/* 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.
#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);
#endif
}
+/* Tests if some arbitrary number of bytes begins in a valid UTF-8 character.
+ * The actual number of bytes in the UTF-8 character will be returned if it
+ * is valid, otherwise 0. */
+int
+Perl_is_utf8_char(pTHX_ U8 *s)
+{
+ U8 u = *s;
+ int slen, len;
+
+ if (!(u & 0x80))
+ return 1;
+
+ if (!(u & 0x40))
+ return 0;
+
+ if (!(u & 0x20)) { len = 2; }
+ else if (!(u & 0x10)) { len = 3; }
+ else if (!(u & 0x08)) { len = 4; }
+ else if (!(u & 0x04)) { len = 5; }
+ else if (!(u & 0x02)) { len = 6; }
+ else if (!(u & 0x01)) { len = 7; }
+ else { len = 13; } /* whoa! */
+
+ slen = len - 1;
+ s++;
+ while (slen--) {
+ if ((*s & 0xc0) != 0x80)
+ return 0;
+ s++;
+ }
+ return len;
+}
+
UV
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;
bool
Perl_is_uni_alnum(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_alnum(tmpbuf);
}
bool
Perl_is_uni_alnumc(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_alnumc(tmpbuf);
}
bool
Perl_is_uni_idfirst(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_idfirst(tmpbuf);
}
bool
Perl_is_uni_alpha(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_alpha(tmpbuf);
}
bool
Perl_is_uni_ascii(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_ascii(tmpbuf);
}
bool
Perl_is_uni_space(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_space(tmpbuf);
}
bool
Perl_is_uni_digit(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_digit(tmpbuf);
}
bool
Perl_is_uni_upper(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_upper(tmpbuf);
}
bool
Perl_is_uni_lower(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_lower(tmpbuf);
}
bool
Perl_is_uni_cntrl(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_cntrl(tmpbuf);
}
bool
Perl_is_uni_graph(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_graph(tmpbuf);
}
bool
Perl_is_uni_print(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_print(tmpbuf);
}
bool
Perl_is_uni_punct(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_punct(tmpbuf);
}
bool
Perl_is_uni_xdigit(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return is_utf8_xdigit(tmpbuf);
}
U32
Perl_to_uni_upper(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return to_utf8_upper(tmpbuf);
}
U32
Perl_to_uni_title(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return to_utf8_title(tmpbuf);
}
U32
Perl_to_uni_lower(pTHX_ U32 c)
{
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
uv_to_utf8(tmpbuf, (UV)c);
return to_utf8_lower(tmpbuf);
}
bool
Perl_is_utf8_alnum(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_alnum)
PL_utf8_alnum = swash_init("utf8", "IsAlnum", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_alnum, p);
bool
Perl_is_utf8_alnumc(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_alnum)
PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_alnum, p);
bool
Perl_is_utf8_alpha(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_alpha)
PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_alpha, p);
bool
Perl_is_utf8_ascii(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_ascii)
PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_ascii, p);
bool
Perl_is_utf8_space(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_space)
PL_utf8_space = swash_init("utf8", "IsSpace", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_space, p);
bool
Perl_is_utf8_digit(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_digit)
PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_digit, p);
bool
Perl_is_utf8_upper(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_upper)
PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_upper, p);
bool
Perl_is_utf8_lower(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_lower)
PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_lower, p);
bool
Perl_is_utf8_cntrl(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_cntrl)
PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_cntrl, p);
bool
Perl_is_utf8_graph(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_graph)
PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_graph, p);
bool
Perl_is_utf8_print(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_print)
PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_print, p);
bool
Perl_is_utf8_punct(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_punct)
PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_punct, p);
bool
Perl_is_utf8_xdigit(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_xdigit)
PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_xdigit, p);
bool
Perl_is_utf8_mark(pTHX_ U8 *p)
{
+ if (!is_utf8_char(p))
+ return FALSE;
if (!PL_utf8_mark)
PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
return swash_fetch(PL_utf8_mark, p);
SV* retval;
char tmpbuf[256];
dSP;
+
+ if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
+ ENTER;
+ Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
+ LEAVE;
+ }
+ SPAGAIN;
PUSHSTACKi(PERLSI_MAGIC);
PUSHMARK(SP);
EXTEND(SP,5);