Staging
v0.8.1
v0.8.1
https://github.com/python/cpython
Revision cbf3b5cb76906fba15dbf59a1e83c540a447b907 authored by Christian Heimes on 03 December 2007, 21:02:03 UTC, committed by Christian Heimes on 03 December 2007, 21:02:03 UTC
svn+ssh://pythondev@svn.python.org/python/trunk !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! NOTE: The merge does NOT contain the modified file Python/import.c from r59288. I can't get it running. Nick, please check in the PEP 366 manually. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! ........ r59279 | georg.brandl | 2007-12-02 19:17:50 +0100 (Sun, 02 Dec 2007) | 2 lines Fix a sentence I missed before. Do not merge to 3k. ........ r59281 | georg.brandl | 2007-12-02 22:58:54 +0100 (Sun, 02 Dec 2007) | 3 lines Add documentation for PySys_* functions. Written by Charlie Shepherd for GHOP. Also fixes #1245. ........ r59288 | nick.coghlan | 2007-12-03 13:55:17 +0100 (Mon, 03 Dec 2007) | 1 line Implement PEP 366 ........ r59290 | christian.heimes | 2007-12-03 14:47:29 +0100 (Mon, 03 Dec 2007) | 3 lines Applied my patch #1455 with some extra fixes for VS 2005 The new msvc9compiler module supports VS 2005 and VS 2008. I've also fixed build_ext to support PCbuild8 and PCbuild9 and backported my fix for xxmodule.c from py3k. The old code msvccompiler is still in place in case somebody likes to build an extension with VS 2003 or earlier. I've also updated the cygwin compiler module for VS 2005 and VS 2008. It works with VS 2005 but I'm unable to test it with VS 2008. We have to wait for a new version of cygwin. ........ r59291 | christian.heimes | 2007-12-03 14:55:16 +0100 (Mon, 03 Dec 2007) | 1 line Added comment to Misc/NEWS for r59290 ........ r59292 | christian.heimes | 2007-12-03 15:28:04 +0100 (Mon, 03 Dec 2007) | 1 line I followed MA Lemberg's suggestion and added comments to the late initialization of the type slots. ........ r59293 | facundo.batista | 2007-12-03 17:29:52 +0100 (Mon, 03 Dec 2007) | 3 lines Speedup and cleaning of __str__. Thanks Mark Dickinson. ........ r59294 | facundo.batista | 2007-12-03 18:55:00 +0100 (Mon, 03 Dec 2007) | 4 lines Faster _fix function, and some reordering for a more elegant coding. Thanks Mark Dickinson. ........ r59295 | martin.v.loewis | 2007-12-03 20:20:02 +0100 (Mon, 03 Dec 2007) | 5 lines Issue #1727780: Support loading pickles of random.Random objects created on 32-bit systems on 64-bit systems, and vice versa. As a consequence of the change, Random pickles created by Python 2.6 cannot be loaded in Python 2.5. ........ r59297 | facundo.batista | 2007-12-03 20:49:54 +0100 (Mon, 03 Dec 2007) | 3 lines Two small fixes. Issue 1547. ........ r59299 | georg.brandl | 2007-12-03 20:57:02 +0100 (Mon, 03 Dec 2007) | 2 lines #1548: fix apostroph placement. ........ r59300 | christian.heimes | 2007-12-03 21:01:02 +0100 (Mon, 03 Dec 2007) | 3 lines Patch #1537 from Chad Austin Change GeneratorExit's base class from Exception to BaseException (This time I'm applying the patch to the correct sandbox.) ........ r59302 | georg.brandl | 2007-12-03 21:03:46 +0100 (Mon, 03 Dec 2007) | 3 lines Add examples to the xmlrpclib docs. Written for GHOP by Josip Dzolonga. ........
1 parent f929077
Tip revision: cbf3b5cb76906fba15dbf59a1e83c540a447b907 authored by Christian Heimes on 03 December 2007, 21:02:03 UTC
Merged revisions 59275-59303 via svnmerge from
Merged revisions 59275-59303 via svnmerge from
Tip revision: cbf3b5c
dumbdbm.py
"""A dumb and slow but simple dbm clone.
For database spam, spam.dir contains the index (a text file),
spam.bak *may* contain a backup of the index (also a text file),
while spam.dat contains the data (a binary file).
XXX TO DO:
- seems to contain a bug when updating...
- reclaim free space (currently, space once occupied by deleted or expanded
items is never reused)
- support concurrent access (currently, if two processes take turns making
updates, they can mess up the index)
- support efficient access to large databases (currently, the whole index
is read when the database is opened, and some updates rewrite the whole index)
- support opening for read-only (flag = 'm')
"""
import io as _io
import os as _os
import UserDict
_BLOCKSIZE = 512
error = IOError # For anydbm
class _Database(UserDict.DictMixin):
# The on-disk directory and data files can remain in mutually
# inconsistent states for an arbitrarily long time (see comments
# at the end of __setitem__). This is only repaired when _commit()
# gets called. One place _commit() gets called is from __del__(),
# and if that occurs at program shutdown time, module globals may
# already have gotten rebound to None. Since it's crucial that
# _commit() finish successfully, we can't ignore shutdown races
# here, and _commit() must not reference any globals.
_os = _os # for _commit()
_io = _io # for _commit()
def __init__(self, filebasename, mode):
self._mode = mode
# The directory file is a text file. Each line looks like
# "%r, (%d, %d)\n" % (key, pos, siz)
# where key is the string key, pos is the offset into the dat
# file of the associated value's first byte, and siz is the number
# of bytes in the associated value.
self._dirfile = filebasename + '.dir'
# The data file is a binary file pointed into by the directory
# file, and holds the values associated with keys. Each value
# begins at a _BLOCKSIZE-aligned byte offset, and is a raw
# binary 8-bit string value.
self._datfile = filebasename + '.dat'
self._bakfile = filebasename + '.bak'
# The index is an in-memory dict, mirroring the directory file.
self._index = None # maps keys to (pos, siz) pairs
# Mod by Jack: create data file if needed
try:
f = _io.open(self._datfile, 'r')
except IOError:
f = _io.open(self._datfile, 'w')
self._chmod(self._datfile)
f.close()
self._update()
# Read directory file into the in-memory index dict.
def _update(self):
self._index = {}
try:
f = _io.open(self._dirfile, 'r')
except IOError:
pass
else:
for line in f:
line = line.rstrip()
key, pos_and_siz_pair = eval(line)
self._index[key] = pos_and_siz_pair
f.close()
# Write the index dict to the directory file. The original directory
# file (if any) is renamed with a .bak extension first. If a .bak
# file currently exists, it's deleted.
def _commit(self):
# CAUTION: It's vital that _commit() succeed, and _commit() can
# be called from __del__(). Therefore we must never reference a
# global in this routine.
if self._index is None:
return # nothing to do
try:
self._os.unlink(self._bakfile)
except self._os.error:
pass
try:
self._os.rename(self._dirfile, self._bakfile)
except self._os.error:
pass
f = self._io.open(self._dirfile, 'w')
self._chmod(self._dirfile)
for key, pos_and_siz_pair in self._index.items():
f.write("%r, %r\n" % (key, pos_and_siz_pair))
f.close()
sync = _commit
def __getitem__(self, key):
key = key.decode("latin-1")
pos, siz = self._index[key] # may raise KeyError
f = _io.open(self._datfile, 'rb')
f.seek(pos)
dat = f.read(siz)
f.close()
return dat
# Append val to the data file, starting at a _BLOCKSIZE-aligned
# offset. The data file is first padded with NUL bytes (if needed)
# to get to an aligned offset. Return pair
# (starting offset of val, len(val))
def _addval(self, val):
f = _io.open(self._datfile, 'rb+')
f.seek(0, 2)
pos = int(f.tell())
npos = ((pos + _BLOCKSIZE - 1) // _BLOCKSIZE) * _BLOCKSIZE
f.write(b'\0'*(npos-pos))
pos = npos
f.write(val)
f.close()
return (pos, len(val))
# Write val to the data file, starting at offset pos. The caller
# is responsible for ensuring that there's enough room starting at
# pos to hold val, without overwriting some other value. Return
# pair (pos, len(val)).
def _setval(self, pos, val):
f = _io.open(self._datfile, 'rb+')
f.seek(pos)
f.write(val)
f.close()
return (pos, len(val))
# key is a new key whose associated value starts in the data file
# at offset pos and with length siz. Add an index record to
# the in-memory index dict, and append one to the directory file.
def _addkey(self, key, pos_and_siz_pair):
self._index[key] = pos_and_siz_pair
f = _io.open(self._dirfile, 'a')
self._chmod(self._dirfile)
f.write("%r, %r\n" % (key, pos_and_siz_pair))
f.close()
def __setitem__(self, key, val):
if not isinstance(key, bytes):
raise TypeError("keys must be bytes")
key = key.decode("latin-1") # hashable bytes
if not isinstance(val, (bytes, bytearray)):
raise TypeError("values must be byte strings")
if key not in self._index:
self._addkey(key, self._addval(val))
else:
# See whether the new value is small enough to fit in the
# (padded) space currently occupied by the old value.
pos, siz = self._index[key]
oldblocks = (siz + _BLOCKSIZE - 1) // _BLOCKSIZE
newblocks = (len(val) + _BLOCKSIZE - 1) // _BLOCKSIZE
if newblocks <= oldblocks:
self._index[key] = self._setval(pos, val)
else:
# The new value doesn't fit in the (padded) space used
# by the old value. The blocks used by the old value are
# forever lost.
self._index[key] = self._addval(val)
# Note that _index may be out of synch with the directory
# file now: _setval() and _addval() don't update the directory
# file. This also means that the on-disk directory and data
# files are in a mutually inconsistent state, and they'll
# remain that way until _commit() is called. Note that this
# is a disaster (for the database) if the program crashes
# (so that _commit() never gets called).
def __delitem__(self, key):
key = key.decode("latin-1")
# The blocks used by the associated value are lost.
del self._index[key]
# XXX It's unclear why we do a _commit() here (the code always
# XXX has, so I'm not changing it). _setitem__ doesn't try to
# XXX keep the directory file in synch. Why should we? Or
# XXX why shouldn't __setitem__?
self._commit()
def keys(self):
return [key.encode("latin-1") for key in self._index.keys()]
def items(self):
return [(key.encode("latin-1"), self[key.encode("latin-1")])
for key in self._index.keys()]
def __contains__(self, key):
key = key.decode("latin-1")
return key in self._index
def iterkeys(self):
return iter(self._index.keys())
__iter__ = iterkeys
def __len__(self):
return len(self._index)
def close(self):
self._commit()
self._index = self._datfile = self._dirfile = self._bakfile = None
__del__ = close
def _chmod (self, file):
if hasattr(self._os, 'chmod'):
self._os.chmod(file, self._mode)
def open(file, flag=None, mode=0o666):
"""Open the database file, filename, and return corresponding object.
The flag argument, used to control how the database is opened in the
other DBM implementations, is ignored in the dumbdbm module; the
database is always opened for update, and will be created if it does
not exist.
The optional mode argument is the UNIX mode of the file, used only when
the database has to be created. It defaults to octal code 0o666 (and
will be modified by the prevailing umask).
"""
# flag argument is currently ignored
# Modify mode depending on the umask
try:
um = _os.umask(0)
_os.umask(um)
except AttributeError:
pass
else:
# Turn off any bits that are set in the umask
mode = mode & (~um)
return _Database(file, mode)
Computing file changes ...