py-cube is a package to build and publish cubes as defined by cube.link, describing a schema to describe structured data from tables in RDF. It allows for an alternative to the Cube-Creator. Currently this project is heavily linked to the LINDAS the Swiss Federal Linked Data Service.
For further information, please refer to our Wiki
There are two ways to install this package, locally or through the Python Package Index (PyPI).
Clone this repository and cd into the directory. You can now install this package locally on your machine - we advise to use a virtual environment to avoid conflicts with other projects. Additionally, install all dependencies as described in requirements.txt
pip install -e .
pip install -r requirements.txtNOT yet implemented Once Published, you'll be able to install this package through pip without cloning the repository.
pip install py-cubeIf you wish to contribute to this project, feel free to clone this repository and open a pull request to be reviewed and merged.
Alternatively feel free to open an issue with a suggestion on what we could implement. We laid out a rough road map for the features ahead on our Timetable
To avoid the feeling of a black box, our philosophy is to make the construction of cubes modular. The process will take place in multiple steps, outlined below.
cube = pycube.Cube(dataframe: pd.Dataframe, cube_yaml: dict, shape_yaml: dict)This step sets some need background information about the cube up.
cube.prepare_data()Adds observation URIs and applies the mappings as described in the shape yaml.
cube:Cubecube.write_cube()Writes the cube:Cube.
cube:Observationcube.write_observations()Writes the cube:Observations and the cube:ObservationSet. The URI for the observations are written as <cube_URI/observations/[list_of_key_dimensions]>. This should avoid the possibilities of conflicts in their uniqueness.
cube:ObersvationConstraintcube.write_shape()Writes the cube:ObservationConstraint.
# Write the cube
cube = pycube.Cube(dataframe: pd.DataFrame, cube_yaml: dict, shape_yaml: dict)
cube.apply_mapping()
cube.write_cube()
cube.write_observations()
cube.write_shape()
# Upload the cube
cube.upload(endpoint: str, named_graph: str)For an upload, use cube.upload(endpoint: str, named_graph: str) with the proper endpoint as well as named_graph.
A lindas.ini file is read for this step, containing these information as well as a password. It contains the structure:
[TEST]
endpoint=https://stardog-test.cluster.ldbar.ch
username=a-lindas-user-name
password=something-you-don't-need-to-see;)With additional information for the other environments.
If you wish, a command line utility is present, that expects an opinionated way to store the data and the description in a directory. It then helps you to perform common operations.
The directory should be structured as follows:
data.csv: This file contains the observations.description.json or description.yml: This file contains the cube and dimension descriptions.For example, to serialize the data, use:
python cli.py serialize <input_directory> <output_ttl_file>For additional help and options, you can use:
python cli.py --helpThere is the possibility to download datasets from other data sources. Right now, the functionality is basic, but it could be possible in the future to extend it.
See Frictionless for more information on Frictionless.
python fetch.py 'https://data.europa.eu/data/datasets/fc49eebf-3750-4c9c-a29e-6696eb644362?locale=en' example/corona/Multiple cube example are ready in the example directory.
$ python cli.py example list
corona: Corona Numbers Timeline
kita: Number of kids in day care facilities
wind: Wind turbines — operated WKA per year in Schleswig-HolsteinTo load an example in a Fuseki database, you can use the load subcommand of the example command.
$ python cli.py example load kitaThere is a start-fuseki command that can be used to start a Fuseki server containing data
from the examples.
$ python cli.py example start-fuseki