arkindex-base-worker 0.5.1b5__tar.gz → 0.5.1rc1__tar.gz

{arkindex_base_worker-0.5.1b5 → arkindex_base_worker-0.5.1rc1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: arkindex-base-worker
-Version: 0.5.1b5
+Version: 0.5.1rc1
 Summary: Base Worker to easily build Arkindex ML workflows
 Author-email: Teklia <contact@teklia.com>
 Maintainer-email: Teklia <contact@teklia.com>

{arkindex_base_worker-0.5.1b5 → arkindex_base_worker-0.5.1rc1}/arkindex_base_worker.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: arkindex-base-worker
-Version: 0.5.1b5
+Version: 0.5.1rc1
 Summary: Base Worker to easily build Arkindex ML workflows
 Author-email: Teklia <contact@teklia.com>
 Maintainer-email: Teklia <contact@teklia.com>

{arkindex_base_worker-0.5.1b5 → arkindex_base_worker-0.5.1rc1}/arkindex_worker/cache.py

@@ -73,6 +73,7 @@ class CachedImage(Model):
     width = IntegerField()
     height = IntegerField()
     url = TextField()
+    version = IntegerField(default=2)
 
     class Meta:
         database = db
@@ -157,6 +158,10 @@ class CachedElement(Model):
             else:
                 resize = f"{max_width or ''},{max_height or ''}"
 
+        # Use `max` instead of `full` for IIIF 3, since `full` was deprecated in 2.1 then removed in 3.0
+        if self.image.version == 3 and resize == "full":
+            resize = "max"
+
         url = self.image.url
         if not url.endswith("/"):
             url += "/"
@@ -259,7 +264,7 @@ MODELS = [
     CachedDataset,
     CachedDatasetElement,
 ]
-SQL_VERSION = 4
+SQL_VERSION = 5
 
 
 def init_cache_db(path: Path):

{arkindex_base_worker-0.5.1b5 → arkindex_base_worker-0.5.1rc1}/arkindex_worker/image.py

@@ -366,6 +366,10 @@ def download_tiles(url: str) -> Image:
     logger.debug("Downloading image information")
     info = _retried_request(url + "info.json").json()
 
+    # Use `max` instead of `full` for IIIF 3, since `full` was deprecated in 2.1 then removed in 3.0
+    # With IIIF 3, the image's ID will be at `id`, while IIIF 2 will use `@id``
+    resize = "max" if "id" in info else "full"
+
     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"), (
@@ -391,7 +395,7 @@ def download_tiles(url: str) -> Image:
 
             logger.debug(f"Downloading tile {tile_x},{tile_y}")
             resp = _retried_request(
-                f"{url}{region_x},{region_y},{region_width},{region_height}/full/0/default.jpg"
+                f"{url}{region_x},{region_y},{region_width},{region_height}/{resize}/0/default.jpg"
             )
 
             tile_img = Image.open(BytesIO(resp.content))

{arkindex_base_worker-0.5.1b5 → arkindex_base_worker-0.5.1rc1}/arkindex_worker/models.py

@@ -87,6 +87,11 @@ class Element(MagicDict):
         url = self.zone.image.get("s3_url")
         if url:
             return url
+
+        # Use `max` instead of `full` for IIIF 3, since `full` was deprecated in 2.1 then removed in 3.0
+        if self.zone.image.server.get("version", 2) == 3 and size == "full":
+            size = "max"
+
         url = self.zone.image.url
         if not url.endswith("/"):
             url += "/"

{arkindex_base_worker-0.5.1b5 → arkindex_base_worker-0.5.1rc1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "arkindex-base-worker"
-version = "0.5.1b5"
+version = "0.5.1rc1"
 description = "Base Worker to easily build Arkindex ML workflows"
 license = { file = "LICENSE" }
 dependencies = [

{arkindex_base_worker-0.5.1b5 → arkindex_base_worker-0.5.1rc1}/tests/test_cache.py

@@ -60,7 +60,7 @@ def test_create_tables(tmp_path):
 CREATE TABLE "dataset_elements" ("id" TEXT NOT NULL PRIMARY KEY, "element_id" TEXT NOT NULL, "dataset_id" TEXT NOT NULL, "set_name" VARCHAR(255) NOT NULL, FOREIGN KEY ("element_id") REFERENCES "elements" ("id"), FOREIGN KEY ("dataset_id") REFERENCES "datasets" ("id"))
 CREATE TABLE "datasets" ("id" TEXT NOT NULL PRIMARY KEY, "name" VARCHAR(255) NOT NULL, "state" VARCHAR(255) NOT NULL DEFAULT 'open', "sets" TEXT NOT NULL)
 CREATE TABLE "elements" ("id" TEXT NOT NULL PRIMARY KEY, "parent_id" TEXT, "type" VARCHAR(50) NOT NULL, "image_id" TEXT, "polygon" text, "rotation_angle" INTEGER NOT NULL, "mirrored" INTEGER NOT NULL, "initial" INTEGER NOT NULL, "worker_version_id" TEXT, "worker_run_id" TEXT, "confidence" REAL, FOREIGN KEY ("image_id") REFERENCES "images" ("id"))
-CREATE TABLE "images" ("id" TEXT NOT NULL PRIMARY KEY, "width" INTEGER NOT NULL, "height" INTEGER NOT NULL, "url" TEXT NOT NULL)
+CREATE TABLE "images" ("id" TEXT NOT NULL PRIMARY KEY, "width" INTEGER NOT NULL, "height" INTEGER NOT NULL, "url" TEXT NOT NULL, "version" INTEGER NOT NULL)
 CREATE TABLE "transcription_entities" ("transcription_id" TEXT NOT NULL, "type" VARCHAR(50) NOT NULL, "offset" INTEGER NOT NULL CHECK (offset >= 0), "length" INTEGER NOT NULL CHECK (length > 0), "worker_run_id" TEXT, "confidence" REAL, PRIMARY KEY ("transcription_id", "type"), FOREIGN KEY ("transcription_id") REFERENCES "transcriptions" ("id"))
 CREATE TABLE "transcriptions" ("id" TEXT NOT NULL PRIMARY KEY, "element_id" TEXT NOT NULL, "text" TEXT NOT NULL, "confidence" REAL, "orientation" VARCHAR(50) NOT NULL, "worker_version_id" TEXT, "worker_run_id" TEXT, FOREIGN KEY ("element_id") REFERENCES "elements" ("id"))"""
 

{arkindex_base_worker-0.5.1b5 → arkindex_base_worker-0.5.1rc1}/tests/test_element.py

@@ -5,26 +5,36 @@ from arkindex_worker.cache import CachedElement
 from arkindex_worker.models import Element
 
 
-def test_no_image_url():
+@pytest.mark.parametrize(
+    ("zone", "expected_url"),
+    [
+        (None, None),
+        (
+            {"image": {"url": "http://something/", "server": {"version": 2}}},
+            "http://something/full/full/0/default.jpg",
+        ),
+        (
+            {"image": {"url": "http://something", "server": {"version": 2}}},
+            "http://something/full/full/0/default.jpg",
+        ),
+        (
+            {"image": {"url": "http://something/", "server": {"version": 3}}},
+            "http://something/full/max/0/default.jpg",
+        ),
+    ],
+)
+def test_image_url(zone, expected_url):
     url = Element({"zone": None}).image_url()
     assert not url
 
 
-def test_image_url_iiif():
-    url = Element({"zone": {"image": {"url": "http://something/"}}}).image_url()
-    assert url == "http://something/full/full/0/default.jpg"
-
-
 def test_image_url_iiif_resize():
-    url = Element({"zone": {"image": {"url": "http://something/"}}}).image_url(500)
+    url = Element(
+        {"zone": {"image": {"url": "http://something/", "server": {"version": 2}}}}
+    ).image_url(500)
     assert url == "http://something/full/500/0/default.jpg"
 
 
-def test_image_url_iiif_append_slash():
-    url = Element({"zone": {"image": {"url": "http://something"}}}).image_url()
-    assert url == "http://something/full/full/0/default.jpg"
-
-
 def test_image_url_s3():
     url = Element(
         {
@@ -418,6 +428,36 @@ def test_open_image_rotation_mirror(mocker):
     )
 
 
+def test_open_image_iiif_3(mocker):
+    open_mock = mocker.patch(
+        "arkindex_worker.image.open_image", return_value="an image!"
+    )
+    elt = Element(
+        {
+            "zone": {
+                "image": {
+                    "url": "http://something",
+                    "server": {
+                        "max_width": None,
+                        "max_height": None,
+                        "version": 3,
+                    },
+                },
+                "polygon": [[0, 0], [181, 0], [181, 240], [0, 240], [0, 0]],
+            },
+            "rotation_angle": 0,
+            "mirrored": False,
+        },
+    )
+    assert elt.open_image(use_full_image=True) == "an image!"
+    assert open_mock.call_count == 1
+    assert open_mock.call_args == mocker.call(
+        "http://something/full/max/0/default.jpg",
+        rotation_angle=0,
+        mirrored=False,
+    )
+
+
 def test_setattr_setitem():
     element = Element({"name": "something"})
     element.type = "page"

{arkindex_base_worker-0.5.1b5 → arkindex_base_worker-0.5.1rc1}/tests/test_image.py

@@ -113,21 +113,37 @@ def test_update_pillow_image_size_limit(max_image_pixels, expected_image_pixels)
     assert Image.MAX_IMAGE_PIXELS == MAX_IMAGE_PIXELS
 
 
-def test_download_tiles(responses):
+@pytest.mark.parametrize(
+    ("id_key", "resize"),
+    [
+        # IIIF version 2
+        ("@id", "full"),
+        # IIIF version 3
+        ("id", "max"),
+    ],
+)
+def test_download_tiles(responses, id_key, resize):
     expected = Image.open(FULL_IMAGE).convert("RGB")
     tile_bytes = TILE.read_bytes()
 
     responses.add(
         responses.GET,
         "http://nowhere/info.json",
-        json={"width": 543, "height": 720, "tiles": [{"width": 181, "height": 240}]},
+        json={
+            id_key: "http://nowhere",
+            "width": 543,
+            "height": 720,
+            "tiles": [
+                {"width": 181, "height": 240},
+            ],
+        },
     )
 
     for x in (0, 181, 362):
         for y in (0, 240, 480):
             responses.add(
                 responses.GET,
-                f"http://nowhere/{x},{y},181,240/full/0/default.jpg",
+                f"http://nowhere/{x},{y},181,240/{resize}/0/default.jpg",
                 body=tile_bytes,
             )