fake_fastnumbers now supports long unicode decimal strings.

The behavior of fake_fastnumbers was not matching that of
fastnumbers, which was to allow conversion of strings such as
"۱۲" to 12. This has been corrected.

1 files changed, 8 insertions, 4 deletions

diff --git a/natsort/compat/fake_fastnumbers.py b/natsort/compat/fake_fastnumbers.py
index 0af60b5..6eee532 100644
--- a/natsort/compat/fake_fastnumbers.py
+++ b/natsort/compat/fake_fastnumbers.py
@@ -13,6 +13,7 @@ from __future__ import (
 
 # Std. lib imports.
 import unicodedata
+from natsort.unicode_numbers import decimal_chars
 from natsort.compat.py23 import PY_VERSION
 if PY_VERSION >= 3:
     long = int
@@ -22,16 +23,18 @@ NAN_INF = ['INF', 'INf', 'Inf', 'inF', 'iNF', 'InF', 'inf', 'iNf',
            'NAN', 'nan', 'NaN', 'nAn', 'naN', 'NAn', 'nAN', 'Nan']
 NAN_INF.extend(['+'+x[:2] for x in NAN_INF] + ['-'+x[:2] for x in NAN_INF])
 NAN_INF = frozenset(NAN_INF)
+ASCII_NUMS = '0123456789+-'
 
 
 def fast_float(x, key=lambda x: x, nan=None,
-               uni=unicodedata.numeric, nan_inf=NAN_INF):
+               uni=unicodedata.numeric, nan_inf=NAN_INF,
+               _first_char=frozenset(decimal_chars + list(ASCII_NUMS + '.'))):
     """\
     Convert a string to a float quickly, return input as-is if not possible.
     We don't need to accept all input that the real fast_int accepts because
     the input will be controlled by the splitting algorithm.
     """
-    if x[0] in '0123456789+-.' or x.lstrip()[:3] in nan_inf:
+    if x[0] in _first_char or x.lstrip()[:3] in nan_inf:
         try:
             x = float(x)
             return nan if nan is not None and x != x else x
@@ -47,13 +50,14 @@ def fast_float(x, key=lambda x: x, nan=None,
             return key(x)
 
 
-def fast_int(x, key=lambda x: x, nan=None, uni=unicodedata.digit):
+def fast_int(x, key=lambda x: x, nan=None, uni=unicodedata.digit,
+             _first_char=frozenset(decimal_chars + list(ASCII_NUMS))):
     """\
     Convert a string to a int quickly, return input as-is if not possible.
     We don't need to accept all input that the real fast_int accepts because
     the input will be controlled by the splitting algorithm.
     """
-    if x[0] in '0123456789+-':
+    if x[0] in _first_char:
         try:
             return long(x)
         except ValueError: