def du(summarize, human_readable, s3_paths):
"""
display disk usage statistics
"""
size_formatter = magic.human_bytes if human_readable else lambda s: s
for s3_path in s3_paths:
totals = {}
# In recursive tallying of totals this key is the stop condition.
stop_path = '/' + s3_util.bucket_and_key(s3_path)[1].rstrip('/')
def tally(path_segment, size):
if not summarize or path_segment == stop_path:
# If the summarize option is present, we only care about the grand total.
if path_segment not in totals:
totals[path_segment] = 0
totals[path_segment] += size
if path_segment != stop_path:
parent_dir = path_segment.rsplit('/', 1)[0]
if parent_dir == '':
# Edge case when totalling the entire bucket.
parent_dir = '/'
tally(parent_dir, size)
for obj in s3_util.keys(s3_path):
# usage for all given paths, and recursively for directories (excludes individual files)
dbg('adding {}', obj)
dir_key = '/' + obj['key'].rsplit('/', 1)[0]
tally(dir_key, obj['len'])
for path, total in sorted(totals.items()):
out('{}\t{}', size_formatter(total), path)
def keys(s3_paths):
"""
Prefix key scan.
Outputs keys in the format s3://bucket/key
"""
if not len(s3_paths):
s3_paths = ['']
for s3_path in s3_paths:
for obj in s3_util.keys(s3_path):
out('s3://{}/{}', obj['bucket'], obj['key'])
