Update libs
This commit is contained in:
1 parent
2d05aafc04
commit
88b71c8116
157 files changed
+15686
-2017
No files matched your search
@@ -0,0 +1,155 @@
|
||||
# -*- coding: utf-8 -*-
|
||||
#
|
||||
# Copyright 2011 Sybren A. Stüvel <sybren@stuvel.eu>
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
|
||||
'''VARBLOCK file support
|
||||
|
||||
The VARBLOCK file format is as follows, where || denotes byte concatenation:
|
||||
|
||||
FILE := VERSION || BLOCK || BLOCK ...
|
||||
|
||||
BLOCK := LENGTH || DATA
|
||||
|
||||
LENGTH := varint-encoded length of the subsequent data. Varint comes from
|
||||
Google Protobuf, and encodes an integer into a variable number of bytes.
|
||||
Each byte uses the 7 lowest bits to encode the value. The highest bit set
|
||||
to 1 indicates the next byte is also part of the varint. The last byte will
|
||||
have this bit set to 0.
|
||||
|
||||
This file format is called the VARBLOCK format, in line with the varint format
|
||||
used to denote the block sizes.
|
||||
|
||||
'''
|
||||
|
||||
from rsa._compat import byte, b
|
||||
|
||||
|
||||
ZERO_BYTE = b('\x00')
|
||||
VARBLOCK_VERSION = 1
|
||||
|
||||
def read_varint(infile):
|
||||
'''Reads a varint from the file.
|
||||
|
||||
When the first byte to be read indicates EOF, (0, 0) is returned. When an
|
||||
EOF occurs when at least one byte has been read, an EOFError exception is
|
||||
raised.
|
||||
|
||||
@param infile: the file-like object to read from. It should have a read()
|
||||
method.
|
||||
@returns (varint, length), the read varint and the number of read bytes.
|
||||
'''
|
||||
|
||||
varint = 0
|
||||
read_bytes = 0
|
||||
|
||||
while True:
|
||||
char = infile.read(1)
|
||||
if len(char) == 0:
|
||||
if read_bytes == 0:
|
||||
return (0, 0)
|
||||
raise EOFError('EOF while reading varint, value is %i so far' %
|
||||
varint)
|
||||
|
||||
byte = ord(char)
|
||||
varint += (byte & 0x7F) << (7 * read_bytes)
|
||||
|
||||
read_bytes += 1
|
||||
|
||||
if not byte & 0x80:
|
||||
return (varint, read_bytes)
|
||||
|
||||
|
||||
def write_varint(outfile, value):
|
||||
'''Writes a varint to a file.
|
||||
|
||||
@param outfile: the file-like object to write to. It should have a write()
|
||||
method.
|
||||
@returns the number of written bytes.
|
||||
'''
|
||||
|
||||
# there is a big difference between 'write the value 0' (this case) and
|
||||
# 'there is nothing left to write' (the false-case of the while loop)
|
||||
|
||||
if value == 0:
|
||||
outfile.write(ZERO_BYTE)
|
||||
return 1
|
||||
|
||||
written_bytes = 0
|
||||
while value > 0:
|
||||
to_write = value & 0x7f
|
||||
value = value >> 7
|
||||
|
||||
if value > 0:
|
||||
to_write |= 0x80
|
||||
|
||||
outfile.write(byte(to_write))
|
||||
written_bytes += 1
|
||||
|
||||
return written_bytes
|
||||
|
||||
|
||||
def yield_varblocks(infile):
|
||||
'''Generator, yields each block in the input file.
|
||||
|
||||
@param infile: file to read, is expected to have the VARBLOCK format as
|
||||
described in the module's docstring.
|
||||
@yields the contents of each block.
|
||||
'''
|
||||
|
||||
# Check the version number
|
||||
first_char = infile.read(1)
|
||||
if len(first_char) == 0:
|
||||
raise EOFError('Unable to read VARBLOCK version number')
|
||||
|
||||
version = ord(first_char)
|
||||
if version != VARBLOCK_VERSION:
|
||||
raise ValueError('VARBLOCK version %i not supported' % version)
|
||||
|
||||
while True:
|
||||
(block_size, read_bytes) = read_varint(infile)
|
||||
|
||||
# EOF at block boundary, that's fine.
|
||||
if read_bytes == 0 and block_size == 0:
|
||||
break
|
||||
|
||||
block = infile.read(block_size)
|
||||
|
||||
read_size = len(block)
|
||||
if read_size != block_size:
|
||||
raise EOFError('Block size is %i, but could read only %i bytes' %
|
||||
(block_size, read_size))
|
||||
|
||||
yield block
|
||||
|
||||
|
||||
def yield_fixedblocks(infile, blocksize):
|
||||
'''Generator, yields each block of ``blocksize`` bytes in the input file.
|
||||
|
||||
:param infile: file to read and separate in blocks.
|
||||
:returns: a generator that yields the contents of each block
|
||||
'''
|
||||
|
||||
while True:
|
||||
block = infile.read(blocksize)
|
||||
|
||||
read_bytes = len(block)
|
||||
if read_bytes == 0:
|
||||
break
|
||||
|
||||
yield block
|
||||
|
||||
if read_bytes < blocksize:
|
||||
break
|
||||
|
||||
Reference in new issue
Block a user