diff Include/unicodeobject.h @ 45056:c8fca9b5cd3c

Merged revisions 60408-60440 via svnmerge from svn+ssh://pythondev@svn.python.org/python/trunk ........ r60425 | raymond.hettinger | 2008-01-29 20:52:09 +0100 (Tue, 29 Jan 2008) | 1 line CallMethod is faster with a NULL third-argument than with an empty format string. ........ r60431 | raymond.hettinger | 2008-01-30 01:01:07 +0100 (Wed, 30 Jan 2008) | 1 line Add isdisjoint() to the Set/MutableSet ABCs. ........ r60432 | raymond.hettinger | 2008-01-30 01:08:31 +0100 (Wed, 30 Jan 2008) | 1 line MutableSets support a remove() method. ........ r60433 | raymond.hettinger | 2008-01-30 01:51:58 +0100 (Wed, 30 Jan 2008) | 1 line Demonstrate new except/as syntax. ........ r60440 | christian.heimes | 2008-01-30 12:32:37 +0100 (Wed, 30 Jan 2008) | 1 line Patch #1970 by Antoine Pitrou: Speedup unicode whitespace and linebreak detection. The speedup is about 25% for split() (571 / 457 usec) and 35% (175 / 127 usec )for splitlines() ........
author Christian Heimes <christian@cheimes.de>
date Wed, 30 Jan 2008 11:58:22 +0000
parents 7c02182e2979
children 7c4c46342137
line wrap: on
line diff
--- a/Include/unicodeobject.h	Wed Jan 30 11:57:58 2008 +0000
+++ b/Include/unicodeobject.h	Wed Jan 30 11:58:22 2008 +0000
@@ -358,7 +358,14 @@
 
 #else
 
-#define Py_UNICODE_ISSPACE(ch) _PyUnicode_IsWhitespace(ch)
+/* Since splitting on whitespace is an important use case, and whitespace
+   in most situations is solely ASCII whitespace, we optimize for the common
+   case by using a quick look-up table with an inlined check.
+ */
+extern const unsigned char _Py_ascii_whitespace[];
+
+#define Py_UNICODE_ISSPACE(ch) \
+	((ch) < 128U ? _Py_ascii_whitespace[(ch)] : _PyUnicode_IsWhitespace(ch))
 
 #define Py_UNICODE_ISLOWER(ch) _PyUnicode_IsLowercase(ch)
 #define Py_UNICODE_ISUPPER(ch) _PyUnicode_IsUppercase(ch)