def get_key(path):
if path == "/" or path == "":
raise CellxgeneException(
"No matching dataset found.", status.HTTP_404_NOT_FOUND
)
trimmed = path[:-1] if path[-1] == "/" else path
try:
# valid paths come in three forms:
if trimmed.endswith('.h5ad') and data_file_exists(trimmed):
# 1) somedir/dataset.h5ad: a dataset
return CacheKey(trimmed, trimmed, None)
elif trimmed.endswith('.csv'):
# 2) somedir/dataset_annotations/saldaal1-T5HMVBNV.csv : an actual annotations file.
annotations_dir = os.path.split(trimmed)[0]
dataset = make_h5ad(annotations_dir)
if data_file_exists(dataset):
data_dir_ensure(annotations_dir)
return CacheKey(trimmed, dataset, trimmed)
elif trimmed.endswith('_annotations') and data_dir_exists(trimmed):
# 3) somedir/dataset_annotations: an annotation directory. The corresponding h5ad must exist, but the directory may not.
dataset = make_h5ad(trimmed)
if data_file_exists(dataset):
return CacheKey(trimmed, dataset, '')
except CellxgeneException:
pass
split = os.path.split(trimmed)
return get_key(split[0])
def recurse_dir(path):
if not os.path.exists(path):
raise CellxgeneException("The given path does not exist.",
status.HTTP_400_BAD_REQUEST)
all_entries = os.listdir(path)
def is_h5ad(el):
return el.endswith('.h5ad') and os.path.isfile(os.path.join(path, el))
h5ad_entries = [x for x in all_entries if is_h5ad(x)]
annotation_dir_entries = [
x for x in all_entries
if x.endswith(annotations_suffix) and make_h5ad(x) in h5ad_entries
]
def list_annotations(el):
full_path = os.path.join(path, el)
if not os.path.isdir(full_path):
entries = []
else:
entries = [{
"name":
x[:-13] if (len(x) > 13 and x[-13] in ['-', '_']) else
(x[:-4] if x.endswith('.csv') else x),
"path":
os.path.join(full_path, x).replace(env.cellxgene_data, ""),
} for x in os.listdir(full_path) if x.endswith('.csv')
and os.path.isfile(os.path.join(full_path, x))]
return [{
"name": 'new',
"class": 'new',
"path": full_path.replace(env.cellxgene_data, "")
}] + entries
def make_entry(el):
full_path = os.path.join(path, el)
if el in h5ad_entries:
return {
"path": full_path.replace(env.cellxgene_data, ""),
"name": el,
"type": "file",
"annotations": list_annotations(make_annotations(el)),
}
elif os.path.isdir(full_path) and el not in annotation_dir_entries:
return {
"path": full_path.replace(env.cellxgene_data, ""),
"name": el,
"type": "directory",
"children": recurse_dir(full_path),
}
else:
return {
"path": full_path,
"name": el,
"type": "neither",
}
return [make_entry(x) for x in os.listdir(path)]
def test_GIVEN_annotation_dir_THEN_returns_h5ad(self):
self.assertEqual(make_h5ad("pbmc_annotations"), "pbmc.h5ad")