This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Revert code change from #32148
[perl5.git] / regcharclass.h
index 08b9eae..f9a1558 100644 (file)
@@ -1,19 +1,23 @@
-/*********************** WARNING WARNING WARNING ************************
-
-Do not modify this code directly: This file was autogenerated by 
-
-        Porting/regcharclass.pl 
-        
-from data contained within the script.  Change the script instead.
-
-Generated at: Mon Apr 23 11:33:52 2007 GMT
-
-************************ WARNING WARNING WARNING ************************/
+/*  -*- buffer-read-only: t -*-
+ *
+ *    regcharclass.h
+ *
+ *    Copyright (C) 2007, by Larry Wall and others
+ *
+ *    You may distribute under the terms of either the GNU General Public
+ *    License or the Artistic License, as specified in the README file.
+ *
+ * !!!!!!!   DO NOT EDIT THIS FILE   !!!!!!!
+ * This file is built by Porting/regcharclass.pl.
+ * 
+ * Any changes made here will be lost!
+ *
+ */
 
 /*
        LNBREAK: Line Break: \R
 
-       "\x0D\x0A"      # CRLF - Network (Windows) line ending 
+       "\x0D\x0A"      # CRLF - Network (Windows) line ending
        0x0A            # LF  | LINE FEED
        0x0B            # VT  | VERTICAL TAB
        0x0C            # FF  | FORM FEED
@@ -24,88 +28,94 @@ Generated at: Mon Apr 23 11:33:52 2007 GMT
 */
 /*** GENERATED CODE ***/
 #define is_LNBREAK(s,is_utf8)                                               \
-( ( ((U8*)s)[0]==13 ) ?                                                     \
-    ( ( ((U8*)s)[1]==10 ) ? 2 : 1 ) :                                       \
-  ( (10<=((U8*)s)[0] && ((U8*)s)[0]<=12) ? 1 :                              \
-( (is_utf8) ?                                                               \
-  ( ( ((U8*)s)[0]==194 ) ?                                                  \
-    ( ( ((U8*)s)[1]==133 ) ? 2 : 0 ) :                                      \
-  (((( ((U8*)s)[0]==226 ) && ( ((U8*)s)[1]==128 )) && ( ((U8*)s)[2]==168 || ((U8*)s)[2]==169 )) ? 3 : 0) ) :\
-  ( ((U8*)s)[0]==133 ) ) ) )                                                
+( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0C ) ? 1                        \
+: ( 0x0D == ((U8*)s)[0] ) ?                                                 \
+    ( ( 0x0A == ((U8*)s)[1] ) ? 2 : 1 )                                     \
+: ( is_utf8 ) ?                                                             \
+    ( ( 0xC2 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x85 == ((U8*)s)[1] ) ? 2 : 0 )                                 \
+    : ( 0xE2 == ((U8*)s)[0] ) ?                                             \
+       ( ( ( 0x80 == ((U8*)s)[1] ) && ( 0xA8 == ((U8*)s)[2] || 0xA9 == ((U8*)s)[2] ) ) ? 3 : 0 )\
+    : 0 )                                                                   \
+: ( 0x85 == ((U8*)s)[0] ) )
 
 /*** GENERATED CODE ***/
 #define is_LNBREAK_safe(s,e,is_utf8)                                        \
-( ( (e) - (s) > 2 ) ?                                                       \
-  ( ( ((U8*)s)[0]==13 ) ?                                                   \
-    ( ( ((U8*)s)[1]==10 ) ? 2 : 1 ) :                                       \
-  ( (10<=((U8*)s)[0] && ((U8*)s)[0]<=12) ? 1 :                              \
-( (is_utf8) ?                                                               \
-  ( ( ((U8*)s)[0]==194 ) ?                                                  \
-    ( ( ((U8*)s)[1]==133 ) ? 2 : 0 ) :                                      \
-  (((( ((U8*)s)[0]==226 ) && ( ((U8*)s)[1]==128 )) && ( ((U8*)s)[2]==168 || ((U8*)s)[2]==169 )) ? 3 : 0) ) :\
-  ( ((U8*)s)[0]==133 ) ) ) ) :                                              \
-( ( (e) - (s) > 1 ) ?                                                       \
-  ( ( ((U8*)s)[0]==13 ) ?                                                   \
-    ( ( ((U8*)s)[1]==10 ) ? 2 : 1 ) :                                       \
-  ( (10<=((U8*)s)[0] && ((U8*)s)[0]<=12) ? 1 :                              \
-( (is_utf8) ?                                                               \
-  ((( ((U8*)s)[0]==194 ) && ( ((U8*)s)[1]==133 )) ? 2 : 0) :                \
-  ( ((U8*)s)[0]==133 ) ) ) ) :                                              \
-( ( (e) - (s) > 0 ) ?                                                       \
-  ( (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) ? 1 :                              \
-( (!is_utf8) ?                                                              \
-  ( ((U8*)s)[0]==133 ) : 0 ) ) : 0 ) ) )                                    
+( ((e)-(s) > 2) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0C ) ? 1                    \
+    : ( 0x0D == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x0A == ((U8*)s)[1] ) ? 2 : 1 )                                 \
+    : ( is_utf8 ) ?                                                         \
+       ( ( 0xC2 == ((U8*)s)[0] ) ?                                         \
+           ( ( 0x85 == ((U8*)s)[1] ) ? 2 : 0 )                             \
+       : ( 0xE2 == ((U8*)s)[0] ) ?                                         \
+           ( ( ( 0x80 == ((U8*)s)[1] ) && ( 0xA8 == ((U8*)s)[2] || 0xA9 == ((U8*)s)[2] ) ) ? 3 : 0 )\
+       : 0 )                                                               \
+    : ( 0x85 == ((U8*)s)[0] ) )                                             \
+: ((e)-(s) > 1) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0C ) ? 1                    \
+    : ( 0x0D == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x0A == ((U8*)s)[1] ) ? 2 : 1 )                                 \
+    : ( is_utf8 ) ?                                                         \
+       ( ( ( 0xC2 == ((U8*)s)[0] ) && ( 0x85 == ((U8*)s)[1] ) ) ? 2 : 0 )  \
+    : ( 0x85 == ((U8*)s)[0] ) )                                             \
+: ((e)-(s) > 0) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D ) ? 1                    \
+    : ( !( is_utf8 ) ) ?                                                    \
+       ( 0x85 == ((U8*)s)[0] )                                             \
+    : 0 )                                                                   \
+: 0 )
 
 /*** GENERATED CODE ***/
 #define is_LNBREAK_utf8(s)                                                  \
-( ( ((U8*)s)[0]==194 ) ?                                                    \
-    ( ( ((U8*)s)[1]==133 ) ? 2 : 0 ) :                                      \
-  ( ( ((U8*)s)[0]==226 ) ?                                                  \
-    ((( ((U8*)s)[1]==128 ) && ( ((U8*)s)[2]==168 || ((U8*)s)[2]==169 )) ? 3 : 0) :\
-  ( ( ((U8*)s)[0]==13 ) ?                                                   \
-    ( ( ((U8*)s)[1]==10 ) ? 2 : 1 ) :                                       \
-  (10<=((U8*)s)[0] && ((U8*)s)[0]<=12) ) ) )                                
+( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0C ) ? 1                        \
+: ( 0x0D == ((U8*)s)[0] ) ?                                                 \
+    ( ( 0x0A == ((U8*)s)[1] ) ? 2 : 1 )                                     \
+: ( 0xC2 == ((U8*)s)[0] ) ?                                                 \
+    ( ( 0x85 == ((U8*)s)[1] ) ? 2 : 0 )                                     \
+: ( 0xE2 == ((U8*)s)[0] ) ?                                                 \
+    ( ( ( 0x80 == ((U8*)s)[1] ) && ( 0xA8 == ((U8*)s)[2] || 0xA9 == ((U8*)s)[2] ) ) ? 3 : 0 )\
+: 0 )
 
 /*** GENERATED CODE ***/
 #define is_LNBREAK_utf8_safe(s,e)                                           \
-( ( (e) - (s) > 2 ) ?                                                       \
-  ( ( ((U8*)s)[0]==194 ) ?                                                  \
-    ( ( ((U8*)s)[1]==133 ) ? 2 : 0 ) :                                      \
-  ( ( ((U8*)s)[0]==226 ) ?                                                  \
-    ((( ((U8*)s)[1]==128 ) && ( ((U8*)s)[2]==168 || ((U8*)s)[2]==169 )) ? 3 : 0) :\
-  ( ( ((U8*)s)[0]==13 ) ?                                                   \
-    ( ( ((U8*)s)[1]==10 ) ? 2 : 1 ) :                                       \
-  (10<=((U8*)s)[0] && ((U8*)s)[0]<=12) ) ) ) :                              \
-( ( (e) - (s) > 1 ) ?                                                       \
-  ( ( ((U8*)s)[0]==194 ) ?                                                  \
-    ( ( ((U8*)s)[1]==133 ) ? 2 : 0 ) :                                      \
-  ( ( ((U8*)s)[0]==13 ) ?                                                   \
-    ( ( ((U8*)s)[1]==10 ) ? 2 : 1 ) :                                       \
-  (10<=((U8*)s)[0] && ((U8*)s)[0]<=12) ) ) :                                \
-( ( (e) - (s) > 0 ) ?                                                       \
-  (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) : 0 ) ) )                            
+( ((e)-(s) > 2) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0C ) ? 1                    \
+    : ( 0x0D == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x0A == ((U8*)s)[1] ) ? 2 : 1 )                                 \
+    : ( 0xC2 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x85 == ((U8*)s)[1] ) ? 2 : 0 )                                 \
+    : ( 0xE2 == ((U8*)s)[0] ) ?                                             \
+       ( ( ( 0x80 == ((U8*)s)[1] ) && ( 0xA8 == ((U8*)s)[2] || 0xA9 == ((U8*)s)[2] ) ) ? 3 : 0 )\
+    : 0 )                                                                   \
+: ((e)-(s) > 1) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0C ) ? 1                    \
+    : ( 0x0D == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x0A == ((U8*)s)[1] ) ? 2 : 1 )                                 \
+    : ( 0xC2 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x85 == ((U8*)s)[1] ) ? 2 : 0 )                                 \
+    : 0 )                                                                   \
+: ((e)-(s) > 0) ?                                                           \
+    ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D )                          \
+: 0 )
 
 /*** GENERATED CODE ***/
 #define is_LNBREAK_latin1(s)                                                \
-( ( ((U8*)s)[0]==13 ) ?                                                     \
-    ( ( ((U8*)s)[1]==10 ) ? 2 : 1 ) :                                       \
-  ( (10<=((U8*)s)[0] && ((U8*)s)[0]<=12) || ((U8*)s)[0]==133 ) )            
+( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0C ) ? 1                        \
+: ( 0x0D == ((U8*)s)[0] ) ?                                                 \
+    ( ( 0x0A == ((U8*)s)[1] ) ? 2 : 1 )                                     \
+: ( 0x85 == ((U8*)s)[0] ) )
 
 /*** GENERATED CODE ***/
 #define is_LNBREAK_latin1_safe(s,e)                                         \
-( ( (e) - (s) > 1 ) ?                                                       \
-  ( ( ((U8*)s)[0]==13 ) ?                                                   \
-    ( ( ((U8*)s)[1]==10 ) ? 2 : 1 ) :                                       \
-  ( (10<=((U8*)s)[0] && ((U8*)s)[0]<=12) || ((U8*)s)[0]==133 ) ) :          \
-( ( (e) - (s) > 0 ) ?                                                       \
-  ( (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) || ((U8*)s)[0]==133 ) : 0 ) )      
-
-/*** GENERATED CODE ***/
-#define is_LNBREAK_cp(cp)                                                   \
-( (10<=cp && cp<=13) || ( cp > 13 &&                                        \
-( cp==133 || ( cp > 133 &&                                                  \
-( cp==8232 || ( cp > 8232 &&                                                \
-cp==8233 ) ) ) ) ) )                                                        
+( ((e)-(s) > 1) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0C ) ? 1                    \
+    : ( 0x0D == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x0A == ((U8*)s)[1] ) ? 2 : 1 )                                 \
+    : ( 0x85 == ((U8*)s)[0] ) )                                             \
+: ((e)-(s) > 0) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D ) || 0x85 == ((U8*)s)[0] )\
+: 0 )
 
 /*
        HORIZWS: Horizontal Whitespace: \h \H
@@ -132,114 +142,141 @@ cp==8233 ) ) ) ) ) )
 */
 /*** GENERATED CODE ***/
 #define is_HORIZWS(s,is_utf8)                                               \
-( ( ((U8*)s)[0]==9 || ((U8*)s)[0]==32 ) ? 1 :                               \
-( (is_utf8) ?                                                               \
-  ( ( ((U8*)s)[0]==194 ) ?                                                  \
-    ( ( ((U8*)s)[1]==160 ) ? 2 : 0 ) :                                      \
-  ( ( ((U8*)s)[0]==225 ) ?                                                  \
-    ( ( ((U8*)s)[1]==160 ) ?                                                \
-      ( ( ((U8*)s)[2]==142 ) ? 3 : 0 ) :                                    \
-    ((( ((U8*)s)[1]==154 ) && ( ((U8*)s)[2]==128 )) ? 3 : 0) ) :            \
-  ( ( ((U8*)s)[0]==226 ) ?                                                  \
-    ( ( ((U8*)s)[1]==129 ) ?                                                \
-      ( ( ((U8*)s)[2]==159 ) ? 3 : 0 ) :                                    \
-    ((( ((U8*)s)[1]==128 ) && ( (128<=((U8*)s)[2] && ((U8*)s)[2]<=138) || ((U8*)s)[2]==175 )) ? 3 : 0) ) :\
-  (((( ((U8*)s)[0]==227 ) && ( ((U8*)s)[1]==128 )) && ( ((U8*)s)[2]==128 )) ? 3 : 0) ) ) ) :\
-  ( ((U8*)s)[0]==160 ) ) )                                                  
+( ( 0x09 == ((U8*)s)[0] || 0x20 == ((U8*)s)[0] ) ? 1                        \
+: ( is_utf8 ) ?                                                             \
+    ( ( 0xC2 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0xA0 == ((U8*)s)[1] ) ? 2 : 0 )                                 \
+    : ( 0xE1 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x9A == ((U8*)s)[1] ) ?                                         \
+           ( ( 0x80 == ((U8*)s)[2] ) ? 3 : 0 )                             \
+       : ( 0xA0 == ((U8*)s)[1] ) ?                                         \
+           ( ( 0x8E == ((U8*)s)[2] ) ? 3 : 0 )                             \
+       : 0 )                                                               \
+    : ( 0xE2 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x80 == ((U8*)s)[1] ) ?                                         \
+           ( ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0x8A ) || 0xAF == ((U8*)s)[2] ) ? 3 : 0 )\
+       : ( 0x81 == ((U8*)s)[1] ) ?                                         \
+           ( ( 0x9F == ((U8*)s)[2] ) ? 3 : 0 )                             \
+       : 0 )                                                               \
+    : ( 0xE3 == ((U8*)s)[0] ) ?                                             \
+       ( ( ( 0x80 == ((U8*)s)[1] ) && ( 0x80 == ((U8*)s)[2] ) ) ? 3 : 0 )  \
+    : 0 )                                                                   \
+: ( 0xA0 == ((U8*)s)[0] ) )
 
 /*** GENERATED CODE ***/
 #define is_HORIZWS_safe(s,e,is_utf8)                                        \
-( ( (e) - (s) > 2 ) ?                                                       \
-  ( ( ((U8*)s)[0]==9 || ((U8*)s)[0]==32 ) ? 1 :                             \
-( (is_utf8) ?                                                               \
-  ( ( ((U8*)s)[0]==194 ) ?                                                  \
-    ( ( ((U8*)s)[1]==160 ) ? 2 : 0 ) :                                      \
-  ( ( ((U8*)s)[0]==225 ) ?                                                  \
-    ( ( ((U8*)s)[1]==160 ) ?                                                \
-      ( ( ((U8*)s)[2]==142 ) ? 3 : 0 ) :                                    \
-    ((( ((U8*)s)[1]==154 ) && ( ((U8*)s)[2]==128 )) ? 3 : 0) ) :            \
-  ( ( ((U8*)s)[0]==226 ) ?                                                  \
-    ( ( ((U8*)s)[1]==129 ) ?                                                \
-      ( ( ((U8*)s)[2]==159 ) ? 3 : 0 ) :                                    \
-    ((( ((U8*)s)[1]==128 ) && ( (128<=((U8*)s)[2] && ((U8*)s)[2]<=138) || ((U8*)s)[2]==175 )) ? 3 : 0) ) :\
-  (((( ((U8*)s)[0]==227 ) && ( ((U8*)s)[1]==128 )) && ( ((U8*)s)[2]==128 )) ? 3 : 0) ) ) ) :\
-  ( ((U8*)s)[0]==160 ) ) ) :                                                \
-( ( (e) - (s) > 1 ) ?                                                       \
-  ( ( ((U8*)s)[0]==9 || ((U8*)s)[0]==32 ) ? 1 :                             \
-( (is_utf8) ?                                                               \
-  ((( ((U8*)s)[0]==194 ) && ( ((U8*)s)[1]==160 )) ? 2 : 0) :                \
-  ( ((U8*)s)[0]==160 ) ) ) :                                                \
-( ( (e) - (s) > 0 ) ?                                                       \
-  ( ( ((U8*)s)[0]==9 || ((U8*)s)[0]==32 ) ? 1 :                             \
-( (!is_utf8) ?                                                              \
-  ( ((U8*)s)[0]==160 ) : 0 ) ) : 0 ) ) )                                    
+( ((e)-(s) > 2) ?                                                           \
+    ( ( 0x09 == ((U8*)s)[0] || 0x20 == ((U8*)s)[0] ) ? 1                    \
+    : ( is_utf8 ) ?                                                         \
+       ( ( 0xC2 == ((U8*)s)[0] ) ?                                         \
+           ( ( 0xA0 == ((U8*)s)[1] ) ? 2 : 0 )                             \
+       : ( 0xE1 == ((U8*)s)[0] ) ?                                         \
+           ( ( 0x9A == ((U8*)s)[1] ) ?                                     \
+               ( ( 0x80 == ((U8*)s)[2] ) ? 3 : 0 )                         \
+           : ( 0xA0 == ((U8*)s)[1] ) ?                                     \
+               ( ( 0x8E == ((U8*)s)[2] ) ? 3 : 0 )                         \
+           : 0 )                                                           \
+       : ( 0xE2 == ((U8*)s)[0] ) ?                                         \
+           ( ( 0x80 == ((U8*)s)[1] ) ?                                     \
+               ( ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0x8A ) || 0xAF == ((U8*)s)[2] ) ? 3 : 0 )\
+           : ( 0x81 == ((U8*)s)[1] ) ?                                     \
+               ( ( 0x9F == ((U8*)s)[2] ) ? 3 : 0 )                         \
+           : 0 )                                                           \
+       : ( 0xE3 == ((U8*)s)[0] ) ?                                         \
+           ( ( ( 0x80 == ((U8*)s)[1] ) && ( 0x80 == ((U8*)s)[2] ) ) ? 3 : 0 )\
+       : 0 )                                                               \
+    : ( 0xA0 == ((U8*)s)[0] ) )                                             \
+: ((e)-(s) > 1) ?                                                           \
+    ( ( 0x09 == ((U8*)s)[0] || 0x20 == ((U8*)s)[0] ) ? 1                    \
+    : ( is_utf8 ) ?                                                         \
+       ( ( ( 0xC2 == ((U8*)s)[0] ) && ( 0xA0 == ((U8*)s)[1] ) ) ? 2 : 0 )  \
+    : ( 0xA0 == ((U8*)s)[0] ) )                                             \
+: ((e)-(s) > 0) ?                                                           \
+    ( ( 0x09 == ((U8*)s)[0] || 0x20 == ((U8*)s)[0] ) ? 1                    \
+    : ( !( is_utf8 ) ) ?                                                    \
+       ( 0xA0 == ((U8*)s)[0] )                                             \
+    : 0 )                                                                   \
+: 0 )
 
 /*** GENERATED CODE ***/
 #define is_HORIZWS_utf8(s)                                                  \
-( ( ((U8*)s)[0]==194 ) ?                                                    \
-    ( ( ((U8*)s)[1]==160 ) ? 2 : 0 ) :                                      \
-  ( ( ((U8*)s)[0]==225 ) ?                                                  \
-    ( ( ((U8*)s)[1]==160 ) ?                                                \
-      ( ( ((U8*)s)[2]==142 ) ? 3 : 0 ) :                                    \
-    ((( ((U8*)s)[1]==154 ) && ( ((U8*)s)[2]==128 )) ? 3 : 0) ) :            \
-  ( ( ((U8*)s)[0]==226 ) ?                                                  \
-    ( ( ((U8*)s)[1]==129 ) ?                                                \
-      ( ( ((U8*)s)[2]==159 ) ? 3 : 0 ) :                                    \
-    ((( ((U8*)s)[1]==128 ) && ( (128<=((U8*)s)[2] && ((U8*)s)[2]<=138) || ((U8*)s)[2]==175 )) ? 3 : 0) ) :\
-  ( ( ((U8*)s)[0]==227 ) ?                                                  \
-    ((( ((U8*)s)[1]==128 ) && ( ((U8*)s)[2]==128 )) ? 3 : 0) :              \
-  ( ((U8*)s)[0]==9 || ((U8*)s)[0]==32 ) ) ) ) )                             
+( ( 0x09 == ((U8*)s)[0] || 0x20 == ((U8*)s)[0] ) ? 1                        \
+: ( 0xC2 == ((U8*)s)[0] ) ?                                                 \
+    ( ( 0xA0 == ((U8*)s)[1] ) ? 2 : 0 )                                     \
+: ( 0xE1 == ((U8*)s)[0] ) ?                                                 \
+    ( ( 0x9A == ((U8*)s)[1] ) ?                                             \
+       ( ( 0x80 == ((U8*)s)[2] ) ? 3 : 0 )                                 \
+    : ( 0xA0 == ((U8*)s)[1] ) ?                                             \
+       ( ( 0x8E == ((U8*)s)[2] ) ? 3 : 0 )                                 \
+    : 0 )                                                                   \
+: ( 0xE2 == ((U8*)s)[0] ) ?                                                 \
+    ( ( 0x80 == ((U8*)s)[1] ) ?                                             \
+       ( ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0x8A ) || 0xAF == ((U8*)s)[2] ) ? 3 : 0 )\
+    : ( 0x81 == ((U8*)s)[1] ) ?                                             \
+       ( ( 0x9F == ((U8*)s)[2] ) ? 3 : 0 )                                 \
+    : 0 )                                                                   \
+: ( 0xE3 == ((U8*)s)[0] ) ?                                                 \
+    ( ( ( 0x80 == ((U8*)s)[1] ) && ( 0x80 == ((U8*)s)[2] ) ) ? 3 : 0 )      \
+: 0 )
 
 /*** GENERATED CODE ***/
 #define is_HORIZWS_utf8_safe(s,e)                                           \
-( ( (e) - (s) > 2 ) ?                                                       \
-  ( ( ((U8*)s)[0]==194 ) ?                                                  \
-    ( ( ((U8*)s)[1]==160 ) ? 2 : 0 ) :                                      \
-  ( ( ((U8*)s)[0]==225 ) ?                                                  \
-    ( ( ((U8*)s)[1]==160 ) ?                                                \
-      ( ( ((U8*)s)[2]==142 ) ? 3 : 0 ) :                                    \
-    ((( ((U8*)s)[1]==154 ) && ( ((U8*)s)[2]==128 )) ? 3 : 0) ) :            \
-  ( ( ((U8*)s)[0]==226 ) ?                                                  \
-    ( ( ((U8*)s)[1]==129 ) ?                                                \
-      ( ( ((U8*)s)[2]==159 ) ? 3 : 0 ) :                                    \
-    ((( ((U8*)s)[1]==128 ) && ( (128<=((U8*)s)[2] && ((U8*)s)[2]<=138) || ((U8*)s)[2]==175 )) ? 3 : 0) ) :\
-  ( ( ((U8*)s)[0]==227 ) ?                                                  \
-    ((( ((U8*)s)[1]==128 ) && ( ((U8*)s)[2]==128 )) ? 3 : 0) :              \
-  ( ((U8*)s)[0]==9 || ((U8*)s)[0]==32 ) ) ) ) ) :                           \
-( ( (e) - (s) > 1 ) ?                                                       \
-  ( ( ((U8*)s)[0]==194 ) ?                                                  \
-    ( ( ((U8*)s)[1]==160 ) ? 2 : 0 ) :                                      \
-  ( ((U8*)s)[0]==9 || ((U8*)s)[0]==32 ) ) :                                 \
-( ( (e) - (s) > 0 ) ?                                                       \
-  ( ((U8*)s)[0]==9 || ((U8*)s)[0]==32 ) : 0 ) ) )                           
+( ((e)-(s) > 2) ?                                                           \
+    ( ( 0x09 == ((U8*)s)[0] || 0x20 == ((U8*)s)[0] ) ? 1                    \
+    : ( 0xC2 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0xA0 == ((U8*)s)[1] ) ? 2 : 0 )                                 \
+    : ( 0xE1 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x9A == ((U8*)s)[1] ) ?                                         \
+           ( ( 0x80 == ((U8*)s)[2] ) ? 3 : 0 )                             \
+       : ( 0xA0 == ((U8*)s)[1] ) ?                                         \
+           ( ( 0x8E == ((U8*)s)[2] ) ? 3 : 0 )                             \
+       : 0 )                                                               \
+    : ( 0xE2 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x80 == ((U8*)s)[1] ) ?                                         \
+           ( ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0x8A ) || 0xAF == ((U8*)s)[2] ) ? 3 : 0 )\
+       : ( 0x81 == ((U8*)s)[1] ) ?                                         \
+           ( ( 0x9F == ((U8*)s)[2] ) ? 3 : 0 )                             \
+       : 0 )                                                               \
+    : ( 0xE3 == ((U8*)s)[0] ) ?                                             \
+       ( ( ( 0x80 == ((U8*)s)[1] ) && ( 0x80 == ((U8*)s)[2] ) ) ? 3 : 0 )  \
+    : 0 )                                                                   \
+: ((e)-(s) > 1) ?                                                           \
+    ( ( 0x09 == ((U8*)s)[0] || 0x20 == ((U8*)s)[0] ) ? 1                    \
+    : ( 0xC2 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0xA0 == ((U8*)s)[1] ) ? 2 : 0 )                                 \
+    : 0 )                                                                   \
+: ((e)-(s) > 0) ?                                                           \
+    ( 0x09 == ((U8*)s)[0] || 0x20 == ((U8*)s)[0] )                          \
+: 0 )
 
 /*** GENERATED CODE ***/
 #define is_HORIZWS_latin1(s)                                                \
-( ((U8*)s)[0]==9 || ((U8*)s)[0]==32 || ((U8*)s)[0]==160 )                   
+( 0x09 == ((U8*)s)[0] || 0x20 == ((U8*)s)[0] || 0xA0 == ((U8*)s)[0] )
 
 /*** GENERATED CODE ***/
 #define is_HORIZWS_latin1_safe(s,e)                                         \
-( ( (e) - (s) > 0 ) ?                                                       \
-  ( ((U8*)s)[0]==9 || ((U8*)s)[0]==32 || ((U8*)s)[0]==160 ) : 0 )           
+( ((e)-(s) > 0) ?                                                           \
+    ( 0x09 == ((U8*)s)[0] || 0x20 == ((U8*)s)[0] || 0xA0 == ((U8*)s)[0] )   \
+: 0 )
 
 /*** GENERATED CODE ***/
 #define is_HORIZWS_cp(cp)                                                   \
-( cp==9 || ( cp > 9 &&                                                      \
-( cp==32 || ( cp > 32 &&                                                    \
-( cp==160 || ( cp > 160 &&                                                  \
-( cp==5760 || ( cp > 5760 &&                                                \
-( cp==6158 || ( cp > 6158 &&                                                \
-( (8192<=cp && cp<=8202) || ( cp > 8202 &&                                  \
-( cp==8239 || ( cp > 8239 &&                                                \
-( cp==8287 || ( cp > 8287 &&                                                \
-cp==12288 ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) )                                   
+( 0x09 == cp || ( 0x09 < cp &&                                              \
+( 0x20 == cp || ( 0x20 < cp &&                                              \
+( 0xA0 == cp || ( 0xA0 < cp &&                                              \
+( 0x1680 == cp || ( 0x1680 < cp &&                                          \
+( 0x180E == cp || ( 0x180E < cp &&                                          \
+( ( 0x2000 <= cp && cp <= 0x200A ) || ( 0x200A < cp &&                      \
+( 0x202F == cp || ( 0x202F < cp &&                                          \
+( 0x205F == cp || ( 0x205F < cp &&                                          \
+0x3000 == cp ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) )
 
 /*
        VERTWS: Vertical Whitespace: \v \V
 
        0x0A            # LF
        0x0B            # VT
-       0x0C            # FF  
+       0x0C            # FF
        0x0D            # CR
        0x85            # NEL
        0x2028          # LINE SEPARATOR
@@ -247,68 +284,171 @@ cp==12288 ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) )
 */
 /*** GENERATED CODE ***/
 #define is_VERTWS(s,is_utf8)                                                \
-( (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) ? 1 :                                \
-( (is_utf8) ?                                                               \
-  ( ( ((U8*)s)[0]==194 ) ?                                                  \
-    ( ( ((U8*)s)[1]==133 ) ? 2 : 0 ) :                                      \
-  (((( ((U8*)s)[0]==226 ) && ( ((U8*)s)[1]==128 )) && ( ((U8*)s)[2]==168 || ((U8*)s)[2]==169 )) ? 3 : 0) ) :\
-  ( ((U8*)s)[0]==133 ) ) )                                                  
+( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D ) ? 1                        \
+: ( is_utf8 ) ?                                                             \
+    ( ( 0xC2 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x85 == ((U8*)s)[1] ) ? 2 : 0 )                                 \
+    : ( 0xE2 == ((U8*)s)[0] ) ?                                             \
+       ( ( ( 0x80 == ((U8*)s)[1] ) && ( 0xA8 == ((U8*)s)[2] || 0xA9 == ((U8*)s)[2] ) ) ? 3 : 0 )\
+    : 0 )                                                                   \
+: ( 0x85 == ((U8*)s)[0] ) )
 
 /*** GENERATED CODE ***/
 #define is_VERTWS_safe(s,e,is_utf8)                                         \
-( ( (e) - (s) > 2 ) ?                                                       \
-  ( (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) ? 1 :                              \
-( (is_utf8) ?                                                               \
-  ( ( ((U8*)s)[0]==194 ) ?                                                  \
-    ( ( ((U8*)s)[1]==133 ) ? 2 : 0 ) :                                      \
-  (((( ((U8*)s)[0]==226 ) && ( ((U8*)s)[1]==128 )) && ( ((U8*)s)[2]==168 || ((U8*)s)[2]==169 )) ? 3 : 0) ) :\
-  ( ((U8*)s)[0]==133 ) ) ) :                                                \
-( ( (e) - (s) > 1 ) ?                                                       \
-  ( (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) ? 1 :                              \
-( (is_utf8) ?                                                               \
-  ((( ((U8*)s)[0]==194 ) && ( ((U8*)s)[1]==133 )) ? 2 : 0) :                \
-  ( ((U8*)s)[0]==133 ) ) ) :                                                \
-( ( (e) - (s) > 0 ) ?                                                       \
-  ( (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) ? 1 :                              \
-( (!is_utf8) ?                                                              \
-  ( ((U8*)s)[0]==133 ) : 0 ) ) : 0 ) ) )                                    
+( ((e)-(s) > 2) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D ) ? 1                    \
+    : ( is_utf8 ) ?                                                         \
+       ( ( 0xC2 == ((U8*)s)[0] ) ?                                         \
+           ( ( 0x85 == ((U8*)s)[1] ) ? 2 : 0 )                             \
+       : ( 0xE2 == ((U8*)s)[0] ) ?                                         \
+           ( ( ( 0x80 == ((U8*)s)[1] ) && ( 0xA8 == ((U8*)s)[2] || 0xA9 == ((U8*)s)[2] ) ) ? 3 : 0 )\
+       : 0 )                                                               \
+    : ( 0x85 == ((U8*)s)[0] ) )                                             \
+: ((e)-(s) > 1) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D ) ? 1                    \
+    : ( is_utf8 ) ?                                                         \
+       ( ( ( 0xC2 == ((U8*)s)[0] ) && ( 0x85 == ((U8*)s)[1] ) ) ? 2 : 0 )  \
+    : ( 0x85 == ((U8*)s)[0] ) )                                             \
+: ((e)-(s) > 0) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D ) ? 1                    \
+    : ( !( is_utf8 ) ) ?                                                    \
+       ( 0x85 == ((U8*)s)[0] )                                             \
+    : 0 )                                                                   \
+: 0 )
 
 /*** GENERATED CODE ***/
 #define is_VERTWS_utf8(s)                                                   \
-( ( ((U8*)s)[0]==194 ) ?                                                    \
-    ( ( ((U8*)s)[1]==133 ) ? 2 : 0 ) :                                      \
-  ( ( ((U8*)s)[0]==226 ) ?                                                  \
-    ((( ((U8*)s)[1]==128 ) && ( ((U8*)s)[2]==168 || ((U8*)s)[2]==169 )) ? 3 : 0) :\
-  (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) ) )                                  
+( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D ) ? 1                        \
+: ( 0xC2 == ((U8*)s)[0] ) ?                                                 \
+    ( ( 0x85 == ((U8*)s)[1] ) ? 2 : 0 )                                     \
+: ( 0xE2 == ((U8*)s)[0] ) ?                                                 \
+    ( ( ( 0x80 == ((U8*)s)[1] ) && ( 0xA8 == ((U8*)s)[2] || 0xA9 == ((U8*)s)[2] ) ) ? 3 : 0 )\
+: 0 )
 
 /*** GENERATED CODE ***/
 #define is_VERTWS_utf8_safe(s,e)                                            \
-( ( (e) - (s) > 2 ) ?                                                       \
-  ( ( ((U8*)s)[0]==194 ) ?                                                  \
-    ( ( ((U8*)s)[1]==133 ) ? 2 : 0 ) :                                      \
-  ( ( ((U8*)s)[0]==226 ) ?                                                  \
-    ((( ((U8*)s)[1]==128 ) && ( ((U8*)s)[2]==168 || ((U8*)s)[2]==169 )) ? 3 : 0) :\
-  (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) ) ) :                                \
-( ( (e) - (s) > 1 ) ?                                                       \
-  ( ( ((U8*)s)[0]==194 ) ?                                                  \
-    ( ( ((U8*)s)[1]==133 ) ? 2 : 0 ) :                                      \
-  (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) ) :                                  \
-( ( (e) - (s) > 0 ) ?                                                       \
-  (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) : 0 ) ) )                            
+( ((e)-(s) > 2) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D ) ? 1                    \
+    : ( 0xC2 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x85 == ((U8*)s)[1] ) ? 2 : 0 )                                 \
+    : ( 0xE2 == ((U8*)s)[0] ) ?                                             \
+       ( ( ( 0x80 == ((U8*)s)[1] ) && ( 0xA8 == ((U8*)s)[2] || 0xA9 == ((U8*)s)[2] ) ) ? 3 : 0 )\
+    : 0 )                                                                   \
+: ((e)-(s) > 1) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D ) ? 1                    \
+    : ( 0xC2 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x85 == ((U8*)s)[1] ) ? 2 : 0 )                                 \
+    : 0 )                                                                   \
+: ((e)-(s) > 0) ?                                                           \
+    ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D )                          \
+: 0 )
 
 /*** GENERATED CODE ***/
 #define is_VERTWS_latin1(s)                                                 \
-( (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) || ((U8*)s)[0]==133 )                
+( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D ) || 0x85 == ((U8*)s)[0] )
 
 /*** GENERATED CODE ***/
 #define is_VERTWS_latin1_safe(s,e)                                          \
-( ( (e) - (s) > 0 ) ?                                                       \
-  ( (10<=((U8*)s)[0] && ((U8*)s)[0]<=13) || ((U8*)s)[0]==133 ) : 0 )        
+( ((e)-(s) > 0) ?                                                           \
+    ( ( 0x0A <= ((U8*)s)[0] && ((U8*)s)[0] <= 0x0D ) || 0x85 == ((U8*)s)[0] )\
+: 0 )
 
 /*** GENERATED CODE ***/
 #define is_VERTWS_cp(cp)                                                    \
-( (10<=cp && cp<=13) || ( cp > 13 &&                                        \
-( cp==133 || ( cp > 133 &&                                                  \
-( cp==8232 || ( cp > 8232 &&                                                \
-cp==8233 ) ) ) ) ) )                                                        
+( ( 0x0A <= cp && cp <= 0x0D ) || ( 0x0D < cp &&                            \
+( 0x85 == cp || ( 0x85 < cp &&                                              \
+( 0x2028 == cp || ( 0x2028 < cp &&                                          \
+0x2029 == cp ) ) ) ) ) )
+
+/*
+       TRICKYFOLD: Problematic fold case letters.
+
+       0x00DF  # LATIN1 SMALL LETTER SHARP S
+       0x0390  # GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
+       0x03B0  # GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS
+*/
+/*** GENERATED CODE ***/
+#define is_TRICKYFOLD(s,is_utf8)                                            \
+( ( is_utf8 ) ?                                                             \
+    ( ( 0xC3 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x9F == ((U8*)s)[1] ) ? 2 : 0 )                                 \
+    : ( 0xCE == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x90 == ((U8*)s)[1] || 0xB0 == ((U8*)s)[1] ) ? 2 : 0 )          \
+    : 0 )                                                                   \
+: ( 0xDF == ((U8*)s)[0] ) )
+
+/*** GENERATED CODE ***/
+#define is_TRICKYFOLD_safe(s,e,is_utf8)                                     \
+( ((e)-(s) > 1) ?                                                           \
+    ( ( is_utf8 ) ?                                                         \
+       ( ( 0xC3 == ((U8*)s)[0] ) ?                                         \
+           ( ( 0x9F == ((U8*)s)[1] ) ? 2 : 0 )                             \
+       : ( 0xCE == ((U8*)s)[0] ) ?                                         \
+           ( ( 0x90 == ((U8*)s)[1] || 0xB0 == ((U8*)s)[1] ) ? 2 : 0 )      \
+       : 0 )                                                               \
+    : ( 0xDF == ((U8*)s)[0] ) )                                             \
+: ((e)-(s) > 0) ?                                                           \
+    ( ( !( is_utf8 ) ) ?                                                    \
+       ( 0xDF == ((U8*)s)[0] )                                             \
+    : 0 )                                                                   \
+: 0 )
+
+/*** GENERATED CODE ***/
+#define is_TRICKYFOLD_cp(cp)                                                \
+( 0xDF == cp || ( 0xDF < cp &&                                              \
+( 0x390 == cp || ( 0x390 < cp &&                                            \
+0x3B0 == cp ) ) ) )
+
+/*** GENERATED CODE ***/
+#define what_TRICKYFOLD(s,is_utf8)                                          \
+( ( is_utf8 ) ?                                                             \
+    ( ( 0xC3 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x9F == ((U8*)s)[1] ) ? 0xDF : 0 )                              \
+    : ( 0xCE == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x90 == ((U8*)s)[1] ) ? 0x390                                   \
+       : ( 0xB0 == ((U8*)s)[1] ) ? 0x3B0 : 0 )                             \
+    : 0 )                                                                   \
+: ( 0xDF == ((U8*)s)[0] ) ? 0xDF : 0 )
+
+/*** GENERATED CODE ***/
+#define what_TRICKYFOLD_safe(s,e,is_utf8)                                   \
+( ((e)-(s) > 1) ?                                                           \
+    ( ( is_utf8 ) ?                                                         \
+       ( ( 0xC3 == ((U8*)s)[0] ) ?                                         \
+           ( ( 0x9F == ((U8*)s)[1] ) ? 0xDF : 0 )                          \
+       : ( 0xCE == ((U8*)s)[0] ) ?                                         \
+           ( ( 0x90 == ((U8*)s)[1] ) ? 0x390                               \
+           : ( 0xB0 == ((U8*)s)[1] ) ? 0x3B0 : 0 )                         \
+       : 0 )                                                               \
+    : ( 0xDF == ((U8*)s)[0] ) ? 0xDF : 0 )                                  \
+: ((e)-(s) > 0) ?                                                           \
+    ( ( ( !( is_utf8 ) ) && ( 0xDF == ((U8*)s)[0] ) ) ? 0xDF : 0 )          \
+: 0 )
+
+/*** GENERATED CODE ***/
+#define what_len_TRICKYFOLD(s,is_utf8,len)                                  \
+( ( is_utf8 ) ?                                                             \
+    ( ( 0xC3 == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x9F == ((U8*)s)[1] ) ? len=2, 0xDF : 0 )                       \
+    : ( 0xCE == ((U8*)s)[0] ) ?                                             \
+       ( ( 0x90 == ((U8*)s)[1] ) ? len=2, 0x390                            \
+       : ( 0xB0 == ((U8*)s)[1] ) ? len=2, 0x3B0 : 0 )                      \
+    : 0 )                                                                   \
+: ( 0xDF == ((U8*)s)[0] ) ? len=1, 0xDF : 0 )
+
+/*** GENERATED CODE ***/
+#define what_len_TRICKYFOLD_safe(s,e,is_utf8,len)                           \
+( ((e)-(s) > 1) ?                                                           \
+    ( ( is_utf8 ) ?                                                         \
+       ( ( 0xC3 == ((U8*)s)[0] ) ?                                         \
+           ( ( 0x9F == ((U8*)s)[1] ) ? len=2, 0xDF : 0 )                   \
+       : ( 0xCE == ((U8*)s)[0] ) ?                                         \
+           ( ( 0x90 == ((U8*)s)[1] ) ? len=2, 0x390                        \
+           : ( 0xB0 == ((U8*)s)[1] ) ? len=2, 0x3B0 : 0 )                  \
+       : 0 )                                                               \
+    : ( 0xDF == ((U8*)s)[0] ) ? len=1, 0xDF : 0 )                           \
+: ((e)-(s) > 0) ?                                                           \
+    ( ( ( !( is_utf8 ) ) && ( 0xDF == ((U8*)s)[0] ) ) ? len=1, 0xDF : 0 )   \
+: 0 )
 
+/* ex: set ro: */