You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
bazarr/libs/fcache/cache.py

305 lines
11 KiB

6 years ago
import codecs
import logging
import os
import pickle
import shutil
import tempfile
import appdirs
try:
from collections.abc import MutableMapping
unicode = str
except ImportError:
# Python 2 imports
from collections import MutableMapping
FileNotFoundError = IOError
from .posixemulation import rename
logger = logging.getLogger(__name__)
class FileCache(MutableMapping):
"""A persistent file cache that is dictionary-like and has a write buffer.
*appname* is passed to `appdirs <https://pypi.python.org/pypi/appdirs/>`_
to determine a system-appropriate location for the cache files. The cache
directory used is available via :data:`cache_dir`.
By default, a write buffer is used, so writing to cache files is not done
until :meth:`sync` is explicitly called. This behavior can be changed using
the optional *flag* argument.
.. NOTE::
Keys and values are always stored as :class:`bytes` objects. If data
serialization is enabled, keys are returned as :class:`str` or
:class:`unicode` objects.
If data serialization is disabled, keys are returned as a
:class:`bytes` object.
:param str appname: The app/script the cache should be associated with.
:param str flag: How the cache should be opened. See below for details.
:param mode: The Unix mode for the cache files.
:param str keyencoding: The encoding the keys use, defaults to 'utf-8'.
This is used if *serialize* is ``False``; the keys are treated as
:class:`bytes` objects.
:param bool serialize: Whether or not to (de)serialize the values. If a
cache is used with a :class:`~shelve.Shelf`, set this to ``False``.
:param str app_cache_dir: absolute path to root cache directory to be
used in place of system-appropriate location determined by appdirs
The optional *flag* argument can be:
+---------+-------------------------------------------+
| Value | Meaning |
+=========+===========================================+
| ``'r'`` | Open existing cache for reading only |
+---------+-------------------------------------------+
| ``'w'`` | Open existing cache for reading and |
| | writing |
+---------+-------------------------------------------+
| ``'c'`` | Open cache for reading and writing, |
| | creating it if it doesn't exist (default) |
+---------+-------------------------------------------+
| ``'n'`` | Always create a new, empty cache, open |
| | for reading and writing |
+---------+-------------------------------------------+
If a ``'s'`` is appended to the *flag* argument, the cache will be opened
in sync mode. Writing to the cache will happen immediately and will not be
buffered.
If an application needs to use more than one cache, then it should use
subcaches. To create a subcache, append a series of one or more names
separated by periods to the application name when creating a
:class:`FileCache` object (e.g. ``'appname.subcache'`` or
``'appname.subcache.subcache'``).
Subcaches are a way for an application to use more than one cache without
polluting a user's cache directory. All caches -- main caches or subcaches
-- are totally independent. The only aspect in which they are linked is
that all of an application's caches exist in the same system directory.
Because each cache is independent of every other cache, calling
:meth:`delete` on an application's main cache will not delete data in
its subcaches.
"""
def __init__(self, appname, flag='c', mode=0o666, keyencoding='utf-8',
5 years ago
serialize=True, app_cache_dir=None):
6 years ago
"""Initialize a :class:`FileCache` object."""
if not isinstance(flag, str):
raise TypeError("flag must be str not '{}'".format(type(flag)))
elif flag[0] not in 'rwcn':
raise ValueError("invalid flag: '{}', first flag must be one of "
"'r', 'w', 'c' or 'n'".format(flag))
elif len(flag) > 1 and flag[1] != 's':
raise ValueError("invalid flag: '{}', second flag must be "
"'s'".format(flag))
appname, subcache = self._parse_appname(appname)
if 'cache' in subcache:
raise ValueError("invalid subcache name: 'cache'.")
self._is_subcache = bool(subcache)
if not app_cache_dir:
app_cache_dir = appdirs.user_cache_dir(appname, appname)
subcache_dir = os.path.join(app_cache_dir, *subcache)
self.cache_dir = os.path.join(subcache_dir, 'cache')
exists = os.path.exists(self.cache_dir)
if len(flag) > 1 and flag[1] == 's':
self._sync = True
else:
self._sync = False
self._buffer = {}
if exists and 'n' in flag:
self.clear()
self.create()
elif not exists and ('c' in flag or 'n' in flag):
self.create()
elif not exists:
raise FileNotFoundError("no such directory: '{}'".format(
self.cache_dir))
self._flag = 'rb' if 'r' in flag else 'wb'
self._mode = mode
self._keyencoding = keyencoding
self._serialize = serialize
def _parse_appname(self, appname):
"""Splits an appname into the appname and subcache components."""
components = appname.split('.')
return components[0], components[1:]
def create(self):
"""Create the write buffer and cache directory."""
if not self._sync and not hasattr(self, '_buffer'):
self._buffer = {}
if not os.path.exists(self.cache_dir):
os.makedirs(self.cache_dir)
def clear(self):
"""Remove all items from the write buffer and cache.
The write buffer object and cache directory are not deleted.
"""
self.delete()
self.create()
def delete(self):
"""Delete the write buffer and cache directory."""
if not self._sync:
del self._buffer
shutil.rmtree(self.cache_dir)
def close(self):
"""Sync the write buffer, then close the cache.
If a closed :class:`FileCache` object's methods are called, a
:exc:`ValueError` will be raised.
"""
self.sync()
self.sync = self.create = self.delete = self._closed
self._write_to_file = self._read_to_file = self._closed
self._key_to_filename = self._filename_to_key = self._closed
self.__getitem__ = self.__setitem__ = self.__delitem__ = self._closed
self.__iter__ = self.__len__ = self.__contains__ = self._closed
def sync(self):
"""Sync the write buffer with the cache files and clear the buffer.
If the :class:`FileCache` object was opened with the optional ``'s'``
*flag* argument, then calling :meth:`sync` will do nothing.
"""
if self._sync:
return # opened in sync mode, so skip the manual sync
self._sync = True
for ekey in self._buffer:
filename = self._key_to_filename(ekey)
5 years ago
self._write_to_file(filename, self._buffer[ekey])
6 years ago
self._buffer.clear()
self._sync = False
def _closed(self, *args, **kwargs):
"""Filler method for closed cache methods."""
raise ValueError("invalid operation on closed cache")
def _encode_key(self, key):
5 years ago
"""Encode key using *hex_codec* for constructing a cache filename.
6 years ago
Keys are implicitly converted to :class:`bytes` if passed as
:class:`str`.
"""
if isinstance(key, str) or isinstance(key, unicode):
key = key.encode(self._keyencoding)
elif not isinstance(key, bytes):
raise TypeError("key must be bytes or str")
5 years ago
return codecs.encode(key, 'hex_codec').decode(self._keyencoding)
6 years ago
def _decode_key(self, key):
5 years ago
"""Decode key using hex_codec to retrieve the original key.
6 years ago
Keys are returned as :class:`str` if serialization is enabled.
Keys are returned as :class:`bytes` if serialization is disabled.
"""
5 years ago
bkey = codecs.decode(key.encode(self._keyencoding), 'hex_codec')
6 years ago
return bkey.decode(self._keyencoding) if self._serialize else bkey
def _dumps(self, value):
return value if not self._serialize else pickle.dumps(value)
def _loads(self, value):
return value if not self._serialize else pickle.loads(value)
def _key_to_filename(self, key):
"""Convert an encoded key to an absolute cache filename."""
5 years ago
return os.path.join(self.cache_dir, key)
6 years ago
def _filename_to_key(self, absfilename):
"""Convert an absolute cache filename to a key name."""
5 years ago
return os.path.split(absfilename)[1]
6 years ago
5 years ago
def _all_filenames(self):
6 years ago
"""Return a list of absolute cache filenames"""
try:
5 years ago
return [os.path.join(self.cache_dir, filename) for filename in
os.listdir(self.cache_dir)]
6 years ago
except (FileNotFoundError, OSError):
5 years ago
return []
6 years ago
def _all_keys(self):
"""Return a list of all encoded key names."""
file_keys = [self._filename_to_key(fn) for fn in self._all_filenames()]
if self._sync:
return set(file_keys)
else:
return set(file_keys + list(self._buffer))
5 years ago
def _write_to_file(self, filename, bytesvalue):
6 years ago
"""Write bytesvalue to filename."""
fh, tmp = tempfile.mkstemp()
with os.fdopen(fh, self._flag) as f:
5 years ago
f.write(self._dumps(bytesvalue))
6 years ago
rename(tmp, filename)
os.chmod(filename, self._mode)
def _read_from_file(self, filename):
"""Read data from filename."""
try:
with open(filename, 'rb') as f:
return self._loads(f.read())
except (IOError, OSError):
logger.warning('Error opening file: {}'.format(filename))
raise
6 years ago
def __setitem__(self, key, value):
ekey = self._encode_key(key)
if not self._sync:
self._buffer[ekey] = value
else:
filename = self._key_to_filename(ekey)
self._write_to_file(filename, value)
def __getitem__(self, key):
ekey = self._encode_key(key)
if not self._sync:
try:
return self._buffer[ekey]
except KeyError:
pass
filename = self._key_to_filename(ekey)
5 years ago
if filename not in self._all_filenames():
6 years ago
raise KeyError(key)
return self._read_from_file(filename)
def __delitem__(self, key):
ekey = self._encode_key(key)
filename = self._key_to_filename(ekey)
if not self._sync:
try:
del self._buffer[ekey]
except KeyError:
if filename not in self._all_filenames():
raise KeyError(key)
try:
os.remove(filename)
except (IOError, OSError):
pass
def __iter__(self):
for key in self._all_keys():
yield self._decode_key(key)
def __len__(self):
return len(self._all_keys())
def __contains__(self, key):
ekey = self._encode_key(key)
5 years ago
return ekey in self._all_keys()