def collect_data(
user_handle: str,
parent: str,
data_types: List[str],
filename: str,
create_json: bool,
) -> None:
"""Collect data of a Dataverse installation.
Collect data from a data node down the Dataverse
tree-like data structure.
Collects the complete data of a Dataverse instance in
a tree structure (`tree.json`), containing all
Dataverses, Datasets and Datafiles. The file is
stored in your instance directory (e. g.
`utils/data/instances/dataverse_production`).
"""
if user_handle == "public":
api = NativeApi(config.BASE_URL)
else:
users = read_json(config.USER_FILENAME)
api = NativeApi(config.BASE_URL, users[user_handle]["api-token"])
tree = api.get_children(parent, children_types=data_types)
if not os.path.isdir(os.path.join(ROOT_DIR, "data")):
os.makedirs(os.path.join(ROOT_DIR, "data"))
if not os.path.isdir(os.path.join(ROOT_DIR, "data", "utils")):
os.makedirs(os.path.join(ROOT_DIR, "data", "utils"))
if not os.path.isdir(
os.path.join(ROOT_DIR, "data", "utils", user_handle)):
os.makedirs(
os.path.join(ROOT_DIR, "data", "utils", user_handle))
write_json(os.path.join(UTILS_DATA_DIR, user_handle, filename), tree)
if create_json:
generate_data(tree, user_handle, filename)
def remove_testdata(
user_handle: str,
parent: str,
data_types: List[str] = ["dataverses", "datasets"],
force: bool = False,
parent_data_type: str = "dataverse",
remove_parent: bool = False,
) -> None:
"""Remove testdata.
Removes all data created by `create-testdata`.
It recursively collects all Dataverses and Datasets
from a passed Dataverse down (by default =
`science`). If `PRODUCTION` is `true`, this function
will not execute, as long as you not add `--force`
to the function call. This is to protect from
unwanted changes on a production instance.
"""
if config.PRODUCTION and not force:
print(
"Delete testdata on a PRODUCTION instance not allowed. Use --force to force it."
)
sys.exit()
user = read_json(config.USER_FILENAME)[user_handle]
api = NativeApi(config.BASE_URL, user["api-token"])
# Clean up
data = api.get_children(parent, children_types=data_types)
dataverses, datasets, = dataverse_tree_walker(data)
if parent_data_type == "dataverse" and remove_parent:
dataverses.append({"dataverse_alias": parent})
for ds in datasets:
api.destroy_dataset(ds["pid"])
for dv in dataverses:
api.delete_dataverse(dv["dataverse_alias"])