def fetch_collection(api: AlephAPI,
prefix: Optional[str],
foreign_id: str,
overwrite: bool = False):
path = _fix_path(prefix)
collection = api.get_collection_by_foreign_id(foreign_id)
if collection is None:
return
filters = [("collection_id", collection.get("id"))]
params = {"empty:properties.parent": "true"}
results = api.search("",
filters=filters,
schemata="Document",
params=params)
label = collection.get("label")
log.info("Dataset [%s]: %s (%d children)", path, label, len(results))
for entity in results:
fetch_object(api, path, entity, overwrite=overwrite)
def load_aleph(foreign_id, api_url, api_key):
api = AlephAPI(api_url, api_key)
collection_id = None
if foreign_id is not None:
collection = api.get_collection_by_foreign_id(foreign_id)
if collection is None:
raise click.BadParameter("Cannot find collection: %s" % foreign_id)
collection_id = collection.get('id')
stdout = click.get_text_stream('stdout')
entities = api.stream_entities(collection_id=collection_id,
include=['schema', 'properties'])
for data in entities:
if 'properties' not in data:
continue
entity = model.get_proxy(data)
api_url = urljoin(api.base_url, 'entities/%s' % entity.id)
entity.add('alephUrl', api_url, quiet=True)
write_object(stdout, entity)