Merge pull request #189 from superannotateai/attach_documnet

Added sdk/cli function attach_document_urls_to_project

Author: VaghinakDev

src/superannotate/__init__.py

@@ -45,6 +45,9 @@
 from superannotate.lib.app.interface.sdk_interface import aggregate_annotations_as_df
 from superannotate.lib.app.interface.sdk_interface import assign_folder
 from superannotate.lib.app.interface.sdk_interface import assign_images
+from superannotate.lib.app.interface.sdk_interface import (
+    attach_document_urls_to_project,
+)
 from superannotate.lib.app.interface.sdk_interface import attach_image_urls_to_project
 from superannotate.lib.app.interface.sdk_interface import attach_video_urls_to_project
 from superannotate.lib.app.interface.sdk_interface import benchmark
@@ -239,6 +242,7 @@
     "upload_images_from_folder_to_project",
     "attach_image_urls_to_project",
     "attach_video_urls_to_project",
+    "attach_document_urls_to_project",
     # Video Section
     "upload_videos_from_folder_to_project",
     # Annotation Section

src/superannotate/lib/app/analytics/common.py

@@ -179,7 +179,7 @@ def aggregate_annotations_as_df(
         and "___objects.json" not in json_paths[0].name
     ):
         raise AppException(
-            "The function does not support projects containing videos attached with URLs"
+            "The function does not support projects containing videos / documents attached with URLs"
         )
 
     if verbose:

src/superannotate/lib/app/input_converters/conversion.py

@@ -143,9 +143,9 @@ def export_annotation(
     project_type="Vector",
     task="object_detection",
 ):
-    if project_type == "Video":
+    if project_type == "Video" or project_type == "Text":
         raise AppValidationException(
-            "The function does not support projects containing videos attached with URLs"
+            f"The function does not support projects containing {project_type} attached with URLs"
         )
     """Converts SuperAnnotate annotation formate to the other annotation formats. Currently available (project_type, task) combinations for converter
     presented below:

src/superannotate/lib/app/input_converters/sa_conversion.py

@@ -182,7 +182,7 @@ def sa_convert_project_type(input_dir, output_dir):
         img_names = from_vector_to_pixel(json_paths, output_dir)
     elif ".json" in json_paths[0].name:
         raise AppException(
-            "The function does not support projects containing videos attached with URLs"
+            "The function does not support projects containing videos / document attached with URLs"
         )
     else:
         raise AppException(

src/superannotate/lib/app/interface/cli_interface.py

@@ -14,6 +14,7 @@
 from lib.app.helpers import split_project_path
 from lib.app.input_converters.conversion import import_annotation
 from lib.app.interface.base_interface import BaseInterfaceFacade
+from lib.app.interface.sdk_interface import attach_document_urls_to_project
 from lib.app.interface.sdk_interface import attach_image_urls_to_project
 from lib.app.interface.sdk_interface import attach_video_urls_to_project
 from lib.app.interface.sdk_interface import create_folder
@@ -263,6 +264,17 @@ def attach_video_urls(
         )
         sys.exit(0)
 
+    @staticmethod
+    def attach_document_urls(
+        project: str, attachments: str, annotation_status: Optional[Any] = None
+    ):
+        attach_document_urls_to_project(
+            project=project,
+            attachments=attachments,
+            annotation_status=annotation_status,
+        )
+        sys.exit(0)
+
     def _attach_urls(
         self, project: str, attachments: str, annotation_status: Optional[Any] = None
     ):

src/superannotate/lib/app/interface/sdk_interface.py

@@ -3567,3 +3567,63 @@ def delete_annotations(project: str, image_names: List[str] = None):
     )
     if response.errors:
         raise AppException(response.errors)
+
+
+@Trackable
+@validate_arguments
+def attach_document_urls_to_project(
+    project: Union[NotEmptyStr, dict],
+    attachments: Union[Path, NotEmptyStr],
+    annotation_status: Optional[NotEmptyStr] = "NotStarted",
+):
+    """Link documents on external storage to SuperAnnotate.
+
+    :param project: project name or project folder path
+    :type project: str or dict
+    :param attachments: path to csv file on attachments metadata
+    :type attachments: Path-like (str or Path)
+    :param annotation_status: value to set the annotation statuses of the linked documents: NotStarted InProgress QualityCheck Returned Completed Skipped
+    :type annotation_status: str
+
+    :return: list of attached documents, list of not attached documents, list of skipped documents
+    :rtype: tuple
+    """
+    project_name, folder_name = extract_project_folder(project)
+
+    image_data = pd.read_csv(attachments, dtype=str)
+    image_data = image_data[~image_data["url"].isnull()]
+    if "name" in image_data.columns:
+        image_data["name"] = (
+            image_data["name"]
+            .fillna("")
+            .apply(lambda cell: cell if str(cell).strip() else str(uuid.uuid4()))
+        )
+    else:
+        image_data["name"] = [str(uuid.uuid4()) for _ in range(len(image_data.index))]
+
+    image_data = pd.DataFrame(image_data, columns=["name", "url"])
+    img_names_urls = image_data.rename(columns={"url": "path"}).to_dict(
+        orient="records"
+    )
+    list_of_not_uploaded = []
+    duplicate_images = []
+    for i in range(0, len(img_names_urls), 500):
+        response = controller.attach_urls(
+            project_name=project_name,
+            folder_name=folder_name,
+            files=ImageSerializer.deserialize(
+                img_names_urls[i : i + 500]  # noqa: E203
+            ),
+            annotation_status=annotation_status,
+        )
+        if response.errors:
+            list_of_not_uploaded.append(response.data[0])
+            duplicate_images.append(response.data[1])
+
+    list_of_uploaded = [
+        image["name"]
+        for image in img_names_urls
+        if image["name"] not in list_of_not_uploaded
+    ]
+
+    return list_of_uploaded, list_of_not_uploaded, duplicate_images

src/superannotate/lib/core/enums.py

@@ -32,6 +32,7 @@ class ProjectType(BaseTitledEnum):
     VECTOR = "Vector", 1
     PIXEL = "Pixel", 2
     VIDEO = "Video", 3
+    DOCUMENT = "Document", 4
 
 
 class UserRole(BaseTitledEnum):