datalad.api.get(path=None, *, source=None, dataset=None, recursive=False, recursion_limit=None, get_data=True, description=None, reckless=None, jobs='auto')

Get any dataset content (files/directories/subdatasets).

This command only operates on dataset content. To obtain a new independent dataset from some source use the clone command.

By default this command operates recursively within a dataset, but not across potential subdatasets, i.e. if a directory is provided, all files in the directory are obtained. Recursion into subdatasets is supported too. If enabled, relevant subdatasets are detected and installed in order to fulfill a request.

Known data locations for each requested file are evaluated and data are obtained from some available location (according to git-annex configuration and possibly assigned remote priorities), unless a specific source is specified.

Getting subdatasets

Just as DataLad supports getting file content from more than one location, the same is supported for subdatasets, including a ranking of individual sources for prioritization.

The following location candidates are considered. For each candidate a cost is given in parenthesis, higher values indicate higher cost, and thus lower priority:

  • A datalad URL recorded in .gitmodules (cost 590). This allows for datalad URLs that require additional handling/resolution by datalad, like ria-schemes (ria+http, ria+ssh, etc.)

  • A URL or absolute path recorded for git in .gitmodules (cost 600).

  • URL of any configured superdataset remote that is known to have the desired submodule commit, with the submodule path appended to it. There can be more than one candidate (cost 650).

  • In case .gitmodules contains a relative path instead of a URL, the URL of any configured superdataset remote that is known to have the desired submodule commit, with this relative path appended to it. There can be more than one candidate (cost 650).

  • In case .gitmodules contains a relative path as a URL, the absolute path of the superdataset, appended with this relative path (cost 900).

Additional candidate URLs can be generated based on templates specified as configuration variables with the pattern


where name is an arbitrary identifier. If name starts with three digits (e.g. ‘400myserver’) these will be interpreted as a cost, and the respective candidate will be sorted into the generated candidate list according to this cost. If no cost is given, a default of 700 is used.

A template string assigned to such a variable can utilize the Python format mini language and may reference a number of properties that are inferred from the parent dataset’s knowledge about the target subdataset. Properties include any submodule property specified in the respective .gitmodules record. For convenience, an existing datalad-id record is made available under the shortened name id.

Additionally, the URL of any configured remote that contains the respective submodule commit is available as remoteurl-<name> property, where name is the configured remote name.

Hence, such a template could be{id} or{path}, where {id} and {path} would be replaced by the datalad-id or path entry in the .gitmodules record.

If this config is committed in .datalad/config, a clone of a dataset can look up any subdataset’s URL according to such scheme(s) irrespective of what URL is recorded in .gitmodules.

Lastly, all candidates are sorted according to their cost (lower values first), and duplicate URLs are stripped, while preserving the first item in the candidate list.


Power-user info: This command uses git annex get to fulfill file handles.


Get a single file:

> get('path/to/file')

Get contents of a directory:

> get('path/to/dir/')

Get all contents of the current dataset and its subdatasets:

> get(dataset='.', recursive=True)

Get (clone) a registered subdataset, but don’t retrieve data:

> get('path/to/subds', get_data=False)
  • path (sequence of str or None, optional) – path/name of the requested dataset component. The component must already be known to a dataset. To add new components to a dataset use the add command. [Default: None]

  • source (str or None, optional) – label of the data source to be used to fulfill requests. This can be the name of a dataset sibling or another known source. [Default: None]

  • dataset (Dataset or None, optional) – specify the dataset to perform the add operation on, in which case path arguments are interpreted as being relative to this dataset. If no dataset is given, an attempt is made to identify a dataset for each input path. [Default: None]

  • recursive (bool, optional) – if set, recurse into potential subdatasets. [Default: False]

  • recursion_limit (int or {'existing'} or None, optional) – limit recursion into subdataset to the given number of levels. Alternatively, ‘existing’ will limit recursion to subdatasets that already existed on the filesystem at the start of processing, and prevent new subdatasets from being obtained recursively. [Default: None]

  • get_data (bool, optional) – whether to obtain data for all file handles. If disabled, get operations are limited to dataset handles. [Default: True]

  • description (str or None, optional) – short description to use for a dataset location. Its primary purpose is to help humans to identify a dataset copy (e.g., “mike’s dataset on lab server”). Note that when a dataset is published, this information becomes available on the remote side. [Default: None]

  • reckless ({None, True, False, 'auto', 'ephemeral'} or shared-..., optional) – Obtain a dataset or subdatset and set it up in a potentially unsafe way for performance, or access reasons. Use with care, any dataset is marked as ‘untrusted’. The reckless mode is stored in a dataset’s local configuration under ‘datalad.clone.reckless’, and will be inherited to any of its subdatasets. Supported modes are: [‘auto’]: hard-link files between local clones. In-place modification in any clone will alter original annex content. [‘ephemeral’]: symlink annex to origin’s annex and discard local availability info via git- annex-dead ‘here’ and declares this annex private. Shares an annex between origin and clone w/o git-annex being aware of it. In case of a change in origin you need to update the clone before you’re able to save new content on your end. Alternative to ‘auto’ when hardlinks are not an option, or number of consumed inodes needs to be minimized. Note that this mode can only be used with clones from non-bare repositories or a RIA store! Otherwise two different annex object tree structures (dirhashmixed vs dirhashlower) will be used simultaneously, and annex keys using the respective other structure will be inaccessible. [‘shared-<mode>’]: set up repository and annex permission to enable multi-user access. This disables the standard write protection of annex’ed files. <mode> can be any value support by ‘git init –shared=’, such as ‘group’, or ‘all’. [Default: None]

  • jobs (int or None or {'auto'}, optional) – how many parallel jobs (where possible) to use. “auto” corresponds to the number defined by ‘datalad.runtime.max-annex-jobs’ configuration item. [Default: ‘auto’]

  • on_failure ({'ignore', 'continue', 'stop'}, optional) – behavior to perform on failure: ‘ignore’ any failure is reported, but does not cause an exception; ‘continue’ if any failure occurs an exception will be raised at the end, but processing other actions will continue for as long as possible; ‘stop’: processing will stop on first failure and an exception is raised. A failure is any result with status ‘impossible’ or ‘error’. Raised exception is an IncompleteResultsError that carries the result dictionaries of the failures in its failed attribute. [Default: ‘continue’]

  • result_filter (callable or None, optional) – if given, each to-be-returned status dictionary is passed to this callable, and is only returned if the callable’s return value does not evaluate to False or a ValueError exception is raised. If the given callable supports **kwargs it will additionally be passed the keyword arguments of the original API call. [Default: None]

  • result_renderer – select rendering mode command results. ‘tailored’ enables a command- specific rendering style that is typically tailored to human consumption, if there is one for a specific command, or otherwise falls back on the the ‘generic’ result renderer; ‘generic’ renders each result in one line with key info like action, status, path, and an optional message); ‘json’ a complete JSON line serialization of the full result record; ‘json_pp’ like ‘json’, but pretty-printed spanning multiple lines; ‘disabled’ turns off result rendering entirely; ‘<template>’ reports any value(s) of any result properties in any format indicated by the template (e.g. ‘{path}’, compare with JSON output for all key-value choices). The template syntax follows the Python “format() language”. It is possible to report individual dictionary values, e.g. ‘{metadata[name]}’. If a 2nd-level key contains a colon, e.g. ‘music:Genre’, ‘:’ must be substituted by ‘#’ in the template, like so: ‘{metadata[music#Genre]}’. [Default: ‘tailored’]

  • result_xfm ({'datasets', 'successdatasets-or-none', 'paths', 'relpaths', 'metadata'} or callable or None, optional) – if given, each to-be-returned result status dictionary is passed to this callable, and its return value becomes the result instead. This is different from result_filter, as it can perform arbitrary transformation of the result value. This is mostly useful for top- level command invocations that need to provide the results in a particular format. Instead of a callable, a label for a pre-crafted result transformation can be given. [Default: None]

  • return_type ({'generator', 'list', 'item-or-list'}, optional) – return value behavior switch. If ‘item-or-list’ a single value is returned instead of a one-item return value list, or a list in case of multiple return values. None is return in case of an empty list. [Default: ‘list’]