Improve the user friendliness of the SoftwareMetadata object #423
Description
In issue #376 the SoftwareMetadata class (subclass of ld_dict) is introduced.
It is supposed to provide an abstracted API for data access by clients for example when harvesting.
Data is added by accessing the SoftwareMetadata object like a dictionary for example:
from hermes.model import SoftwareMetadata
data = SoftwareMetadata()
data["schema:field"] = "foo"This is fine for single strings. When writing a harvest plugin one might not consider that this can lead to problems:
from hermes.model import SoftwareMetadata
data = SoftwareMetadata()
data["schema:field"] = "foo"
# something else might get done here
data["schema:field"] = "bar"In this case "schema:field" only contains "bar" by accident instead of ["foo", "bar"].
I therefor suggest adding an "add" method to the SoftwareMetadata or ld_dict class as SoftwareMetadata is a subclass of ld_dict. It should work like this:
from hermes.model import SoftwareMetadata
data = SoftwareMetadata()
data.add(key="schema:field", value="foo")
data.add(key="schema:field", value="bar")
data.add(key="schema:field", value="foobar")
# equivalent to data["schema:field"] = ["foo", "bar", "foobar"]Not yet specified is how lists or dicts should be added. I suggest that this should be done like this for user convenience:
from hermes.model import SoftwareMetadata
from hermes.model.types.ld_list import ld_list
from hermes.model.types.ld_dict import ld_dict
data = SoftwareMetadata()
data["schema:field"] = ["foo", "bar", "foobar"]
# instead of data["schema:field"] = ld_list.from_list(["foo", "bar", "foobar"])
data["schema:other_field"] = {"@type": "schema:Person", "name": "ABC"}
# instead of data["schema:other_field"] = ld_dict.from_dict({"@type": "schema:Person", "name": "ABC"})