Integrate with Sarathy.

[p5sagit/p5-mst-13.2.git] / utf8.c

diff --git a/utf8.c b/utf8.c
index 0e52f21..b14fafe 100644 (file)
--- a/utf8.c
+++ b/utf8.c
@@ -1,6 +1,6 @@
 /*    utf8.c
  *
- *    Copyright (c) 1998-1999, Larry Wall
+ *    Copyright (c) 1998-2000, Larry Wall
  *
  *    You may distribute under the terms of either the GNU General Public
  *    License or the Artistic License, as specified in the README file.
@@ -68,8 +68,8 @@ Perl_uv_to_utf8(pTHX_ U8 *d, UV uv)
        *d++ = (( uv        & 0x3f) | 0x80);
        return d;
     }
-#ifdef Quad_t
-    if (uv < 0x2000000000)
+#ifdef HAS_QUAD
+    if (uv < 0x1000000000LL)
 #endif
     {
        *d++ =                        0xfe;     /* Can't match U+FEFF! */
@@ -81,9 +81,14 @@ Perl_uv_to_utf8(pTHX_ U8 *d, UV uv)
        *d++ = (( uv        & 0x3f) | 0x80);
        return d;
     }
-#ifdef Quad_t
+#ifdef HAS_QUAD
     {
        *d++ =                        0xff;     /* Can't match U+FFFE! */
+       *d++ =                        0x80;     /* 6 Reserved bits */
+       *d++ = (((uv >> 60) & 0x0f) | 0x80);    /* 2 Reserved bits */
+       *d++ = (((uv >> 54) & 0x3f) | 0x80);
+       *d++ = (((uv >> 48) & 0x3f) | 0x80);
+       *d++ = (((uv >> 42) & 0x3f) | 0x80);
        *d++ = (((uv >> 36) & 0x3f) | 0x80);
        *d++ = (((uv >> 30) & 0x3f) | 0x80);
        *d++ = (((uv >> 24) & 0x3f) | 0x80);
@@ -107,7 +112,9 @@ Perl_utf8_to_uv(pTHX_ U8* s, I32* retlen)
        return *s;
     }
     if (!(uv & 0x40)) {
-       Perl_warn(aTHX_ "Malformed UTF-8 character");
+        dTHR;
+       if (ckWARN_d(WARN_UTF8))     
+           Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
        if (retlen)
            *retlen = 1;
        return *s;
@@ -118,8 +125,8 @@ Perl_utf8_to_uv(pTHX_ U8* s, I32* retlen)
     else if (!(uv & 0x08))     { len = 4; uv &= 0x07; }
     else if (!(uv & 0x04))     { len = 5; uv &= 0x03; }
     else if (!(uv & 0x02))     { len = 6; uv &= 0x01; }
-    else if (!(uv & 0x01))     { len = 7; uv &= 0x00; }
-    else                         len = 8;      /* whoa! */
+    else if (!(uv & 0x01))     { len = 7;  uv = 0; }
+    else                       { len = 13; uv = 0; } /* whoa! */
 
     if (retlen)
        *retlen = len;
@@ -127,7 +134,9 @@ Perl_utf8_to_uv(pTHX_ U8* s, I32* retlen)
     s++;
     while (len--) {
        if ((*s & 0xc0) != 0x80) {
-           Perl_warn(aTHX_ "Malformed UTF-8 character");
+            dTHR;
+           if (ckWARN_d(WARN_UTF8))     
+               Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
            if (retlen)
                *retlen -= len + 1;
            return 0xfffd;
@@ -203,9 +212,11 @@ Perl_utf16_to_utf8(pTHX_ U16* p, U8* d, I32 bytelen)
            continue;
        }
        if (uv >= 0xd800 && uv < 0xdbff) {      /* surrogates */
+            dTHR;
            int low = *p++;
            if (low < 0xdc00 || low >= 0xdfff) {
-               Perl_warn(aTHX_ "Malformed UTF-16 surrogate");
+               if (ckWARN_d(WARN_UTF8))     
+                   Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-16 surrogate");
                p--;
                uv = 0xfffd;
            }
@@ -279,6 +290,14 @@ Perl_is_uni_alpha(pTHX_ U32 c)
 }
 
 bool
+Perl_is_uni_ascii(pTHX_ U32 c)
+{
+    U8 tmpbuf[10];
+    uv_to_utf8(tmpbuf, (UV)c);
+    return is_utf8_ascii(tmpbuf);
+}
+
+bool
 Perl_is_uni_space(pTHX_ U32 c)
 {
     U8 tmpbuf[10];
@@ -335,13 +354,21 @@ Perl_is_uni_print(pTHX_ U32 c)
 }
 
 bool
-is_uni_punct(U32 c)
+Perl_is_uni_punct(pTHX_ U32 c)
 {
     U8 tmpbuf[10];
     uv_to_utf8(tmpbuf, (UV)c);
     return is_utf8_punct(tmpbuf);
 }
 
+bool
+Perl_is_uni_xdigit(pTHX_ U32 c)
+{
+    U8 tmpbuf[10];
+    uv_to_utf8(tmpbuf, (UV)c);
+    return is_utf8_xdigit(tmpbuf);
+}
+
 U32
 Perl_to_uni_upper(pTHX_ U32 c)
 {
@@ -393,6 +420,12 @@ Perl_is_uni_alpha_lc(pTHX_ U32 c)
 }
 
 bool
+Perl_is_uni_ascii_lc(pTHX_ U32 c)
+{
+    return is_uni_ascii(c);    /* XXX no locale support yet */
+}
+
+bool
 Perl_is_uni_space_lc(pTHX_ U32 c)
 {
     return is_uni_space(c);    /* XXX no locale support yet */
@@ -440,6 +473,12 @@ Perl_is_uni_punct_lc(pTHX_ U32 c)
     return is_uni_punct(c);    /* XXX no locale support yet */
 }
 
+bool
+Perl_is_uni_xdigit_lc(pTHX_ U32 c)
+{
+    return is_uni_xdigit(c);   /* XXX no locale support yet */
+}
+
 U32
 Perl_to_uni_upper_lc(pTHX_ U32 c)
 {