blob: 5cddfebca0ae813349b9c1bcf038d2541b383b72 [file] [log] [blame]
"""Python 2/3 compat layer."""
from __future__ import print_function, division, absolute_import
import sys
__all__ = ['basestring', 'unicode', 'unichr', 'byteord', 'bytechr', 'BytesIO',
'StringIO', 'UnicodeIO', 'strjoin', 'bytesjoin', 'tobytes', 'tostr',
'tounicode', 'Tag', 'open', 'range', 'xrange', 'round', 'Py23Error',
'SimpleNamespace', 'zip', 'RecursionError']
class Py23Error(NotImplementedError):
pass
PY3 = sys.version_info[0] == 3
PY2 = sys.version_info[0] == 2
try:
basestring = basestring
except NameError:
basestring = str
try:
unicode = unicode
except NameError:
unicode = str
try:
unichr = unichr
if sys.maxunicode < 0x10FFFF:
# workarounds for Python 2 "narrow" builds with UCS2-only support.
_narrow_unichr = unichr
def unichr(i):
"""
Return the unicode character whose Unicode code is the integer 'i'.
The valid range is 0 to 0x10FFFF inclusive.
>>> _narrow_unichr(0xFFFF + 1)
Traceback (most recent call last):
File "<stdin>", line 1, in ?
ValueError: unichr() arg not in range(0x10000) (narrow Python build)
>>> unichr(0xFFFF + 1) == u'\U00010000'
True
>>> unichr(1114111) == u'\U0010FFFF'
True
>>> unichr(0x10FFFF + 1)
Traceback (most recent call last):
File "<stdin>", line 1, in ?
ValueError: unichr() arg not in range(0x110000)
"""
try:
return _narrow_unichr(i)
except ValueError:
try:
padded_hex_str = hex(i)[2:].zfill(8)
escape_str = "\\U" + padded_hex_str
return escape_str.decode("unicode-escape")
except UnicodeDecodeError:
raise ValueError('unichr() arg not in range(0x110000)')
import re
_unicode_escape_RE = re.compile(r'\\U[A-Fa-f0-9]{8}')
def byteord(c):
"""
Given a 8-bit or unicode character, return an integer representing the
Unicode code point of the character. If a unicode argument is given, the
character's code point must be in the range 0 to 0x10FFFF inclusive.
>>> ord(u'\U00010000')
Traceback (most recent call last):
File "<stdin>", line 1, in ?
TypeError: ord() expected a character, but string of length 2 found
>>> byteord(u'\U00010000') == 0xFFFF + 1
True
>>> byteord(u'\U0010FFFF') == 1114111
True
"""
try:
return ord(c)
except TypeError as e:
try:
escape_str = c.encode('unicode-escape')
if not _unicode_escape_RE.match(escape_str):
raise
hex_str = escape_str[3:]
return int(hex_str, 16)
except:
raise TypeError(e)
else:
byteord = ord
bytechr = chr
except NameError:
unichr = chr
def bytechr(n):
return bytes([n])
def byteord(c):
return c if isinstance(c, int) else ord(c)
# the 'io' module provides the same I/O interface on both 2 and 3.
# here we define an alias of io.StringIO to disambiguate it eternally...
from io import BytesIO
from io import StringIO as UnicodeIO
try:
# in python 2, by 'StringIO' we still mean a stream of *byte* strings
from StringIO import StringIO
except ImportError:
# in Python 3, we mean instead a stream of *unicode* strings
StringIO = UnicodeIO
def strjoin(iterable, joiner=''):
return tostr(joiner).join(iterable)
def tobytes(s, encoding='ascii', errors='strict'):
if not isinstance(s, bytes):
return s.encode(encoding, errors)
else:
return s
def tounicode(s, encoding='ascii', errors='strict'):
if not isinstance(s, unicode):
return s.decode(encoding, errors)
else:
return s
if str == bytes:
class Tag(str):
def tobytes(self):
if isinstance(self, bytes):
return self
else:
return self.encode('latin1')
tostr = tobytes
bytesjoin = strjoin
else:
class Tag(str):
@staticmethod
def transcode(blob):
if isinstance(blob, bytes):
blob = blob.decode('latin-1')
return blob
def __new__(self, content):
return str.__new__(self, self.transcode(content))
def __ne__(self, other):
return not self.__eq__(other)
def __eq__(self, other):
return str.__eq__(self, self.transcode(other))
def __hash__(self):
return str.__hash__(self)
def tobytes(self):
return self.encode('latin-1')
tostr = tounicode
def bytesjoin(iterable, joiner=b''):
return tobytes(joiner).join(tobytes(item) for item in iterable)
import os
import io as _io
try:
from msvcrt import setmode as _setmode
except ImportError:
_setmode = None # only available on the Windows platform
def open(file, mode='r', buffering=-1, encoding=None, errors=None,
newline=None, closefd=True, opener=None):
""" Wrapper around `io.open` that bridges the differences between Python 2
and Python 3's built-in `open` functions. In Python 2, `io.open` is a
backport of Python 3's `open`, whereas in Python 3, it is an alias of the
built-in `open` function.
One difference is that the 'opener' keyword argument is only supported in
Python 3. Here we pass the value of 'opener' only when it is not None.
This causes Python 2 to raise TypeError, complaining about the number of
expected arguments, so it must be avoided in py2 or py2-3 contexts.
Another difference between 2 and 3, this time on Windows, has to do with
opening files by name or by file descriptor.
On the Windows C runtime, the 'O_BINARY' flag is defined which disables
the newlines translation ('\r\n' <=> '\n') when reading/writing files.
On both Python 2 and 3 this flag is always set when opening files by name.
This way, the newlines translation at the MSVCRT level doesn't interfere
with the Python io module's own newlines translation.
However, when opening files via fd, on Python 2 the fd is simply copied,
regardless of whether it has the 'O_BINARY' flag set or not.
This becomes a problem in the case of stdout, stdin, and stderr, because on
Windows these are opened in text mode by default (ie. don't have the
O_BINARY flag set).
On Python 3, this issue has been fixed, and all fds are now opened in
binary mode on Windows, including standard streams. Similarly here, I use
the `_setmode` function to ensure that integer file descriptors are
O_BINARY'ed before I pass them on to io.open.
For more info, see: https://bugs.python.org/issue10841
"""
if isinstance(file, int):
# the 'file' argument is an integer file descriptor
fd = file
if fd < 0:
raise ValueError('negative file descriptor')
if _setmode:
# `_setmode` function sets the line-end translation and returns the
# value of the previous mode. AFAIK there's no `_getmode`, so to
# check if the previous mode already had the bit set, I fist need
# to duplicate the file descriptor, set the binary flag on the copy
# and check the returned value.
fdcopy = os.dup(fd)
current_mode = _setmode(fdcopy, os.O_BINARY)
if not (current_mode & os.O_BINARY):
# the binary mode was not set: use the file descriptor's copy
file = fdcopy
if closefd:
# close the original file descriptor
os.close(fd)
else:
# ensure the copy is closed when the file object is closed
closefd = True
else:
# original file descriptor already had binary flag, close copy
os.close(fdcopy)
if opener is not None:
# "opener" is not supported on Python 2, use it at your own risk!
return _io.open(
file, mode, buffering, encoding, errors, newline, closefd,
opener=opener)
else:
return _io.open(
file, mode, buffering, encoding, errors, newline, closefd)
# always use iterator for 'range' and 'zip' on both py 2 and 3
try:
range = xrange
except NameError:
range = range
def xrange(*args, **kwargs):
raise Py23Error("'xrange' is not defined. Use 'range' instead.")
try:
from itertools import izip as zip
except ImportError:
zip = zip
import math as _math
try:
isclose = _math.isclose
except AttributeError:
# math.isclose() was only added in Python 3.5
_isinf = _math.isinf
_fabs = _math.fabs
def isclose(a, b, rel_tol=1e-09, abs_tol=0):
"""
Python 2 implementation of Python 3.5 math.isclose()
https://hg.python.org/cpython/file/v3.5.2/Modules/mathmodule.c#l1993
"""
# sanity check on the inputs
if rel_tol < 0 or abs_tol < 0:
raise ValueError("tolerances must be non-negative")
# short circuit exact equality -- needed to catch two infinities of
# the same sign. And perhaps speeds things up a bit sometimes.
if a == b:
return True
# This catches the case of two infinities of opposite sign, or
# one infinity and one finite number. Two infinities of opposite
# sign would otherwise have an infinite relative tolerance.
# Two infinities of the same sign are caught by the equality check
# above.
if _isinf(a) or _isinf(b):
return False
# Cast to float to allow decimal.Decimal arguments
if not isinstance(a, float):
a = float(a)
if not isinstance(b, float):
b = float(b)
# now do the regular computation
# this is essentially the "weak" test from the Boost library
diff = _fabs(b - a)
result = ((diff <= _fabs(rel_tol * a)) or
(diff <= _fabs(rel_tol * b)) or
(diff <= abs_tol))
return result
try:
_isfinite = _math.isfinite # Python >= 3.2
except AttributeError:
_isfinite = None
_isnan = _math.isnan
_isinf = _math.isinf
def isfinite(f):
"""
>>> isfinite(0.0)
True
>>> isfinite(-0.1)
True
>>> isfinite(1e10)
True
>>> isfinite(float("nan"))
False
>>> isfinite(float("+inf"))
False
>>> isfinite(float("-inf"))
False
"""
if _isfinite is not None:
return _isfinite(f)
else:
return not (_isnan(f) or _isinf(f))
import decimal as _decimal
if PY3:
def round2(number, ndigits=None):
"""
Implementation of Python 2 built-in round() function.
Rounds a number to a given precision in decimal digits (default
0 digits). The result is a floating point number. Values are rounded
to the closest multiple of 10 to the power minus ndigits; if two
multiples are equally close, rounding is done away from 0.
ndigits may be negative.
See Python 2 documentation:
https://docs.python.org/2/library/functions.html?highlight=round#round
"""
if ndigits is None:
ndigits = 0
if ndigits < 0:
exponent = 10 ** (-ndigits)
quotient, remainder = divmod(number, exponent)
if remainder >= exponent//2 and number >= 0:
quotient += 1
return float(quotient * exponent)
else:
exponent = _decimal.Decimal('10') ** (-ndigits)
d = _decimal.Decimal.from_float(number).quantize(
exponent, rounding=_decimal.ROUND_HALF_UP)
return float(d)
if sys.version_info[:2] >= (3, 6):
# in Python 3.6, 'round3' is an alias to the built-in 'round'
round = round3 = round
else:
# in Python3 < 3.6 we need work around the inconsistent behavior of
# built-in round(), whereby floats accept a second None argument,
# while integers raise TypeError. See https://bugs.python.org/issue27936
_round = round
def round3(number, ndigits=None):
return _round(number) if ndigits is None else _round(number, ndigits)
round = round3
else:
# in Python 2, 'round2' is an alias to the built-in 'round' and
# 'round' is shadowed by 'round3'
round2 = round
def round3(number, ndigits=None):
"""
Implementation of Python 3 built-in round() function.
Rounds a number to a given precision in decimal digits (default
0 digits). This returns an int when ndigits is omitted or is None,
otherwise the same type as the number.
Values are rounded to the closest multiple of 10 to the power minus
ndigits; if two multiples are equally close, rounding is done toward
the even choice (aka "Banker's Rounding"). For example, both round(0.5)
and round(-0.5) are 0, and round(1.5) is 2.
ndigits may be negative.
See Python 3 documentation:
https://docs.python.org/3/library/functions.html?highlight=round#round
Derived from python-future:
https://github.com/PythonCharmers/python-future/blob/master/src/future/builtins/newround.py
"""
if ndigits is None:
ndigits = 0
# return an int when called with one argument
totype = int
# shortcut if already an integer, or a float with no decimal digits
inumber = totype(number)
if inumber == number:
return inumber
else:
# return the same type as the number, when called with two arguments
totype = type(number)
m = number * (10 ** ndigits)
# if number is half-way between two multiples, and the mutliple that is
# closer to zero is even, we use the (slow) pure-Python implementation
if isclose(m % 1, .5) and int(m) % 2 == 0:
if ndigits < 0:
exponent = 10 ** (-ndigits)
quotient, remainder = divmod(number, exponent)
half = exponent//2
if remainder > half or (remainder == half and quotient % 2 != 0):
quotient += 1
d = quotient * exponent
else:
exponent = _decimal.Decimal('10') ** (-ndigits) if ndigits != 0 else 1
d = _decimal.Decimal.from_float(number).quantize(
exponent, rounding=_decimal.ROUND_HALF_EVEN)
else:
# else we use the built-in round() as it produces the same results
d = round2(number, ndigits)
return totype(d)
round = round3
try:
from types import SimpleNamespace
except ImportError:
class SimpleNamespace(object):
"""
A backport of Python 3.3's ``types.SimpleNamespace``.
"""
def __init__(self, **kwargs):
self.__dict__.update(kwargs)
def __repr__(self):
keys = sorted(self.__dict__)
items = ("{0}={1!r}".format(k, self.__dict__[k]) for k in keys)
return "{0}({1})".format(type(self).__name__, ", ".join(items))
def __eq__(self, other):
return self.__dict__ == other.__dict__
if sys.version_info[:2] > (3, 4):
from contextlib import redirect_stdout, redirect_stderr
else:
# `redirect_stdout` was added with python3.4, while `redirect_stderr`
# with python3.5. For simplicity, I redefine both for any versions
# less than or equal to 3.4.
# The code below is copied from:
# https://github.com/python/cpython/blob/57161aa/Lib/contextlib.py
class _RedirectStream(object):
_stream = None
def __init__(self, new_target):
self._new_target = new_target
# We use a list of old targets to make this CM re-entrant
self._old_targets = []
def __enter__(self):
self._old_targets.append(getattr(sys, self._stream))
setattr(sys, self._stream, self._new_target)
return self._new_target
def __exit__(self, exctype, excinst, exctb):
setattr(sys, self._stream, self._old_targets.pop())
class redirect_stdout(_RedirectStream):
"""Context manager for temporarily redirecting stdout to another file.
# How to send help() to stderr
with redirect_stdout(sys.stderr):
help(dir)
# How to write help() to a file
with open('help.txt', 'w') as f:
with redirect_stdout(f):
help(pow)
"""
_stream = "stdout"
class redirect_stderr(_RedirectStream):
"""Context manager for temporarily redirecting stderr to another file."""
_stream = "stderr"
try:
RecursionError = RecursionError
except NameError:
RecursionError = RuntimeError
if __name__ == "__main__":
import doctest, sys
sys.exit(doctest.testmod().failed)