unitlab 2.1.3__tar.gz → 2.1.4__tar.gz

{unitlab-2.1.3/src/unitlab.egg-info → unitlab-2.1.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: unitlab
-Version: 2.1.3
+Version: 2.1.4
 Home-page: https://github.com/teamunitlab/unitlab-sdk
 Author: Unitlab Inc.
 Author-email: team@unitlab.ai

{unitlab-2.1.3 → unitlab-2.1.4}/setup.py

@@ -2,7 +2,7 @@ from setuptools import find_packages, setup
 
 setup(
     name="unitlab",
-    version="2.1.3",
+    version="2.1.4",
     license="MIT",
     author="Unitlab Inc.",
     author_email="team@unitlab.ai",

{unitlab-2.1.3 → unitlab-2.1.4}/src/unitlab/client.py

@@ -244,38 +244,13 @@ class UnitlabClient:
 
         asyncio.run(main())
 
-    def create_dataset(self, name, annotation_type, categories, license_id=None):
-        response = self._post(
-            "/api/sdk/datasets/create/",
-            data={
-                "name": name,
-                "annotation_type": annotation_type,
-                "classes": [
-                    {"name": category["name"], "value": category["id"]}
-                    for category in categories
-                ],
-                "license": license_id,
-            },
-        )
-        return response["pk"]
-
-    def finalize_dataset(self, dataset_id):
+    def _finalize_dataset(self, dataset_id):
         return self._post(f"/api/sdk/datasets/{dataset_id}/finalize/")
 
-    def dataset_upload(
-        self,
-        name,
-        annotation_type,
-        annotation_path,
-        data_path,
-        license_id=None,
-        batch_size=15,
+    def _dataset_data_upload(
+        self, dataset_id, upload_handler: DatasetUploadHandler, batch_size=15
     ):
-        handler = DatasetUploadHandler(annotation_type, annotation_path, data_path)
-        dataset_id = self.create_dataset(
-            name, annotation_type, handler.categories, license_id=license_id
-        )
-        image_ids = handler.getImgIds()
+        image_ids = upload_handler.getImgIds()
         url = urllib.parse.urljoin(
             self.api_url, f"/api/sdk/datasets/{dataset_id}/upload/"
         )
@@ -294,7 +269,7 @@ class UnitlabClient:
                                 )
                             ]:
                                 tasks.append(
-                                    handler.upload_image(session, url, image_id)
+                                    upload_handler.upload_image(session, url, image_id)
                                 )
                             for f in asyncio.as_completed(tasks):
                                 try:
@@ -305,4 +280,55 @@ class UnitlabClient:
                         raise e
 
         asyncio.run(main())
-        self.finalize_dataset(dataset_id)
+
+    def dataset_upload(
+        self,
+        name,
+        annotation_type,
+        annotation_path,
+        data_path,
+        license_id=None,
+        batch_size=15,
+    ):
+        upload_handler = DatasetUploadHandler(
+            annotation_type, annotation_path, data_path
+        )
+        dataset_id = self._post(
+            "/api/sdk/datasets/create/",
+            data={
+                "name": name,
+                "annotation_type": annotation_type,
+                "classes": [
+                    {"name": category["name"], "value": category["id"]}
+                    for category in upload_handler.categories
+                ],
+                "license": license_id,
+            },
+        )["pk"]
+        self._dataset_data_upload(dataset_id, upload_handler, batch_size=batch_size)
+        self._finalize_dataset(dataset_id)
+
+    def dataset_update(self, pk, annotation_path, data_path, batch_size=15):
+        dataset = self._get(f"api/sdk/datasets/{pk}/")
+        upload_handler = DatasetUploadHandler(
+            dataset["annotation_type"], annotation_path, data_path
+        )
+        new_dataset = self._post(
+            f"/api/sdk/datasets/{pk}/update/",
+            data={
+                "classes": [
+                    {"name": category["name"], "value": category["id"]}
+                    for category in sorted(
+                        upload_handler.loadCats(upload_handler.getCatIds()),
+                        key=lambda x: x["id"],
+                    )
+                ]
+            },
+        )
+        upload_handler.original_category_referecences = {
+            int(k): v for k, v in new_dataset["original_category_referecences"].items()
+        }
+        self._dataset_data_upload(
+            new_dataset["pk"], upload_handler, batch_size=batch_size
+        )
+        self._finalize_dataset(new_dataset["pk"])

{unitlab-2.1.3 → unitlab-2.1.4}/src/unitlab/dataset.py

@@ -1,4 +1,5 @@
 import asyncio
+import copy
 import itertools
 import json
 import logging
@@ -79,7 +80,9 @@ class COCO:
         self.catToImgs = catToImgs
         self.imgs = imgs
         self.cats = cats
-        self.categories = sorted(self.loadCats(self.getCatIds()), key=lambda x: x["id"])
+        self.categories = sorted(
+            copy.deepcopy(self.loadCats(self.getCatIds())), key=lambda x: x["id"]
+        )
         self.classes = [cat["name"] for cat in self.categories]
         self.original_category_referecences = dict()
         for i, category in enumerate(self.categories):
@@ -321,6 +324,9 @@ class DatasetUploadHandler(COCO):
                             raise SubscriptionError(
                                 "You have reached the maximum number of datasources for your subscription."
                             )
+                        elif response.status == 400:
+                            logger.error(await response.text())
+                            return 0
                         response.raise_for_status()
                         return 1
                 except SubscriptionError as e:

{unitlab-2.1.3 → unitlab-2.1.4}/src/unitlab/main.py

@@ -116,6 +116,18 @@ def dataset_upload(
     )
 
 
+@dataset_app.command(name="update", help="Update dataset")
+def dataset_update(
+    pk: UUID,
+    api_key: API_KEY,
+    annotation_path: Annotated[Path, typer.Option(help="Path to the COCO json file")],
+    data_path: Annotated[
+        Path, typer.Option(help="Directory containing the data to be uploaded")
+    ],
+):
+    get_client(api_key).dataset_update(pk, annotation_path, data_path)
+
+
 @dataset_app.command(name="download", help="Download dataset")
 def dataset_download(
     pk: UUID,

{unitlab-2.1.3 → unitlab-2.1.4/src/unitlab.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: unitlab
-Version: 2.1.3
+Version: 2.1.4
 Home-page: https://github.com/teamunitlab/unitlab-sdk
 Author: Unitlab Inc.
 Author-email: team@unitlab.ai