E2E tests for converting annotations (#949)

.gitignore

@@ -9,7 +9,6 @@ data/
 !tests/darwin/data
 darwin_py.egg-info/PKG-INFO
 
-*.png
 *.jpeg
 *.jpg
 *.bpm

e2e_tests/cli/convert/test_convert.py → e2e_tests/cli/test_convert.py

@@ -4,13 +4,14 @@
 
 import orjson as json
 import pytest
+import xml.etree.ElementTree as ET
 
 from e2e_tests.helpers import assert_cli, run_cli_command
 
 
 class TestExportCli:
     this_file_path = Path(dirname(__file__)).absolute()
-    data_path = (this_file_path / ".." / ".." / "data").resolve()
+    data_path = (this_file_path / ".." / "data" / "convert").resolve()
 
     @pytest.fixture(autouse=True)
     def config(self) -> None:
@@ -33,10 +34,10 @@ def compare_directories(self, path: Path, expected_path: Path) -> None:
                     continue
 
                 # Compare files
-                with file.open("r") as f:
+                with file.open("rb") as f:
                     content = f.read()
 
-                with Path(expected_path / file.name).open() as f:
+                with Path(expected_path / file.name).open("rb") as f:
                     expected_content = f.read()
 
                 if content != expected_content:
@@ -52,6 +53,14 @@ def compare_directories(self, path: Path, expected_path: Path) -> None:
         [
             ("yolo_segmented", data_path / "yolov8/from", data_path / "yolov8/to"),
             ("yolo", data_path / "yolo/from", data_path / "yolo/to"),
+            ("cvat", data_path / "cvat/from", data_path / "cvat/to"),
+            ("pascalvoc", data_path / "pascalvoc/from", data_path / "pascalvoc/to"),
+            ("nifti", data_path / "nifti/from", data_path / "nifti/to"),
+            (
+                "instance_mask",
+                data_path / "instance_mask/from",
+                data_path / "instance_mask/to",
+            ),
             pytest.param(
                 "coco",
                 data_path / "coco/from",
@@ -87,30 +96,64 @@ def test_darwin_convert(
         result = run_cli_command(
             f"darwin convert {format} {str(input_path)} {str(tmp_path)}"
         )
-        if format == "coco":
-            self.patch_coco(tmp_path / "output.json")
+        self.patch_format(format, tmp_path)
         assert_cli(result, 0)
         self.compare_directories(expectation_path, tmp_path)
 
+    def patch_format(self, format: str, path: Path) -> None:
+        """
+        Patch files based on format to match the expected output.
+        """
+        patch_methods = {
+            "coco": self.patch_coco,
+            "cvat": self.patch_cvat,
+        }
+        patch_method = patch_methods.get(format)
+        if patch_method:
+            patch_method(path)
+
     def patch_coco(self, path: Path) -> None:
         """
         Patch coco file to match the expected output, includes changes to year and date_created,
         wrapped in try except so that format errors are still caught later with correct error messages
         """
         try:
-            with open(path, "r") as f:
+            with open(path / "output.json", "r") as f:
                 contents = f.read()
                 temp = json.loads(contents)
                 temp["info"]["year"] = 2023
                 temp["info"]["date_created"] = "2023/12/05"
-            with open(path, "w") as f:
+            with open(path / "output.json", "w") as f:
                 op = json.dumps(
                     temp, option=json.OPT_INDENT_2 | json.OPT_SERIALIZE_NUMPY
                 ).decode("utf-8")
                 f.write(op)
         except Exception:
             print(f"Error patching {path}")
 
+    def patch_cvat(self, path: Path) -> None:
+        """
+        Patch cvat file to match the expected output.
+        """
+        try:
+            tree = ET.parse(path / "output.xml")
+            root = tree.getroot()
+            # Adjust the required fields
+            dumped_elem = root.find(".//meta/dumped")
+            if dumped_elem is not None:
+                dumped_elem.text = "2024-10-25 10:33:01.789498+00:00"
+            created_elem = root.find(".//meta/task/created")
+            if created_elem is not None:
+                created_elem.text = "2024-10-25 10:33:01.789603+00:00"
+            updated_elem = root.find(".//meta/task/updated")
+            if updated_elem is not None:
+                updated_elem.text = "2024-10-25 10:33:01.789608+00:00"
+            tree.write(path / "output.xml")
+        except ET.ParseError:
+            print(f"Error parsing XML in {path}")
+        except Exception as e:
+            print(f"Error patching {path}: {e}")
+
 
 if __name__ == "__main__":
     pytest.main(["-vv", "-s", __file__])

e2e_tests/data/convert/cvat/from/000000021295.json

@@ -0,0 +1,67 @@
+{
+  "version": "2.0",
+  "schema_ref": "https://darwin-public.s3.eu-west-1.amazonaws.com/darwin_json/2.0/schema.json",
+  "item": {
+    "name": "000000021295.jpg",
+    "path": "/",
+    "source_info": {
+      "item_id": "0192c338-2dc8-bb9f-8dde-9c4a8714c957",
+      "dataset": {
+        "name": "tmp",
+        "slug": "tmp",
+        "dataset_management_url": "https://staging.v7labs.com/datasets/426200/dataset-management"
+      },
+      "team": {
+        "name": "E2E Testing",
+        "slug": "e2e-testing"
+      },
+      "workview_url": "https://staging.v7labs.com/workview?dataset=426200&item=0192c338-2dc8-bb9f-8dde-9c4a8714c957"
+    },
+    "slots": [
+      {
+        "type": "image",
+        "slot_name": "0",
+        "width": 640,
+        "height": 427,
+        "thumbnail_url": "https://staging.v7labs.com/api/v2/teams/e2e-testing/files/6c554c31-8bf6-4fc7-9f31-394fd775b2a6/thumbnail",
+        "source_files": [
+          {
+            "file_name": "000000021295.jpg",
+            "url": "https://staging.v7labs.com/api/v2/teams/e2e-testing/uploads/ade5f1e3-a165-4a07-989f-86ee342499c9"
+          }
+        ]
+      }
+    ]
+  },
+  "annotations": [
+    {
+      "bounding_box": {
+        "h": 43.7181,
+        "w": 71.2665,
+        "x": 166.0884,
+        "y": 113.1879
+      },
+      "id": "46b33c9c-7453-4722-8b43-91bbb3fc247f",
+      "name": "test_bounding_box_basic",
+      "properties": [],
+      "slot_names": [
+        "0"
+      ]
+    },
+    {
+      "bounding_box": {
+        "h": 31.7405,
+        "w": 58.0912,
+        "x": 360.1248,
+        "y": 259.913
+      },
+      "id": "2b7c85ed-74b3-4111-a5d6-073da56d0072",
+      "name": "test_bounding_box_basic",
+      "properties": [],
+      "slot_names": [
+        "0"
+      ]
+    }
+  ],
+  "properties": []
+}

e2e_tests/data/convert/cvat/to/output.xml

@@ -0,0 +1 @@
+<annotations><version>1.1</version><meta><dumped>2024-10-25 10:33:01.789498+00:00</dumped><task><id>1</id><name>exported_task_from_darwin</name><size>1</size><mode>annotation</mode><overlapp>0</overlapp><bugtracker>None</bugtracker><flipped>False</flipped><created>2024-10-25 10:33:01.789603+00:00</created><updated>2024-10-25 10:33:01.789608+00:00</updated><labels><label><name>test_bounding_box_basic</name><attributes /></label></labels><segments><segment><id>1</id><start>1</start><end>1</end><url>not applicable</url></segment></segments><owner><username>example_username</username><email>[email protected]</email></owner></task></meta><image id="1" name="000000021295.jpg" width="640" height="427"><box label="test_bounding_box_basic" xtl="166.088" ytl="113.188" xbr="237.35399999999998" ybr="156.906" occluded="0" /><box label="test_bounding_box_basic" xtl="360.125" ytl="259.913" xbr="418.216" ybr="291.654" occluded="0" /></image></annotations>

e2e_tests/data/convert/instance_mask/from/000000021295.json

@@ -0,0 +1,155 @@
+{
+  "version": "2.0",
+  "schema_ref": "https://darwin-public.s3.eu-west-1.amazonaws.com/darwin_json/2.0/schema.json",
+  "item": {
+    "name": "000000021295.jpg",
+    "path": "/",
+    "source_info": {
+      "item_id": "0192c338-2dc8-bb9f-8dde-9c4a8714c957",
+      "team": {
+        "name": "E2E Testing",
+        "slug": "e2e-testing"
+      },
+      "dataset": {
+        "name": "tmp",
+        "slug": "tmp",
+        "dataset_management_url": "https://staging.v7labs.com/datasets/426200/dataset-management"
+      },
+      "workview_url": "https://staging.v7labs.com/workview?dataset=426200&item=0192c338-2dc8-bb9f-8dde-9c4a8714c957"
+    },
+    "slots": [
+      {
+        "type": "image",
+        "slot_name": "0",
+        "width": 640,
+        "height": 427,
+        "thumbnail_url": "https://staging.v7labs.com/api/v2/teams/e2e-testing/files/6c554c31-8bf6-4fc7-9f31-394fd775b2a6/thumbnail",
+        "source_files": [
+          {
+            "file_name": "000000021295.jpg",
+            "url": "https://staging.v7labs.com/api/v2/teams/e2e-testing/uploads/ade5f1e3-a165-4a07-989f-86ee342499c9"
+          }
+        ]
+      }
+    ]
+  },
+  "annotations": [
+    {
+      "bounding_box": {
+        "h": 123.8166,
+        "w": 147.2414,
+        "x": 105.8307,
+        "y": 63.5815
+      },
+      "id": "4f23f7d2-2f83-45bc-927a-d03b9f642619",
+      "name": "test_polygon_basic",
+      "polygon": {
+        "paths": [
+          [
+            {
+              "x": 189.4906,
+              "y": 63.5815
+            },
+            {
+              "x": 105.8307,
+              "y": 182.0439
+            },
+            {
+              "x": 249.0564,
+              "y": 187.3981
+            },
+            {
+              "x": 196.1834,
+              "y": 105.0768
+            },
+            {
+              "x": 253.0721,
+              "y": 99.7226
+            }
+          ]
+        ]
+      },
+      "properties": [],
+      "slot_names": [
+        "0"
+      ]
+    },
+    {
+      "bounding_box": {
+        "h": 99.05329999999998,
+        "w": 155.2727,
+        "x": 201.5376,
+        "y": 287.1207
+      },
+      "id": "b4248df6-5390-488a-9190-edfa397458a6",
+      "name": "test_polygon_basic",
+      "polygon": {
+        "paths": [
+          [
+            {
+              "x": 247.0486,
+              "y": 287.1207
+            },
+            {
+              "x": 201.5376,
+              "y": 386.174
+            },
+            {
+              "x": 293.8981,
+              "y": 386.174
+            },
+            {
+              "x": 356.8103,
+              "y": 384.1661
+            },
+            {
+              "x": 356.8103,
+              "y": 329.9545
+            },
+            {
+              "x": 319.3307,
+              "y": 316.569
+            }
+          ]
+        ]
+      },
+      "properties": [],
+      "slot_names": [
+        "0"
+      ]
+    },
+    {
+      "bounding_box": {
+        "h": 163.3041,
+        "w": 171.33550000000002,
+        "x": 338.0705,
+        "y": 50.1959
+      },
+      "id": "5092f929-ec75-4eb9-8b56-e6bbc472e272",
+      "name": "test_polygon_basic",
+      "polygon": {
+        "paths": [
+          [
+            {
+              "x": 429.7618,
+              "y": 50.1959
+            },
+            {
+              "x": 338.0705,
+              "y": 212.8307
+            },
+            {
+              "x": 509.406,
+              "y": 213.5
+            }
+          ]
+        ]
+      },
+      "properties": [],
+      "slot_names": [
+        "0"
+      ]
+    }
+  ],
+  "properties": []
+}