Skip to content

Latest commit

 

History

History
182 lines (132 loc) · 6.74 KB

CHANGELOG.md

File metadata and controls

182 lines (132 loc) · 6.74 KB
Raw

CHANGELOG

Log of changes in the versions

v1.6.1

  • hotfix skipND when calling dump_jsonld(). Option was not passed correctly to underlying function.

v1.6.0

  • rootgroup as alias for rootparent
  • ZenodoRecord has new property env_name_for_token to define the environment variable name to be used for the Zenodo token
  • bugfix downloading zenodo files
  • allowing higher versions of pymongo

v1.5.2

  • bugfix dumping json-ld

v1.5.1

v1.5.0

  • make compliant with higher pydantic and ontolutils versions
  • concrete version selection for other dependencies

v1.4.1

  • Downloading files will be cached by their checksum and/or URL. This avoids multiple downloads of the same file.
  • RepositoryFile has new abstract property suffix,
  • RepositoryInterface has new abstract method get_jsonld
  • RepositoryInterface has new abstract property identifier and title
  • bugfixes

v1.4.0

  • update package dependency versions
  • minor bugfixes and updates in documentation

v1.4.0rc2

  • using suffix .jsonld instead of .json for JSON-LD files, as it is recommended (see https://www.w3.org/TR/json-ld/#iana-considerations)
  • bugfixes in documentation (links, figures, ...)
  • enhancing zenodo interfaces:
    • removed depr methods, e.g. get() from AbstractZenodoInterface
    • using cached json dict for zenodo records. call refresh() to update the json
    • minor bugfixes
    • introduced property files, which is Dict[str, RepositoryFile]
    • improve url handling by using properties instead of class variables

v1.4.0rc1

  • The repository interface to Zenodo has one single upload method upload_file with the parameter metamapper. It is a callable which extracts meta information from the actual file to be uploaded. This is especially useful and specifically intended for HDF5 files. Unless the value for metamapper is None, the upload_file method will use the built-in hdf5 extraction function automatically on HDF5 files.
  • Clarify abstraction for HDF5 database interfaces. HDF5DBInterface is the top abstraction from which ExtHDF5Interface inherits. ExtHDF5Interface makes use of external databases such as mongoDB.
  • fix issue in online documentation: mongomock is used to run the mongodb jupyter notebook in the documentation
  • codemeta.json file is updated with author and institution ROR ID

v1.3.2

  • calling the RDF accessor on an attribute name will only work if the attribute already exists. If not, an error is raised.
  • Likewise, if an attribute is deleted, the entry in the RDF accessor dictionary is deleted

v1.3.1

  • minor fixes
  • add $in operator to query functions
  • add argument rebase to layout specifications

v1.3.0

  • important changes:
    • improved and consequent support of RDF/JSON-LD. This means, an HDF5 can be created from a JSON-LD file and vice versa. The JSON-LD file contains the structural and contextual metadata of the HDF5 file.
    • namespaces are outsourced to ontolutils
  • minor changes:
    • When a file is opened with a filename which does not exist and mode is None, the file will NOT be created. This was the case in the past, but this may lead to unwanted empty files.
    • Bugfix namespace creation
    • some method renaming and refactoring
    • accessors are refactored and improved (especially shifted away from xarray and fully integrated in hdf)

v1.2.2

  • Hotfix dumping json-ld data (dimension scales were the issue)

v1.2.1

  • Add codemeta namespace
  • Improved json-ld export
  • Updated qudt namespace
  • colab notebook will be managed on a separate branch. the readme link points to the branch

v1.2.0

  • Improved assignment of IRI to attributes
  • Export of a JSON-LD file possible
  • Updated documentation
  • bugfixes

v1.1.1

  • bugfix: Setting a default value for toolbox validators in convention yaml file was not working. Fixed it.

v1.1.0

  • simplified and clean up much code, especially convention sub package
  • added identifier utils
  • updated and improved documentation

v1.0.1

  • fixed unnecessary call in create_dataset, which writes the data twice. Now, the time data is written is comparable to the time h5py needs to write the data (for small datasets h5py is still faster due to the (constant) overhead, h5tbx adds).

v1.0.0

major changes:

  • zenodo is not a dependency anymore but is implemented as a new subpackage of the toolbox
  • zenodo is part or repository which is designed to provide interfaces to different data repositories (however, only zenodo is implemented at the moment)
  • the database architecture is changed similarly, such that it has a more logic structure
  • both above changes follow a more or less strict inheritance structure from abstract classes defining the interface to repositories or databases (databases are meant to be local, like mongoDB, SQL, etc., repositories are online data storage, like zenodo, which allows searching for metadata but not within the raw files.)
  • python 3.8 until 3.12 inclusive are supported
  • IRI as persistent identifier is now supported, which fulfills the F3 requirement of the FAIR principles ("Metadata clearly and explicitly include the identifier of the data they describe", https://www.go-fair.org/fair-principles/)
  • package renaming and reorganization: conventions is now convention, layout is now a module, new is repository
  • usage of IRI (persistent identifier) is now supported

v0.13.0

  • scale and offset is now implemented in the package. it should no longer be defined in a convention.
  • bugfix normalization extension
  • bugfix exporting xr.DataArray built with the toolbox to netCDF
  • support usage of IRI to describe metadata

v0.12.2

  • bugfix requirements
  • add $exist find()-methods inside of HDF files
  • bugfix 0D time data as dimension
  • module query functions (find, ...) can guess the filename from objects that have with hasattr(hdf_filename, 'hdf_filename')

v0.12.1

  • bugfix in zenodo search (did Zenodo change their API?)

v0.12.0

  • 0D data is written to MongoDB
  • new utils like computing file size
  • update to new zenodo_search package due to change in backend at Zenodo.org
  • find, find_one and distinct can be called on HDF files
  • small bugfixes

v0.11.1

  • bugfix standard attribute validation
  • bugfix in EngMeta.ipynb

v0.11.0

  • working with time data is now possible:
    • time data can be created using the high-level method create_time_dataset
    • slicing the above "time datasets" will return xarray data
    • See docs/wrapper/Misc.ipynb
  • fixed issue with user-defined & nested standard attributes

v0.10.0

  • Initial version, published on pypi