feat: add GlobalCaption to db/api for text annotation

apiv2/db_import/importers/annotation.py

@@ -106,7 +106,9 @@ def get_filters(self) -> dict[str, Any]:
     def get_finder_args(self) -> dict[str, Any]:
         return {
             "path": os.path.join(self.tomogram_voxel_spacing.s3_prefix, "Annotations/"),
-            "file_glob": "*/*.json",
+            # Use *[0-9].json to match only annotation metadata files (e.g., foo-1.0.json)
+            # and exclude annotation data files which have a _{shape} suffix (e.g., foo-1.0_globalcaption.json)
+            "file_glob": "*/*[0-9].json",
         }
 
 

apiv2/db_import/tests/conftest.py

@@ -133,5 +133,5 @@ def expected_dataset(http_prefix: str) -> dict[str, Any]:
         "key_photo_url": f"{http_prefix}/{DATASET_ID}/KeyPhoto/snapshot.png",
         "key_photo_thumbnail_url": f"{http_prefix}/{DATASET_ID}/KeyPhoto/thumbnail.png",
         "deposition_id": 300,
-        "file_size": 1374849.0,
+        "file_size": 1375370.0,
     }

apiv2/db_import/tests/test_db_annotation_import.py

@@ -54,6 +54,15 @@ def expected_annotations(http_prefix: str) -> list[dict[str, Any]]:
 def expected_annotation_files(http_prefix: str) -> list[dict[str, Any]]:
     path = f"{DATASET_ID}/RUN1/Reconstructions/VoxelSpacing12.300/Annotations/"
     return [
+        {
+            "tomogram_voxel_spacing_id": TOMOGRAM_VOXEL_ID1,
+            "s3_path": f"s3://test-public-bucket/{path}100-foo-1.0_globalcaption.json",
+            "https_path": f"{http_prefix}/{path}100-foo-1.0_globalcaption.json",
+            "source": "community",
+            "format": "saber",
+            "is_visualization_default": False,
+            "file_size": 0,
+        },
         {
             "tomogram_voxel_spacing_id": TOMOGRAM_VOXEL_ID1,
             "s3_path": f"s3://test-public-bucket/{path}100-foo-1.0_instancesegmask.mrc",

apiv2/schema/schema.yaml

@@ -285,6 +285,9 @@ enums:
       Mesh:
         text: Mesh
         description: A surface mesh volumes
+      GlobalCaption:
+        text: GlobalCaption
+        description: A text caption for the tomogram
   annotation_method_link_type_enum:
     name: annotation_method_link_type_enum
     description: Describes the type of link associated to the annotation method.
@@ -925,11 +928,11 @@ classes:
         annotations:
           cascade_delete: true
       shape_type:
-        description: The shape of the annotation (SegmentationMask, OrientedPoint, Point, InstanceSegmentation, Mesh, InstanceSegmentationMask)
+        description: The shape of the annotation (SegmentationMask, OrientedPoint, Point, InstanceSegmentation, Mesh, InstanceSegmentationMask, GlobalCaption)
         name: shape_type
         from_schema: cdp-dataset-config
         range: annotation_file_shape_type_enum
-        pattern: (^SegmentationMask$)|(^OrientedPoint$)|(^Point$)|(^InstanceSegmentation$)|(^Mesh$)|(^InstanceSegmentationMask$)
+        pattern: (^SegmentationMask$)|(^OrientedPoint$)|(^Point$)|(^InstanceSegmentation$)|(^Mesh$)|(^InstanceSegmentationMask$)|(^GlobalCaption$)
   Annotation:
     name: Annotation
     annotations:

test_infra/test_files/30001/RUN1/Reconstructions/VoxelSpacing12.300/Annotations/100/100-foo-1.0_globalcaption.json

@@ -0,0 +1,16 @@
+{
+    "captions": [
+        {
+            "id": 0,
+            "text": "A caption for tomogram X."
+        },
+        {
+            "id": 1,
+            "text": "A 2nd caption for tomogram X."
+        },
+        {
+            "id": 2,
+            "text": "A 3rd caption for tomogram X."
+        }
+    ]
+}

test_infra/test_files/30001/RUN1/Reconstructions/VoxelSpacing12.300/Annotations/100/foo-1.0.json

@@ -66,6 +66,12 @@
       "path": "30001/RUN1/Reconstructions/VoxelSpacing12.300/Annotations/100-foo-1.0_instancesegmask.zarr",
       "shape": "InstanceSegmentationMask",
       "is_visualization_default": false
+    },
+    {
+      "format": "saber",
+      "path": "30001/RUN1/Reconstructions/VoxelSpacing12.300/Annotations/100-foo-1.0_globalcaption.json",
+      "shape": "GlobalCaption",
+      "is_visualization_default": false
     }
   ],
   "method_links": [