regen/regcharclass.pl: Add optimization

[perl5.git] / regcharclass.h

diff --git a/regcharclass.h b/regcharclass.h
index 59b2fdf..840caf1 100644 (file)
--- a/regcharclass.h
+++ b/regcharclass.h
@@ -189,7 +189,7 @@
     : 0 )                                                                   \
 : ( 0xE2 == ((U8*)s)[0] ) ?                                                 \
     ( ( 0x80 == ((U8*)s)[1] ) ?                                             \
-       ( ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0x8A ) || 0xAF == ((U8*)s)[2] ) ? 3 : 0 )\
+       ( ( ( ((U8*)s)[2] <= 0x8A ) || 0xAF == ((U8*)s)[2] ) ? 3 : 0 )      \
     : ( 0x81 == ((U8*)s)[1] ) ?                                             \
        ( ( 0x9F == ((U8*)s)[2] ) ? 3 : 0 )                                 \
     : 0 )                                                                   \
@@ -341,9 +341,9 @@
 #define is_GCB_L_utf8(s)                                                    \
 ( ( 0xE1 == ((U8*)s)[0] ) ?                                                 \
     ( ( 0x84 == ((U8*)s)[1] ) ?                                             \
-       ( ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ? 3 : 0 )                      \
+       3                                                                   \
     : ( 0x85 == ((U8*)s)[1] ) ?                                             \
-       ( ( ( ((U8*)s)[2] & 0xE0 ) == 0x80 ) ? 3 : 0 )                      \
+       ( ( ((U8*)s)[2] <= 0x9F ) ? 3 : 0 )                                 \
     : 0 )                                                                   \
 : ( 0xEA == ((U8*)s)[0] ) ?                                                 \
     ( ( ( 0xA5 == ((U8*)s)[1] ) && ( 0xA0 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBC ) ) ? 3 : 0 )\
@@ -358,21 +358,23 @@
 #define is_GCB_LV_LVT_V_utf8(s)                                             \
 ( ( 0xE1 == ((U8*)s)[0] ) ?                                                 \
     ( ( 0x85 == ((U8*)s)[1] ) ?                                             \
-       ( ( ( ((U8*)s)[2] & 0xE0 ) == 0xA0 ) ? 3 : 0 )                      \
+       ( ( ((U8*)s)[2] >= 0xA0 ) ? 3 : 0 )                                 \
     : ( 0x86 == ((U8*)s)[1] ) ?                                             \
-       ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xA7 ) ? 3 : 0 )          \
+       ( ( ((U8*)s)[2] <= 0xA7 ) ? 3 : 0 )                                 \
     : 0 )                                                                   \
 : ( 0xEA == ((U8*)s)[0] ) ?                                                 \
-    ( ( ( ( ((U8*)s)[1] & 0xF0 ) == 0xB0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
+    ( ( ((U8*)s)[1] >= 0xB0 ) ?                                             \
+       3                                                                   \
+    : 0 )                                                                   \
 : ( 0xEB == ((U8*)s)[0] || 0xEC == ((U8*)s)[0] ) ?                          \
-    ( ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
+    3                                                                       \
 : ( 0xED == ((U8*)s)[0] ) ?                                                 \
-    ( ( 0x80 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0x9D ) ?                      \
-       ( ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ? 3 : 0 )                      \
+    ( ( ((U8*)s)[1] <= 0x9D ) ?                                             \
+       3                                                                   \
     : ( 0x9E == ((U8*)s)[1] ) ?                                             \
-       ( ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xA3 ) || ( ((U8*)s)[2] & 0xF0 ) == 0xB0 ) ? 3 : 0 )\
+       ( ( ( ((U8*)s)[2] <= 0xA3 ) || ( ((U8*)s)[2] >= 0xB0 ) ) ? 3 : 0 )  \
     : ( 0x9F == ((U8*)s)[1] ) ?                                             \
-       ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0x86 ) ? 3 : 0 )          \
+       ( ( ((U8*)s)[2] <= 0x86 ) ? 3 : 0 )                                 \
     : 0 )                                                                   \
 : 0 )
 
@@ -392,7 +394,7 @@
 */
 /*** GENERATED CODE ***/
 #define is_GCB_RI_utf8(s)                                                   \
-( ( ( ( ( 0xF0 == ((U8*)s)[0] ) && ( 0x9F == ((U8*)s)[1] ) ) && ( 0x87 == ((U8*)s)[2] ) ) && ( 0xA6 <= ((U8*)s)[3] && ((U8*)s)[3] <= 0xBF ) ) ? 4 : 0 )
+( ( ( ( ( 0xF0 == ((U8*)s)[0] ) && ( 0x9F == ((U8*)s)[1] ) ) && ( 0x87 == ((U8*)s)[2] ) ) && ( ((U8*)s)[3] >= 0xA6 ) ) ? 4 : 0 )
 
 /*
        GCB_SPECIAL_BEGIN: Grapheme_Cluster_Break=special_begins
@@ -401,25 +403,28 @@
 */
 /*** GENERATED CODE ***/
 #define is_GCB_SPECIAL_BEGIN_utf8(s)                                        \
-( ( ( 0xE1 == ((U8*)s)[0] ) && ( ( ((U8*)s)[1] & 0xFC ) == 0x84 ) ) ? ( ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ? 3 : 0 )\
+( ( 0xE1 == ((U8*)s)[0] ) ?                                                 \
+    ( ( ( ((U8*)s)[1] & 0xFC ) == 0x84 ) ?                                  \
+       3                                                                   \
+    : 0 )                                                                   \
 : ( 0xEA == ((U8*)s)[0] ) ?                                                 \
     ( ( 0xA5 == ((U8*)s)[1] ) ?                                             \
        ( ( 0xA0 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBC ) ? 3 : 0 )          \
-    : ( ( ((U8*)s)[1] & 0xF0 ) == 0xB0 ) ?                                  \
-       ( ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ? 3 : 0 )                      \
+    : ( ((U8*)s)[1] >= 0xB0 ) ?                                             \
+       3                                                                   \
     : 0 )                                                                   \
 : ( 0xEB == ((U8*)s)[0] || 0xEC == ((U8*)s)[0] ) ?                          \
-    ( ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
+    3                                                                       \
 : ( 0xED == ((U8*)s)[0] ) ?                                                 \
-    ( ( 0x80 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0x9D ) ?                      \
-       ( ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ? 3 : 0 )                      \
+    ( ( ((U8*)s)[1] <= 0x9D ) ?                                             \
+       3                                                                   \
     : ( 0x9E == ((U8*)s)[1] ) ?                                             \
-       ( ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xA3 ) || ( ((U8*)s)[2] & 0xF0 ) == 0xB0 ) ? 3 : 0 )\
+       ( ( ( ((U8*)s)[2] <= 0xA3 ) || ( ((U8*)s)[2] >= 0xB0 ) ) ? 3 : 0 )  \
     : ( 0x9F == ((U8*)s)[1] ) ?                                             \
-       ( ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0x86 ) || ( 0x8B <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBB ) ) ? 3 : 0 )\
+       ( ( ( ((U8*)s)[2] <= 0x86 ) || ( 0x8B <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBB ) ) ? 3 : 0 )\
     : 0 )                                                                   \
 : ( 0xF0 == ((U8*)s)[0] ) ?                                                 \
-    ( ( ( ( 0x9F == ((U8*)s)[1] ) && ( 0x87 == ((U8*)s)[2] ) ) && ( 0xA6 <= ((U8*)s)[3] && ((U8*)s)[3] <= 0xBF ) ) ? 4 : 0 )\
+    ( ( ( ( 0x9F == ((U8*)s)[1] ) && ( 0x87 == ((U8*)s)[2] ) ) && ( ((U8*)s)[3] >= 0xA6 ) ) ? 4 : 0 )\
 : 0 )
 
 /*
@@ -431,9 +436,9 @@
 #define is_GCB_T_utf8(s)                                                    \
 ( ( 0xE1 == ((U8*)s)[0] ) ?                                                 \
     ( ( 0x86 == ((U8*)s)[1] ) ?                                             \
-       ( ( 0xA8 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBF ) ? 3 : 0 )          \
+       ( ( ((U8*)s)[2] >= 0xA8 ) ? 3 : 0 )                                 \
     : ( 0x87 == ((U8*)s)[1] ) ?                                             \
-       ( ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ? 3 : 0 )                      \
+       3                                                                   \
     : 0 )                                                                   \
 : ( 0xED == ((U8*)s)[0] ) ?                                                 \
     ( ( ( 0x9F == ((U8*)s)[1] ) && ( 0x8B <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBB ) ) ? 3 : 0 )\
@@ -448,15 +453,15 @@
 #define is_GCB_V_utf8(s)                                                    \
 ( ( 0xE1 == ((U8*)s)[0] ) ?                                                 \
     ( ( 0x85 == ((U8*)s)[1] ) ?                                             \
-       ( ( ( ((U8*)s)[2] & 0xE0 ) == 0xA0 ) ? 3 : 0 )                      \
+       ( ( ((U8*)s)[2] >= 0xA0 ) ? 3 : 0 )                                 \
     : ( 0x86 == ((U8*)s)[1] ) ?                                             \
-       ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xA7 ) ? 3 : 0 )          \
+       ( ( ((U8*)s)[2] <= 0xA7 ) ? 3 : 0 )                                 \
     : 0 )                                                                   \
 : ( 0xED == ((U8*)s)[0] ) ?                                                 \
     ( ( 0x9E == ((U8*)s)[1] ) ?                                             \
-       ( ( ( ((U8*)s)[2] & 0xF0 ) == 0xB0 ) ? 3 : 0 )                      \
+       ( ( ((U8*)s)[2] >= 0xB0 ) ? 3 : 0 )                                 \
     : ( 0x9F == ((U8*)s)[1] ) ?                                             \
-       ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0x86 ) ? 3 : 0 )          \
+       ( ( ((U8*)s)[2] <= 0x86 ) ? 3 : 0 )                                 \
     : 0 )                                                                   \
 : 0 )
 
@@ -481,35 +486,35 @@
     : 0 )                                                                   \
 : ( 0xE2 == ((U8*)s)[0] ) ?                                                 \
     ( ( 0x80 == ((U8*)s)[1] ) ?                                             \
-       ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBE ) ? 3 : 0 )          \
+       ( ( ((U8*)s)[2] <= 0xBE ) ? 3 : 0 )                                 \
     : ( 0x81 == ((U8*)s)[1] ) ?                                             \
        ( ( ( 0x81 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0x93 ) || ( 0x95 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xAF ) ) ? 3 : 0 )\
     : ( 0x86 == ((U8*)s)[1] ) ?                                             \
-       ( ( 0x90 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBF ) ? 3 : 0 )          \
+       ( ( ((U8*)s)[2] >= 0x90 ) ? 3 : 0 )                                 \
     : ( 0x87 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0x90 ) ?                      \
-       ( ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ? 3 : 0 )                      \
+       3                                                                   \
     : ( 0x91 == ((U8*)s)[1] ) ?                                             \
-       ( ( ( ((U8*)s)[2] & 0xE0 ) == 0x80 ) ? 3 : 0 )                      \
+       ( ( ((U8*)s)[2] <= 0x9F ) ? 3 : 0 )                                 \
     : ( 0x94 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0x9C ) ?                      \
-       ( ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ? 3 : 0 )                      \
+       3                                                                   \
     : ( 0x9D == ((U8*)s)[1] ) ?                                             \
-       ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xB5 ) ? 3 : 0 )          \
+       ( ( ((U8*)s)[2] <= 0xB5 ) ? 3 : 0 )                                 \
     : ( 0x9E == ((U8*)s)[1] ) ?                                             \
-       ( ( 0x94 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBF ) ? 3 : 0 )          \
+       ( ( ((U8*)s)[2] >= 0x94 ) ? 3 : 0 )                                 \
     : ( ( 0x9F <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xAF ) || ( ((U8*)s)[1] & 0xFE ) == 0xB8 ) ?\
-       ( ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ? 3 : 0 )                      \
+       3                                                                   \
     : 0 )                                                                   \
 : ( 0xE3 == ((U8*)s)[0] ) ?                                                 \
     ( ( 0x80 == ((U8*)s)[1] ) ?                                             \
-       ( ( ( ((U8*)s)[2] & 0xFC ) == 0x80 || ( 0x88 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xA0 ) || 0xB0 == ((U8*)s)[2] ) ? 3 : 0 )\
+       ( ( ( ((U8*)s)[2] <= 0x83 ) || ( 0x88 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xA0 ) || 0xB0 == ((U8*)s)[2] ) ? 3 : 0 )\
     : ( 0x85 == ((U8*)s)[1] ) ?                                             \
        ( ( 0xA4 == ((U8*)s)[2] ) ? 3 : 0 )                                 \
     : 0 )                                                                   \
 : ( 0xEF == ((U8*)s)[0] ) ?                                                 \
     ( ( 0xB4 == ((U8*)s)[1] ) ?                                             \
-       ( ( ( ((U8*)s)[2] & 0xFE ) == 0xBE ) ? 3 : 0 )                      \
+       ( ( ((U8*)s)[2] >= 0xBE ) ? 3 : 0 )                                 \
     : ( 0xB8 == ((U8*)s)[1] ) ?                                             \
-       ( ( ( ((U8*)s)[2] & 0xF0 ) == 0x80 ) ? 3 : 0 )                      \
+       ( ( ((U8*)s)[2] <= 0x8F ) ? 3 : 0 )                                 \
     : ( 0xB9 == ((U8*)s)[1] ) ?                                             \
        ( ( 0x85 == ((U8*)s)[2] || 0x86 == ((U8*)s)[2] ) ? 3 : 0 )          \
     : ( 0xBB == ((U8*)s)[1] ) ?                                             \
@@ -522,7 +527,9 @@
 : ( 0xF0 == ((U8*)s)[0] ) ?                                                 \
     ( ( ( ( 0x9D == ((U8*)s)[1] ) && ( 0x85 == ((U8*)s)[2] ) ) && ( 0xB3 <= ((U8*)s)[3] && ((U8*)s)[3] <= 0xBA ) ) ? 4 : 0 )\
 : ( 0xF3 == ((U8*)s)[0] ) ?                                                 \
-    ( ( ( ( 0xA0 == ((U8*)s)[1] ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
+    ( ( 0xA0 == ((U8*)s)[1] ) ?                                             \
+       4                                                                   \
+    : 0 )                                                                   \
 : 0 )