You can create a dataset collection from Python using the init()
function. You need to provided a file system friendly dataset name.
Additional metadata can be added to the collection if you need it.
The standard approach for using dataset is using the from
notation for your import. The examples code assumes you’ve
done this.
from py_dataset import dataset
We need to have a collection name, e.g. things.ds. The extension
“.ds” is a convension and is not enforced by the dataset library.
# We're going to save our collection name for latter
c_name = 'things.ds'
if not dataset.init(c_name):
print(dataset.error_message())
At this point you should have a directory (folder) on your file system called “things.ds” it will contain a collection.json file.
The library will try to guess some metadata about the collection based on the environment where your script is running. E.g. it will use the current date and time to record when the collection was created. On Unix-like systems it’ll also add the creating user as a creator of the collection.
Metadata for the collection is based around the concept in libraries and archival software called “namaste” or Name as text. This was pioneered by the developers at the California Digital Library. See Namaste.
py_dataset supports the following namaste style attributes –
who, what, when, where
dataset.set_who(c_name, "Jane Doe")
You can retreive this metadata with get_who()
who = dataset.get_who(c_name)
“whats”, “when”, “where” work as follows