client.skeleton

Client for interacting with the skeleton service.

Methods:

Name	Description
get_skeleton	Gets one skeleton, blocking for skeletonization if necessary.
get_cache_contents	Mirror CloudFiles.list() for skeletons as a pass-through interface to the underlying service and bucket.
skeletons_exist	Confirm or deny that a set of root ids have H5 skeletons in the cache.
get_bulk_skeletons	Generates skeletons for a list of root ids in a "small" bulk (ten at the time of this writing). Use the async interface for larger bulk requests.
generate_bulk_skeletons_async	Generates skeletons for a list of root ids without retrieving them.
fetch_skeletons	Retrieve already-cached skeletons in bulk, up to 500 at a time.

Attributes:

Name	Type	Description
server_version	Optional[Version]	The version of the service running on the remote server. Note that this

server_version property

The version of the service running on the remote server. Note that this refers to the software running on the server and has nothing to do with the version of the datastack itself.

get_skeleton(root_id, datastack_name=None, skeleton_version=4, output_format='dict', log_warning=True, verbose_level=0)

Gets one skeleton, blocking for skeletonization if necessary.

Parameters:

Name	Type	Description	Default
root_id	int	The root id of the skeleton to retrieve	required
datastack_name	str	The name of the datastack to check	None
skeleton_version	int	The skeleton version to generate and retrieve. Options are documented in SkeletonService. Use 0 for Neuroglancer-compatibility. Use -1 for latest.	4
output_format	string	The format to retrieve. Options are: - 'dict': A dictionary - 'swc': A pandas DataFrame	'dict'

Returns:

Type	Description
	Skeleton of the requested type. See output_format for details.

get_cache_contents(datastack_name=None, skeleton_version=4, root_id_prefixes=0, limit=0, log_warning=True, verbose_level=0)

Mirror CloudFiles.list() for skeletons as a pass-through interface to the underlying service and bucket.

skeletons_exist(datastack_name=None, skeleton_version=4, root_ids=0, log_warning=True, verbose_level=0)

Confirm or deny that a set of root ids have H5 skeletons in the cache.

get_bulk_skeletons(root_ids, datastack_name=None, skeleton_version=4, output_format='dict', generate_missing_skeletons=False, log_warning=True, verbose_level=0)

Generates skeletons for a list of root ids in a "small" bulk (ten at the time of this writing). Use the async interface for larger bulk requests.

Parameters:

Name	Type	Description	Default
root_ids	List	A list of root ids of the skeletons to generate	required
datastack_name	str	The name of the datastack to check	None
skeleton_version	int	The skeleton version to generate. Use 0 for Neuroglancer-compatibility. Use -1 for latest.	4

generate_bulk_skeletons_async(root_ids, datastack_name=None, skeleton_version=None, log_warning=True, verbose_level=0)

Generates skeletons for a list of root ids without retrieving them.

Parameters:

Name	Type	Description	Default
root_ids	List	A list of root ids of the skeletons to generate	required
datastack_name	str	The name of the datastack to check	None
skeleton_version	int	The skeleton version to generate. Use 0 for Neuroglancer-compatibility. Use -1 for latest.	None

Returns:

Type	Description
float	The estimated time in seconds to generate all skeletons (a comparable message will be output to the console prior to return).

fetch_skeletons(root_ids, datastack_name=None, skeleton_version=4, output_format='dict', method='server', generate_missing_skeletons=False, verbose_level=0)

Retrieve already-cached skeletons in bulk, up to 500 at a time.

Unlike :meth:get_bulk_skeletons, this method:

• Accepts up to 500 root IDs per call (vs the 10-skeleton limit of get_bulk_skeletons)
• Skips per-RID chunkedgraph validation, so it never blocks on network calls
• Never generates skeletons inline; only returns what is already in the cache

Root IDs not found in the cache are simply absent from the returned dict.

Parameters:

Name	Type	Description	Default
root_ids	List	Root IDs to retrieve. Truncated to 500 if longer.	required
datastack_name	str	Datastack name. Defaults to the client's configured datastack.	None
skeleton_version	int	Skeleton version. Default is 4 (latest).	4
output_format	'dict' or 'swc'	Output format. Default is "dict". method="gcs" only supports "dict".	'dict'
method	'server' or 'gcs'	How to retrieve skeletons. "server" (default) — POST root IDs to the server; server decodes and returns skeletons. "gcs" — Obtain a short-lived downscoped GCS token (cached client-side), then download and parse H5 files directly from the storage bucket, bypassing the service for data transfer. Significantly faster for large batches. Only supports output_format="dict".	'server'
generate_missing_skeletons	bool	If True, root IDs not found in the cache are queued for async background generation. They will still be absent from the returned dict. Default False.	False
verbose_level	int	Verbosity level for server-side logging.	0

Returns:

Type	Description
dict	Mapping of root_id (str) → skeleton object. Only root IDs successfully retrieved appear in the dict; missing ones are absent.