def calc_dir_sizes(filestat_file, dirstat_file):
lines = FileUtil.read_all(filestat_file).splitlines()
dir2size = {}
for filepath, size in FileUtil.parse_path_sizes(lines):
dir2size[filepath] = size
dirpath = os.path.dirname(filepath)
while dirpath:
try:
dir2size[dirpath] += size
except KeyError:
dir2size[dirpath] = size
dirpath = dirpath.rpartition("/")[0]
iter_lines = FileUtil.combine_path_sizes(sorted(dir2size.items()))
FileUtil.write_all(dirstat_file, iter_lines)
def test_01_file_util(self):
from fileutil import FileUtil
test_file = __file__ + ".test_01.data"
path_sizes = [
("path1", 1),
("path2", 2),
("path3", 3),
]
iter_lines = FileUtil.combine_path_sizes(path_sizes)
FileUtil.write_all(test_file, iter_lines)
data = FileUtil.read_all(test_file)
self.assertEqual("""
path1: 1
path2: 2
path3: 3
""".strip(), data.strip())
path2size = dict([(path, size) for path, size in FileUtil.parse_path_sizes(data.splitlines())])
self.assertEqual(path2size['path2'], 2)
def scan_file_sizes(from_path, stat_file):
iter_path_sizes = FileUtil.walk_path_sizes(from_path)
iter_lines = FileUtil.combine_path_sizes(iter_path_sizes)
FileUtil.write_all(stat_file, iter_lines)