         :param element: The element to list transcriptions on.
         :param element_type: Restrict to transcriptions whose elements have an element type with this slug.
         :param recursive: Include transcriptions of any descendant of this element, recursively.
-        :param worker_version: Restrict to transcriptions created by a worker version with this UUID. Set to False to look for manually created transcriptions.
+        :param worker_version: **Deprecated** Restrict to transcriptions created by a worker version with this UUID. Set to False to look for manually created transcriptions.
+        :param worker_run: Restrict to transcriptions created by a worker run with this UUID. Set to False to look for manually created transcriptions.
         :returns: An iterable of dicts representing each transcription,
            or an iterable of CachedTranscription when cache support is enabled.
         """
@@ -388,6 +397,11 @@ class TranscriptionMixin:
             assert isinstance(recursive, bool), "recursive should be of type bool"
             query_params["recursive"] = recursive
         if worker_version is not None:
+            warn(
+                "`worker_version` usage is deprecated. Consider using `worker_run` instead.",
+                DeprecationWarning,
+                stacklevel=1,
+            )
             assert isinstance(
                 worker_version, str | bool
             ), "worker_version should be of type str or bool"
@@ -396,6 +410,15 @@ class TranscriptionMixin:
                     worker_version is False
                 ), "if of type bool, worker_version can only be set to False"
             query_params["worker_version"] = worker_version
+        if worker_run is not None:
+            assert isinstance(
+                worker_run, str | bool
+            ), "worker_run should be of type str or bool"
+            if isinstance(worker_run, bool):
+                assert (
+                    worker_run is False
+                ), "if of type bool, worker_run can only be set to False"
+            query_params["worker_run"] = worker_run
 
         if self.use_cache:
             if not recursive:
@@ -427,10 +450,27 @@ class TranscriptionMixin:
 
             if worker_version is not None:
                 # If worker_version=False, filter by manual worker_version e.g. None
-                worker_version_id = worker_version if worker_version else None
-                transcriptions = transcriptions.where(
-                    CachedTranscription.worker_version_id == worker_version_id
-                )
+                worker_version_id = worker_version or None
+                if worker_version_id:
+                    transcriptions = transcriptions.where(
+                        CachedTranscription.worker_version_id == worker_version_id
+                    )
+                else:
+                    transcriptions = transcriptions.where(
+                        CachedTranscription.worker_version_id.is_null()
+                    )
+
+            if worker_run is not None:
+                # If worker_run=False, filter by manual worker_run e.g. None
+                worker_run_id = worker_run or None
+                if worker_run_id:
+                    transcriptions = transcriptions.where(
+                        CachedTranscription.worker_run_id == worker_run_id
+                    )
+                else:
+                    transcriptions = transcriptions.where(
+                        CachedTranscription.worker_run_id.is_null()
+                    )
         else:
             transcriptions = self.api_client.paginate(
                 "ListTranscriptions", id=element.id, **query_params

arkindex_worker/worker/version.py

@@ -2,10 +2,27 @@
 ElementsWorker methods for worker versions.
 """
 
+import functools
+from warnings import warn
+
+
+def worker_version_deprecation(func):
+    @functools.wraps(func)
+    def wrapper(self, *args, **kwargs):
+        warn("WorkerVersion usage is deprecated.", DeprecationWarning, stacklevel=2)
+        return func(self, *args, **kwargs)
+
+    return wrapper
+
 
 class WorkerVersionMixin:
+    @worker_version_deprecation
     def get_worker_version(self, worker_version_id: str) -> dict:
         """
+        Warns:
+        ----
+        This method is **deprecated**.
+
         Retrieve a worker version, using the [ElementsWorker][arkindex_worker.worker.ElementsWorker]'s internal cache when possible.
 
         :param worker_version_id: ID of the worker version to retrieve.
@@ -22,8 +39,13 @@ class WorkerVersionMixin:
 
         return worker_version
 
+    @worker_version_deprecation
     def get_worker_version_slug(self, worker_version_id: str) -> str:
         """
+        Warns:
+        ----
+        This method is **deprecated**.
+
         Retrieve the slug of the worker of a worker version, from a worker version UUID.
         Uses a worker version from the internal cache if possible, otherwise makes an API request.
 

hooks/pre_gen_project.py

@@ -0,0 +1,3 @@
+# Normalize the slug to generate __package and __module private variables
+{{cookiecutter.update({"__package": cookiecutter.slug.lower().replace("_", "-")})}}  # noqa: F821
+{{cookiecutter.update({"__module": cookiecutter.slug.lower().replace("-", "_")})}}  # noqa: F821

tests/conftest.py

@@ -22,7 +22,7 @@ from arkindex_worker.cache import (
     create_version_table,
     init_cache_db,
 )
-from arkindex_worker.models import Artifact, Dataset
+from arkindex_worker.models import Artifact, Dataset, Set
 from arkindex_worker.worker import BaseWorker, DatasetWorker, ElementsWorker
 from arkindex_worker.worker.dataset import DatasetState
 from arkindex_worker.worker.transcription import TextOrientation
@@ -93,7 +93,7 @@ def _setup_api(responses, monkeypatch, _cache_yaml):
 
     # Fallback to prod environment
     if schema_url is None:
-        schema_url = "https://arkindex.teklia.com/api/v1/openapi/?format=openapi-json"
+        schema_url = "https://arkindex.teklia.com/api/v1/openapi/?format=json"
         monkeypatch.setenv("ARKINDEX_API_SCHEMA_URL", schema_url)
 
     # Allow accessing remote API schemas
@@ -466,6 +466,7 @@ def _mock_cached_transcriptions(mock_cache_db):
         confidence=0.42,
         orientation=TextOrientation.HorizontalLeftToRight,
         worker_version_id=UUID("56785678-5678-5678-5678-567856785678"),
+        worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
     )
     CachedTranscription.create(
         id=UUID("22222222-2222-2222-2222-222222222222"),
@@ -506,6 +507,7 @@ def _mock_cached_transcriptions(mock_cache_db):
         confidence=0.42,
         orientation=TextOrientation.HorizontalLeftToRight,
         worker_version_id=None,
+        worker_run_id=None,
     )
 
 
@@ -595,11 +597,11 @@ def mock_databases(tmp_path):
 @pytest.fixture()
 def default_dataset():
     return Dataset(
-        **{
+        {
             "id": "dataset_id",
             "name": "My dataset",
             "description": "A super dataset built by me",
-            "sets": ["set_1", "set_2", "set_3"],
+            "sets": ["set_1", "set_2", "set_3", "set_4"],
             "state": DatasetState.Open.value,
             "corpus_id": "corpus_id",
             "creator": "creator@teklia.com",
@@ -610,6 +612,11 @@ def default_dataset():
     )
 
 
+@pytest.fixture()
+def default_train_set(default_dataset):
+    return Set(name="train", dataset=default_dataset)
+
+
 @pytest.fixture()
 def mock_dataset_worker(monkeypatch, mocker, _mock_worker_run_api):
     monkeypatch.setenv("PONOS_TASK", "my_task")
@@ -632,9 +639,10 @@ def mock_dev_dataset_worker(mocker):
         [
             "worker",
             "--dev",
-            "--dataset",
-            "11111111-1111-1111-1111-111111111111",
-            "22222222-2222-2222-2222-222222222222",
+            "--set",
+            "11111111-1111-1111-1111-111111111111:train",
+            "11111111-1111-1111-1111-111111111111:val",
+            "22222222-2222-2222-2222-222222222222:my_set",
         ],
     )
 

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,