EBCDIC: SHARP S is different.

p4raw-id: //depot/perl@14561

diff --git a/regexec.c b/regexec.c
index 830488a..8bd2284 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -924,8 +924,11 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
                STRLEN skip = do_utf8 ? UTF8SKIP(s) : 1;
 
                if (reginclass(c, (U8*)s, do_utf8) ||
-                   (ANYOF_UNICODE_FOLD_SHARP_S(c, s, strend) &&
-                    (skip = 2))) {
+                   (ANYOF_FOLD_SHARP_S(c, s, strend) &&
+                    /* The assignment of 2 is intentional:
+                     * for the sharp s, the skip is 2. */
+                    (skip = SHARP_S_SKIP)
+                    )) {
                    if (tmp && (norun || regtry(prog, s)))
                        goto got_it;
                    else
@@ -2439,8 +2442,8 @@ S_regmatch(pTHX_ regnode *prog)
            /* If we might have the case of the German sharp s
             * in a casefolding Unicode character class. */
 
-           if (ANYOF_UNICODE_FOLD_SHARP_S(scan, locinput, PL_regeol)) {
-                locinput += 2;
+           if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) {
+                locinput += SHARP_S_SKIP;
                 nextchr = UCHARAT(locinput);
            }
            else

diff --git a/utf8.h b/utf8.h
index c69cd86..49531a9 100644 (file)
--- a/utf8.h
+++ b/utf8.h
@@ -194,15 +194,28 @@ END_EXTERN_C
 #define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
 #define UNICODE_GREEK_SMALL_LETTER_SIGMA       0x03C3
 
+#define EBCDIC_LATIN_SMALL_LETTER_SHARP_S      0x0059
+
 #define UNI_DISPLAY_ISPRINT    0x0001
 #define UNI_DISPLAY_BACKSLASH  0x0002
 #define UNI_DISPLAY_QQ         (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
 #define UNI_DISPLAY_REGEX      (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
 
-#define ANYOF_UNICODE_FOLD_SHARP_S(node, input, end)   \
+#ifdef EBCDIC
+#   define ANYOF_FOLD_SHARP_S(node, input, end)        \
+       (ANYOF_BITMAP_TEST(node, EBCDIC_LATIN_SMALL_LETTER_SHARP_S) && \
+        (ANYOF_FLAGS(node) & ANYOF_UNICODE) && \
+        (ANYOF_FLAGS(node) & ANYOF_FOLD) && \
+        ((end) > (input) + 1) && \
+        toLOWER((input)[0]) == 's' && \
+        toLOWER((input)[1]) == 's')
+#else
+#   define ANYOF_FOLD_SHARP_S(node, input, end)        \
        (ANYOF_BITMAP_TEST(node, UNICODE_LATIN_SMALL_LETTER_SHARP_S) && \
         (ANYOF_FLAGS(node) & ANYOF_UNICODE) && \
         (ANYOF_FLAGS(node) & ANYOF_FOLD) && \
         ((end) > (input) + 1) && \
         toLOWER((input)[0]) == 's' && \
         toLOWER((input)[1]) == 's')
+#endif
+#define SHARP_S_SKIP 2