On UCS-2 builds py27 handle low surrogate chars GH342

On UCS-2 builds of Python 2.7-3.2, handle low surrogate characters.

Author: gpshead

cpplint/cpplint.py

@@ -51,6 +51,7 @@
 import string
 import sys
 import unicodedata
+import sysconfig
 
 try:
   xrange          # Python 2
@@ -4291,6 +4292,16 @@ def GetLineWidth(line):
       if unicodedata.east_asian_width(uc) in ('W', 'F'):
         width += 2
       elif not unicodedata.combining(uc):
+        # Issue 337
+        # https://mail.python.org/pipermail/python-list/2012-August/628809.html
+        if (sys.version_info.major, sys.version_info.minor) <= (3, 2):
+          # https://github.com/python/cpython/blob/2.7/Include/unicodeobject.h#L81
+          is_wide_build = sysconfig.get_config_var("Py_UNICODE_SIZE") >= 4
+          # https://github.com/python/cpython/blob/2.7/Objects/unicodeobject.c#L564
+          is_low_surrogate = 0xDC00 <= ord(uc) <= 0xDFFF
+          if not is_wide_build and is_low_surrogate:
+            width -= 1
+          
         width += 1
     return width
   else:

cpplint/cpplint_unittest.py

@@ -321,6 +321,8 @@ def testGetLineWidth(self):
     self.assertEquals(0, cpplint.GetLineWidth(''))
     self.assertEquals(10, cpplint.GetLineWidth(u'x' * 10))
     self.assertEquals(16, cpplint.GetLineWidth(u'都|道|府|県|支庁'))
+    self.assertEquals(5 + 13 + 9, cpplint.GetLineWidth(
+        u'd𝐱/dt' + u'f : t ⨯ 𝐱 → ℝ' + u't ⨯ 𝐱 → ℝ'))
 
   def testGetTextInside(self):
     self.assertEquals('', cpplint._GetTextInside('fun()', r'fun\('))