VERSION

-0.2.1-beta1
+0.2.1-beta2

arkindex_worker/cache.py

@@ -72,19 +72,28 @@ class CachedElement(Model):
         """
         if not self.image_id or not self.polygon:
             raise ValueError(f"Element {self.id} has no image")
+
+        # Always fetch the image from the bounding box when size differs from full image
+        bounding_box = polygon_bounding_box(self.polygon)
+        if (
+            bounding_box.width != self.image.width
+            or bounding_box.height != self.image.height
+        ):
+            box = f"{bounding_box.x},{bounding_box.y},{bounding_box.x + bounding_box.width},{bounding_box.y + bounding_box.height}"
+        else:
+            box = "full"
+
         if max_size is None:
             resize = "full"
         else:
-            bounding_box = polygon_bounding_box(self.polygon)
             # Do not resize for polygons that do not exactly match the images
+            # as the resize is made directly by the IIIF server using the box parameter
             if (
                 bounding_box.width != self.image.width
                 or bounding_box.height != self.image.height
             ):
                 resize = "full"
-                logger.warning(
-                    "Only full size elements covered, downloading full size image"
-                )
+
             # Do not resize when the image is below the maximum size
             elif self.image.width <= max_size and self.image.height <= max_size:
                 resize = "full"
@@ -99,7 +108,7 @@ class CachedElement(Model):
         if not url.endswith("/"):
             url += "/"
 
-        return open_image(f"{url}full/{resize}/0/default.jpg", *args, **kwargs)
+        return open_image(f"{url}{box}/{resize}/0/default.jpg", *args, **kwargs)
 
 
 class CachedTranscription(Model):

arkindex_worker/worker/__init__.py

@@ -86,6 +86,8 @@ class ElementsWorker(
 
     @property
     def store_activity(self):
+        if self.args.dev:
+            return False
         assert (
             self.process_information
         ), "Worker must be configured to access its process activity state"

arkindex_worker/worker/base.py

@@ -61,6 +61,7 @@ class BaseWorker(object):
 
         logger.info(f"Worker will use {self.work_dir} as working directory")
 
+        self.process_information = None
         self.support_cache = support_cache
         # use_cache will be updated in configure() if the cache is supported and if there
         # is at least one available sqlite database either given or in the parent tasks
@@ -68,8 +69,8 @@ class BaseWorker(object):
 
     @property
     def is_read_only(self):
-        """Worker cannot publish anything without a worker version ID"""
-        return self.worker_version_id is None
+        """Worker cannot publish anything in dev mode or without a worker version ID"""
+        return self.args.dev or self.worker_version_id is None
 
     def configure(self):
         """
@@ -95,6 +96,15 @@ class BaseWorker(object):
             action="store_true",
             default=False,
         )
+        self.parser.add_argument(
+            "--dev",
+            help=(
+                "Run worker in developer mode. "
+                "Worker will be in read-only state even if a worker_version is supplied. "
+                "ARKINDEX_PROCESS_ID environment variable is not required with this flag."
+            ),
+            action="store_true",
+        )
 
         # Call potential extra arguments
         self.add_arguments()
@@ -116,13 +126,14 @@ class BaseWorker(object):
         logger.debug(f"Connected as {user['display_name']} - {user['email']}")
         self.features = user["features"]
 
-        # Load process information
-        assert os.environ.get(
-            "ARKINDEX_PROCESS_ID"
-        ), "ARKINDEX_PROCESS_ID environment variable is not defined"
-        self.process_information = self.request(
-            "RetrieveDataImport", id=os.environ["ARKINDEX_PROCESS_ID"]
-        )
+        # Load process information except in developer mode
+        if not self.args.dev:
+            assert os.environ.get(
+                "ARKINDEX_PROCESS_ID"
+            ), "ARKINDEX_PROCESS_ID environment variable is not defined"
+            self.process_information = self.request(
+                "RetrieveDataImport", id=os.environ["ARKINDEX_PROCESS_ID"]
+            )
 
         if self.worker_version_id:
             # Retrieve initial configuration from API

requirements.txt

 arkindex-client==1.0.6
 peewee==3.14.4
-Pillow==8.1.0
-python-gitlab==2.6.0
+Pillow==8.2.0
+python-gitlab==2.7.1
 python-gnupg==0.4.7
-sh==1.14.1
+sh==1.14.2
 tenacity==7.0.0

tests-requirements.txt

-pytest==6.2.3
-pytest-mock==3.5.1
-pytest-responses==0.4.0
+pytest==6.2.4
+pytest-mock==3.6.1
+pytest-responses==0.5.0

tests/test_base_worker.py

@@ -119,6 +119,20 @@ def test_cli_arg_verbose_given(mocker, mock_config_api):
     logger.setLevel(logging.NOTSET)
 
 
+def test_configure_dev_mode(mocker, mock_user_api, mock_worker_version_api):
+    """
+    Configuring a worker in developer mode avoid retrieving process information
+    """
+    worker = BaseWorker()
+    mocker.patch.object(sys, "argv", ["worker", "--dev"])
+    worker.configure()
+
+    assert worker.args.dev is True
+    assert worker.process_information is None
+    assert worker.worker_version_id == "12341234-1234-1234-1234-123412341234"
+    assert worker.is_read_only is True
+
+
 def test_load_missing_secret():
     worker = BaseWorker()
     worker.api_client = MockApiClient()

tests/test_cache.py

@@ -78,13 +78,15 @@ CREATE TABLE "transcriptions" ("id" TEXT NOT NULL PRIMARY KEY, "element_id" TEXT
     [
         # No max_size: no resize
         (400, 600, 400, 600, None, "http://something/full/full/0/default.jpg"),
+        # No max_size: resize on bbox
+        (400, 600, 200, 100, None, "http://something/0,0,200,100/full/0/default.jpg"),
         # max_size equal to the image size, no resize
         (400, 600, 400, 600, 600, "http://something/full/full/0/default.jpg"),
         (600, 400, 600, 400, 600, "http://something/full/full/0/default.jpg"),
         (400, 400, 400, 400, 400, "http://something/full/full/0/default.jpg"),
         # max_size is smaller than the image, resize
         (400, 600, 400, 600, 400, "http://something/full/266,400/0/default.jpg"),
-        (400, 600, 200, 600, 400, "http://something/full/266,400/0/default.jpg"),
+        (400, 600, 200, 600, 400, "http://something/0,0,200,600/full/0/default.jpg"),
         (600, 400, 600, 400, 400, "http://something/full/400,266/0/default.jpg"),
         (400, 400, 400, 400, 200, "http://something/full/200,200/0/default.jpg"),
         # max_size above the image size, no resize
@@ -118,9 +120,9 @@ def test_element_open_image(
         image=image,
         polygon=[
             [0, 0],
-            [image_width, 0],
-            [image_width, image_height],
-            [0, image_height],
+            [polygon_width, 0],
+            [polygon_width, polygon_height],
+            [0, polygon_height],
             [0, 0],
         ],
     )

tests/test_elements_worker/test_elements.py

@@ -91,6 +91,7 @@ def test_list_elements_element_arg(mocker, mock_elements_worker):
             verbose=False,
             elements_list=None,
             database=None,
+            dev=False,
         ),
     )
 
@@ -121,6 +122,7 @@ def test_list_elements_both_args_error(mocker, mock_elements_worker):
             verbose=False,
             elements_list=open(path),
             database=None,
+            dev=False,
         ),
     )
 
@@ -144,6 +146,7 @@ def test_database_arg(mocker, mock_elements_worker, tmp_path):
             verbose=False,
             elements_list=None,
             database=str(database_path),
+            dev=False,
         ),
     )
 

tests/test_elements_worker/test_worker.py

@@ -100,6 +100,20 @@ def test_activities_disabled(
     ] == BASE_API_CALLS
 
 
+def test_activities_dev_mode(mocker, mock_user_api, mock_worker_version_api):
+    """
+    Worker activities are not stored in dev mode
+    """
+    worker = ElementsWorker()
+    mocker.patch.object(sys, "argv", ["worker", "--dev"])
+    worker.configure()
+
+    assert worker.args.dev is True
+    assert worker.process_information is None
+    assert worker.is_read_only is True
+    assert worker.store_activity is False
+
+
 def test_update_call(responses, mock_elements_worker, mock_process_api):
     """Test an update call with feature enabled triggers an API call"""
     responses.add(

tests/test_git.py

@@ -435,7 +435,7 @@ def test_create_merge_request__no_retry_5xx_error(
 def test_create_merge_request__retry_5xx_error(
     fake_responses, fake_gitlab_helper_factory
 ):
-    request_url = f"https://gitlab.com/api/v4/projects/{PROJECT_ID}/merge_requests?retry_transient_errors=True"
+    request_url = f"https://gitlab.com/api/v4/projects/{PROJECT_ID}/merge_requests"
 
     fake_responses.add(
         fake_responses.POST,

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

-arkindex-base-worker==0.1.14
+arkindex-base-worker==0.2.0