diff -r a3c94a7e628d Include/unicodeobject.h
--- a/Include/unicodeobject.h	Tue Jan 29 13:18:50 2008 +0100
+++ b/Include/unicodeobject.h	Wed Jan 30 01:36:34 2008 +0100
@@ -358,7 +358,14 @@ typedef PY_UNICODE_TYPE Py_UNICODE;
 
 #else
 
-#define Py_UNICODE_ISSPACE(ch) _PyUnicode_IsWhitespace(ch)
+extern const unsigned char _ascii_whitespace[];
+
+/* 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.
+ */
+#define Py_UNICODE_ISSPACE(ch) \
+	((ch) < 128U ? _ascii_whitespace[(ch)] : _PyUnicode_IsWhitespace(ch))
 
 #define Py_UNICODE_ISLOWER(ch) _PyUnicode_IsLowercase(ch)
 #define Py_UNICODE_ISUPPER(ch) _PyUnicode_IsUppercase(ch)
diff -r a3c94a7e628d Objects/unicodeobject.c
--- a/Objects/unicodeobject.c	Tue Jan 29 13:18:50 2008 +0100
+++ b/Objects/unicodeobject.c	Wed Jan 30 01:36:34 2008 +0100
@@ -125,6 +125,64 @@ static PyUnicodeObject *unicode_latin1[2
 */
 static const char unicode_default_encoding[] = "utf-8";
 
+/* Fast detection of the most frequent whitespace characters */
+const unsigned char _ascii_whitespace[] = {
+	0, 0, 0, 0, 0, 0, 0, 0,
+//     case 0x0009: /* HORIZONTAL TABULATION */
+//     case 0x000A: /* LINE FEED */
+//     case 0x000B: /* VERTICAL TABULATION */
+//     case 0x000C: /* FORM FEED */
+//     case 0x000D: /* CARRIAGE RETURN */
+	0, 1, 1, 1, 1, 1, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+//     case 0x001C: /* FILE SEPARATOR */
+//     case 0x001D: /* GROUP SEPARATOR */
+//     case 0x001E: /* RECORD SEPARATOR */
+//     case 0x001F: /* UNIT SEPARATOR */
+	0, 0, 0, 0, 1, 1, 1, 1,
+//     case 0x0020: /* SPACE */
+	1, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0
+};
+
+/* Same for linebreaks */
+const unsigned char _ascii_linebreak[] = {
+	0, 0, 0, 0, 0, 0, 0, 0,
+//         0x000A, /* LINE FEED */
+//         0x000D, /* CARRIAGE RETURN */
+	0, 0, 1, 0, 0, 1, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+//         0x001C, /* FILE SEPARATOR */
+//         0x001D, /* GROUP SEPARATOR */
+//         0x001E, /* RECORD SEPARATOR */
+	0, 0, 0, 0, 1, 1, 1, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0,
+	0, 0, 0, 0, 0, 0, 0, 0
+};
+
+
 Py_UNICODE
 PyUnicode_GetMax(void)
 {
@@ -151,8 +209,9 @@ static BLOOM_MASK bloom_linebreak;
 
 #define BLOOM(mask, ch) ((mask & (1 << ((ch) & 0x1F))))
 
-#define BLOOM_LINEBREAK(ch)\
-    (BLOOM(bloom_linebreak, (ch)) && Py_UNICODE_ISLINEBREAK((ch)))
+#define BLOOM_LINEBREAK(ch) \
+	((ch) < 128U ? _ascii_linebreak[(ch)] : \
+    (BLOOM(bloom_linebreak, (ch)) && Py_UNICODE_ISLINEBREAK(ch)))
 
 Py_LOCAL_INLINE(BLOOM_MASK) make_bloom_mask(Py_UNICODE* ptr, Py_ssize_t len)
 {
@@ -5602,25 +5661,26 @@ PyObject *split_whitespace(PyUnicodeObje
     register Py_ssize_t j;
     Py_ssize_t len = self->length;
     PyObject *str;
+	const register Py_UNICODE *buf = self->str;
 
     for (i = j = 0; i < len; ) {
 	/* find a token */
-	while (i < len && Py_UNICODE_ISSPACE(self->str[i]))
+	while (i < len && Py_UNICODE_ISSPACE(buf[i]))
 	    i++;
 	j = i;
-	while (i < len && !Py_UNICODE_ISSPACE(self->str[i]))
+	while (i < len && !Py_UNICODE_ISSPACE(buf[i]))
 	    i++;
 	if (j < i) {
 	    if (maxcount-- <= 0)
 		break;
-	    SPLIT_APPEND(self->str, j, i);
-	    while (i < len && Py_UNICODE_ISSPACE(self->str[i]))
+	    SPLIT_APPEND(buf, j, i);
+	    while (i < len && Py_UNICODE_ISSPACE(buf[i]))
 		i++;
 	    j = i;
 	}
     }
     if (j < len) {
-	SPLIT_APPEND(self->str, j, len);
+	SPLIT_APPEND(buf, j, len);
     }
     return list;
 
@@ -5693,18 +5753,19 @@ PyObject *split_char(PyUnicodeObject *se
     register Py_ssize_t j;
     Py_ssize_t len = self->length;
     PyObject *str;
+	register const Py_UNICODE *buf = self->str;
 
     for (i = j = 0; i < len; ) {
-	if (self->str[i] == ch) {
+	if (buf[i] == ch) {
 	    if (maxcount-- <= 0)
 		break;
-	    SPLIT_APPEND(self->str, j, i);
+	    SPLIT_APPEND(buf, j, i);
 	    i = j = i + 1;
 	} else
 	    i++;
     }
     if (j <= len) {
-	SPLIT_APPEND(self->str, j, len);
+	SPLIT_APPEND(buf, j, len);
     }
     return list;
 
@@ -5753,25 +5814,26 @@ PyObject *rsplit_whitespace(PyUnicodeObj
     register Py_ssize_t j;
     Py_ssize_t len = self->length;
     PyObject *str;
+	register const Py_UNICODE *buf = self->str;
 
     for (i = j = len - 1; i >= 0; ) {
 	/* find a token */
-	while (i >= 0 && Py_UNICODE_ISSPACE(self->str[i]))
+	while (i >= 0 && Py_UNICODE_ISSPACE(buf[i]))
 	    i--;
 	j = i;
-	while (i >= 0 && !Py_UNICODE_ISSPACE(self->str[i]))
+	while (i >= 0 && !Py_UNICODE_ISSPACE(buf[i]))
 	    i--;
 	if (j > i) {
 	    if (maxcount-- <= 0)
 		break;
-	    SPLIT_APPEND(self->str, i + 1, j + 1);
-	    while (i >= 0 && Py_UNICODE_ISSPACE(self->str[i]))
+	    SPLIT_APPEND(buf, i + 1, j + 1);
+	    while (i >= 0 && Py_UNICODE_ISSPACE(buf[i]))
 		i--;
 	    j = i;
 	}
     }
     if (j >= 0) {
-	SPLIT_APPEND(self->str, 0, j + 1);
+	SPLIT_APPEND(buf, 0, j + 1);
     }
     if (PyList_Reverse(list) < 0)
         goto onError;
@@ -5792,18 +5854,19 @@ PyObject *rsplit_char(PyUnicodeObject *s
     register Py_ssize_t j;
     Py_ssize_t len = self->length;
     PyObject *str;
+	register const Py_UNICODE *buf = self->str;
 
     for (i = j = len - 1; i >= 0; ) {
-	if (self->str[i] == ch) {
+	if (buf[i] == ch) {
 	    if (maxcount-- <= 0)
 		break;
-	    SPLIT_APPEND(self->str, i + 1, j + 1);
+	    SPLIT_APPEND(buf, i + 1, j + 1);
 	    j = i = i - 1;
 	} else
 	    i--;
     }
     if (j >= -1) {
-	SPLIT_APPEND(self->str, 0, j + 1);
+	SPLIT_APPEND(buf, 0, j + 1);
     }
     if (PyList_Reverse(list) < 0)
         goto onError;