Avoid unnecessary malformed checking

regen/regcharclass.pl can create macros for use where we need to worry
about the possibility of malformed UTF-8, and for where we don't.  In
the case of looking at regex patterns, the Perl core has complete
control over generating them, and hence isn't generally going to create
too short a buffer;  if it does, it's a bug that will show up and get
fixed.  This commit changes to generate and use the faster macros that
don't do bounds checking.

diff --git a/regcharclass.h b/regcharclass.h
index 2c482ce..868edbc 100644 (file)
--- a/regcharclass.h
+++ b/regcharclass.h
@@ -605,234 +605,88 @@
        &regcharclass_multi_char_folds::multi_char_folds(1)
 */
 /*** GENERATED CODE ***/
-#define is_MULTI_CHAR_FOLD_utf8_safe_part0(s,e)                             \
-( ( 0x73 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x74 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ? 2 : 0 )
-
-
-/*** GENERATED CODE ***/
-#define is_MULTI_CHAR_FOLD_utf8_safe_part1(s,e)                             \
-( ( 0x74 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x88 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x77 == NATIVE_TO_LATIN1(((U8*)s)[0]) || 0x79 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?\
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x8A == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0xC5 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( ( 0xBF == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xC5 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0xBF == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-    : ( 0xCA == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xBC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x6E == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xFD ) == 0xAC ) ?            \
-           ( ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0xB1 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0xB7 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?\
-           ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                   \
-               ( ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ?               \
-                   ( ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[4]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[5]) ) ) ? 6 : 4 )\
-               : 0 )                                                       \
-           : ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                   \
-               ( ( 0x88 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ?               \
-                   ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[4]) ) ?           \
-                       ( ( ( NATIVE_TO_LATIN1(((U8*)s)[5]) & 0xFE ) == 0x80 ) ? 6 : 0 )\
-                   : ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[4]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[5]) ) ) ? 6 : 0 )\
-               : 0 )                                                       \
-           : ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : 0 )                                                               \
-    : ( 0xCF == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0x81 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x93 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0x85 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                   \
-               ( ( 0x88 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ?               \
-                   ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[4]) ) ?           \
-                       ( ( ( NATIVE_TO_LATIN1(((U8*)s)[5]) & 0xFE ) == 0x80 ) ? 6 : 0 )\
-                   : ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[4]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[5]) ) ) ? 6 : 0 )\
-               : ( 0x93 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ?               \
-                   ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[4]) ) ?           \
-                       ( ( ( NATIVE_TO_LATIN1(((U8*)s)[5]) & 0xFE ) == 0x80 ) ? 6 : 4 )\
-                   : ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[4]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[5]) ) ) ? 6 : 4 )\
-               : 0 )                                                       \
-           : ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0x89 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                   \
-               ( ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ?               \
-                   ( ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[4]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[5]) ) ) ? 6 : 4 )\
-               : 0 )                                                       \
-           : ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( ( ( 0x8E == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-    : ( 0xD5 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0xA5 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( ( 0xD6 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0xB4 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( ( 0xD5 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( ( ( NATIVE_TO_LATIN1(((U8*)s)[3]) & 0xF7 ) == 0xA5 ) || NATIVE_TO_LATIN1(((U8*)s)[3]) == 0xAB || NATIVE_TO_LATIN1(((U8*)s)[3]) == 0xB6 ) ) ? 4 : 0 )\
-       : ( ( ( 0xBE == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xD5 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0xB6 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-    : ( 0xE1 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0xBC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( ( ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xD8 ) == 0x80 ) && ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[4]) ) ) ? 5 : 0 )\
-       : ( ( ( ( 0xBD == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xF8 ) == 0xA0 ) || ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xFB ) == 0xB0 ) || NATIVE_TO_LATIN1(((U8*)s)[2]) == 0xBC ) ) && ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[4]) ) ) ? 5 : 0 )\
-    : 0 )
-
-
-/*** GENERATED CODE ***/
-#define is_MULTI_CHAR_FOLD_utf8_safe_part2(s,e)                             \
-( ( 0x61 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
-       ( ( ( 0xCA == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xBE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x66 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0x66 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[2]) || 0x6C == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ? 3 : 2 )\
-       : ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x6C == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ? 2 : 0 )\
-    : ( 0x68 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xB1 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x87 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x6A == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x8C == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x73 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0x73 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x74 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ? 2 : 0 )\
-    : ( 0x74 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x88 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x77 == NATIVE_TO_LATIN1(((U8*)s)[0]) || 0x79 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?\
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x8A == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0xC5 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( ( 0xBF == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xC5 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0xBF == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-    : ( 0xCA == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xBC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x6E == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xFD ) == 0xAC ) ?            \
-           ( ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0xB1 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0xB7 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?\
-           ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                   \
-               ( ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ? 4 : 0 )       \
-           : ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( ( ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-    : ( 0xCF == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0x81 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x93 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0x85 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                   \
-               ( ( 0x93 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ? 4 : 0 )       \
-           : ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0x89 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                   \
-               ( ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ? 4 : 0 )       \
-           : ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( ( ( 0x8E == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-    : ( 0xD5 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0xA5 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( ( 0xD6 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0xB4 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( ( 0xD5 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( ( ( NATIVE_TO_LATIN1(((U8*)s)[3]) & 0xF7 ) == 0xA5 ) || NATIVE_TO_LATIN1(((U8*)s)[3]) == 0xAB || NATIVE_TO_LATIN1(((U8*)s)[3]) == 0xB6 ) ) ? 4 : 0 )\
-       : ( ( ( 0xBE == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xD5 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0xB6 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-    : 0 )
-
-
-/*** GENERATED CODE ***/
-#define is_MULTI_CHAR_FOLD_utf8_safe_part3(s,e)                             \
-( ((e)-(s) > 2) ?                                                           \
-    ( ( 0x61 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCA == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xBE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x66 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0x66 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[2]) || 0x6C == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ? 3 : 2 )\
-       : ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x6C == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ? 2 : 0 )\
-    : ( 0x68 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xB1 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x87 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x6A == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x8C == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x73 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0x73 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x74 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ? 2 : 0 )\
-    : ( 0x74 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x88 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x77 == NATIVE_TO_LATIN1(((U8*)s)[0]) || 0x79 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?\
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x8A == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( ( ( 0xCA == NATIVE_TO_LATIN1(((U8*)s)[0]) ) && ( 0xBC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ) && ( 0x6E == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-: ((e)-(s) > 1) ?                                                           \
-    ( ( 0x66 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0x66 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x69 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x6C == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ? 2 : 0 )\
-    : ( ( 0x73 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) && ( 0x73 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x74 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ) ? 2 : 0 )\
+#define is_MULTI_CHAR_FOLD_utf8_part0(s)                                    \
+( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xB1 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )
+
+
+/*** GENERATED CODE ***/
+#define is_MULTI_CHAR_FOLD_utf8_part1(s)                                    \
+( ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
+    ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x87 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
+: ( 0x6A == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
+    ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x8C == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
+: ( 0x73 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
+    ( ( 0x73 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x74 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ? 2 : 0 )\
+: ( 0x74 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
+    ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x88 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
+: ( 0x77 == NATIVE_TO_LATIN1(((U8*)s)[0]) || 0x79 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?\
+    ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x8A == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
+: ( 0xC5 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
+    ( ( ( ( 0xBF == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xC5 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0xBF == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
+: ( 0xCA == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
+    ( ( ( 0xBC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x6E == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
+: ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
+    ( ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xFD ) == 0xAC ) ?                \
+       ( ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
+    : ( 0xB1 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0xB7 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?\
+       ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                       \
+           ( ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ?                   \
+               ( ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[4]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[5]) ) ) ? 6 : 4 )\
+           : 0 )                                                           \
+       : ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
+    : ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                           \
+       ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                       \
+           ( ( 0x88 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ?                   \
+               ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[4]) ) ?               \
+                   ( ( NATIVE_TO_LATIN1(((U8*)s)[5]) <= 0x81 ) ? 6 : 0 )   \
+               : ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[4]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[5]) ) ) ? 6 : 0 )\
+           : 0 )                                                           \
+       : ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
+    : 0 )                                                                   \
+: ( 0xCF == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
+    ( ( 0x81 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                           \
+       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x93 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
+    : ( 0x85 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                           \
+       ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                       \
+           ( ( 0x88 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ?                   \
+               ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[4]) ) ?               \
+                   ( ( NATIVE_TO_LATIN1(((U8*)s)[5]) <= 0x81 ) ? 6 : 0 )   \
+               : ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[4]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[5]) ) ) ? 6 : 0 )\
+           : ( 0x93 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ?                   \
+               ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[4]) ) ?               \
+                   ( ( NATIVE_TO_LATIN1(((U8*)s)[5]) <= 0x81 ) ? 6 : 4 )   \
+               : ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[4]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[5]) ) ) ? 6 : 4 )\
+           : 0 )                                                           \
+       : ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
+    : ( 0x89 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                           \
+       ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                       \
+           ( ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ?                   \
+               ( ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[4]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[5]) ) ) ? 6 : 4 )\
+           : 0 )                                                           \
+       : ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
+    : ( ( ( 0x8E == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
+: ( 0xD5 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
+    ( ( 0xA5 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                           \
+       ( ( ( 0xD6 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
+    : ( 0xB4 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                           \
+       ( ( ( 0xD5 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( ( ( NATIVE_TO_LATIN1(((U8*)s)[3]) & 0xF7 ) == 0xA5 ) || NATIVE_TO_LATIN1(((U8*)s)[3]) == 0xAB || NATIVE_TO_LATIN1(((U8*)s)[3]) == 0xB6 ) ) ? 4 : 0 )\
+    : ( ( ( 0xBE == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xD5 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0xB6 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
+: ( 0xE1 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
+    ( ( 0xBC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                           \
+       ( ( ( ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xD8 ) == 0x80 ) && ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[4]) ) ) ? 5 : 0 )\
+    : ( ( ( ( 0xBD == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xF8 ) == 0xA0 ) || ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xFB ) == 0xB0 ) || NATIVE_TO_LATIN1(((U8*)s)[2]) == 0xBC ) ) && ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[4]) ) ) ? 5 : 0 )\
 : 0 )
 
 
 /*** GENERATED CODE ***/
-#define is_MULTI_CHAR_FOLD_utf8_safe_part4(s,e)                             \
+#define is_MULTI_CHAR_FOLD_utf8(s)                                          \
 ( ( 0x61 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
-       ( ( ( 0xCA == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xBE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x66 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0x66 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[2]) || 0x6C == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ? 3 : 2 )\
-       : ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x6C == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ? 2 : 0 )\
-    : ( 0x68 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xB1 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x87 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x6A == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x8C == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x73 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ? is_MULTI_CHAR_FOLD_utf8_safe_part0(s,e) : is_MULTI_CHAR_FOLD_utf8_safe_part1(s,e) )
-
-
-/*** GENERATED CODE ***/
-#define is_MULTI_CHAR_FOLD_utf8_safe_part5(s,e)                             \
-( ((e)-(s) > 4) ?                                                           \
-    ( ( 0x61 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCA == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xBE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x66 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0x66 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[2]) || 0x6C == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ? 3 : 2 )\
-       : ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x6C == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ? 2 : 0 )\
-    : ( 0x68 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xB1 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x87 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x6A == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x8C == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x73 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0x73 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x74 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ? 2 : 0 )\
-    : ( 0x74 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x88 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0x77 == NATIVE_TO_LATIN1(((U8*)s)[0]) || 0x79 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?\
-       ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x8A == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0xC5 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( ( 0xBF == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xC5 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0xBF == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-    : ( 0xCA == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( 0xBC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0x6E == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
-    : ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xFD ) == 0xAC ) ?            \
-           ( ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0xB1 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0xB7 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?\
-           ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                   \
-               ( ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ? 4 : 0 )       \
-           : ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( ( ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-    : ( 0xCF == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0x81 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x93 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0x85 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( 0xCC == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                   \
-               ( ( 0x93 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ? 4 : 0 )       \
-           : ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0x89 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( 0xCD == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ?                   \
-               ( ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ? 4 : 0 )       \
-           : ( ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( ( ( 0x8E == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-    : ( 0xD5 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0xA5 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( ( 0xD6 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( 0x82 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-       : ( 0xB4 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( ( 0xD5 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) && ( ( ( NATIVE_TO_LATIN1(((U8*)s)[3]) & 0xF7 ) == 0xA5 ) || NATIVE_TO_LATIN1(((U8*)s)[3]) == 0xAB || NATIVE_TO_LATIN1(((U8*)s)[3]) == 0xB6 ) ) ? 4 : 0 )\
-       : ( ( ( 0xBE == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xD5 == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) && ( 0xB6 == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) ? 4 : 0 )\
-    : ( 0xE1 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                           \
-       ( ( 0xBC == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                       \
-           ( ( ( ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xD8 ) == 0x80 ) && ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[4]) ) ) ? 5 : 0 )\
-       : ( ( ( ( 0xBD == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xF8 ) == 0xA0 ) || ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xFB ) == 0xB0 ) || NATIVE_TO_LATIN1(((U8*)s)[2]) == 0xBC ) ) && ( 0xCE == NATIVE_TO_LATIN1(((U8*)s)[3]) ) ) && ( 0xB9 == NATIVE_TO_LATIN1(((U8*)s)[4]) ) ) ? 5 : 0 )\
-    : 0 )                                                                   \
-: ((e)-(s) > 3) ? is_MULTI_CHAR_FOLD_utf8_safe_part2(s,e) : is_MULTI_CHAR_FOLD_utf8_safe_part3(s,e) )
-
-
-/*** GENERATED CODE ***/
-#define is_MULTI_CHAR_FOLD_utf8_safe(s,e)                                   \
-( ((e)-(s) > 5) ? is_MULTI_CHAR_FOLD_utf8_safe_part4(s,e) : is_MULTI_CHAR_FOLD_utf8_safe_part5(s,e) )
+    ( ( ( 0xCA == NATIVE_TO_LATIN1(((U8*)s)[1]) ) && ( 0xBE == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ) ? 3 : 0 )\
+: ( 0x66 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ?                               \
+    ( ( 0x66 == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ?                           \
+       ( ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[2]) || 0x6C == NATIVE_TO_LATIN1(((U8*)s)[2]) ) ? 3 : 2 )\
+    : ( 0x69 == NATIVE_TO_LATIN1(((U8*)s)[1]) || 0x6C == NATIVE_TO_LATIN1(((U8*)s)[1]) ) ? 2 : 0 )\
+: ( 0x68 == NATIVE_TO_LATIN1(((U8*)s)[0]) ) ? is_MULTI_CHAR_FOLD_utf8_part0(s) : is_MULTI_CHAR_FOLD_utf8_part1(s) )
 
 /*
        MULTI_CHAR_FOLD: multi-char strings that are folded to by a single character
@@ -840,18 +694,12 @@
        &regcharclass_multi_char_folds::multi_char_folds(0)
 */
 /*** GENERATED CODE ***/
-#define is_MULTI_CHAR_FOLD_latin1_safe(s,e)                                 \
-( ((e)-(s) > 2) ?                                                           \
-    ( ( ( NATIVE_TO_LATIN1(((U8*)s)[0]) & 0xDF ) == 0x46 ) ?                \
-       ( ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x46 ) ?            \
-           ( ( ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xDF ) == 0x49 ) || ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xDF ) == 0x4C ) ) ? 3 : 2 )\
-       : ( ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x49 ) || ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x4C ) ) ? 2 : 0 )\
-    : ( ( ( NATIVE_TO_LATIN1(((U8*)s)[0]) & 0xDF ) == 0x53 ) && ( ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x53 ) || ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x54 ) ) ) ? 2 : 0 )\
-: ((e)-(s) > 1) ?                                                           \
-    ( ( ( NATIVE_TO_LATIN1(((U8*)s)[0]) & 0xDF ) == 0x46 ) ?                \
-       ( ( ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x46 ) || ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x49 ) || ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x4C ) ) ? 2 : 0 )\
-    : ( ( ( NATIVE_TO_LATIN1(((U8*)s)[0]) & 0xDF ) == 0x53 ) && ( ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x53 ) || ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x54 ) ) ) ? 2 : 0 )\
-: 0 )
+#define is_MULTI_CHAR_FOLD_latin1(s)                                        \
+( ( ( NATIVE_TO_LATIN1(((U8*)s)[0]) & 0xDF ) == 0x46 ) ?                    \
+    ( ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x46 ) ?                \
+       ( ( ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xDF ) == 0x49 ) || ( ( NATIVE_TO_LATIN1(((U8*)s)[2]) & 0xDF ) == 0x4C ) ) ? 3 : 2 )\
+    : ( ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x49 ) || ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x4C ) ) ? 2 : 0 )\
+: ( ( ( NATIVE_TO_LATIN1(((U8*)s)[0]) & 0xDF ) == 0x53 ) && ( ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x53 ) || ( ( NATIVE_TO_LATIN1(((U8*)s)[1]) & 0xDF ) == 0x54 ) ) ) ? 2 : 0 )
 
 /*
        FOLDS_TO_MULTI: characters that fold to multi-char strings

diff --git a/regcomp.c b/regcomp.c
index 152cac9..6be379e 100644 (file)
--- a/regcomp.c
+++ b/regcomp.c
@@ -2104,7 +2104,7 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch,
                         foldlen -= UTF8SKIP(uc);
                     }
                     else {
-                        foldlen = is_MULTI_CHAR_FOLD_utf8_safe(uc, e);
+                        foldlen = is_MULTI_CHAR_FOLD_utf8(uc);
                         minbytes++;
                     }
                 }
@@ -2121,7 +2121,7 @@ S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch,
                         foldlen--;
                     }
                     else {
-                        foldlen = is_MULTI_CHAR_FOLD_latin1_safe(uc, e);
+                        foldlen = is_MULTI_CHAR_FOLD_latin1(uc);
                         minbytes++;
                     }
                 }
@@ -3288,7 +3288,7 @@ S_join_exact(pTHX_ RExC_state_t *pRExC_state, regnode *scan,
                                      length sequence we are looking for is 2 */
            {
                 int count = 0;  /* How many characters in a multi-char fold */
-                int len = is_MULTI_CHAR_FOLD_utf8_safe(s, s_end);
+                int len = is_MULTI_CHAR_FOLD_utf8(s);
                 if (! len) {    /* Not a multi-char fold: get next char */
                     s += UTF8SKIP(s);
                     continue;
@@ -3377,7 +3377,7 @@ S_join_exact(pTHX_ RExC_state_t *pRExC_state, regnode *scan,
            const U8* upper = (OP(scan) == EXACTF) ? s_end : s_end -1;
 
            while (s < upper) {
-                int len = is_MULTI_CHAR_FOLD_latin1_safe(s, s_end);
+                int len = is_MULTI_CHAR_FOLD_latin1(s);
                 if (! len) {    /* Not a multi-char fold. */
                     if (*s == LATIN_SMALL_LETTER_SHARP_S && OP(scan) == EXACTF)
                     {

diff --git a/regen/regcharclass.pl b/regen/regcharclass.pl
index d6e0010..8aee7c4 100755 (executable)
--- a/regen/regcharclass.pl
+++ b/regen/regcharclass.pl
@@ -1508,14 +1508,14 @@ QUOTEMETA: Meta-characters that \Q should quote
 \p{_Perl_Quotemeta}
 
 MULTI_CHAR_FOLD: multi-char strings that are folded to by a single character
-=> UTF8 :safe
+=> UTF8 :fast
 do regen/regcharclass_multi_char_folds.pl
 
 # 1 => All folds
 &regcharclass_multi_char_folds::multi_char_folds(1)
 
 MULTI_CHAR_FOLD: multi-char strings that are folded to by a single character
-=> LATIN1 :safe
+=> LATIN1 :fast
 
 &regcharclass_multi_char_folds::multi_char_folds(0)
 # 0 => Latin1-only

diff --git a/regexec.c b/regexec.c
index 4b586be..b8e709f 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -3445,12 +3445,8 @@ S_setup_EXACTISH_ST_c1_c2(pTHX_ const regnode * const text_node, int *c1p,
         }
     }
     else /* an EXACTFish node */
-         if ((is_utf8_pat
-                    && is_MULTI_CHAR_FOLD_utf8_safe(pat,
-                                                    pat + STR_LEN(text_node)))
-             || (!is_utf8_pat
-                    && is_MULTI_CHAR_FOLD_latin1_safe(pat,
-                                                    pat + STR_LEN(text_node))))
+        if ((is_utf8_pat && is_MULTI_CHAR_FOLD_utf8(pat))
+             || (!is_utf8_pat && is_MULTI_CHAR_FOLD_latin1(pat)))
     {
         /* Multi-character folds require more context to sort out.  Also
          * PL_utf8_foldclosures used below doesn't handle them, so have to be