clean all: really calculate disk usage

While we mention "disk usage" in the related messages and docs, we
display apparent size of the files instead of actual disk usage.  This
commit fixes that (produces the same output as the du(1) tool).

cli.py

@@ -1781,18 +1781,15 @@ def cleanCli(self, userlist):
  col = 0
  else:
  col = 1
- # Recursively gather all files in this repodir
- files = yum.misc.getFileList(path, '', [])
  else:
  # Ordinary file (such as timedhosts)
  col = 3
- files = [path]
- usage = sum(map(os.path.getsize, files))
+ usage = yum.misc.disk_usage(path)
  if usage > 0:
  table[col].append((usage, path))
 
  # Print the table (verbose mode only)
- lines = [_('Disk usage of %s after cleanup:') % cacheglob]
+ lines = [_('Disk usage under %s after cleanup:') % cacheglob]
  headers = ('enabled repos', 'disabled repos', 'untracked repos',
  'other data')
  totals = [0, 0, 0, 0]

yum/misc.py

@@ -1310,3 +1310,15 @@ def validate_repoid(repoid):
  return char
  else:
  return None
+
+def disk_usage(path):
+ """Return disk usage of the given filename, recursively for dirs."""
+ def usage(path):
+ return os.stat(path).st_blocks * 512
+ total = usage(path)
+ if not os.path.isdir(path):
+ return total
+ for root, dirs, files in os.walk(path):
+ paths = (os.path.join(root, entry) for entry in dirs + files)
+ total += sum(usage(path) for path in paths)
+ return total