datalad.api.aggregate_metadata(path=None, *, dataset=None, recursive=False, recursion_limit=None, update_mode='target', incremental=False, force_extraction=False, save=True)

Aggregate metadata of one or more datasets for later query.

Metadata aggregation refers to a procedure that extracts metadata present in a dataset into a portable representation that is stored a single standardized format. Moreover, metadata aggregation can also extract metadata in this format from one dataset and store it in another (super)dataset. Based on such collections of aggregated metadata it is possible to discover particular datasets and specific parts of their content, without having to obtain the target datasets first (see the DataLad ‘search’ command).

To enable aggregation of metadata that are contained in files of a dataset, one has to enable one or more metadata extractor for a dataset. DataLad supports a number of common metadata standards, such as the Exchangeable Image File Format (EXIF), Adobe’s Extensible Metadata Platform (XMP), and various audio file metadata systems like ID3. DataLad extension packages can provide metadata data extractors for additional metadata sources. For example, the neuroimaging extension provides extractors for scientific (meta)data standards like BIDS, DICOM, and NIfTI1. Some metadata extractors depend on particular 3rd-party software. The list of metadata extractors available to a particular DataLad installation is reported by the ‘wtf’ command (‘datalad wtf’).

Enabling a metadata extractor for a dataset is done by adding its name to the ‘datalad.metadata.nativetype’ configuration variable – typically in the dataset’s configuration file (.datalad/config), e.g.:

[datalad "metadata"]
  nativetype = exif
  nativetype = xmp

If an enabled metadata extractor is not available in a particular DataLad installation, metadata extraction will not succeed in order to avoid inconsistent aggregation results.

Enabling multiple extractors is supported. In this case, metadata are extracted by each extractor individually, and stored alongside each other. Metadata aggregation will also extract DataLad’s own metadata (extractors ‘datalad_core’, and ‘annex’).

Metadata aggregation can be performed recursively, in order to aggregate all metadata across all subdatasets, for example, to be able to search across any content in any dataset of a collection. Aggregation can also be performed for subdatasets that are not available locally. In this case, pre-aggregated metadata from the closest available superdataset will be considered instead.

Depending on the versatility of the present metadata and the number of dataset or files, aggregated metadata can grow prohibitively large. A number of configuration switches are provided to mitigate such issues.


If set to false, content metadata aggregation will not be performed for the named metadata extractor (a potential underscore ‘_’ in the extractor name must be replaced by a dash ‘-‘). This can substantially reduce the runtime for metadata extraction, and also reduce the size of the generated metadata aggregate. Note, however, that some extractors may not produce any metadata when this is disabled, because their metadata might come from individual file headers only. ‘’ might be a more appropriate setting in such cases.


Any metadata key matching any regular expression in this configuration setting is removed prior to generating the dataset-level metadata summary (keys and their unique values across all dataset content), and from the dataset metadata itself. This switch can also be used to filter out sensitive information prior aggregation.


If set to false, DataLad will not auto-generate a summary of unique content metadata values for a particular extractor as part of the dataset-global metadata (a potential underscore ‘_’ in the extractor name must be replaced by a dash ‘-‘). This can be useful if such a summary is bloated due to minor uninformative (e.g. numerical) differences, or when a particular extractor already provides a carefully designed content metadata summary.


Any metadata value that exceeds the size threshold given by this configuration setting (in bytes/characters) is removed.

If set, extracted content metadata are still used to generate a dataset-level summary of present metadata (all keys and their unique values across all files in a dataset are determined and stored as part of the dataset-level metadata aggregate, see datalad.metadata.generate-unique-<extractor-name>), but metadata on individual files are not stored. This switch can be used to avoid prohibitively large metadata files. Discovery of datasets containing content matching particular metadata properties will still be possible, but such datasets would have to be obtained first in order to discover which particular files in them match these properties.

  • path (sequence of str or None, optional) – path to datasets that shall be aggregated. When a given path is pointing into a dataset, the metadata of the containing dataset will be aggregated. If no paths given, current dataset metadata is aggregated. [Default: None]

  • dataset (Dataset or None, optional) – topmost dataset metadata will be aggregated into. All dataset between this dataset and any given path will receive updated aggregated metadata from all given paths. [Default: None]

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

  • recursion_limit (int or None, optional) – limit recursion into subdatasets to the given number of levels. [Default: None]

  • update_mode ({'all', 'target'}, optional) – which datasets to update with newly aggregated metadata: all datasets from any leaf dataset to the top-level target dataset including all intermediate datasets (all), or just the top-level target dataset (target). [Default: ‘target’]

  • incremental (bool, optional) – If set, all information on metadata records of subdatasets that have not been (re-)aggregated in this run will be kept unchanged. This is useful when (re-)aggregation only a subset of a dataset hierarchy, for example, because not all subdatasets are locally available. [Default: False]

  • force_extraction (bool, optional) – If set, all enabled extractors will be engaged regardless of whether change detection indicates that metadata has already been extracted for a given dataset state. [Default: False]

  • save (bool, optional) – by default all modifications to a dataset are immediately saved. Giving this option will disable this behavior. [Default: True]

  • 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’]