utf8.c revision 7c478bd95313f5f23a4c958a745db2134aa03244
/* utf8.c
*
* Copyright (c) 1998-2001, 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.
*
*/
/*
* 'What a fix!' said Sam. 'That's the one place in all the lands we've ever
* heard of that we don't want to see any closer; and that's the one place
* we're trying to get to! And that's just where we can't get, nohow.'
*
* 'Well do I understand your speech,' he answered in the same language;
* 'yet few strangers do so. Why then do you not speak in the Common Tongue,
* as is the custom in the West, if you wish to be answered?'
*
* ...the travellers perceived that the floor was paved with stones of many
* hues; branching runes and strange devices intertwined beneath their feet.
*/
#include "EXTERN.h"
#define PERL_IN_UTF8_C
#include "perl.h"
/* Unicode support */
/*
=for apidoc A|U8*|uv_to_utf8|U8 *d|UV uv
Adds the UTF8 representation of the Unicode codepoint C<uv> to the end
of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free
bytes available. The return value is the pointer to the byte after the
end of the new character. In other words,
d = uv_to_utf8(d, uv);
is the recommended Unicode-aware way of saying
*(d++) = uv;
=cut
*/
U8 *
{
if (uv < 0x80) {
*d++ = uv;
return d;
}
if (uv < 0x800) {
return d;
}
if (uv < 0x10000) {
return d;
}
if (uv < 0x200000) {
return d;
}
if (uv < 0x4000000) {
return d;
}
if (uv < 0x80000000) {
return d;
}
#ifdef HAS_QUAD
if (uv < UTF8_QUAD_MAX)
#endif
{
*d++ = 0xfe; /* Can't match U+FEFF! */
return d;
}
#ifdef HAS_QUAD
{
*d++ = 0xff; /* Can't match U+FFFE! */
*d++ = 0x80; /* 6 Reserved bits */
return d;
}
#endif
}
/*
=for apidoc A|STRLEN|is_utf8_char|U8 *s
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.
=cut
*/
{
U8 u = *s;
if (UTF8_IS_ASCII(u))
return 1;
if (!UTF8_IS_START(u))
return 0;
return 0;
s++;
uv = u;
while (slen--) {
if (!UTF8_IS_CONTINUATION(*s))
return 0;
return 0;
s++;
}
return 0;
return len;
}
/*
=for apidoc A|bool|is_utf8_string|U8 *s|STRLEN len
Returns true if first C<len> bytes of the given string form valid a UTF8
string, false otherwise.
=cut
*/
bool
{
U8* x = s;
STRLEN c;
if (!len)
while (x < send) {
c = is_utf8_char(x);
if (!c)
return FALSE;
x += c;
}
if (x != send)
return FALSE;
return TRUE;
}
/*
=for apidoc A|U8* s|utf8_to_uv|STRLEN curlen|STRLEN *retlen|U32 flags
Returns the character value of the first character in the string C<s>
which is assumed to be in UTF8 encoding and no longer than C<curlen>;
C<retlen> will be set to the length, in bytes, of that character.
If C<s> does not point to a well-formed UTF8 character, the behaviour
is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
it is assumed that the caller will raise a warning, and this function
will silently just set C<retlen> to C<-1> and return zero. If the
C<flags> does not contain UTF8_CHECK_ONLY, warnings about
malformations will be given, C<retlen> will be set to the expected
length of the UTF-8 character in bytes, and zero will be returned.
The C<flags> can also contain various flags to allow deviations from
the strict UTF-8 encoding (see F<utf8.h>).
=cut */
{
#ifdef EBCDIC
bool dowarn = 0;
#else
#endif
/* This list is a superset of the UTF8_ALLOW_XXX. */
#define UTF8_WARN_EMPTY 1
#define UTF8_WARN_CONTINUATION 2
#define UTF8_WARN_NON_CONTINUATION 3
#define UTF8_WARN_FE_FF 4
#define UTF8_WARN_SHORT 5
#define UTF8_WARN_OVERFLOW 6
#define UTF8_WARN_SURROGATE 7
#define UTF8_WARN_BOM 8
#define UTF8_WARN_LONG 9
#define UTF8_WARN_FFFF 10
if (curlen == 0 &&
!(flags & UTF8_ALLOW_EMPTY)) {
goto malformed;
}
if (UTF8_IS_ASCII(uv)) {
if (retlen)
*retlen = 1;
return *s;
}
if (UTF8_IS_CONTINUATION(uv) &&
!(flags & UTF8_ALLOW_CONTINUATION)) {
goto malformed;
}
!(flags & UTF8_ALLOW_NON_CONTINUATION)) {
goto malformed;
}
!(flags & UTF8_ALLOW_FE_FF)) {
goto malformed;
}
if (retlen)
!(flags & UTF8_ALLOW_SHORT)) {
goto malformed;
}
len--;
s++;
while (len--) {
if (!UTF8_IS_CONTINUATION(*s) &&
!(flags & UTF8_ALLOW_NON_CONTINUATION)) {
s--;
goto malformed;
}
else
/* These cannot be allowed. */
if (!(flags & UTF8_ALLOW_LONG)) {
goto malformed;
}
}
else { /* uv < ouv */
/* This cannot be allowed. */
goto malformed;
}
}
s++;
}
if (UNICODE_IS_SURROGATE(uv) &&
!(flags & UTF8_ALLOW_SURROGATE)) {
goto malformed;
} else if (UNICODE_IS_BYTE_ORDER_MARK(uv) &&
!(flags & UTF8_ALLOW_BOM)) {
goto malformed;
!(flags & UTF8_ALLOW_LONG)) {
goto malformed;
} else if (UNICODE_IS_ILLEGAL(uv) &&
!(flags & UTF8_ALLOW_FFFF)) {
goto malformed;
}
return uv;
if (flags & UTF8_CHECK_ONLY) {
if (retlen)
*retlen = -1;
return 0;
}
if (dowarn) {
switch (warning) {
case 0: /* Intentionally empty. */ break;
case UTF8_WARN_EMPTY:
break;
case UTF8_WARN_CONTINUATION:
break;
Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")",
break;
case UTF8_WARN_FE_FF:
break;
case UTF8_WARN_SHORT:
break;
case UTF8_WARN_OVERFLOW:
ouv, *s);
break;
case UTF8_WARN_SURROGATE:
break;
case UTF8_WARN_BOM:
break;
case UTF8_WARN_LONG:
break;
case UTF8_WARN_FFFF:
break;
default:
break;
}
if (warning) {
if (PL_op)
else
}
}
if (retlen)
return 0;
}
/*
=for apidoc A|U8* s|utf8_to_uv_simple|STRLEN *retlen
Returns the character value of the first character in the string C<s>
which is assumed to be in UTF8 encoding; C<retlen> will be set to the
length, in bytes, of that character.
If C<s> does not point to a well-formed UTF8 character, zero is
returned and retlen is set, if possible, to -1.
=cut
*/
{
}
/*
=for apidoc A|STRLEN|utf8_length|U8* s|U8 *e
Return the length of the UTF-8 char encoded string C<s> in characters.
Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
up past C<e>, croaks.
=cut
*/
{
/* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
* the bitops (especially ~) can create illegal UTF-8.
* In other words: in Perl UTF-8 is not just for Unicode. */
if (e < s)
while (s < e) {
if (e - s < t)
s += t;
len++;
}
return len;
}
/*
=for apidoc A|IV|utf8_distance|U8 *a|U8 *b
Returns the number of UTF8 characters between the UTF-8 pointers C<a>
and C<b>.
WARNING: use only if you *know* that the pointers point inside the
same UTF-8 buffer.
=cut */
{
/* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
* the bitops (especially ~) can create illegal UTF-8.
* In other words: in Perl UTF-8 is not just for Unicode. */
if (a < b) {
while (a < b) {
if (b - a < c)
a += c;
off--;
}
}
else {
while (b < a) {
if (a - b < c)
b += c;
off++;
}
}
return off;
}
/*
=for apidoc A|U8*|utf8_hop|U8 *s|I32 off
Return the UTF-8 pointer C<s> displaced by C<off> characters, either
forward or backward.
WARNING: do not use the following unless you *know* C<off> is within
the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
on the first byte of character or just after the last byte of a character.
=cut */
U8 *
{
/* Note: cannot use UTF8_IS_...() too eagerly here since e.g
* the bitops (especially ~) can create illegal UTF-8.
* In other words: in Perl UTF-8 is not just for Unicode. */
if (off >= 0) {
while (off--)
s += UTF8SKIP(s);
}
else {
while (off++) {
s--;
while (UTF8_IS_CONTINUATION(*s))
s--;
}
}
return s;
}
/*
=for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
Converts a string C<s> of length C<len> from UTF8 into byte encoding.
Unlike C<bytes_to_utf8>, this over-writes the original string, and
updates len to contain the new length.
Returns zero on failure, setting C<len> to -1.
=cut
*/
U8 *
{
U8 *d;
/* ensure valid UTF8 and chars < 256 before updating string */
U8 c = *s++;
if (c >= 0x80 &&
((s >= send) ||
((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
*len = -1;
return 0;
}
}
d = s = save;
while (s < send) {
if (UTF8_IS_ASCII(*s)) {
*d++ = *s++;
}
else {
s += ulen;
}
}
*d = '\0';
return save;
}
/*
=for apidoc A|U8 *|bytes_from_utf8|U8 *s|STRLEN *len|bool *is_utf8
Converts a string C<s> of length C<len> from UTF8 into byte encoding.
Unlike <utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
the newly-created string, and updates C<len> to contain the new
length. Returns the original string if no conversion occurs, C<len>
is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
0 if C<s> is converted or contains all 7bit characters.
=cut */
U8 *
{
U8 *d;
if (!*is_utf8)
return start;
/* ensure valid UTF8 and chars < 256 before converting string */
U8 c = *s++;
if (!UTF8_IS_ASCII(c)) {
if (UTF8_IS_CONTINUATION(c) || s >= send ||
!UTF8_IS_CONTINUATION(*s) || UTF8_IS_DOWNGRADEABLE_START(c))
return start;
s++, count++;
}
}
*is_utf8 = 0;
if (!count)
return start;
while (s < send) {
U8 c = *s++;
if (UTF8_IS_ASCII(c))
*d++ = c;
else
*d++ = UTF8_ACCUMULATE(c, *s++);
}
*d = '\0';
return start;
}
/*
=for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
Returns a pointer to the newly-created string, and sets C<len> to
reflect the new length.
=cut
*/
U8*
{
U8 *d;
dst = d;
while (s < send) {
if (UTF8_IS_ASCII(*s))
*d++ = *s++;
else {
*d++ = UTF8_EIGHT_BIT_HI(uv);
*d++ = UTF8_EIGHT_BIT_LO(uv);
}
}
*d = '\0';
return dst;
}
/*
* Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
*
* Destination must be pre-extended to 3/2 source. Do not use in-place.
* We optimize for native, for obvious reasons. */
U8*
{
if (bytelen & 1)
while (p < pend) {
p += 2;
if (uv < 0x80) {
*d++ = uv;
continue;
}
if (uv < 0x800) {
continue;
}
}
if (uv < 0x10000) {
continue;
}
else {
continue;
}
}
return d;
}
/* Note: this one is slightly destructive of the source. */
U8*
{
while (s < send) {
s[0] = s[1];
s[1] = tmp;
s += 2;
}
}
/* for now these are all defined (inefficiently) in terms of the utf8 versions */
bool
{
return is_utf8_alnum(tmpbuf);
}
bool
{
return is_utf8_alnumc(tmpbuf);
}
bool
{
return is_utf8_idfirst(tmpbuf);
}
bool
{
return is_utf8_alpha(tmpbuf);
}
bool
{
return is_utf8_ascii(tmpbuf);
}
bool
{
return is_utf8_space(tmpbuf);
}
bool
{
return is_utf8_digit(tmpbuf);
}
bool
{
return is_utf8_upper(tmpbuf);
}
bool
{
return is_utf8_lower(tmpbuf);
}
bool
{
return is_utf8_cntrl(tmpbuf);
}
bool
{
return is_utf8_graph(tmpbuf);
}
bool
{
return is_utf8_print(tmpbuf);
}
bool
{
return is_utf8_punct(tmpbuf);
}
bool
{
return is_utf8_xdigit(tmpbuf);
}
{
return to_utf8_upper(tmpbuf);
}
{
return to_utf8_title(tmpbuf);
}
{
return to_utf8_lower(tmpbuf);
}
/* for now these all assume no locale info available for Unicode > 255 */
bool
{
return is_uni_alnum(c); /* XXX no locale support yet */
}
bool
{
return is_uni_alnumc(c); /* XXX no locale support yet */
}
bool
{
return is_uni_idfirst(c); /* XXX no locale support yet */
}
bool
{
return is_uni_alpha(c); /* XXX no locale support yet */
}
bool
{
return is_uni_ascii(c); /* XXX no locale support yet */
}
bool
{
return is_uni_space(c); /* XXX no locale support yet */
}
bool
{
return is_uni_digit(c); /* XXX no locale support yet */
}
bool
{
return is_uni_upper(c); /* XXX no locale support yet */
}
bool
{
return is_uni_lower(c); /* XXX no locale support yet */
}
bool
{
return is_uni_cntrl(c); /* XXX no locale support yet */
}
bool
{
return is_uni_graph(c); /* XXX no locale support yet */
}
bool
{
return is_uni_print(c); /* XXX no locale support yet */
}
bool
{
return is_uni_punct(c); /* XXX no locale support yet */
}
bool
{
return is_uni_xdigit(c); /* XXX no locale support yet */
}
{
return to_uni_upper(c); /* XXX no locale support yet */
}
{
return to_uni_title(c); /* XXX no locale support yet */
}
{
return to_uni_lower(c); /* XXX no locale support yet */
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_alnum)
/* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
* descendant of isalnum(3), in other words, it doesn't
* contain the '_'. --jhi */
return swash_fetch(PL_utf8_alnum, p);
/* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
#ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
if (!PL_utf8_alnum)
return swash_fetch(PL_utf8_alnum, p);
#endif
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_alnum)
return swash_fetch(PL_utf8_alnum, p);
/* return is_utf8_alpha(p) || is_utf8_digit(p); */
#ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
if (!PL_utf8_alnum)
return swash_fetch(PL_utf8_alnum, p);
#endif
}
bool
{
return *p == '_' || is_utf8_alpha(p);
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_alpha)
return swash_fetch(PL_utf8_alpha, p);
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_ascii)
return swash_fetch(PL_utf8_ascii, p);
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_space)
return swash_fetch(PL_utf8_space, p);
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_digit)
return swash_fetch(PL_utf8_digit, p);
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_upper)
return swash_fetch(PL_utf8_upper, p);
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_lower)
return swash_fetch(PL_utf8_lower, p);
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_cntrl)
return swash_fetch(PL_utf8_cntrl, p);
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_graph)
return swash_fetch(PL_utf8_graph, p);
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_print)
return swash_fetch(PL_utf8_print, p);
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_punct)
return swash_fetch(PL_utf8_punct, p);
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_xdigit)
return swash_fetch(PL_utf8_xdigit, p);
}
bool
{
if (!is_utf8_char(p))
return FALSE;
if (!PL_utf8_mark)
return swash_fetch(PL_utf8_mark, p);
}
{
if (!PL_utf8_toupper)
}
{
if (!PL_utf8_totitle)
}
{
if (!PL_utf8_tolower)
}
/* a "swash" is a swatch hash */
SV*
{
dSP;
}
PL_hints = 0;
if (PL_curcop == &PL_compiling)
/* XXX ought to be handled by lex_start */
else
retval = &PL_sv_undef;
if (PL_curcop == &PL_compiling) {
}
return retval;
}
{
/*
* This single-entry cache saves about 1/3 of the utf8 overhead in test
* suite. (That is, only 7-8% overall over just a hash cache. Still,
* it's nothing to sniff at.) Pity we usually come through at least
* two function calls to get here...
*
* NB: this code assumes that swatches are never modified, once generated!
*/
if (hv == PL_last_swash_hv &&
klen == PL_last_swash_klen &&
{
}
else {
/* Try our second-level swatch cache, kept in a hash. */
/* If not cached, generate it via utf8::SWASHGET */
dSP;
else
retval = &PL_sv_undef;
if (PL_curcop == &PL_compiling)
}
if (klen)
}
case 1:
off >>= 3;
case 8:
case 16:
off <<= 1;
case 32:
off <<= 2;
}
return 0;
}