3 * Copyright (c) 1998-2000, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * 'What a fix!' said Sam. 'That's the one place in all the lands we've ever
12 * heard of that we don't want to see any closer; and that's the one place
13 * we're trying to get to! And that's just where we can't get, nohow.'
15 * 'Well do I understand your speech,' he answered in the same language;
16 * 'yet few strangers do so. Why then do you not speak in the Common Tongue,
17 * as is the custom in the West, if you wish to be answered?'
19 * ...the travellers perceived that the floor was paved with stones of many
20 * hues; branching runes and strange devices intertwined beneath their feet.
24 #define PERL_IN_UTF8_C
30 Perl_uv_to_utf8(pTHX_ U8 *d, UV uv) /* the d must be UTF8_MAXLEN+1 deep */
38 *d++ = (( uv >> 6) | 0xc0);
39 *d++ = (( uv & 0x3f) | 0x80);
44 *d++ = (( uv >> 12) | 0xe0);
45 *d++ = (((uv >> 6) & 0x3f) | 0x80);
46 *d++ = (( uv & 0x3f) | 0x80);
51 *d++ = (( uv >> 18) | 0xf0);
52 *d++ = (((uv >> 12) & 0x3f) | 0x80);
53 *d++ = (((uv >> 6) & 0x3f) | 0x80);
54 *d++ = (( uv & 0x3f) | 0x80);
59 *d++ = (( uv >> 24) | 0xf8);
60 *d++ = (((uv >> 18) & 0x3f) | 0x80);
61 *d++ = (((uv >> 12) & 0x3f) | 0x80);
62 *d++ = (((uv >> 6) & 0x3f) | 0x80);
63 *d++ = (( uv & 0x3f) | 0x80);
67 if (uv < 0x80000000) {
68 *d++ = (( uv >> 30) | 0xfc);
69 *d++ = (((uv >> 24) & 0x3f) | 0x80);
70 *d++ = (((uv >> 18) & 0x3f) | 0x80);
71 *d++ = (((uv >> 12) & 0x3f) | 0x80);
72 *d++ = (((uv >> 6) & 0x3f) | 0x80);
73 *d++ = (( uv & 0x3f) | 0x80);
78 if (uv < UTF8_QUAD_MAX)
81 *d++ = 0xfe; /* Can't match U+FEFF! */
82 *d++ = (((uv >> 30) & 0x3f) | 0x80);
83 *d++ = (((uv >> 24) & 0x3f) | 0x80);
84 *d++ = (((uv >> 18) & 0x3f) | 0x80);
85 *d++ = (((uv >> 12) & 0x3f) | 0x80);
86 *d++ = (((uv >> 6) & 0x3f) | 0x80);
87 *d++ = (( uv & 0x3f) | 0x80);
93 *d++ = 0xff; /* Can't match U+FFFE! */
94 *d++ = 0x80; /* 6 Reserved bits */
95 *d++ = (((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
96 *d++ = (((uv >> 54) & 0x3f) | 0x80);
97 *d++ = (((uv >> 48) & 0x3f) | 0x80);
98 *d++ = (((uv >> 42) & 0x3f) | 0x80);
99 *d++ = (((uv >> 36) & 0x3f) | 0x80);
100 *d++ = (((uv >> 30) & 0x3f) | 0x80);
101 *d++ = (((uv >> 24) & 0x3f) | 0x80);
102 *d++ = (((uv >> 18) & 0x3f) | 0x80);
103 *d++ = (((uv >> 12) & 0x3f) | 0x80);
104 *d++ = (((uv >> 6) & 0x3f) | 0x80);
105 *d++ = (( uv & 0x3f) | 0x80);
112 /* Tests if some arbitrary number of bytes begins in a valid UTF-8 character.
113 * The actual number of bytes in the UTF-8 character will be returned if it
114 * is valid, otherwise 0. */
116 Perl_is_utf8_char(pTHX_ U8 *s)
125 if (u >= 0x80 && u <= 0xbf)
130 if (len < 2 || (u >= 0xc0 && u <= 0xfd && s[1] < 0x80))
138 if ((*s & 0xc0) != 0x80)
140 uv = (uv << 6) | (*s & 0x3f);
147 if (UNISKIP(uv) < len)
154 =for apidoc Am|is_utf8_string|U8 *s|STRLEN len
156 Returns true if first C<len> bytes of the given string form valid a UTF8
157 string, false otherwise.
163 Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len)
182 =for apidoc Am|U8* s|utf8_to_uv|STRLEN curlen|STRLEN *retlen|U32 flags
184 Returns the character value of the first character in the string C<s>
185 which is assumed to be in UTF8 encoding and no longer than C<curlen>;
186 C<retlen> will be set to the length, in bytes, of that character,
187 and the pointer C<s> will be advanced to the end of the character.
189 If C<s> does not point to a well-formed UTF8 character, the behaviour
190 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
191 it is assumed that the caller will raise a warning, and this function
192 will set C<retlen> to C<-1> and return. The C<flags> can also contain
193 various flags to allow deviations from the strict UTF-8 encoding
199 Perl_utf8_to_uv(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags)
207 bool dowarn = ckWARN_d(WARN_UTF8);
209 STRLEN expectlen = 0;
213 Perl_warner(aTHX_ WARN_UTF8,
214 "Malformed UTF-8 character (an empty string)");
218 if (uv <= 0x7f) { /* Pure ASCII. */
224 if ((uv >= 0x80 && uv <= 0xbf) &&
225 !(flags & UTF8_ALLOW_CONTINUATION)) {
227 Perl_warner(aTHX_ WARN_UTF8,
228 "Malformed UTF-8 character (unexpected continuation byte 0x%02"UVxf")",
233 if ((uv >= 0xc0 && uv <= 0xfd && curlen > 1 && s[1] < 0x80) &&
234 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
236 Perl_warner(aTHX_ WARN_UTF8,
237 "Malformed UTF-8 character (unexpected non-continuation byte 0x%02"UVxf" after byte 0x%02"UVxf")",
242 if ((uv == 0xfe || uv == 0xff) &&
243 !(flags & UTF8_ALLOW_FE_FF)) {
245 Perl_warner(aTHX_ WARN_UTF8,
246 "Malformed UTF-8 character (byte 0x%02"UVxf")",
251 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
252 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
253 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
254 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
255 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
256 else if (!(uv & 0x01)) { len = 7; uv = 0; }
257 else { len = 13; uv = 0; } /* whoa! */
264 if ((curlen < expectlen) &&
265 !(flags & UTF8_ALLOW_SHORT)) {
267 Perl_warner(aTHX_ WARN_UTF8,
268 "Malformed UTF-8 character (%d byte%s, need %d)",
269 curlen, curlen == 1 ? "" : "s", expectlen);
278 if ((*s & 0xc0) != 0x80) {
280 Perl_warner(aTHX_ WARN_UTF8,
281 "Malformed UTF-8 character (unexpected continuation byte 0x%02x)",
286 uv = (uv << 6) | (*s & 0x3f);
288 /* This cannot be allowed. */
290 Perl_warner(aTHX_ WARN_UTF8,
291 "Malformed UTF-8 character (overflow at 0x%"UVxf", byte 0x%02x)",
299 if ((uv >= 0xd800 && uv <= 0xdfff) &&
300 !(flags & UTF8_ALLOW_SURROGATE)) {
302 Perl_warner(aTHX_ WARN_UTF8,
303 "Malformed UTF-8 character (UTF-16 surrogate 0x%04"UVxf")",
306 } else if ((uv == 0xfffe) &&
307 !(flags & UTF8_ALLOW_BOM)) {
309 Perl_warner(aTHX_ WARN_UTF8,
310 "Malformed UTF-8 character (byte order mark 0x%04"UVxf")",
313 } else if ((uv == 0xffff) &&
314 !(flags & UTF8_ALLOW_FFFF)) {
316 Perl_warner(aTHX_ WARN_UTF8,
317 "Malformed UTF-8 character (character 0x%04"UVxf")",
320 } else if ((expectlen > UNISKIP(uv)) &&
321 !(flags & UTF8_ALLOW_LONG)) {
323 Perl_warner(aTHX_ WARN_UTF8,
324 "Malformed UTF-8 character (%d byte%s, need %d)",
325 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
333 if (flags & UTF8_CHECK_ONLY) {
340 *retlen = expectlen ? expectlen : len;
342 return UNICODE_REPLACEMENT_CHARACTER;
346 =for apidoc Am|U8* s|utf8_to_uv_simple|STRLEN *retlen
348 Returns the character value of the first character in the string C<s>
349 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
350 length, in bytes, of that character, and the pointer C<s> will be
351 advanced to the end of the character.
353 If C<s> does not point to a well-formed UTF8 character, zero is
354 returned and retlen is set, if possible, to -1.
360 Perl_utf8_to_uv_simple(pTHX_ U8* s, STRLEN* retlen)
362 return Perl_utf8_to_uv(aTHX_ s, UTF8_MAXLEN, retlen, 0);
366 =for apidoc|utf8_length|U8 *s|U8 *e
368 Return the length of the UTF-8 char encoded string C<s> in characters.
369 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
370 up past C<e>, croaks.
376 Perl_utf8_length(pTHX_ U8* s, U8* e)
381 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
386 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
394 /* utf8_distance(a,b) returns the number of UTF8 characters between
395 the pointers a and b */
398 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
407 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
417 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
426 /* WARNING: do not use the following unless you *know* off is within bounds */
429 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
439 while ((*s & 0xc0) == 0x80)
448 =for apidoc Am|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
450 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
451 Unlike C<bytes_to_utf8>, this over-writes the original string, and
452 updates len to contain the new length.
453 Returns zero on failure, setting C<len> to -1.
459 Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len)
465 /* ensure valid UTF8 and chars < 256 before updating string */
466 for (send = s + *len; s < send; ) {
471 ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
484 *d++ = (U8)utf8_to_uv_simple(s, &ulen);
494 =for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
496 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
497 Returns a pointer to the newly-created string, and sets C<len> to
498 reflect the new length.
504 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
512 Newz(801, d, (*len) * 2 + 1, U8);
520 *d++ = (( uv >> 6) | 0xc0);
521 *d++ = (( uv & 0x3f) | 0x80);
530 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
532 * Destination must be pre-extended to 3/2 source. Do not use in-place.
533 * We optimize for native, for obvious reasons. */
536 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
542 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
547 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
554 *d++ = (( uv >> 6) | 0xc0);
555 *d++ = (( uv & 0x3f) | 0x80);
558 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
561 if (low < 0xdc00 || low >= 0xdfff)
562 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
563 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
566 *d++ = (( uv >> 12) | 0xe0);
567 *d++ = (((uv >> 6) & 0x3f) | 0x80);
568 *d++ = (( uv & 0x3f) | 0x80);
572 *d++ = (( uv >> 18) | 0xf0);
573 *d++ = (((uv >> 12) & 0x3f) | 0x80);
574 *d++ = (((uv >> 6) & 0x3f) | 0x80);
575 *d++ = (( uv & 0x3f) | 0x80);
579 *newlen = d - dstart;
583 /* Note: this one is slightly destructive of the source. */
586 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
589 U8* send = s + bytelen;
596 return utf16_to_utf8(p, d, bytelen, newlen);
599 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
602 Perl_is_uni_alnum(pTHX_ U32 c)
604 U8 tmpbuf[UTF8_MAXLEN+1];
605 uv_to_utf8(tmpbuf, (UV)c);
606 return is_utf8_alnum(tmpbuf);
610 Perl_is_uni_alnumc(pTHX_ U32 c)
612 U8 tmpbuf[UTF8_MAXLEN+1];
613 uv_to_utf8(tmpbuf, (UV)c);
614 return is_utf8_alnumc(tmpbuf);
618 Perl_is_uni_idfirst(pTHX_ U32 c)
620 U8 tmpbuf[UTF8_MAXLEN+1];
621 uv_to_utf8(tmpbuf, (UV)c);
622 return is_utf8_idfirst(tmpbuf);
626 Perl_is_uni_alpha(pTHX_ U32 c)
628 U8 tmpbuf[UTF8_MAXLEN+1];
629 uv_to_utf8(tmpbuf, (UV)c);
630 return is_utf8_alpha(tmpbuf);
634 Perl_is_uni_ascii(pTHX_ U32 c)
636 U8 tmpbuf[UTF8_MAXLEN+1];
637 uv_to_utf8(tmpbuf, (UV)c);
638 return is_utf8_ascii(tmpbuf);
642 Perl_is_uni_space(pTHX_ U32 c)
644 U8 tmpbuf[UTF8_MAXLEN+1];
645 uv_to_utf8(tmpbuf, (UV)c);
646 return is_utf8_space(tmpbuf);
650 Perl_is_uni_digit(pTHX_ U32 c)
652 U8 tmpbuf[UTF8_MAXLEN+1];
653 uv_to_utf8(tmpbuf, (UV)c);
654 return is_utf8_digit(tmpbuf);
658 Perl_is_uni_upper(pTHX_ U32 c)
660 U8 tmpbuf[UTF8_MAXLEN+1];
661 uv_to_utf8(tmpbuf, (UV)c);
662 return is_utf8_upper(tmpbuf);
666 Perl_is_uni_lower(pTHX_ U32 c)
668 U8 tmpbuf[UTF8_MAXLEN+1];
669 uv_to_utf8(tmpbuf, (UV)c);
670 return is_utf8_lower(tmpbuf);
674 Perl_is_uni_cntrl(pTHX_ U32 c)
676 U8 tmpbuf[UTF8_MAXLEN+1];
677 uv_to_utf8(tmpbuf, (UV)c);
678 return is_utf8_cntrl(tmpbuf);
682 Perl_is_uni_graph(pTHX_ U32 c)
684 U8 tmpbuf[UTF8_MAXLEN+1];
685 uv_to_utf8(tmpbuf, (UV)c);
686 return is_utf8_graph(tmpbuf);
690 Perl_is_uni_print(pTHX_ U32 c)
692 U8 tmpbuf[UTF8_MAXLEN+1];
693 uv_to_utf8(tmpbuf, (UV)c);
694 return is_utf8_print(tmpbuf);
698 Perl_is_uni_punct(pTHX_ U32 c)
700 U8 tmpbuf[UTF8_MAXLEN+1];
701 uv_to_utf8(tmpbuf, (UV)c);
702 return is_utf8_punct(tmpbuf);
706 Perl_is_uni_xdigit(pTHX_ U32 c)
708 U8 tmpbuf[UTF8_MAXLEN+1];
709 uv_to_utf8(tmpbuf, (UV)c);
710 return is_utf8_xdigit(tmpbuf);
714 Perl_to_uni_upper(pTHX_ U32 c)
716 U8 tmpbuf[UTF8_MAXLEN+1];
717 uv_to_utf8(tmpbuf, (UV)c);
718 return to_utf8_upper(tmpbuf);
722 Perl_to_uni_title(pTHX_ U32 c)
724 U8 tmpbuf[UTF8_MAXLEN+1];
725 uv_to_utf8(tmpbuf, (UV)c);
726 return to_utf8_title(tmpbuf);
730 Perl_to_uni_lower(pTHX_ U32 c)
732 U8 tmpbuf[UTF8_MAXLEN+1];
733 uv_to_utf8(tmpbuf, (UV)c);
734 return to_utf8_lower(tmpbuf);
737 /* for now these all assume no locale info available for Unicode > 255 */
740 Perl_is_uni_alnum_lc(pTHX_ U32 c)
742 return is_uni_alnum(c); /* XXX no locale support yet */
746 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
748 return is_uni_alnumc(c); /* XXX no locale support yet */
752 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
754 return is_uni_idfirst(c); /* XXX no locale support yet */
758 Perl_is_uni_alpha_lc(pTHX_ U32 c)
760 return is_uni_alpha(c); /* XXX no locale support yet */
764 Perl_is_uni_ascii_lc(pTHX_ U32 c)
766 return is_uni_ascii(c); /* XXX no locale support yet */
770 Perl_is_uni_space_lc(pTHX_ U32 c)
772 return is_uni_space(c); /* XXX no locale support yet */
776 Perl_is_uni_digit_lc(pTHX_ U32 c)
778 return is_uni_digit(c); /* XXX no locale support yet */
782 Perl_is_uni_upper_lc(pTHX_ U32 c)
784 return is_uni_upper(c); /* XXX no locale support yet */
788 Perl_is_uni_lower_lc(pTHX_ U32 c)
790 return is_uni_lower(c); /* XXX no locale support yet */
794 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
796 return is_uni_cntrl(c); /* XXX no locale support yet */
800 Perl_is_uni_graph_lc(pTHX_ U32 c)
802 return is_uni_graph(c); /* XXX no locale support yet */
806 Perl_is_uni_print_lc(pTHX_ U32 c)
808 return is_uni_print(c); /* XXX no locale support yet */
812 Perl_is_uni_punct_lc(pTHX_ U32 c)
814 return is_uni_punct(c); /* XXX no locale support yet */
818 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
820 return is_uni_xdigit(c); /* XXX no locale support yet */
824 Perl_to_uni_upper_lc(pTHX_ U32 c)
826 return to_uni_upper(c); /* XXX no locale support yet */
830 Perl_to_uni_title_lc(pTHX_ U32 c)
832 return to_uni_title(c); /* XXX no locale support yet */
836 Perl_to_uni_lower_lc(pTHX_ U32 c)
838 return to_uni_lower(c); /* XXX no locale support yet */
842 Perl_is_utf8_alnum(pTHX_ U8 *p)
844 if (!is_utf8_char(p))
847 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
848 * descendant of isalnum(3), in other words, it doesn't
849 * contain the '_'. --jhi */
850 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
851 return swash_fetch(PL_utf8_alnum, p);
852 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
853 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
855 PL_utf8_alnum = swash_init("utf8", "",
856 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
857 return swash_fetch(PL_utf8_alnum, p);
862 Perl_is_utf8_alnumc(pTHX_ U8 *p)
864 if (!is_utf8_char(p))
867 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
868 return swash_fetch(PL_utf8_alnum, p);
869 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
870 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
872 PL_utf8_alnum = swash_init("utf8", "",
873 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
874 return swash_fetch(PL_utf8_alnum, p);
879 Perl_is_utf8_idfirst(pTHX_ U8 *p)
881 return *p == '_' || is_utf8_alpha(p);
885 Perl_is_utf8_alpha(pTHX_ U8 *p)
887 if (!is_utf8_char(p))
890 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
891 return swash_fetch(PL_utf8_alpha, p);
895 Perl_is_utf8_ascii(pTHX_ U8 *p)
897 if (!is_utf8_char(p))
900 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
901 return swash_fetch(PL_utf8_ascii, p);
905 Perl_is_utf8_space(pTHX_ U8 *p)
907 if (!is_utf8_char(p))
910 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
911 return swash_fetch(PL_utf8_space, p);
915 Perl_is_utf8_digit(pTHX_ U8 *p)
917 if (!is_utf8_char(p))
920 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
921 return swash_fetch(PL_utf8_digit, p);
925 Perl_is_utf8_upper(pTHX_ U8 *p)
927 if (!is_utf8_char(p))
930 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
931 return swash_fetch(PL_utf8_upper, p);
935 Perl_is_utf8_lower(pTHX_ U8 *p)
937 if (!is_utf8_char(p))
940 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
941 return swash_fetch(PL_utf8_lower, p);
945 Perl_is_utf8_cntrl(pTHX_ U8 *p)
947 if (!is_utf8_char(p))
950 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
951 return swash_fetch(PL_utf8_cntrl, p);
955 Perl_is_utf8_graph(pTHX_ U8 *p)
957 if (!is_utf8_char(p))
960 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
961 return swash_fetch(PL_utf8_graph, p);
965 Perl_is_utf8_print(pTHX_ U8 *p)
967 if (!is_utf8_char(p))
970 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
971 return swash_fetch(PL_utf8_print, p);
975 Perl_is_utf8_punct(pTHX_ U8 *p)
977 if (!is_utf8_char(p))
980 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
981 return swash_fetch(PL_utf8_punct, p);
985 Perl_is_utf8_xdigit(pTHX_ U8 *p)
987 if (!is_utf8_char(p))
990 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
991 return swash_fetch(PL_utf8_xdigit, p);
995 Perl_is_utf8_mark(pTHX_ U8 *p)
997 if (!is_utf8_char(p))
1000 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1001 return swash_fetch(PL_utf8_mark, p);
1005 Perl_to_utf8_upper(pTHX_ U8 *p)
1009 if (!PL_utf8_toupper)
1010 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
1011 uv = swash_fetch(PL_utf8_toupper, p);
1012 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1016 Perl_to_utf8_title(pTHX_ U8 *p)
1020 if (!PL_utf8_totitle)
1021 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
1022 uv = swash_fetch(PL_utf8_totitle, p);
1023 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1027 Perl_to_utf8_lower(pTHX_ U8 *p)
1031 if (!PL_utf8_tolower)
1032 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
1033 uv = swash_fetch(PL_utf8_tolower, p);
1034 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1037 /* a "swash" is a swatch hash */
1040 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1046 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
1048 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1052 PUSHSTACKi(PERLSI_MAGIC);
1055 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1056 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1058 PUSHs(sv_2mortal(newSViv(minbits)));
1059 PUSHs(sv_2mortal(newSViv(none)));
1065 if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */
1066 strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf);
1067 if (call_method("SWASHNEW", G_SCALAR))
1068 retval = newSVsv(*PL_stack_sp--);
1070 retval = &PL_sv_undef;
1073 if (PL_curcop == &PL_compiling) {
1074 strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf);
1075 PL_curcop->op_private = PL_hints;
1077 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1078 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1083 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1085 HV* hv = (HV*)SvRV(sv);
1086 U32 klen = UTF8SKIP(ptr) - 1;
1087 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1089 STRLEN needents = (klen ? 64 : 128);
1095 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1096 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1097 * it's nothing to sniff at.) Pity we usually come through at least
1098 * two function calls to get here...
1100 * NB: this code assumes that swatches are never modified, once generated!
1103 if (hv == PL_last_swash_hv &&
1104 klen == PL_last_swash_klen &&
1105 (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
1107 tmps = PL_last_swash_tmps;
1108 slen = PL_last_swash_slen;
1111 /* Try our second-level swatch cache, kept in a hash. */
1112 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1114 /* If not cached, generate it via utf8::SWASHGET */
1115 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1120 PUSHSTACKi(PERLSI_MAGIC);
1124 PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, UTF8_MAXLEN, 0, 0) & ~(needents - 1))));
1125 PUSHs(sv_2mortal(newSViv(needents)));
1127 if (call_method("SWASHGET", G_SCALAR))
1128 retval = newSVsv(*PL_stack_sp--);
1130 retval = &PL_sv_undef;
1134 if (PL_curcop == &PL_compiling)
1135 PL_curcop->op_private = PL_hints;
1137 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1139 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1140 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1143 PL_last_swash_hv = hv;
1144 PL_last_swash_klen = klen;
1145 PL_last_swash_tmps = tmps;
1146 PL_last_swash_slen = slen;
1148 Copy(ptr, PL_last_swash_key, klen, U8);
1151 switch ((int)((slen << 3) / needents)) {
1153 bit = 1 << (off & 7);
1155 return (tmps[off] & bit) != 0;
1160 return (tmps[off] << 8) + tmps[off + 1] ;
1163 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1165 Perl_croak(aTHX_ "panic: swash_fetch");