*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! */
*d++ = (( uv & 0x3f) | 0x80);
return d;
}
-#ifdef Quad_t
+#ifdef HAS_QUAD
{
*d++ = 0xff; /* Can't match U+FFFE! */
*d++ = (((uv >> 36) & 0x3f) | 0x80);
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;
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;
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;
}
}
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];
}
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)
{
}
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 */
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)
{