X7ROOT File Manager
Current Path:
/usr/lib64/python2.7/site-packages/sepolgen
usr
/
lib64
/
python2.7
/
site-packages
/
sepolgen
/
📁
..
📄
__init__.py
(0 B)
📄
__init__.pyc
(142 B)
📄
__init__.pyo
(142 B)
📄
access.py
(12.33 KB)
📄
access.pyc
(13.18 KB)
📄
access.pyo
(13.18 KB)
📄
audit.py
(23.64 KB)
📄
audit.pyc
(22.33 KB)
📄
audit.pyo
(22.33 KB)
📄
classperms.py
(2.75 KB)
📄
classperms.pyc
(3.16 KB)
📄
classperms.pyo
(3.16 KB)
📄
defaults.py
(2.69 KB)
📄
defaults.pyc
(2.87 KB)
📄
defaults.pyo
(2.87 KB)
📄
interfaces.py
(16.21 KB)
📄
interfaces.pyc
(15.86 KB)
📄
interfaces.pyo
(15.86 KB)
📄
lex.py
(32.86 KB)
📄
lex.pyc
(19.24 KB)
📄
lex.pyo
(19.24 KB)
📄
matching.py
(8.45 KB)
📄
matching.pyc
(7.4 KB)
📄
matching.pyo
(7.4 KB)
📄
module.py
(7.12 KB)
📄
module.pyc
(8.4 KB)
📄
module.pyo
(8.4 KB)
📄
objectmodel.py
(6.37 KB)
📄
objectmodel.pyc
(4.64 KB)
📄
objectmodel.pyo
(4.64 KB)
📄
output.py
(5.01 KB)
📄
output.pyc
(4.51 KB)
📄
output.pyo
(4.51 KB)
📄
policygen.py
(16.31 KB)
📄
policygen.pyc
(15.25 KB)
📄
policygen.pyo
(15.18 KB)
📄
refparser.py
(30.2 KB)
📄
refparser.pyc
(36.27 KB)
📄
refparser.pyo
(36.27 KB)
📄
refpolicy.py
(27.31 KB)
📄
refpolicy.pyc
(46.45 KB)
📄
refpolicy.pyo
(46.45 KB)
📄
sepolgeni18n.py
(912 B)
📄
sepolgeni18n.pyc
(410 B)
📄
sepolgeni18n.pyo
(410 B)
📄
util.py
(5.41 KB)
📄
util.pyc
(7.87 KB)
📄
util.pyo
(7.87 KB)
📄
yacc.py
(80.04 KB)
📄
yacc.pyc
(41.4 KB)
📄
yacc.pyo
(41.4 KB)
Editing: util.py
# Authors: Karl MacMillan <kmacmillan@mentalrootkit.com> # # Copyright (C) 2006 Red Hat # see file 'COPYING' for use and warranty information # # This program is free software; you can redistribute it and/or # modify it under the terms of the GNU General Public License as # published by the Free Software Foundation; version 2 only # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this program; if not, write to the Free Software # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # import locale import sys PY3 = sys.version_info[0] == 3 if PY3: bytes_type=bytes string_type=str else: bytes_type=str string_type=unicode class ConsoleProgressBar: def __init__(self, out, steps=100, indicator='#'): self.blocks = 0 self.current = 0 self.steps = steps self.indicator = indicator self.out = out self.done = False def start(self, message=None): self.done = False if message: self.out.write('\n%s:\n' % message) self.out.write('%--10---20---30---40---50---60---70---80---90--100\n') def step(self, n=1): self.current += n old = self.blocks self.blocks = int(round(self.current / float(self.steps) * 100) / 2) if self.blocks > 50: self.blocks = 50 new = self.blocks - old self.out.write(self.indicator * new) self.out.flush() if self.blocks == 50 and not self.done: self.done = True self.out.write("\n") def set_to_list(s): l = [] l.extend(s) return l def first(s, sorted=False): """ Return the first element of a set. It sometimes useful to return the first element from a set but, because sets are not indexable, this is rather hard. This function will return the first element from a set. If sorted is True, then the set will first be sorted (making this an expensive operation). Otherwise a random element will be returned (as sets are not ordered). """ if not len(s): raise IndexError("empty containter") if sorted: l = set_to_list(s) l.sort() return l[0] else: for x in s: return x def encode_input(text): import locale """Encode given text via preferred system encoding""" # locale will often find out the correct encoding encoding = locale.getpreferredencoding() try: encoded_text = text.encode(encoding) except UnicodeError: # if it fails to find correct encoding then ascii is used # which may lead to UnicodeError if `text` contains non ascii signs # utf-8 is our guess to fix the situation encoded_text = text.encode('utf-8') return encoded_text def decode_input(text): import locale """Decode given text via preferred system encoding""" # locale will often find out the correct encoding encoding = locale.getpreferredencoding() try: decoded_text = text.decode(encoding) except UnicodeError: # if it fails to find correct encoding then ascii is used # which may lead to UnicodeError if `text` contains non ascii signs # utf-8 is our guess to fix the situation decoded_text = text.decode('utf-8') return decoded_text class Comparison(): """Class used when implementing rich comparison. Inherit from this class if you want to have a rich comparison withing the class, afterwards implement _compare function within your class.""" def _compare(self, other, method): raise NotImplemented def __eq__(self, other): return self._compare(other, lambda a, b: a == b) def __lt__(self, other): return self._compare(other, lambda a, b: a < b) def __le__(self, other): return self._compare(other, lambda a, b: a <= b) def __ge__(self, other): return self._compare(other, lambda a, b: a >= b) def __gt__(self, other): return self._compare(other, lambda a, b: a > b) def __ne__(self, other): return self._compare(other, lambda a, b: a != b) if sys.version_info < (2,7): # cmp_to_key function is missing in python2.6 def cmp_to_key(mycmp): 'Convert a cmp= function into a key= function' class K: def __init__(self, obj, *args): self.obj = obj def __lt__(self, other): return mycmp(self.obj, other.obj) < 0 def __gt__(self, other): return mycmp(self.obj, other.obj) > 0 def __eq__(self, other): return mycmp(self.obj, other.obj) == 0 def __le__(self, other): return mycmp(self.obj, other.obj) <= 0 def __ge__(self, other): return mycmp(self.obj, other.obj) >= 0 def __ne__(self, other): return mycmp(self.obj, other.obj) != 0 return K else: from functools import cmp_to_key def cmp(first, second): return (first > second) - (second > first) if __name__ == "__main__": import sys import time p = ConsoleProgressBar(sys.stdout, steps=999) p.start("computing pi") for i in range(999): p.step() time.sleep(0.001)
Upload File
Create Folder