pysize/core/deletion.py

   1 import os
   2
   3 from pysize.core.pysize_global_fs_cache import get_dev_ino
   4
   5 # ['/path/to/file1', '/path/to/file2']
   6 TO_DELETE = []
   7
   8 # {device: {inode: (size_to_subtract|DELETED)}}
   9 DEV_INO_TO_SUBTRACT = {}
  10
  11 DELETED = 'deleted'
  12
  13 def _get_subtracted(path):
  14     dev, ino = get_dev_ino(path)
  15     return DEV_INO_TO_SUBTRACT.setdefault(dev, {}).get(ino, None)
  16
  17 def _iter_to_the_root(path):
  18     while path not in ('/', ''):
  19         yield get_dev_ino(path), path
  20         path = os.path.dirname(path)
  21
  22 def get_subtracted(path):
  23     if DEV_INO_TO_SUBTRACT:
  24         for dev_ino, tmp_path in _iter_to_the_root(path):
  25             if _get_subtracted(tmp_path) is DELETED:
  26                 return DELETED
  27         subtracted = _get_subtracted(path)
  28         if subtracted is not None:
  29             return subtracted
  30     return 0
  31
  32 def _add_hash(path):
  33     from pysize.core import compute_size
  34     size = compute_size.slow(path)
  35     dev, ino = get_dev_ino(path)
  36     DEV_INO_TO_SUBTRACT.setdefault(dev, {})[ino] = DELETED
  37     for (dev, ino), path in _iter_to_the_root(os.path.dirname(path)):
  38         inodes = DEV_INO_TO_SUBTRACT.setdefault(dev, {})
  39         subtract = inodes.get(ino, 0)
  40         subtract += size
  41         inodes[ino] = subtract
  42
  43 # TODO: Deletion of hard links
  44 def schedule(nodes):
  45     for node in nodes:
  46         for path in node.get_fullpaths():
  47             subtract = get_subtracted(path)
  48             if subtract is not DELETED:
  49                 TO_DELETE.append(path)
  50                 _add_hash(path)
  51     _prune_list()
  52     _rebuild_hash()
  53
  54 def _prune_list():
  55     """Remove children if we have the parent"""
  56     while True:
  57         for i in xrange(len(TO_DELETE)):
  58             parent = os.path.dirname(TO_DELETE[i])
  59             if get_subtracted(parent) is DELETED:
  60                 del TO_DELETE[i]
  61                 break
  62         else:
  63             break
  64
  65 def _rebuild_hash():
  66     DEV_INO_TO_SUBTRACT.clear()
  67     for path in TO_DELETE:
  68         _add_hash(path)
  69
  70 def filter_deleted(paths):
  71     if DEV_INO_TO_SUBTRACT:
  72         def is_not_deleted(path):
  73             return get_subtracted(path) is not DELETED
  74         return filter(is_not_deleted, paths)
  75     return paths
  76
  77 def get_deleted():
  78     return TO_DELETE[:]
  79
  80 def restore(path):
  81     from pysize.core import compute_size
  82     TO_DELETE.remove(path)
  83     dev, ino = get_dev_ino(path)
  84     del DEV_INO_TO_SUBTRACT[dev][ino]
  85     size = compute_size.slow(path, account_deletion=False)
  86     for (dev, ino), path in _iter_to_the_root(os.path.dirname(path)):
  87         inodes = DEV_INO_TO_SUBTRACT[dev]
  88         if inodes[ino] == size:
  89             del inodes[ino]
  90         else:
  91             inodes[ino] -= size