arkindex-base-worker 0.4.0rc6__py3-none-any.whl → 0.5.0a2__py3-none-any.whl

{arkindex_base_worker-0.4.0rc6.dist-info → arkindex_base_worker-0.5.0a2.dist-info}/METADATA

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.2
 Name: arkindex-base-worker
-Version: 0.4.0rc6
+Version: 0.5.0a2
 Summary: Base Worker to easily build Arkindex ML workflows
 Author-email: Teklia <contact@teklia.com>
 Maintainer-email: Teklia <contact@teklia.com>
@@ -46,7 +46,7 @@ Requires-Dist: peewee~=3.17
 Requires-Dist: Pillow==11.0.0
 Requires-Dist: python-gnupg==0.5.3
 Requires-Dist: shapely==2.0.6
-Requires-Dist: teklia-toolbox==0.1.7b1
+Requires-Dist: teklia-toolbox==0.1.7
 Requires-Dist: zstandard==0.23.0
 Provides-Extra: docs
 Requires-Dist: black==24.10.0; extra == "docs"

{arkindex_base_worker-0.4.0rc6.dist-info → arkindex_base_worker-0.5.0a2.dist-info}/RECORD

@@ -1,32 +1,32 @@
-arkindex_worker/__init__.py,sha256=OlgCtTC9MaWeejviY0a3iQpALcRQGMVArFVVYwTF6I8,162
-arkindex_worker/cache.py,sha256=qTblc_zKdYC47Wip6_O9Jf5qBkQW2ozQQrg-nsx1WuY,11221
-arkindex_worker/image.py,sha256=oEgVCrSHiGh3D5-UXfM6PvT17TttSxC0115irpvB3Dw,18581
+arkindex_worker/__init__.py,sha256=Sdt5KXn8EgURb2MurYVrUWaHbH3iFA1XLRo0Lc5AJ44,250
+arkindex_worker/cache.py,sha256=NpCsYFnqBmyBrACqeV7c3P6j6YrTtyi-HgtewwxUpxc,11221
+arkindex_worker/image.py,sha256=-oBhLqzTRsxJoXkzDYFw1Ic4JxQtpmNtzGxe1zOHotw,20980
 arkindex_worker/models.py,sha256=bPQzGZNs5a6z6DEcygsa8T33VOqPlMUbwKzHqlKzwbw,9923
-arkindex_worker/utils.py,sha256=q1EeLdC6ebYIH-C0LOAqw2cNpjCjVoP-Vbr-39mF4w0,9884
-arkindex_worker/worker/__init__.py,sha256=0_YHeOe31KR_8ynbnYMIMwnSQTVbKkkeLGmnlTMhFx0,16234
-arkindex_worker/worker/base.py,sha256=7Pmw-UQSxV-xkW8NO5cXsxJ8W8szzyppMaNjq_az81A,19844
-arkindex_worker/worker/classification.py,sha256=zECSNzGCZFzoPoDVZN4kuGYRNLzMQLBaRt3q1jnBSaA,10952
-arkindex_worker/worker/corpus.py,sha256=0TQFOwZ6Te-CZi6lgkZY1wzyJ5wO9LAmcVQtqHvZpPk,2291
-arkindex_worker/worker/dataset.py,sha256=LwzKwNFX4FqfLxh29LSvJydPwRw3VHaB1wjuFhUshsE,5267
-arkindex_worker/worker/element.py,sha256=Qvvq9kJnAHNATHW7zi96eIY1x-0MsR-T5rrSJg6e9Y4,45309
-arkindex_worker/worker/entity.py,sha256=s5wjX6_JfTyk4qfMoV0OWfOXUx6T-9WpOiEpaoaCEFM,14808
-arkindex_worker/worker/image.py,sha256=t_Az6IGnj0EZyvcA4XxfPikOUjn_pztgsyxTkFZhaXU,621
-arkindex_worker/worker/metadata.py,sha256=VRajtd2kaBvar9GercX4knvR6l1WFYjoCdJWU9ccKgk,7291
+arkindex_worker/utils.py,sha256=MbbJT8oh8DMHHR-vidFeXdUH0TSXGWm7ZDGWzrRXoEY,9933
+arkindex_worker/worker/__init__.py,sha256=3adK1-BDv6uCKUmY0cqaz7LuEJChDHyNmRVPCA1y8lA,16238
+arkindex_worker/worker/base.py,sha256=MbZW9WmSjwh0yKeMckyjm6WxHI9xpfFhWsx5AOzQ0aY,19844
+arkindex_worker/worker/classification.py,sha256=gu_xAkGgvoebbG1xCZ4P7DqYe6cHwTGiHUc9nG0__8A,10996
+arkindex_worker/worker/corpus.py,sha256=MeIMod7jkWyX0frtD0a37rhumnMV3p9ZOC1xwAoXrAA,2291
+arkindex_worker/worker/dataset.py,sha256=tVaPx43vaH-KTtx4w5V06e26ha8XPfiJTRzBXlu928Y,5273
+arkindex_worker/worker/element.py,sha256=0zwODtutkX4AIeSKe0wV9cmNeIZ5cRwTqnuFrVQmKOw,45403
+arkindex_worker/worker/entity.py,sha256=5OaP3HBrA-jbP-3xF-_TpaYh5WxMlQAUUaGLmhynpEE,14833
+arkindex_worker/worker/image.py,sha256=L6Ikuf0Z0RxJk7JarY5PggJGrYSHLaPK0vn0dy0CIaQ,623
+arkindex_worker/worker/metadata.py,sha256=mb9hVU-nRw3drCN-0AvtZ0nPY-4tD-ye9_mVy6icbk4,7309
 arkindex_worker/worker/process.py,sha256=9TEHpMcBax1wc6PrWMMrdXe2uNfqyVj7n_dAYZRBGnY,1854
-arkindex_worker/worker/task.py,sha256=r1j7_qbdNu2Z8H8HbGzO3P3qdx-2N1pBbUPFDca0rqg,1519
-arkindex_worker/worker/training.py,sha256=H8FmCdzGcDW-WMMwcgvmZPlN5tPHwGo0BXn12qmzj8g,10875
-arkindex_worker/worker/transcription.py,sha256=52RY9kYsiR1sz9FxOigyo12Ker3VDbQ4U42gK9DpR3g,21146
+arkindex_worker/worker/task.py,sha256=nYfMSFm_d-4t8y4PO4HjFBnLsZf7IsDjkS7-A2Pgnac,1525
+arkindex_worker/worker/training.py,sha256=tyQOHcwv--_wdYz6CgLEe1YM7kwwwKN30LvGTsnWd78,10923
+arkindex_worker/worker/transcription.py,sha256=sw718R119tsLNY8inPMVeIilvFJo94fMbMtYgH0zTM8,21250
 arkindex_worker/worker/version.py,sha256=JIT7OI3Mo7RPkNrjOB9hfqrsG-FYygz_zi4l8PbkuAo,1960
 hooks/pre_gen_project.py,sha256=xQJERv3vv9VzIqcBHI281eeWLWREXUF4mMw7PvJHHXM,269
 tests/__init__.py,sha256=DG--S6IpGl399rzSAjDdHL76CkOIeZIjajCcyUSDhOQ,241
-tests/conftest.py,sha256=2ocZ2x-mZQrNe9zvWwhWk2_4ExdaBHIB74SvtDlExRE,21580
+tests/conftest.py,sha256=Z9amrKmVtFltzTUUm07fGDrT4m540biaTpjedmplyzc,21536
 tests/test_base_worker.py,sha256=2EIYcd_3f9O0zB5WiGIQV0Cn9wndLvnEnSfcAE1qWWU,30607
 tests/test_cache.py,sha256=ii0gyr0DrG7ChEs7pmT8hMdSguAOAcCze4bRMiFQxuk,10640
-tests/test_dataset_worker.py,sha256=gApYz0LArHr1cNn079_fa_BQABF6RVQYuM1Tc4m3NsQ,22089
+tests/test_dataset_worker.py,sha256=z8ydliUlwW2j-irgLAotJMacgJXkVvF5TgsWLyCn1Jo,22087
 tests/test_element.py,sha256=2G9M15TLxQRmvrWM9Kw2ucnElh4kSv_oF_5FYwwAxTY,13181
-tests/test_image.py,sha256=03E24JVa7TZJfuwQyfVEBe3RAq3R993IMl1AHXRr7zY,25497
+tests/test_image.py,sha256=mYyRfDXGLLzcQQtmaM7GR3jt7ScxsLLog16pUVHrH3M,27824
 tests/test_merge.py,sha256=TuOeUS0UCz66DPOQFFhc4NQBxIjZL9f5czi4XnvGrr4,8270
-tests/test_utils.py,sha256=_WJUPnt-pM_TQ0er4yjPZy-u_LePrHq1lxwk_teky7M,2544
+tests/test_utils.py,sha256=nYL1s2ViZoLoMiNpLGDaWwxf8dJ1D8aT522AO-PVaEQ,3607
 tests/test_elements_worker/__init__.py,sha256=Fh4nkbbyJSMv_VtjQxnWrOqTnxXaaWI8S9WU0VrzCHs,179
 tests/test_elements_worker/test_classification.py,sha256=nya7veSPR_O9G41Enodp2-o6AifMBcaSTWJP2vXSSJ4,30133
 tests/test_elements_worker/test_cli.py,sha256=a23i1pUDbXi23MUtbWwGEcLLrmc_YlrbDgOG3h66wLM,2620
@@ -43,19 +43,19 @@ tests/test_elements_worker/test_image.py,sha256=BljMNKgec_9a5bzNzFpYZIvSbuvwsWDf
 tests/test_elements_worker/test_metadata.py,sha256=Xfggy-vxw5DZ3hFKx3sB7OYb2d1tu1RiNK8fvKJIaBs,22294
 tests/test_elements_worker/test_process.py,sha256=y4RoVhPfyHzR795fw7-_FXElBcKo3fy4Ew_HI-kxJic,3088
 tests/test_elements_worker/test_task.py,sha256=wTUWqN9UhfKmJn3IcFY75EW4I1ulRhisflmY1kmP47s,5574
-tests/test_elements_worker/test_training.py,sha256=Qxi9EzGr_uKcn2Fh5aE6jNrq1K8QKLiOiSew4upASPs,8721
+tests/test_elements_worker/test_training.py,sha256=qgK7BLucddRzc8ePbQtY75x17QvGDEq5XCwgyyvmAJE,8717
 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_version.py,sha256=xqCgcgukTFJzkMgYfQG-8mTbu0o2fdYjWC07FktThfw,2125
-tests/test_elements_worker/test_worker.py,sha256=pLUgjyrrXrzVD6T-kdH1ppk5Yn_iDuI8JdFGweTEMXE,25156
+tests/test_elements_worker/test_worker.py,sha256=HDw_UQdiMUzlBd4-jRvC-B3pNrZmmpps4sfZ9a87JVY,25378
 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-0.4.0rc6.dist-info/LICENSE,sha256=NVshRi1efwVezMfW7xXYLrdDr2Li1AfwfGOd5WuH1kQ,1063
-arkindex_base_worker-0.4.0rc6.dist-info/METADATA,sha256=gJd_0X7A26nuBe2EsIPHwap1XV2KnJBq2QwjBBB3Wi0,3339
-arkindex_base_worker-0.4.0rc6.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
-arkindex_base_worker-0.4.0rc6.dist-info/top_level.txt,sha256=58NuslgxQC2vT4DiqZEgO4JqJRrYa2yeNI9QvkbfGQU,40
-arkindex_base_worker-0.4.0rc6.dist-info/RECORD,,
+arkindex_base_worker-0.5.0a2.dist-info/LICENSE,sha256=NVshRi1efwVezMfW7xXYLrdDr2Li1AfwfGOd5WuH1kQ,1063
+arkindex_base_worker-0.5.0a2.dist-info/METADATA,sha256=MAvsl683tKCx2pqnw1JmQXydhj3JSfcC1Kb3n37R4Kw,3336
+arkindex_base_worker-0.5.0a2.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+arkindex_base_worker-0.5.0a2.dist-info/top_level.txt,sha256=58NuslgxQC2vT4DiqZEgO4JqJRrYa2yeNI9QvkbfGQU,40
+arkindex_base_worker-0.5.0a2.dist-info/RECORD,,

{arkindex_base_worker-0.4.0rc6.dist-info → arkindex_base_worker-0.5.0a2.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (75.6.0)
+Generator: setuptools (75.8.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

arkindex_worker/__init__.py

@@ -1,3 +1,4 @@
+import importlib.metadata
 import logging
 
 logging.basicConfig(
@@ -5,3 +6,5 @@ logging.basicConfig(
     format="%(asctime)s %(levelname)s/%(name)s: %(message)s",
 )
 logger = logging.getLogger(__name__)
+
+VERSION = importlib.metadata.version("arkindex-base-worker")

arkindex_worker/cache.py

@@ -327,9 +327,9 @@ def check_version(cache_path: str | Path):
         except OperationalError:
             version = None
 
-        assert (
-            version == SQL_VERSION
-        ), f"The SQLite database {cache_path} does not have the correct cache version, it should be {SQL_VERSION}"
+        assert version == SQL_VERSION, (
+            f"The SQLite database {cache_path} does not have the correct cache version, it should be {SQL_VERSION}"
+        )
 
 
 def merge_parents_cache(paths: list, current_database: Path):

arkindex_worker/image.py

@@ -2,6 +2,7 @@
 Helper methods to download and open IIIF images, and manage polygons.
 """
 
+import base64
 import functools
 import os
 import re
@@ -14,6 +15,7 @@ from pathlib import Path
 from typing import TYPE_CHECKING
 
 import humanize
+import numpy as np
 import requests
 from PIL import Image
 from shapely.affinity import rotate, scale, translate
@@ -25,7 +27,7 @@ from tenacity import (
     wait_exponential,
 )
 
-from arkindex_worker import logger
+from arkindex_worker import VERSION, logger
 from arkindex_worker.utils import pluralize
 from teklia_toolbox.requests import should_verify_cert
 
@@ -39,14 +41,16 @@ DOWNLOAD_TIMEOUT = (30, 60)
 
 BoundingBox = namedtuple("BoundingBox", ["x", "y", "width", "height"])
 
+# Specific User-Agent to bypass potential server limitations
+IIIF_USER_AGENT = f"Teklia/Workers {VERSION}"
 # To parse IIIF Urls
 IIIF_URL = re.compile(r"\w+:\/{2}.+\/.+\/.+\/.+\/(?P<size>.+)\/!?\d+\/\w+\.\w+")
 # Full size of the region
 IIIF_FULL = "full"
 # Maximum size available
 IIIF_MAX = "max"
-# Ratio to resize image
-IMAGE_RATIO = [1, 0.9, 0.85, 0.80, 0.75, 0.70, 0.60, 0.50, 0.40, 0.30]
+# Ratios to resize images: 1.0, 0.95, [...], 0.1, 0.05
+IMAGE_RATIOS = np.arange(1, 0, -0.05).round(2).tolist()
 
 
 def update_pillow_image_size_limit(func):
@@ -206,44 +210,81 @@ def upload_image(image: Image, url: str) -> requests.Response:
 def resized_images(
     *args,
     element: "Element",
-    max_pixels: int | None = None,
+    max_pixels_short: int | None = None,
+    max_pixels_long: int | None = None,
     max_bytes: int | None = None,
+    use_base64: bool = False,
     **kwargs,
-) -> Iterator[Generator[tempfile.NamedTemporaryFile, None, None]]:
+) -> Iterator[Generator[tempfile._TemporaryFileWrapper | str]]:
     """
-    Build resized images according to the pixel and byte limits.
+    Build resized images according to pixel and byte limits.
 
     :param *args: Positional arguments passed to [arkindex_worker.models.Element.open_image_tempfile][].
     :param element: Element whose image needs to be resized.
-    :param max_pixels: Maximum pixel size of the resized images.
+    :param max_pixels_short: Maximum pixel size of the resized images' short side.
+    :param max_pixels_long: Maximum pixel size of the resized images' long side.
     :param max_bytes: Maximum byte size of the resized images.
+    :param use_base64: Whether or not to encode resized images in base64 before calculating their size.
     :param **kwargs: Keyword arguments passed to [arkindex_worker.models.Element.open_image_tempfile][].
-    :returns: An iterator of the temporary file of the resized image.
+    :returns: An iterator of temporary files for resized images OR an iterator of base64-encoded strings if `use_base64` is set.
     """
     _, _, element_width, element_height = polygon_bounding_box(element.polygon)
+    logger.info(
+        f"This element's image dimensions are ({element_width} x {element_height})."
+    )
+
+    portrait_format = element_width <= element_height
+    max_pixels_width, max_pixels_height = (
+        (max_pixels_short, max_pixels_long)
+        if portrait_format
+        else (max_pixels_long, max_pixels_short)
+    )
+
+    # The image dimension is already within the pixel limitation, no need to resize the image
+    if max_pixels_width and max_pixels_width >= element_width:
+        max_pixels_width = None
+    if max_pixels_height and max_pixels_height >= element_height:
+        max_pixels_height = None
 
-    logger.info(f"This element's image sizes are ({element_width} x {element_height}).")
-    if max_pixels and max(element_width, element_height) > max_pixels:
+    if (max_pixels_width and element_width > max_pixels_width) or (
+        max_pixels_height and element_height > max_pixels_height
+    ):
         logger.warning(
-            f"Maximum image input size supported is ({max_pixels} x {max_pixels})."
+            f"Maximum image dimensions supported are ({max_pixels_width or element_width} x {max_pixels_height or element_height})."
         )
         logger.warning("The image will be resized.")
 
-    element_pixel, param = (
-        (element_width, "max_width")
-        if element_width > element_height
-        else (element_height, "max_height")
-    )
+    # No limitations provided, we keep the image initial dimensions
+    if max_pixels_width is None and max_pixels_height is None:
+        open_image_param, max_value = (
+            ("max_height", element_height)
+            if portrait_format
+            else ("max_width", element_width)
+        )
+    # A limitation is only given for the height, we resize it
+    elif max_pixels_width is None:
+        open_image_param, max_value = ("max_height", max_pixels_height)
+    # A limitation is only given for the width, we resize it
+    elif max_pixels_height is None:
+        open_image_param, max_value = ("max_width", max_pixels_width)
+    # Limitations are provided for both sides:
+    # - we resize only the one with the biggest scale factor
+    # - the remaining one will automatically fall within the other limitation
+    else:
+        width_rescaling_factor = element_width / max_pixels_width
+        height_rescaling_factor = element_height / max_pixels_height
+        open_image_param, max_value = (
+            ("max_height", max_pixels_height)
+            if height_rescaling_factor > width_rescaling_factor
+            else ("max_width", max_pixels_width)
+        )
 
-    for resized_pixel in sorted(
-        set(
-            min(round(ratio * element_pixel), max_pixels or element_pixel)
-            for ratio in IMAGE_RATIO
-        ),
-        reverse=True,
-    ):
+    resized_pixels = set(
+        min(round(ratio * max_value), max_value) for ratio in IMAGE_RATIOS
+    )
+    for resized_pixel in sorted(resized_pixels, reverse=True):
         with element.open_image_tempfile(
-            *args, **{**kwargs, param: resized_pixel}
+            *args, **{**kwargs, open_image_param: resized_pixel}
         ) as image:
             pillow_image = Image.open(image)
             if (
@@ -254,8 +295,12 @@ def resized_images(
                     f"The image was resized to ({pillow_image.width} x {pillow_image.height})."
                 )
 
-            # The image is still too large
             image_size = Path(image.name).stat().st_size
+            if use_base64:
+                image = base64.b64encode(Path(image.name).read_bytes()).decode("utf-8")
+                image_size = len(image)
+
+            # The image is still too heavy
             if max_bytes and image_size > max_bytes:
                 logger.warning(f"The image size is {humanize.naturalsize(image_size)}.")
                 logger.warning(
@@ -283,7 +328,7 @@ def polygon_bounding_box(polygon: list[list[int | float]]) -> BoundingBox:
 def _retry_log(retry_state, *args, **kwargs):
     logger.warning(
         f"Request to {retry_state.args[0]} failed ({repr(retry_state.outcome.exception())}), "
-        f'retrying in {retry_state.idle_for} {pluralize("second", retry_state.idle_for)}'
+        f"retrying in {retry_state.idle_for} {pluralize('second', retry_state.idle_for)}"
     )
 
 
@@ -296,7 +341,12 @@ def _retry_log(retry_state, *args, **kwargs):
 )
 def _retried_request(url, *args, method=requests.get, **kwargs):
     resp = method(
-        url, *args, timeout=DOWNLOAD_TIMEOUT, verify=should_verify_cert(url), **kwargs
+        url,
+        *args,
+        headers={"User-Agent": IIIF_USER_AGENT},
+        timeout=DOWNLOAD_TIMEOUT,
+        verify=should_verify_cert(url),
+        **kwargs,
     )
     resp.raise_for_status()
     return resp
@@ -316,9 +366,9 @@ def download_tiles(url: str) -> Image:
 
     image_width, image_height = info.get("width"), info.get("height")
     assert image_width and image_height, "Missing image dimensions in info.json"
-    assert info.get(
-        "tiles"
-    ), "Image cannot be retrieved at full size and tiles are not supported"
+    assert info.get("tiles"), (
+        "Image cannot be retrieved at full size and tiles are not supported"
+    )
 
     # Take the biggest available tile size
     tile = sorted(info["tiles"], key=lambda tile: tile.get("width", 0), reverse=True)[0]
@@ -392,15 +442,15 @@ def trim_polygon(
        is entirely outside of the image's bounds.
     """
 
-    assert isinstance(
-        polygon, list | tuple
-    ), "Input polygon must be a valid list or tuple of points."
-    assert all(
-        isinstance(point, list | tuple) for point in polygon
-    ), "Polygon points must be tuples or lists."
-    assert all(
-        len(point) == 2 for point in polygon
-    ), "Polygon points must be tuples or lists of 2 elements."
+    assert isinstance(polygon, list | tuple), (
+        "Input polygon must be a valid list or tuple of points."
+    )
+    assert all(isinstance(point, list | tuple) for point in polygon), (
+        "Polygon points must be tuples or lists."
+    )
+    assert all(len(point) == 2 for point in polygon), (
+        "Polygon points must be tuples or lists of 2 elements."
+    )
     assert all(
         isinstance(point[0], int) and isinstance(point[1], int) for point in polygon
     ), "Polygon point coordinates must be integers."
@@ -451,12 +501,12 @@ def revert_orientation(
     from arkindex_worker.cache import CachedElement
     from arkindex_worker.models import Element
 
-    assert element and isinstance(
-        element, Element | CachedElement
-    ), "element shouldn't be null and should be an Element or CachedElement"
-    assert polygon and isinstance(
-        polygon, list
-    ), "polygon shouldn't be null and should be a list"
+    assert element and isinstance(element, Element | CachedElement), (
+        "element shouldn't be null and should be an Element or CachedElement"
+    )
+    assert polygon and isinstance(polygon, list), (
+        "polygon shouldn't be null and should be a list"
+    )
     assert isinstance(reverse, bool), "Reverse should be a bool"
     # Rotating with Pillow can cause it to move the image around, as the image cannot have negative coordinates
     # and must be a rectangle.  This means the origin point of any coordinates from an image is invalid, and the
@@ -464,9 +514,9 @@ def revert_orientation(
     # To properly undo the mirroring and rotation implicitly applied by open_image, we first need to find the center
     # of the rotated bounding box.
     if isinstance(element, Element):
-        assert (
-            element.zone and element.zone.polygon
-        ), "element should have a zone and a polygon"
+        assert element.zone and element.zone.polygon, (
+            "element should have a zone and a polygon"
+        )
         parent_ring = LinearRing(element.zone.polygon)
     elif isinstance(element, CachedElement):
         assert element.polygon, "cached element should have a polygon"

arkindex_worker/utils.py

@@ -243,11 +243,12 @@ def batch_publication(func: Callable) -> Callable:
         bound_func.apply_defaults()
         batch_size = bound_func.arguments.get("batch_size")
         assert (
-            batch_size and isinstance(batch_size, int) and batch_size > 0
+            batch_size is not None and isinstance(batch_size, int) and batch_size > 0
         ), "batch_size shouldn't be null and should be a strictly positive integer"
 
         return func(self, *args, **kwargs)
 
+    wrapper.__name__ = func.__name__
     return wrapper
 
 

arkindex_worker/worker/__init__.py

@@ -82,9 +82,9 @@ class ElementsWorker(
         :return: An iterable of [CachedElement][arkindex_worker.cache.CachedElement] when cache support is enabled,
            or a list of strings representing element IDs otherwise.
         """
-        assert not (
-            self.args.elements_list and self.args.element
-        ), "elements-list and element CLI args shouldn't be both set"
+        assert not (self.args.elements_list and self.args.element), (
+            "elements-list and element CLI args shouldn't be both set"
+        )
 
         def invalid_element_id(value: str) -> bool:
             """
@@ -125,9 +125,9 @@ class ElementsWorker(
             return {item["id"] for item in self.list_process_elements()}
 
         invalid_element_ids = list(filter(invalid_element_id, out))
-        assert (
-            not invalid_element_ids
-        ), f"These element IDs are invalid: {', '.join(invalid_element_ids)}"
+        assert not invalid_element_ids, (
+            f"These element IDs are invalid: {', '.join(invalid_element_ids)}"
+        )
 
         return out
 
@@ -144,9 +144,9 @@ class ElementsWorker(
             # Worker activities are also disabled when running an ElementsWorker in a Dataset process
             # and when running export processes.
             return False
-        assert (
-            self.process_information
-        ), "Worker must be configured to access its process activity state"
+        assert self.process_information, (
+            "Worker must be configured to access its process activity state"
+        )
         return self.process_information.get("activity_state") == "ready"
 
     def run(self):
@@ -221,7 +221,7 @@ class ElementsWorker(
                     with contextlib.suppress(Exception):
                         self.update_activity(element.id, ActivityState.Error)
 
-        message = f'Ran on {count} {pluralize("element", count)}: {count - failed} completed, {failed} failed'
+        message = f"Ran on {count} {pluralize('element', count)}: {count - failed} completed, {failed} failed"
         if failed:
             logger.error(message)
             if failed >= count:  # Everything failed!
@@ -256,9 +256,9 @@ class ElementsWorker(
             )
             return True
 
-        assert element_id and isinstance(
-            element_id, uuid.UUID | str
-        ), "element_id shouldn't be null and should be an UUID or str"
+        assert element_id and isinstance(element_id, uuid.UUID | str), (
+            "element_id shouldn't be null and should be an UUID or str"
+        )
         assert isinstance(state, ActivityState), "state should be an ActivityState"
 
         try:
@@ -382,9 +382,9 @@ class DatasetWorker(DatasetMixin, BaseWorker, TaskMixin):
         failed = 0
         for i, dataset_set in enumerate(dataset_sets, start=1):
             try:
-                assert (
-                    dataset_set.dataset.state == DatasetState.Complete.value
-                ), "When processing a set, its dataset state should be Complete."
+                assert dataset_set.dataset.state == DatasetState.Complete.value, (
+                    "When processing a set, its dataset state should be Complete."
+                )
 
                 logger.info(f"Retrieving data for {dataset_set} ({i}/{count})")
                 self.download_dataset_artifact(dataset_set.dataset)
@@ -405,7 +405,7 @@ class DatasetWorker(DatasetMixin, BaseWorker, TaskMixin):
         # Cleanup the latest downloaded dataset artifact
         self.cleanup_downloaded_artifact()
 
-        message = f'Ran on {count} {pluralize("set", count)}: {count - failed} completed, {failed} failed'
+        message = f"Ran on {count} {pluralize('set', count)}: {count - failed} completed, {failed} failed"
         if failed:
             logger.error(message)
             if failed >= count:  # Everything failed!

arkindex_worker/worker/base.py

@@ -305,9 +305,9 @@ class BaseWorker:
 
         if self.use_cache:
             if self.args.database is not None:
-                assert (
-                    self.args.database.is_file()
-                ), f"Database in {self.args.database} does not exist"
+                assert self.args.database.is_file(), (
+                    f"Database in {self.args.database} does not exist"
+                )
                 self.cache_path = self.args.database
             else:
                 cache_dir = self.task_data_dir / self.task_id
@@ -378,9 +378,9 @@ class BaseWorker:
                 gpg = gnupg.GPG()
                 with path.open("rb") as gpg_file:
                     decrypted = gpg.decrypt_file(gpg_file)
-                assert (
-                    decrypted.ok
-                ), f"GPG error: {decrypted.status} - {decrypted.stderr}"
+                assert decrypted.ok, (
+                    f"GPG error: {decrypted.status} - {decrypted.stderr}"
+                )
                 secret = decrypted.data.decode("utf-8")
                 logging.info(f"Loaded local secret {name}")
             except Exception as e:

arkindex_worker/worker/classification.py

@@ -27,7 +27,7 @@ class ClassificationMixin:
         )
         self.classes = {ml_class["name"]: ml_class["id"] for ml_class in corpus_classes}
         logger.info(
-            f'Loaded {len(self.classes)} ML {pluralize("class", len(self.classes))} in corpus ({self.corpus_id})'
+            f"Loaded {len(self.classes)} ML {pluralize('class', len(self.classes))} in corpus ({self.corpus_id})"
         )
 
     def get_ml_class_id(self, ml_class: str) -> str:
@@ -60,9 +60,9 @@ class ClassificationMixin:
                     f"Reloading corpus classes to see if {ml_class} already exists"
                 )
                 self.load_corpus_classes()
-                assert (
-                    ml_class in self.classes
-                ), "Missing class {ml_class} even after reloading"
+                assert ml_class in self.classes, (
+                    "Missing class {ml_class} even after reloading"
+                )
                 ml_class_id = self.classes[ml_class]
 
         return ml_class_id
@@ -86,9 +86,9 @@ class ClassificationMixin:
             ),
             None,
         )
-        assert (
-            ml_class_name is not None
-        ), f"Missing class with id ({ml_class_id}) in corpus ({self.corpus_id})"
+        assert ml_class_name is not None, (
+            f"Missing class with id ({ml_class_id}) in corpus ({self.corpus_id})"
+        )
         return ml_class_name
 
     def create_classification(
@@ -107,18 +107,18 @@ class ClassificationMixin:
         :param high_confidence: Whether or not the classification is of high confidence.
         :returns: The created classification, as returned by the ``CreateClassification`` API endpoint.
         """
-        assert element and isinstance(
-            element, Element | CachedElement
-        ), "element shouldn't be null and should be an Element or CachedElement"
-        assert ml_class and isinstance(
-            ml_class, str
-        ), "ml_class shouldn't be null and should be of type str"
-        assert (
-            isinstance(confidence, float) and 0 <= confidence <= 1
-        ), "confidence shouldn't be null and should be a float in [0..1] range"
-        assert isinstance(
-            high_confidence, bool
-        ), "high_confidence shouldn't be null and should be of type bool"
+        assert element and isinstance(element, Element | CachedElement), (
+            "element shouldn't be null and should be an Element or CachedElement"
+        )
+        assert ml_class and isinstance(ml_class, str), (
+            "ml_class shouldn't be null and should be of type str"
+        )
+        assert isinstance(confidence, float) and 0 <= confidence <= 1, (
+            "confidence shouldn't be null and should be a float in [0..1] range"
+        )
+        assert isinstance(high_confidence, bool), (
+            "high_confidence shouldn't be null and should be of type bool"
+        )
         if self.is_read_only:
             logger.warning(
                 "Cannot create classification as this worker is in read-only mode"
@@ -198,31 +198,33 @@ class ClassificationMixin:
         :returns: List of created classifications, as returned in the ``classifications`` field by
            the ``CreateClassifications`` API endpoint.
         """
-        assert element and isinstance(
-            element, Element | CachedElement
-        ), "element shouldn't be null and should be an Element or CachedElement"
-        assert classifications and isinstance(
-            classifications, list
-        ), "classifications shouldn't be null and should be of type list"
+        assert element and isinstance(element, Element | CachedElement), (
+            "element shouldn't be null and should be an Element or CachedElement"
+        )
+        assert classifications and isinstance(classifications, list), (
+            "classifications shouldn't be null and should be of type list"
+        )
 
         for index, classification in enumerate(classifications):
             ml_class = classification.get("ml_class")
-            assert (
-                ml_class and isinstance(ml_class, str)
-            ), f"Classification at index {index} in classifications: ml_class shouldn't be null and should be of type str"
+            assert ml_class and isinstance(ml_class, str), (
+                f"Classification at index {index} in classifications: ml_class shouldn't be null and should be of type str"
+            )
 
             confidence = classification.get("confidence")
             assert (
                 confidence is not None
                 and isinstance(confidence, float)
                 and 0 <= confidence <= 1
-            ), f"Classification at index {index} in classifications: confidence shouldn't be null and should be a float in [0..1] range"
+            ), (
+                f"Classification at index {index} in classifications: confidence shouldn't be null and should be a float in [0..1] range"
+            )
 
             high_confidence = classification.get("high_confidence")
             if high_confidence is not None:
-                assert isinstance(
-                    high_confidence, bool
-                ), f"Classification at index {index} in classifications: high_confidence should be of type bool"
+                assert isinstance(high_confidence, bool), (
+                    f"Classification at index {index} in classifications: high_confidence should be of type bool"
+                )
 
         if self.is_read_only:
             logger.warning(

arkindex_worker/worker/corpus.py

@@ -76,9 +76,9 @@ class CorpusMixin:
             key=itemgetter("updated"),
             reverse=True,
         )
-        assert (
-            len(exports) > 0
-        ), f'No available exports found for the corpus ({self.corpus_id}) with state "{CorpusExportState.Done.value.capitalize()}".'
+        assert len(exports) > 0, (
+            f'No available exports found for the corpus ({self.corpus_id}) with state "{CorpusExportState.Done.value.capitalize()}".'
+        )
 
         # Download latest export
         export_id: str = exports[0]["id"]