Merge pull request #2078 from FedML-AI/bhargav191098/adding_tags_to_d…

…ata_storage

Adding tags to data storage object and the corresponding pretty table…

python/fedml/api/__init__.py

@@ -179,13 +179,12 @@ def cluster_killall(api_key=None) -> bool:
     return cluster.kill(cluster_names=(), api_key=api_key)
 
 
-def upload(data_path, api_key=None, service="R2", name=None, description=None, metadata=None, show_progress=False,
+def upload(data_path, api_key=None, tag_list=[], service="R2", name=None, description=None, metadata=None, show_progress=False,
            out_progress_to_err=True, progress_desc=None) -> FedMLResponse:
-    return storage.upload(data_path=data_path, api_key=api_key, name=name, description=description,
+    return storage.upload(data_path=data_path, api_key=api_key, name=name, description=description, tag_list =tag_list,
                           service=service, progress_desc=progress_desc, show_progress=show_progress,
                           out_progress_to_err=out_progress_to_err, metadata=metadata)
 
-
 def get_storage_user_defined_metadata(data_name, api_key=None) -> FedMLResponse:
     return storage.get_user_metadata(data_name=data_name, api_key=api_key)
 

python/fedml/api/modules/storage.py

@@ -18,12 +18,13 @@ def __init__(self, data: dict):
         self.createdAt = data.get("createTime", None)
         self.updatedAt = data.get("updateTime", None)
         self.size = _get_size(data.get("fileSize",None))
+        self.tag_list = data.get("tags", None)
 
 
 # Todo (alaydshah): Store service name in metadata
 # Todo (alaydshah): If data already exists, don't upload again. Instead suggest to use update command
 
-def upload(data_path, api_key, name, description, service, show_progress, out_progress_to_err, progress_desc,
+def upload(data_path, api_key, name, description, tag_list, service, show_progress, out_progress_to_err, progress_desc,
            metadata) -> FedMLResponse:
     api_key = authenticate(api_key)
 
@@ -58,7 +59,7 @@ def upload(data_path, api_key, name, description, service, show_progress, out_pr
         "description": description,
         "fileSize": file_size,
         "fileUrl": file_uploaded_url,
-        "tagNameList": [],
+        "tagNameList": tag_list,
     }
 
     try:

python/fedml/cli/modules/storage.py

@@ -53,6 +53,7 @@ def validate_argument(ctx, param, value):
 @click.option("--user_metadata", "-um", type=str, help="User-defined metadata in the form of a dictionary, for instance, "
                                                        " {'name':'value'} within double quotes. "" "
                                                        "Defaults to None.")
+@click.option("--tags", "-t", type=str, help="Add tags to your data to store. Give tags in comma separated form like 'cv,unet,segmentation' If not provided, the tags will be empty.")
 @click.option('--service', "-s", type=click.Choice(['R2']), default="R2", help="Storage service for object storage. "
                                                                                "Only R2 is supported as of now")
 @click.option(
@@ -65,10 +66,11 @@ def validate_argument(ctx, param, value):
     default="release",
     help=version_help,
 )
-def upload(data_path: str, name: str, user_metadata: str, description: str, version: str, api_key: str, service):
+def upload(data_path: str, name: str, user_metadata: str, description: str, version: str, api_key: str, tags:str, service):
     metadata = _parse_metadata(user_metadata)
+    tag_list = _parse_tags(tags)
     fedml.set_env_version(version)
-    response = fedml.api.upload(data_path=data_path, api_key=api_key, name=name, service=service, show_progress=True,
+    response = fedml.api.upload(data_path=data_path, api_key=api_key, name=name, tag_list = tag_list, service=service, show_progress=True,
                                 description=description, metadata=metadata)
     if response.code == ResponseCode.SUCCESS:
         click.echo(f"Data uploaded successfully. | url: {response.data}")
@@ -96,10 +98,10 @@ def list_data(version, api_key):
         if not response.data:
             click.echo(f"No stored objects found for account linked with apikey: {api_key}")
             return
-        object_list_table = PrettyTable(["Data Name", "Data Size", "Description", "Created At", "Updated At"])
+        object_list_table = PrettyTable(["Data Name", "Data Size", "Description", "Data Tags","Created At", "Updated At"])
         for stored_object in response.data:
             object_list_table.add_row(
-                [stored_object.dataName, stored_object.size, stored_object.description, stored_object.createdAt, stored_object.updatedAt])
+                [stored_object.dataName, stored_object.size, stored_object.description, stored_object.tag_list,stored_object.createdAt, stored_object.updatedAt])
         click.echo(object_list_table)
     else:
         click.echo(f"Failed to list stored objects for account linked with apikey {api_key}. "
@@ -157,8 +159,8 @@ def get_metadata(data_name, version, api_key):
             return
         click.echo(f"Successfully fetched metadata for object {data_name}:")
         # Todo (alaydshah): Add file size and tags
-        metadata_table = PrettyTable(["Data Name","Data Size","Description", "Created At", "Updated At"])
-        metadata_table.add_row([metadata.dataName,metadata.size,metadata.description, metadata.createdAt, metadata.updatedAt])
+        metadata_table = PrettyTable(["Data Name","Data Size","Description","Data Tags","Created At", "Updated At"])
+        metadata_table.add_row([metadata.dataName,metadata.size,metadata.description,metadata.tag_list,metadata.createdAt, metadata.updatedAt])
         click.echo(metadata_table)
         click.echo("")
     else:
@@ -238,3 +240,9 @@ def _parse_metadata(metadata: str):
         click.echo(
             f"Input metadata cannot be evaluated. Please make sure metadata is in the correct format. Error: {e}.")
         exit()
+
+def _parse_tags(tags:str):
+    if not tags:
+        return []
+    tag_list = tags.split(",")
+    return tag_list