def parse(self, response):
g = None
try:
g = sotools.loadSOGraphFromHtml(response.body, response.url)
except Exception as e:
self.logger.warning(e)
self.logger.info(
f"response url = {response.url} has a Dataset = {sotools.isDataset(g)}"
)
identifiers = sotools.getDatasetIdentifiers(g)
print(str(identifiers))
metadata = sotools.getDatasetMetadataLinks(g)
res = {
"source": response.url,
"identifiers": identifiers,
"metadata": metadata
}
yield res
# Get links to metadata documents referenced from a SO:dataset
import sotools
from pprint import pprint
json_source = "examples/data/ds_m_subjectof.json"
g = sotools.loadSOGraph(filename=json_source,
publicID="https://my.server.net/data/")
links = sotools.getDatasetMetadataLinks(g)
pprint(links, indent=2)