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 zero. If the C<flags> does not
193 contain UTF8_CHECK_ONLY, the UNICODE_REPLACEMENT (0xFFFD) will be
194 returned, and C<retlen> will be set to the expected length of the
195 UTF-8 character in bytes. The C<flags> can also contain various flags
196 to allow deviations from the strict UTF-8 encoding (see F<utf8.h>).
201 Perl_utf8_to_uv(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags)
208 bool dowarn = ckWARN_d(WARN_UTF8);
210 STRLEN expectlen = 0;
214 Perl_warner(aTHX_ WARN_UTF8,
215 "Malformed UTF-8 character (an empty string)");
219 if (UTF8_IS_ASCII(uv)) {
225 if (UTF8_IS_CONTINUATION(uv) &&
226 !(flags & UTF8_ALLOW_CONTINUATION)) {
228 Perl_warner(aTHX_ WARN_UTF8,
229 "Malformed UTF-8 character (unexpected continuation byte 0x%02"UVxf")",
234 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
235 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
237 Perl_warner(aTHX_ WARN_UTF8,
238 "Malformed UTF-8 character (unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")",
243 if ((uv == 0xfe || uv == 0xff) &&
244 !(flags & UTF8_ALLOW_FE_FF)) {
246 Perl_warner(aTHX_ WARN_UTF8,
247 "Malformed UTF-8 character (byte 0x%02"UVxf")",
252 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
253 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
254 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
255 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
256 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
257 else if (!(uv & 0x01)) { len = 7; uv = 0; }
258 else { len = 13; uv = 0; } /* whoa! */
265 if ((curlen < expectlen) &&
266 !(flags & UTF8_ALLOW_SHORT)) {
268 Perl_warner(aTHX_ WARN_UTF8,
269 "Malformed UTF-8 character (%d byte%s, need %d)",
270 curlen, curlen == 1 ? "" : "s", expectlen);
279 if (!UTF8_IS_CONTINUATION(*s) &&
280 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
282 Perl_warner(aTHX_ WARN_UTF8,
283 "Malformed UTF-8 character (unexpected non-continuation byte 0x%02x)",
288 uv = (uv << 6) | (*s & 0x3f);
290 /* This cannot be allowed. */
292 Perl_warner(aTHX_ WARN_UTF8,
293 "Malformed UTF-8 character (overflow at 0x%"UVxf", byte 0x%02x)",
301 if (UNICODE_IS_SURROGATE(uv) &&
302 !(flags & UTF8_ALLOW_SURROGATE)) {
304 Perl_warner(aTHX_ WARN_UTF8,
305 "Malformed UTF-8 character (UTF-16 surrogate 0x%04"UVxf")",
308 } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) &&
309 !(flags & UTF8_ALLOW_BOM)) {
311 Perl_warner(aTHX_ WARN_UTF8,
312 "Malformed UTF-8 character (byte order mark 0x%04"UVxf")",
315 } else if ((expectlen > UNISKIP(uv)) &&
316 !(flags & UTF8_ALLOW_LONG)) {
318 Perl_warner(aTHX_ WARN_UTF8,
319 "Malformed UTF-8 character (%d byte%s, need %d)",
320 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
322 } else if (UNICODE_IS_ILLEGAL(uv) &&
323 !(flags & UTF8_ALLOW_FFFF)) {
325 Perl_warner(aTHX_ WARN_UTF8,
326 "Malformed UTF-8 character (character 0x%04"UVxf")",
335 if (flags & UTF8_CHECK_ONLY) {
344 return UNICODE_REPLACEMENT;
348 =for apidoc Am|U8* s|utf8_to_uv_simple|STRLEN *retlen
350 Returns the character value of the first character in the string C<s>
351 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
352 length, in bytes, of that character, and the pointer C<s> will be
353 advanced to the end of the character.
355 If C<s> does not point to a well-formed UTF8 character, zero is
356 returned and retlen is set, if possible, to -1.
362 Perl_utf8_to_uv_simple(pTHX_ U8* s, STRLEN* retlen)
364 return Perl_utf8_to_uv(aTHX_ s, UTF8_MAXLEN, retlen, 0);
368 =for apidoc Am|STRLEN|utf8_length|U8* s|U8 *e
370 Return the length of the UTF-8 char encoded string C<s> in characters.
371 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
372 up past C<e>, croaks.
378 Perl_utf8_length(pTHX_ U8* s, U8* e)
383 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
388 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
397 =for apidoc Am|IV|utf8_distance|U8 *a|U8 *b
399 Returns the number of UTF8 characters between the UTF-8 pointers C<a>
402 WARNING: use only if you *know* that the pointers point inside the
408 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
417 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
427 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
437 =for apidoc Am|U8*|utf8_hop|U8 *s|I32 off
439 Move the C<s> pointing to UTF-8 data by C<off> characters, either forward
442 WARNING: do not use the following unless you *know* C<off> is within
443 the UTF-8 buffer pointed to by C<s>.
448 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
458 while ((*s & 0xc0) == 0x80)
467 =for apidoc Am|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
469 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
470 Unlike C<bytes_to_utf8>, this over-writes the original string, and
471 updates len to contain the new length.
472 Returns zero on failure, setting C<len> to -1.
478 Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len)
484 /* ensure valid UTF8 and chars < 256 before updating string */
485 for (send = s + *len; s < send; ) {
490 ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
503 *d++ = (U8)utf8_to_uv_simple(s, &ulen);
513 =for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
515 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
516 Returns a pointer to the newly-created string, and sets C<len> to
517 reflect the new length.
523 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
530 Newz(801, d, (*len) * 2 + 1, U8);
538 *d++ = (( uv >> 6) | 0xc0);
539 *d++ = (( uv & 0x3f) | 0x80);
548 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
550 * Destination must be pre-extended to 3/2 source. Do not use in-place.
551 * We optimize for native, for obvious reasons. */
554 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
560 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
565 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
572 *d++ = (( uv >> 6) | 0xc0);
573 *d++ = (( uv & 0x3f) | 0x80);
576 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
578 if (low < 0xdc00 || low >= 0xdfff)
579 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
580 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
583 *d++ = (( uv >> 12) | 0xe0);
584 *d++ = (((uv >> 6) & 0x3f) | 0x80);
585 *d++ = (( uv & 0x3f) | 0x80);
589 *d++ = (( uv >> 18) | 0xf0);
590 *d++ = (((uv >> 12) & 0x3f) | 0x80);
591 *d++ = (((uv >> 6) & 0x3f) | 0x80);
592 *d++ = (( uv & 0x3f) | 0x80);
596 *newlen = d - dstart;
600 /* Note: this one is slightly destructive of the source. */
603 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
606 U8* send = s + bytelen;
613 return utf16_to_utf8(p, d, bytelen, newlen);
616 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
619 Perl_is_uni_alnum(pTHX_ U32 c)
621 U8 tmpbuf[UTF8_MAXLEN+1];
622 uv_to_utf8(tmpbuf, (UV)c);
623 return is_utf8_alnum(tmpbuf);
627 Perl_is_uni_alnumc(pTHX_ U32 c)
629 U8 tmpbuf[UTF8_MAXLEN+1];
630 uv_to_utf8(tmpbuf, (UV)c);
631 return is_utf8_alnumc(tmpbuf);
635 Perl_is_uni_idfirst(pTHX_ U32 c)
637 U8 tmpbuf[UTF8_MAXLEN+1];
638 uv_to_utf8(tmpbuf, (UV)c);
639 return is_utf8_idfirst(tmpbuf);
643 Perl_is_uni_alpha(pTHX_ U32 c)
645 U8 tmpbuf[UTF8_MAXLEN+1];
646 uv_to_utf8(tmpbuf, (UV)c);
647 return is_utf8_alpha(tmpbuf);
651 Perl_is_uni_ascii(pTHX_ U32 c)
653 U8 tmpbuf[UTF8_MAXLEN+1];
654 uv_to_utf8(tmpbuf, (UV)c);
655 return is_utf8_ascii(tmpbuf);
659 Perl_is_uni_space(pTHX_ U32 c)
661 U8 tmpbuf[UTF8_MAXLEN+1];
662 uv_to_utf8(tmpbuf, (UV)c);
663 return is_utf8_space(tmpbuf);
667 Perl_is_uni_digit(pTHX_ U32 c)
669 U8 tmpbuf[UTF8_MAXLEN+1];
670 uv_to_utf8(tmpbuf, (UV)c);
671 return is_utf8_digit(tmpbuf);
675 Perl_is_uni_upper(pTHX_ U32 c)
677 U8 tmpbuf[UTF8_MAXLEN+1];
678 uv_to_utf8(tmpbuf, (UV)c);
679 return is_utf8_upper(tmpbuf);
683 Perl_is_uni_lower(pTHX_ U32 c)
685 U8 tmpbuf[UTF8_MAXLEN+1];
686 uv_to_utf8(tmpbuf, (UV)c);
687 return is_utf8_lower(tmpbuf);
691 Perl_is_uni_cntrl(pTHX_ U32 c)
693 U8 tmpbuf[UTF8_MAXLEN+1];
694 uv_to_utf8(tmpbuf, (UV)c);
695 return is_utf8_cntrl(tmpbuf);
699 Perl_is_uni_graph(pTHX_ U32 c)
701 U8 tmpbuf[UTF8_MAXLEN+1];
702 uv_to_utf8(tmpbuf, (UV)c);
703 return is_utf8_graph(tmpbuf);
707 Perl_is_uni_print(pTHX_ U32 c)
709 U8 tmpbuf[UTF8_MAXLEN+1];
710 uv_to_utf8(tmpbuf, (UV)c);
711 return is_utf8_print(tmpbuf);
715 Perl_is_uni_punct(pTHX_ U32 c)
717 U8 tmpbuf[UTF8_MAXLEN+1];
718 uv_to_utf8(tmpbuf, (UV)c);
719 return is_utf8_punct(tmpbuf);
723 Perl_is_uni_xdigit(pTHX_ U32 c)
725 U8 tmpbuf[UTF8_MAXLEN+1];
726 uv_to_utf8(tmpbuf, (UV)c);
727 return is_utf8_xdigit(tmpbuf);
731 Perl_to_uni_upper(pTHX_ U32 c)
733 U8 tmpbuf[UTF8_MAXLEN+1];
734 uv_to_utf8(tmpbuf, (UV)c);
735 return to_utf8_upper(tmpbuf);
739 Perl_to_uni_title(pTHX_ U32 c)
741 U8 tmpbuf[UTF8_MAXLEN+1];
742 uv_to_utf8(tmpbuf, (UV)c);
743 return to_utf8_title(tmpbuf);
747 Perl_to_uni_lower(pTHX_ U32 c)
749 U8 tmpbuf[UTF8_MAXLEN+1];
750 uv_to_utf8(tmpbuf, (UV)c);
751 return to_utf8_lower(tmpbuf);
754 /* for now these all assume no locale info available for Unicode > 255 */
757 Perl_is_uni_alnum_lc(pTHX_ U32 c)
759 return is_uni_alnum(c); /* XXX no locale support yet */
763 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
765 return is_uni_alnumc(c); /* XXX no locale support yet */
769 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
771 return is_uni_idfirst(c); /* XXX no locale support yet */
775 Perl_is_uni_alpha_lc(pTHX_ U32 c)
777 return is_uni_alpha(c); /* XXX no locale support yet */
781 Perl_is_uni_ascii_lc(pTHX_ U32 c)
783 return is_uni_ascii(c); /* XXX no locale support yet */
787 Perl_is_uni_space_lc(pTHX_ U32 c)
789 return is_uni_space(c); /* XXX no locale support yet */
793 Perl_is_uni_digit_lc(pTHX_ U32 c)
795 return is_uni_digit(c); /* XXX no locale support yet */
799 Perl_is_uni_upper_lc(pTHX_ U32 c)
801 return is_uni_upper(c); /* XXX no locale support yet */
805 Perl_is_uni_lower_lc(pTHX_ U32 c)
807 return is_uni_lower(c); /* XXX no locale support yet */
811 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
813 return is_uni_cntrl(c); /* XXX no locale support yet */
817 Perl_is_uni_graph_lc(pTHX_ U32 c)
819 return is_uni_graph(c); /* XXX no locale support yet */
823 Perl_is_uni_print_lc(pTHX_ U32 c)
825 return is_uni_print(c); /* XXX no locale support yet */
829 Perl_is_uni_punct_lc(pTHX_ U32 c)
831 return is_uni_punct(c); /* XXX no locale support yet */
835 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
837 return is_uni_xdigit(c); /* XXX no locale support yet */
841 Perl_to_uni_upper_lc(pTHX_ U32 c)
843 return to_uni_upper(c); /* XXX no locale support yet */
847 Perl_to_uni_title_lc(pTHX_ U32 c)
849 return to_uni_title(c); /* XXX no locale support yet */
853 Perl_to_uni_lower_lc(pTHX_ U32 c)
855 return to_uni_lower(c); /* XXX no locale support yet */
859 Perl_is_utf8_alnum(pTHX_ U8 *p)
861 if (!is_utf8_char(p))
864 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
865 * descendant of isalnum(3), in other words, it doesn't
866 * contain the '_'. --jhi */
867 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
868 return swash_fetch(PL_utf8_alnum, p);
869 /* return *p == '_' || 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_alnumc(pTHX_ U8 *p)
881 if (!is_utf8_char(p))
884 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
885 return swash_fetch(PL_utf8_alnum, p);
886 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
887 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
889 PL_utf8_alnum = swash_init("utf8", "",
890 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
891 return swash_fetch(PL_utf8_alnum, p);
896 Perl_is_utf8_idfirst(pTHX_ U8 *p)
898 return *p == '_' || is_utf8_alpha(p);
902 Perl_is_utf8_alpha(pTHX_ U8 *p)
904 if (!is_utf8_char(p))
907 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
908 return swash_fetch(PL_utf8_alpha, p);
912 Perl_is_utf8_ascii(pTHX_ U8 *p)
914 if (!is_utf8_char(p))
917 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
918 return swash_fetch(PL_utf8_ascii, p);
922 Perl_is_utf8_space(pTHX_ U8 *p)
924 if (!is_utf8_char(p))
927 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
928 return swash_fetch(PL_utf8_space, p);
932 Perl_is_utf8_digit(pTHX_ U8 *p)
934 if (!is_utf8_char(p))
937 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
938 return swash_fetch(PL_utf8_digit, p);
942 Perl_is_utf8_upper(pTHX_ U8 *p)
944 if (!is_utf8_char(p))
947 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
948 return swash_fetch(PL_utf8_upper, p);
952 Perl_is_utf8_lower(pTHX_ U8 *p)
954 if (!is_utf8_char(p))
957 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
958 return swash_fetch(PL_utf8_lower, p);
962 Perl_is_utf8_cntrl(pTHX_ U8 *p)
964 if (!is_utf8_char(p))
967 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
968 return swash_fetch(PL_utf8_cntrl, p);
972 Perl_is_utf8_graph(pTHX_ U8 *p)
974 if (!is_utf8_char(p))
977 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
978 return swash_fetch(PL_utf8_graph, p);
982 Perl_is_utf8_print(pTHX_ U8 *p)
984 if (!is_utf8_char(p))
987 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
988 return swash_fetch(PL_utf8_print, p);
992 Perl_is_utf8_punct(pTHX_ U8 *p)
994 if (!is_utf8_char(p))
997 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
998 return swash_fetch(PL_utf8_punct, p);
1002 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1004 if (!is_utf8_char(p))
1006 if (!PL_utf8_xdigit)
1007 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1008 return swash_fetch(PL_utf8_xdigit, p);
1012 Perl_is_utf8_mark(pTHX_ U8 *p)
1014 if (!is_utf8_char(p))
1017 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1018 return swash_fetch(PL_utf8_mark, p);
1022 Perl_to_utf8_upper(pTHX_ U8 *p)
1026 if (!PL_utf8_toupper)
1027 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
1028 uv = swash_fetch(PL_utf8_toupper, p);
1029 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1033 Perl_to_utf8_title(pTHX_ U8 *p)
1037 if (!PL_utf8_totitle)
1038 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
1039 uv = swash_fetch(PL_utf8_totitle, p);
1040 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1044 Perl_to_utf8_lower(pTHX_ U8 *p)
1048 if (!PL_utf8_tolower)
1049 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
1050 uv = swash_fetch(PL_utf8_tolower, p);
1051 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1054 /* a "swash" is a swatch hash */
1057 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1063 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
1065 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1069 PUSHSTACKi(PERLSI_MAGIC);
1072 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1073 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1075 PUSHs(sv_2mortal(newSViv(minbits)));
1076 PUSHs(sv_2mortal(newSViv(none)));
1082 if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */
1083 strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf);
1084 if (call_method("SWASHNEW", G_SCALAR))
1085 retval = newSVsv(*PL_stack_sp--);
1087 retval = &PL_sv_undef;
1090 if (PL_curcop == &PL_compiling) {
1091 strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf);
1092 PL_curcop->op_private = PL_hints;
1094 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1095 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1100 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1102 HV* hv = (HV*)SvRV(sv);
1103 U32 klen = UTF8SKIP(ptr) - 1;
1104 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1106 STRLEN needents = (klen ? 64 : 128);
1112 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1113 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1114 * it's nothing to sniff at.) Pity we usually come through at least
1115 * two function calls to get here...
1117 * NB: this code assumes that swatches are never modified, once generated!
1120 if (hv == PL_last_swash_hv &&
1121 klen == PL_last_swash_klen &&
1122 (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
1124 tmps = PL_last_swash_tmps;
1125 slen = PL_last_swash_slen;
1128 /* Try our second-level swatch cache, kept in a hash. */
1129 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1131 /* If not cached, generate it via utf8::SWASHGET */
1132 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1137 PUSHSTACKi(PERLSI_MAGIC);
1141 PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, UTF8_MAXLEN, 0, 0) & ~(needents - 1))));
1142 PUSHs(sv_2mortal(newSViv(needents)));
1144 if (call_method("SWASHGET", G_SCALAR))
1145 retval = newSVsv(*PL_stack_sp--);
1147 retval = &PL_sv_undef;
1151 if (PL_curcop == &PL_compiling)
1152 PL_curcop->op_private = PL_hints;
1154 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1156 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1157 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1160 PL_last_swash_hv = hv;
1161 PL_last_swash_klen = klen;
1162 PL_last_swash_tmps = tmps;
1163 PL_last_swash_slen = slen;
1165 Copy(ptr, PL_last_swash_key, klen, U8);
1168 switch ((int)((slen << 3) / needents)) {
1170 bit = 1 << (off & 7);
1172 return (tmps[off] & bit) != 0;
1177 return (tmps[off] << 8) + tmps[off + 1] ;
1180 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1182 Perl_croak(aTHX_ "panic: swash_fetch");