unitlab 2.3.47__tar.gz → 2.3.48__tar.gz

{unitlab-2.3.47/src/unitlab.egg-info → unitlab-2.3.48}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: unitlab
-Version: 2.3.47
+Version: 2.3.48
 Home-page: https://github.com/teamunitlab/unitlab-sdk
 Author: Unitlab Inc.
 Author-email: team@unitlab.ai

{unitlab-2.3.47 → unitlab-2.3.48}/setup.py

@@ -2,7 +2,7 @@ from setuptools import find_packages, setup
 
 setup(
     name="unitlab",
-    version="2.3.47",
+    version="2.3.48",
     license="MIT",
     author="Unitlab Inc.",
     author_email="team@unitlab.ai",

{unitlab-2.3.47 → unitlab-2.3.48}/src/unitlab/client.py

@@ -117,7 +117,12 @@ class UnitlabClient:
         return self._get(f"/api/sdk/projects/{project_id}/upload-info/")
 
     def project_upload_data(
-        self, project_id, directory, batch_size=100, sentences_per_chunk=10
+        self,
+        project_id,
+        directory,
+        batch_size=100,
+        sentences_per_chunk=10,
+        fps=1.0,
     ):
         if not os.path.isdir(directory):
             raise ValueError(f"Directory {directory} does not exist")
@@ -143,7 +148,8 @@ class UnitlabClient:
             if file_size_bytes > max_file_size_bytes:
                 file_size_mb = file_size_bytes / 1024 / 1024
                 logger.warning(
-                    f"File {file} is too large ({file_size_mb:.4f} MB) skipping, max size is {max_file_size_mb:.2f} MB"
+                    f"File {file} is too large ({file_size_mb:.4f} MB) skipping, "
+                    f"max size is {max_file_size_mb:.2f} MB"
                 )
                 continue
             filtered_files.append(file)
@@ -159,6 +165,8 @@ class UnitlabClient:
                 )
                 if generic_type == "text":
                     form_data.add_field("sentences_per_chunk", str(sentences_per_chunk))
+                elif generic_type == "video":
+                    form_data.add_field("fps", str(fps))
 
                 try:
                     await asyncio.sleep(0.1)
@@ -221,16 +229,44 @@ class UnitlabClient:
             return os.path.abspath(filename)
 
     def dataset_download_files(self, dataset_id):
+        """Download files from a dataset.
+
+        Args:
+            dataset_id: UUID of the dataset/release to download.
+
+        Returns:
+            Path to the download folder.
+        """
         response = self._post(
             f"/api/sdk/datasets/{dataset_id}/", data={"download_type": "files"}
         )
-        folder = f"dataset-files-{dataset_id}"
-        os.makedirs(folder, exist_ok=True)
-        dataset_files = [
-            dataset_file
-            for dataset_file in response
-            if not os.path.isfile(os.path.join(folder, dataset_file["file_name"]))
-        ]
+        base_folder = f"dataset-files-{dataset_id}"
+        os.makedirs(base_folder, exist_ok=True)
+
+        files_to_download = []
+        for dataset_file in response:
+            file_name = dataset_file["file_name"]
+            file_path = os.path.join(base_folder, file_name)
+
+            parent_dir = os.path.dirname(file_path)
+            if parent_dir:
+                os.makedirs(parent_dir, exist_ok=True)
+
+            if "content" in dataset_file:
+                if not os.path.isfile(file_path):
+                    with open(file_path, "w", encoding="utf-8") as f:
+                        f.write(dataset_file["content"])
+            elif "source" in dataset_file and not os.path.isfile(file_path):
+                files_to_download.append(
+                    {
+                        "file_name": file_name,
+                        "source": dataset_file["source"],
+                        "file_path": file_path,
+                    }
+                )
+
+        if not files_to_download:
+            return base_folder
 
         async def download_file(session: aiohttp.ClientSession, dataset_file: dict):
             async with session.get(url=dataset_file["source"]) as r:
@@ -241,21 +277,20 @@ class UnitlabClient:
                         f"Error downloading file {dataset_file['file_name']} - {e}"
                     )
                     return 0
-                async with aiofiles.open(
-                    os.path.join(folder, dataset_file["file_name"]), "wb"
-                ) as f:
+                async with aiofiles.open(dataset_file["file_path"], "wb") as f:
                     async for chunk in r.content.iter_any():
                         await f.write(chunk)
                     return 1
 
         async def main():
-            with tqdm.tqdm(total=len(dataset_files), ncols=80) as pbar:
+            with tqdm.tqdm(total=len(files_to_download), ncols=80) as pbar:
                 async with aiohttp.ClientSession() as session:
                     tasks = [
-                        download_file(session=session, dataset_file=dataset_file)
-                        for dataset_file in dataset_files
+                        download_file(session=session, dataset_file=df)
+                        for df in files_to_download
                     ]
                     for f in asyncio.as_completed(tasks):
                         pbar.update(await f)
 
         asyncio.run(main())
+        return base_folder

{unitlab-2.3.47 → unitlab-2.3.48}/src/unitlab/main.py

@@ -78,9 +78,18 @@ def upload(
             help="Number of sentences per chunk for text projects (default: 10, range: 3-50)"
         ),
     ] = 10,
+    fps: Annotated[
+        float,
+        typer.Option(
+            help="Frames per second for video projects (default: 1.0, min: 0.25)"
+        ),
+    ] = 1.0,
 ):
     get_client(api_key).project_upload_data(
-        str(pk), directory=directory, sentences_per_chunk=sentences_per_chunk
+        str(pk),
+        directory=directory,
+        sentences_per_chunk=sentences_per_chunk,
+        fps=fps,
     )
 
 
@@ -113,8 +122,9 @@ def dataset_download(
             raise typer.BadParameter(
                 "Split type is required when download type is annotation"
             )
-        return get_client(api_key).dataset_download(pk, export_type, split_type)
-    get_client(api_key).dataset_download_files(pk)
+        print(get_client(api_key).dataset_download(pk, export_type, split_type))
+    else:
+        print(get_client(api_key).dataset_download_files(pk))
 
 
 if __name__ == "__main__":

{unitlab-2.3.47 → unitlab-2.3.48/src/unitlab.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: unitlab
-Version: 2.3.47
+Version: 2.3.48
 Home-page: https://github.com/teamunitlab/unitlab-sdk
 Author: Unitlab Inc.
 Author-email: team@unitlab.ai