arkindex-base-worker 0.3.5rc6__py3-none-any.whl → 0.3.6rc2__py3-none-any.whl

arkindex_worker/worker/metadata.py

@@ -1,10 +1,8 @@
-# -*- coding: utf-8 -*-
 """
 ElementsWorker methods for metadata.
 """
 
 from enum import Enum
-from typing import Dict, List, Optional, Union
 
 from arkindex_worker import logger
 from arkindex_worker.cache import CachedElement
@@ -57,14 +55,14 @@ class MetaType(Enum):
     """
 
 
-class MetaDataMixin(object):
+class MetaDataMixin:
     def create_metadata(
         self,
-        element: Union[Element, CachedElement],
+        element: Element | CachedElement,
         type: MetaType,
         name: str,
         value: str,
-        entity: Optional[str] = None,
+        entity: str | None = None,
     ) -> str:
         """
         Create a metadata on the given element through API.
@@ -77,7 +75,7 @@ class MetaDataMixin(object):
         :returns: UUID of the created metadata.
         """
         assert element and isinstance(
-            element, (Element, CachedElement)
+            element, Element | CachedElement
         ), "element shouldn't be null and should be of type Element or CachedElement"
         assert type and isinstance(
             type, MetaType
@@ -110,26 +108,22 @@ class MetaDataMixin(object):
 
     def create_metadatas(
         self,
-        element: Union[Element, CachedElement],
-        metadatas: List[
-            Dict[
-                str, Union[MetaType, str, Union[str, Union[int, float]], Optional[str]]
-            ]
-        ],
-    ) -> List[Dict[str, str]]:
+        element: Element | CachedElement,
+        metadatas: list[dict[str, MetaType | str | int | float | None]],
+    ) -> list[dict[str, str]]:
         """
-        Create multiple metadatas on an existing element.
+        Create multiple metadata on an existing element.
         This method does not support cache.
 
-        :param element Element: The element to create multiple metadata on.
-        :param metadata_list List(Dict): The list of dict whose keys are the following:
-            - type : MetaType
-            - name : str
-            - value : Union[str, Union[int, float]]
-            - entity_id : Union[str, None]
+        :param element: The element to create multiple metadata on.
+        :param metadatas: The list of dict whose keys are the following:
+            - type: MetaType
+            - name: str
+            - value: str | int | float
+            - entity_id: str | None
         """
         assert element and isinstance(
-            element, (Element, CachedElement)
+            element, Element | CachedElement
         ), "element shouldn't be null and should be of type Element or CachedElement"
 
         assert metadatas and isinstance(
@@ -152,7 +146,7 @@ class MetaDataMixin(object):
             ), "name shouldn't be null and should be of type str"
 
             assert metadata.get("value") is not None and isinstance(
-                metadata.get("value"), (str, float, int)
+                metadata.get("value"), str | float | int
             ), "value shouldn't be null and should be of type (str or float or int)"
 
             assert metadata.get("entity_id") is None or isinstance(
@@ -172,7 +166,7 @@ class MetaDataMixin(object):
             logger.warning("Cannot create metadata as this worker is in read-only mode")
             return
 
-        created_metadatas = self.request(
+        created_metadata_list = self.request(
             "CreateMetaDataBulk",
             id=element.id,
             body={
@@ -181,11 +175,11 @@ class MetaDataMixin(object):
             },
         )["metadata_list"]
 
-        return created_metadatas
+        return created_metadata_list
 
     def list_element_metadata(
-        self, element: Union[Element, CachedElement]
-    ) -> List[Dict[str, str]]:
+        self, element: Element | CachedElement
+    ) -> list[dict[str, str]]:
         """
         List all metadata linked to an element.
         This method does not support cache.
@@ -193,7 +187,7 @@ class MetaDataMixin(object):
         :param element: The element to list metadata on.
         """
         assert element and isinstance(
-            element, (Element, CachedElement)
+            element, Element | CachedElement
         ), "element shouldn't be null and should be of type Element or CachedElement"
 
         return self.api_client.paginate("ListElementMetaData", id=element.id)

arkindex_worker/worker/task.py

@@ -1,17 +1,16 @@
-# -*- coding: utf-8 -*-
 """
 BaseWorker methods for tasks.
 """
 
 import uuid
-from typing import Iterator
+from collections.abc import Iterator
 
 from apistar.compat import DownloadedFile
 
 from arkindex_worker.models import Artifact
 
 
-class TaskMixin(object):
+class TaskMixin:
     def list_artifacts(self, task_id: uuid.UUID) -> Iterator[Artifact]:
         """
         List artifacts associated to a task.

arkindex_worker/worker/training.py

@@ -1,4 +1,3 @@
-# -*- coding: utf-8 -*-
 """
 BaseWorker methods for training.
 """
@@ -6,7 +5,7 @@ BaseWorker methods for training.
 import functools
 from contextlib import contextmanager
 from pathlib import Path
-from typing import NewType, Optional, Tuple, Union
+from typing import NewType
 from uuid import UUID
 
 import requests
@@ -26,7 +25,7 @@ FileSize = NewType("FileSize", int)
 
 
 @contextmanager
-def create_archive(path: DirPath) -> Tuple[Path, Hash, FileSize, Hash]:
+def create_archive(path: DirPath) -> tuple[Path, Hash, FileSize, Hash]:
     """
     Create a tar archive from the files at the given location then compress it to a zst archive.
 
@@ -37,15 +36,15 @@ def create_archive(path: DirPath) -> Tuple[Path, Hash, FileSize, Hash]:
     """
     assert path.is_dir(), "create_archive needs a directory"
 
-    zstd_descriptor, zstd_archive, archive_hash, content_hash = create_tar_zst_archive(
+    zst_descriptor, zst_archive, archive_hash, content_hash = create_tar_zst_archive(
         path
     )
 
     # Get content hash, archive size and hash
-    yield zstd_archive, content_hash, zstd_archive.stat().st_size, archive_hash
+    yield zst_archive, content_hash, zst_archive.stat().st_size, archive_hash
 
-    # Remove the zstd archive
-    close_delete_file(zstd_descriptor, zstd_archive)
+    # Remove the zst archive
+    close_delete_file(zst_descriptor, zst_archive)
 
 
 def build_clean_payload(**kwargs):
@@ -72,7 +71,7 @@ def skip_if_read_only(func):
     return wrapper
 
 
-class TrainingMixin(object):
+class TrainingMixin:
     """
     A mixin helper to create a new model version easily.
     You may use `publish_model_version` to publish a ready model version directly, or
@@ -87,10 +86,10 @@ class TrainingMixin(object):
         self,
         model_path: DirPath,
         model_id: str,
-        tag: Optional[str] = None,
-        description: Optional[str] = None,
-        configuration: Optional[dict] = {},
-        parent: Optional[Union[str, UUID]] = None,
+        tag: str | None = None,
+        description: str | None = None,
+        configuration: dict | None = None,
+        parent: str | UUID | None = None,
     ):
         """
         Publish a unique version of a model in Arkindex, identified by its hash.
@@ -105,6 +104,7 @@ class TrainingMixin(object):
         :param parent: ID of the parent model version
         """
 
+        configuration = configuration or {}
         if not self.model_version:
             self.create_model_version(
                 model_id=model_id,
@@ -161,10 +161,10 @@ class TrainingMixin(object):
     def create_model_version(
         self,
         model_id: str,
-        tag: Optional[str] = None,
-        description: Optional[str] = None,
-        configuration: Optional[dict] = {},
-        parent: Optional[Union[str, UUID]] = None,
+        tag: str | None = None,
+        description: str | None = None,
+        configuration: dict | None = None,
+        parent: str | UUID | None = None,
     ):
         """
         Create a new version of the specified model with its base attributes.
@@ -176,6 +176,8 @@ class TrainingMixin(object):
         :param parent: ID of the parent model version
         """
         assert not self.model_version, "A model version has already been created."
+
+        configuration = configuration or {}
         self.model_version = self.request(
             "CreateModelVersion",
             id=model_id,
@@ -186,6 +188,7 @@ class TrainingMixin(object):
                 parent=parent,
             ),
         )
+
         logger.info(
             f"Model version ({self.model_version['id']}) was successfully created"
         )
@@ -193,10 +196,10 @@ class TrainingMixin(object):
     @skip_if_read_only
     def update_model_version(
         self,
-        tag: Optional[str] = None,
-        description: Optional[str] = None,
-        configuration: Optional[dict] = None,
-        parent: Optional[Union[str, UUID]] = None,
+        tag: str | None = None,
+        description: str | None = None,
+        configuration: dict | None = None,
+        parent: str | UUID | None = None,
     ):
         """
         Update the current model version with the given attributes.
@@ -235,9 +238,7 @@ class TrainingMixin(object):
         ), "The model is already marked as available."
 
         s3_put_url = self.model_version.get("s3_put_url")
-        assert (
-            s3_put_url
-        ), "S3 PUT URL is not set, please ensure you have the right to validate a model version."
+        assert s3_put_url, "S3 PUT URL is not set, please ensure you have the right to validate a model version."
 
         logger.info("Uploading to s3...")
         # Upload the archive on s3
@@ -263,9 +264,7 @@ class TrainingMixin(object):
         :param size: The size of the uploaded archive
         :param archive_hash: MD5 hash of the uploaded archive
         """
-        assert (
-            self.model_version
-        ), "You must create the model version and upload its archive before validating it."
+        assert self.model_version, "You must create the model version and upload its archive before validating it."
         try:
             self.model_version = self.request(
                 "ValidateModelVersion",

arkindex_worker/worker/transcription.py

@@ -1,10 +1,9 @@
-# -*- coding: utf-8 -*-
 """
 ElementsWorker methods for transcriptions.
 """
 
+from collections.abc import Iterable
 from enum import Enum
-from typing import Dict, Iterable, List, Optional, Union
 
 from peewee import IntegrityError
 
@@ -40,14 +39,14 @@ class TextOrientation(Enum):
     """
 
 
-class TranscriptionMixin(object):
+class TranscriptionMixin:
     def create_transcription(
         self,
-        element: Union[Element, CachedElement],
+        element: Element | CachedElement,
         text: str,
         confidence: float,
         orientation: TextOrientation = TextOrientation.HorizontalLeftToRight,
-    ) -> Optional[Dict[str, Union[str, float]]]:
+    ) -> dict[str, str | float] | None:
         """
         Create a transcription on the given element through the API.
 
@@ -59,7 +58,7 @@ class TranscriptionMixin(object):
            or None if the worker is in read-only mode.
         """
         assert element and isinstance(
-            element, (Element, CachedElement)
+            element, Element | CachedElement
         ), "element shouldn't be null and should be an Element or CachedElement"
         assert text and isinstance(
             text, str
@@ -111,8 +110,8 @@ class TranscriptionMixin(object):
 
     def create_transcriptions(
         self,
-        transcriptions: List[Dict[str, Union[str, float, Optional[TextOrientation]]]],
-    ) -> List[Dict[str, Union[str, float]]]:
+        transcriptions: list[dict[str, str | float | TextOrientation | None]],
+    ) -> list[dict[str, str | float]]:
         """
         Create multiple transcriptions at once on existing elements through the API,
         and creates [CachedTranscription][arkindex_worker.cache.CachedTranscription] instances if cache support is enabled.
@@ -140,13 +139,13 @@ class TranscriptionMixin(object):
 
         for index, transcription in enumerate(transcriptions_payload):
             element_id = transcription.get("element_id")
-            assert element_id and isinstance(
-                element_id, str
+            assert (
+                element_id and isinstance(element_id, str)
             ), f"Transcription at index {index} in transcriptions: element_id shouldn't be null and should be of type str"
 
             text = transcription.get("text")
-            assert text and isinstance(
-                text, str
+            assert (
+                text and isinstance(text, str)
             ), f"Transcription at index {index} in transcriptions: text shouldn't be null and should be of type str"
 
             confidence = transcription.get("confidence")
@@ -159,8 +158,8 @@ class TranscriptionMixin(object):
             orientation = transcription.get(
                 "orientation", TextOrientation.HorizontalLeftToRight
             )
-            assert orientation and isinstance(
-                orientation, TextOrientation
+            assert (
+                orientation and isinstance(orientation, TextOrientation)
             ), f"Transcription at index {index} in transcriptions: orientation shouldn't be null and should be of type TextOrientation"
             if orientation:
                 transcription["orientation"] = orientation.value
@@ -203,10 +202,10 @@ class TranscriptionMixin(object):
 
     def create_element_transcriptions(
         self,
-        element: Union[Element, CachedElement],
+        element: Element | CachedElement,
         sub_element_type: str,
-        transcriptions: List[Dict[str, Union[str, float]]],
-    ) -> Dict[str, Union[str, bool]]:
+        transcriptions: list[dict[str, str | float]],
+    ) -> dict[str, str | bool]:
         """
         Create multiple elements and transcriptions at once on a single parent element through the API.
 
@@ -228,7 +227,7 @@ class TranscriptionMixin(object):
         :returns: A list of dicts as returned by the ``CreateElementTranscriptions`` API endpoint.
         """
         assert element and isinstance(
-            element, (Element, CachedElement)
+            element, Element | CachedElement
         ), "element shouldn't be null and should be an Element or CachedElement"
         assert sub_element_type and isinstance(
             sub_element_type, str
@@ -242,8 +241,8 @@ class TranscriptionMixin(object):
 
         for index, transcription in enumerate(transcriptions_payload):
             text = transcription.get("text")
-            assert text and isinstance(
-                text, str
+            assert (
+                text and isinstance(text, str)
             ), f"Transcription at index {index} in transcriptions: text shouldn't be null and should be of type str"
 
             confidence = transcription.get("confidence")
@@ -256,15 +255,15 @@ class TranscriptionMixin(object):
             orientation = transcription.get(
                 "orientation", TextOrientation.HorizontalLeftToRight
             )
-            assert orientation and isinstance(
-                orientation, TextOrientation
+            assert (
+                orientation and isinstance(orientation, TextOrientation)
             ), f"Transcription at index {index} in transcriptions: orientation shouldn't be null and should be of type TextOrientation"
             if orientation:
                 transcription["orientation"] = orientation.value
 
             polygon = transcription.get("polygon")
-            assert polygon and isinstance(
-                polygon, list
+            assert (
+                polygon and isinstance(polygon, list)
             ), f"Transcription at index {index} in transcriptions: polygon shouldn't be null and should be of type list"
             assert (
                 len(polygon) >= 3
@@ -273,12 +272,16 @@ class TranscriptionMixin(object):
                 isinstance(point, list) and len(point) == 2 for point in polygon
             ), f"Transcription at index {index} in transcriptions: polygon points should be lists of two items"
             assert all(
-                isinstance(coord, (int, float)) for point in polygon for coord in point
+                isinstance(coord, int | float) for point in polygon for coord in point
             ), f"Transcription at index {index} in transcriptions: polygon points should be lists of two numbers"
 
             element_confidence = transcription.get("element_confidence")
-            assert element_confidence is None or (
-                isinstance(element_confidence, float) and 0 <= element_confidence <= 1
+            assert (
+                element_confidence is None
+                or (
+                    isinstance(element_confidence, float)
+                    and 0 <= element_confidence <= 1
+                )
             ), f"Transcription at index {index} in transcriptions: element_confidence should be either null or a float in [0..1] range"
 
         if self.is_read_only:
@@ -359,11 +362,11 @@ class TranscriptionMixin(object):
 
     def list_transcriptions(
         self,
-        element: Union[Element, CachedElement],
-        element_type: Optional[str] = None,
-        recursive: Optional[bool] = None,
-        worker_version: Optional[Union[str, bool]] = None,
-    ) -> Union[Iterable[dict], Iterable[CachedTranscription]]:
+        element: Element | CachedElement,
+        element_type: str | None = None,
+        recursive: bool | None = None,
+        worker_version: str | bool | None = None,
+    ) -> Iterable[dict] | Iterable[CachedTranscription]:
         """
         List transcriptions on an element.
 
@@ -375,7 +378,7 @@ class TranscriptionMixin(object):
            or an iterable of CachedTranscription when cache support is enabled.
         """
         assert element and isinstance(
-            element, (Element, CachedElement)
+            element, Element | CachedElement
         ), "element shouldn't be null and should be an Element or CachedElement"
         query_params = {}
         if element_type:
@@ -386,7 +389,7 @@ class TranscriptionMixin(object):
             query_params["recursive"] = recursive
         if worker_version is not None:
             assert isinstance(
-                worker_version, (str, bool)
+                worker_version, str | bool
             ), "worker_version should be of type str or bool"
             if isinstance(worker_version, bool):
                 assert (

arkindex_worker/worker/version.py

@@ -1,10 +1,9 @@
-# -*- coding: utf-8 -*-
 """
 ElementsWorker methods for worker versions.
 """
 
 
-class WorkerVersionMixin(object):
+class WorkerVersionMixin:
     def get_worker_version(self, worker_version_id: str) -> dict:
         """
         Retrieve a worker version, using the [ElementsWorker][arkindex_worker.worker.ElementsWorker]'s internal cache when possible.