After much rewriting we are now pretty much
Jarkko Hietaniemi [Mon, 18 Feb 2002 02:24:31 +0000 (02:24 +0000)]
back to where we started.

p4raw-id: //depot/perl@14737

utf8.c

diff --git a/utf8.c b/utf8.c
index b33e3db..fa562fe 100644 (file)
--- a/utf8.c
+++ b/utf8.c
@@ -1288,10 +1288,9 @@ UV
 Perl_to_utf8_case(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp, char *normal, char *special)
 {
     UV uv0, uv1, uv2;
-    U8 tmpbuf[UTF8_MAXLEN_FOLD+1], *d;
+    U8 tmpbuf[UTF8_MAXLEN_FOLD+1];
     char *s = NULL;
     STRLEN len;
-    bool has_utf8 = FALSE;
 
     if (!*swashp)
         *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
@@ -1304,8 +1303,9 @@ Perl_to_utf8_case(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp, char *norma
     uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
     if (uv2) {
         /* It was "normal" (a single character mapping). */
-         d = uvuni_to_utf8(ustrp, uv2);
-        has_utf8 = !UNI_IS_INVARIANT(uv2);
+         UV uv3 = UNI_TO_NATIVE(uv2);
+       
+         len = uvuni_to_utf8(ustrp, uv2) - ustrp;
     }
     else {
         /* It might be "special" (sometimes, but not always,
@@ -1314,61 +1314,58 @@ Perl_to_utf8_case(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp, char *norma
         SV *keysv;
         HE *he;
         SV *val;
-
+       
         if ((hv    = get_hv(special, FALSE)) &&
             (keysv = sv_2mortal(Perl_newSVpvf(aTHX_ "%04"UVXf, uv1))) &&
             (he    = hv_fetch_ent(hv, keysv, FALSE, 0)) &&
             (val   = HeVAL(he))) {
-
+              U8* d;
+       
              s = SvPV(val, len);
-             if (len == 1)
+             if (len == 1) {
                   d = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s));
+                  len = d - ustrp;
+              }
              else {
+#ifdef EBCDIC
+                  /* If we have EBCDIC we need to remap the characters
+                   * since any characters in the low 256 are Unicode
+                   * code points, not EBCDIC. */
+                  U8 *t = (U8*)s, *tend = t + len;
+               
+                  d = tmpbuf;
+                  if (SvUTF8(val)) {
+                       STRLEN tlen = 0;
+                       
+                       while (t < tend) {
+                            UV c = utf8_to_uvchr(t, &tlen);
+                            if (tlen > 0) {
+                                 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
+                                 t += tlen;
+                            }
+                            else
+                                 break;
+                       }
+                  }
+                  else {
+                       while (t < tend) {
+                            d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
+                            t++;
+                       }
+                  }
+                  len = d - tmpbuf;
+                  Copy(tmpbuf, ustrp, len, U8);
+#else
                   Copy(s, ustrp, len, U8);
-                  d = ustrp + len;
              }
-             if (SvUTF8(val))
-                  has_utf8 = TRUE;
+#endif
         }
         else {
              /* It was not "special", either. */
-             d = uvuni_to_utf8(ustrp, uv1);
-             has_utf8 = !UNI_IS_INVARIANT(uv1);
+             len = uvchr_to_utf8(ustrp, uv0) - ustrp;
         }
     }
 
-    len = d - ustrp;
-
-#ifdef EBCDIC
-    {
-        /* If we have EBCDIC we need to remap the characters since
-         * any characters in the low 256 are in Unicode code points,
-         * not EBCDIC. */
-        U8 *t, *tend;
-        
-        d = tmpbuf;
-        if (has_utf8) {
-             STRLEN tlen = 0;
-             
-             for (t = ustrp, tend = t + len;
-                  t < tend; t += tlen) {
-                  UV c = utf8_to_uvchr(t, &tlen);
-                  
-                  if (tlen > 0)
-                       d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
-                  else
-                       break;
-             }
-        } else {
-             for (t = ustrp, tend = t + len;
-                  t < tend; t++)
-                  d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
-        }
-        len = d - tmpbuf; 
-        Copy(tmpbuf, ustrp, len, U8);
-    }
-#endif
-
     if (lenp)
         *lenp = len;