def _assert_raise_mke_with_keys(exception_keys: List[str],
*args,
**kwargs):
try:
meta_add(*args, **kwargs)
raise RuntimeError("MetadataKeyException not raised")
except MetadataKeyException as mke:
eq_(mke.keys, exception_keys)
def test_object_parameter():
with \
patch("datalad_metalad.add.add_file_metadata") as fp, \
patch("datalad_metalad.add.add_dataset_metadata") as dp:
meta_add(
metadata={
**metadata_template,
"type": "file",
"path": "d1/d1.1./f1.1.1"
})
assert_true(fp.call_count == 1)
assert_true(dp.call_count == 0)
def test_add_file_end_to_end(file_name):
test_path = "d_0/d_0.0/f_0.0.0"
json.dump({
**metadata_template,
"type": "file",
"path": test_path
}, open(file_name, "tw"))
with tempfile.TemporaryDirectory() as temp_dir:
git_repo = GitRepo(temp_dir)
res = meta_add(metadata=file_name, metadata_store=git_repo.path)
assert_result_count(res, 1)
assert_result_count(res, 1, type='file')
assert_result_count(res, 0, type='dataset')
# Verify file level metadata was added
tree_version_list, uuid_set, mrr = _get_top_nodes(
git_repo,
UUID(metadata_template["dataset_id"]),
metadata_template["dataset_version"])
file_tree = mrr.get_file_tree()
assert_is_not_none(file_tree)
assert_true(test_path in file_tree)
metadata = file_tree.get_metadata(MetadataPath(test_path))
metadata_content = _get_metadata_content(metadata)
eq_(metadata_content, metadata_template["extracted_metadata"])
def test_add_dataset_end_to_end(file_name):
json.dump({
**metadata_template,
"type": "dataset"
},
open(file_name, "tw"))
with tempfile.TemporaryDirectory() as temp_dir:
git_repo = GitRepo(temp_dir)
res = meta_add(metadata=file_name, metadata_store=git_repo.path)
assert_result_count(res, 1)
assert_result_count(res, 1, type='dataset')
assert_result_count(res, 0, type='file')
# Verify dataset level metadata was added
tree_version_list, uuid_set, mrr = _get_top_nodes(
git_repo,
UUID(metadata_template["dataset_id"]),
metadata_template["dataset_version"])
metadata = mrr.get_dataset_level_metadata()
metadata_content = _get_metadata_content(metadata)
eq_(metadata_content, metadata_template["extracted_metadata"])
def test_unknown_key_allowed(file_name):
json.dump({
**metadata_template,
"type": "dataset",
"strange_key_name": "some value"
},
open(file_name, "tw"))
with \
patch("datalad_metalad.add.add_file_metadata") as fp, \
patch("datalad_metalad.add.add_dataset_metadata") as dp:
meta_add(
metadata=file_name,
allow_unknown=True)
assert_true(fp.call_count == 0)
assert_true(dp.call_count == 1)
def test_override_key_allowed(file_name):
json.dump({
**metadata_template,
"type": "dataset"
},
open(file_name, "tw"))
with \
patch("datalad_metalad.add.add_file_metadata") as fp, \
patch("datalad_metalad.add.add_dataset_metadata") as dp:
meta_add(
metadata=file_name,
additionalvalues=json.dumps(
{"dataset_id": "a1010203-1011-2021-3031-404142434445"}),
allow_override=True)
assert_true(fp.call_count == 0)
assert_true(dp.call_count == 1)
def test_optional_keys(file_name):
json.dump({
**metadata_template,
"type": "file",
"path": "d1/d1.1./f1.1.1"
},
open(file_name, "tw"))
with \
patch("datalad_metalad.add.add_file_metadata") as fp, \
patch("datalad_metalad.add.add_dataset_metadata") as dp:
meta_add(
metadata=file_name,
allow_unknown=True)
assert_true(fp.call_count == 1)
assert_true(dp.call_count == 0)
def test_subdataset_add_file_end_to_end(file_name):
test_path = "d_1/d_1.0/f_1.0.0"
json.dump({
**metadata_template,
**additional_keys_template,
"type": "file",
"path": test_path
}, open(file_name, "tw"))
with tempfile.TemporaryDirectory() as temp_dir:
git_repo = GitRepo(temp_dir)
res = meta_add(metadata=file_name, metadata_store=git_repo.path)
assert_result_count(res, 1)
assert_result_count(res, 1, type='file')
assert_result_count(res, 0, type='dataset')
# Verify dataset level metadata was added
root_dataset_id = UUID(additional_keys_template["root_dataset_id"])
root_dataset_version = additional_keys_template["root_dataset_version"]
dataset_tree_path = MetadataPath(
additional_keys_template["dataset_path"])
tree_version_list, uuid_set, mrr = _get_top_nodes(
git_repo,
root_dataset_id,
root_dataset_version)
_, dataset_tree = tree_version_list.get_dataset_tree(
root_dataset_version)
mrr = dataset_tree.get_metadata_root_record(dataset_tree_path)
eq_(mrr.dataset_identifier, UUID(metadata_template["dataset_id"]))
file_tree = mrr.get_file_tree()
assert_is_not_none(file_tree)
assert_true(test_path in file_tree)
metadata = file_tree.get_metadata(MetadataPath(test_path))
metadata_content = _get_metadata_content(metadata)
eq_(metadata_content, metadata_template["extracted_metadata"])