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)
206 bool dowarn = ckWARN_d(WARN_UTF8);
208 STRLEN expectlen = 0;
212 Perl_warner(aTHX_ WARN_UTF8,
213 "Malformed UTF-8 character (an empty string)");
217 if (uv <= 0x7f) { /* Pure ASCII. */
223 if ((uv >= 0x80 && uv <= 0xbf) &&
224 !(flags & UTF8_ALLOW_CONTINUATION)) {
226 Perl_warner(aTHX_ WARN_UTF8,
227 "Malformed UTF-8 character (unexpected continuation byte 0x%02"UVxf")",
232 if ((uv >= 0xc0 && uv <= 0xfd && curlen > 1 && s[1] < 0x80) &&
233 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
235 Perl_warner(aTHX_ WARN_UTF8,
236 "Malformed UTF-8 character (unexpected non-continuation byte 0x%02"UVxf" after byte 0x%02"UVxf")",
241 if ((uv == 0xfe || uv == 0xff) &&
242 !(flags & UTF8_ALLOW_FE_FF)) {
244 Perl_warner(aTHX_ WARN_UTF8,
245 "Malformed UTF-8 character (byte 0x%02"UVxf")",
250 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
251 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
252 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
253 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
254 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
255 else if (!(uv & 0x01)) { len = 7; uv = 0; }
256 else { len = 13; uv = 0; } /* whoa! */
263 if ((curlen < expectlen) &&
264 !(flags & UTF8_ALLOW_SHORT)) {
266 Perl_warner(aTHX_ WARN_UTF8,
267 "Malformed UTF-8 character (%d byte%s, need %d)",
268 curlen, curlen == 1 ? "" : "s", expectlen);
277 if ((*s & 0xc0) != 0x80) {
279 Perl_warner(aTHX_ WARN_UTF8,
280 "Malformed UTF-8 character (unexpected continuation byte 0x%02x)",
285 uv = (uv << 6) | (*s & 0x3f);
287 /* This cannot be allowed. */
289 Perl_warner(aTHX_ WARN_UTF8,
290 "Malformed UTF-8 character (overflow at 0x%"UVxf", byte 0x%02x)",
298 if ((uv >= 0xd800 && uv <= 0xdfff) &&
299 !(flags & UTF8_ALLOW_SURROGATE)) {
301 Perl_warner(aTHX_ WARN_UTF8,
302 "Malformed UTF-8 character (UTF-16 surrogate 0x%04"UVxf")",
305 } else if ((uv == 0xfffe) &&
306 !(flags & UTF8_ALLOW_BOM)) {
308 Perl_warner(aTHX_ WARN_UTF8,
309 "Malformed UTF-8 character (byte order mark 0x%04"UVxf")",
312 } else if ((expectlen > UNISKIP(uv)) &&
313 !(flags & UTF8_ALLOW_LONG)) {
315 Perl_warner(aTHX_ WARN_UTF8,
316 "Malformed UTF-8 character (%d byte%s, need %d)",
317 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
319 } else if ((uv == 0xffff) &&
320 !(flags & UTF8_ALLOW_FFFF)) {
322 Perl_warner(aTHX_ WARN_UTF8,
323 "Malformed UTF-8 character (character 0x%04"UVxf")",
332 if (flags & UTF8_CHECK_ONLY) {
339 *retlen = expectlen ? expectlen : len;
341 return UNICODE_REPLACEMENT_CHARACTER;
345 =for apidoc Am|U8* s|utf8_to_uv_simple|STRLEN *retlen
347 Returns the character value of the first character in the string C<s>
348 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
349 length, in bytes, of that character, and the pointer C<s> will be
350 advanced to the end of the character.
352 If C<s> does not point to a well-formed UTF8 character, zero is
353 returned and retlen is set, if possible, to -1.
359 Perl_utf8_to_uv_simple(pTHX_ U8* s, STRLEN* retlen)
361 return Perl_utf8_to_uv(aTHX_ s, UTF8_MAXLEN, retlen, 0);
365 =for apidoc Am|STRLEN|utf8_length|U8* s|U8 *e
367 Return the length of the UTF-8 char encoded string C<s> in characters.
368 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
369 up past C<e>, croaks.
375 Perl_utf8_length(pTHX_ U8* s, U8* e)
380 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
385 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
394 =for apidoc Am|IV|utf8_distance|U8 *a|U8 *b
396 Returns the number of UTF8 characters between the UTF-8 pointers C<a>
399 WARNING: use only if you *know* that the pointers point inside the
405 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
414 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
424 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
434 =for apidoc Am|U8*|utf8_hop|U8 *s|I32 off
436 Move the C<s> pointing to UTF-8 data by C<off> characters, either forward
439 WARNING: do not use the following unless you *know* C<off> is within
440 the UTF-8 buffer pointed to by C<s>.
445 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
455 while ((*s & 0xc0) == 0x80)
464 =for apidoc Am|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
466 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
467 Unlike C<bytes_to_utf8>, this over-writes the original string, and
468 updates len to contain the new length.
469 Returns zero on failure, setting C<len> to -1.
475 Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len)
481 /* ensure valid UTF8 and chars < 256 before updating string */
482 for (send = s + *len; s < send; ) {
487 ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
500 *d++ = (U8)utf8_to_uv_simple(s, &ulen);
510 =for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
512 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
513 Returns a pointer to the newly-created string, and sets C<len> to
514 reflect the new length.
520 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
527 Newz(801, d, (*len) * 2 + 1, U8);
535 *d++ = (( uv >> 6) | 0xc0);
536 *d++ = (( uv & 0x3f) | 0x80);
545 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
547 * Destination must be pre-extended to 3/2 source. Do not use in-place.
548 * We optimize for native, for obvious reasons. */
551 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
557 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
562 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
569 *d++ = (( uv >> 6) | 0xc0);
570 *d++ = (( uv & 0x3f) | 0x80);
573 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
575 if (low < 0xdc00 || low >= 0xdfff)
576 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
577 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
580 *d++ = (( uv >> 12) | 0xe0);
581 *d++ = (((uv >> 6) & 0x3f) | 0x80);
582 *d++ = (( uv & 0x3f) | 0x80);
586 *d++ = (( uv >> 18) | 0xf0);
587 *d++ = (((uv >> 12) & 0x3f) | 0x80);
588 *d++ = (((uv >> 6) & 0x3f) | 0x80);
589 *d++ = (( uv & 0x3f) | 0x80);
593 *newlen = d - dstart;
597 /* Note: this one is slightly destructive of the source. */
600 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
603 U8* send = s + bytelen;
610 return utf16_to_utf8(p, d, bytelen, newlen);
613 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
616 Perl_is_uni_alnum(pTHX_ U32 c)
618 U8 tmpbuf[UTF8_MAXLEN+1];
619 uv_to_utf8(tmpbuf, (UV)c);
620 return is_utf8_alnum(tmpbuf);
624 Perl_is_uni_alnumc(pTHX_ U32 c)
626 U8 tmpbuf[UTF8_MAXLEN+1];
627 uv_to_utf8(tmpbuf, (UV)c);
628 return is_utf8_alnumc(tmpbuf);
632 Perl_is_uni_idfirst(pTHX_ U32 c)
634 U8 tmpbuf[UTF8_MAXLEN+1];
635 uv_to_utf8(tmpbuf, (UV)c);
636 return is_utf8_idfirst(tmpbuf);
640 Perl_is_uni_alpha(pTHX_ U32 c)
642 U8 tmpbuf[UTF8_MAXLEN+1];
643 uv_to_utf8(tmpbuf, (UV)c);
644 return is_utf8_alpha(tmpbuf);
648 Perl_is_uni_ascii(pTHX_ U32 c)
650 U8 tmpbuf[UTF8_MAXLEN+1];
651 uv_to_utf8(tmpbuf, (UV)c);
652 return is_utf8_ascii(tmpbuf);
656 Perl_is_uni_space(pTHX_ U32 c)
658 U8 tmpbuf[UTF8_MAXLEN+1];
659 uv_to_utf8(tmpbuf, (UV)c);
660 return is_utf8_space(tmpbuf);
664 Perl_is_uni_digit(pTHX_ U32 c)
666 U8 tmpbuf[UTF8_MAXLEN+1];
667 uv_to_utf8(tmpbuf, (UV)c);
668 return is_utf8_digit(tmpbuf);
672 Perl_is_uni_upper(pTHX_ U32 c)
674 U8 tmpbuf[UTF8_MAXLEN+1];
675 uv_to_utf8(tmpbuf, (UV)c);
676 return is_utf8_upper(tmpbuf);
680 Perl_is_uni_lower(pTHX_ U32 c)
682 U8 tmpbuf[UTF8_MAXLEN+1];
683 uv_to_utf8(tmpbuf, (UV)c);
684 return is_utf8_lower(tmpbuf);
688 Perl_is_uni_cntrl(pTHX_ U32 c)
690 U8 tmpbuf[UTF8_MAXLEN+1];
691 uv_to_utf8(tmpbuf, (UV)c);
692 return is_utf8_cntrl(tmpbuf);
696 Perl_is_uni_graph(pTHX_ U32 c)
698 U8 tmpbuf[UTF8_MAXLEN+1];
699 uv_to_utf8(tmpbuf, (UV)c);
700 return is_utf8_graph(tmpbuf);
704 Perl_is_uni_print(pTHX_ U32 c)
706 U8 tmpbuf[UTF8_MAXLEN+1];
707 uv_to_utf8(tmpbuf, (UV)c);
708 return is_utf8_print(tmpbuf);
712 Perl_is_uni_punct(pTHX_ U32 c)
714 U8 tmpbuf[UTF8_MAXLEN+1];
715 uv_to_utf8(tmpbuf, (UV)c);
716 return is_utf8_punct(tmpbuf);
720 Perl_is_uni_xdigit(pTHX_ U32 c)
722 U8 tmpbuf[UTF8_MAXLEN+1];
723 uv_to_utf8(tmpbuf, (UV)c);
724 return is_utf8_xdigit(tmpbuf);
728 Perl_to_uni_upper(pTHX_ U32 c)
730 U8 tmpbuf[UTF8_MAXLEN+1];
731 uv_to_utf8(tmpbuf, (UV)c);
732 return to_utf8_upper(tmpbuf);
736 Perl_to_uni_title(pTHX_ U32 c)
738 U8 tmpbuf[UTF8_MAXLEN+1];
739 uv_to_utf8(tmpbuf, (UV)c);
740 return to_utf8_title(tmpbuf);
744 Perl_to_uni_lower(pTHX_ U32 c)
746 U8 tmpbuf[UTF8_MAXLEN+1];
747 uv_to_utf8(tmpbuf, (UV)c);
748 return to_utf8_lower(tmpbuf);
751 /* for now these all assume no locale info available for Unicode > 255 */
754 Perl_is_uni_alnum_lc(pTHX_ U32 c)
756 return is_uni_alnum(c); /* XXX no locale support yet */
760 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
762 return is_uni_alnumc(c); /* XXX no locale support yet */
766 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
768 return is_uni_idfirst(c); /* XXX no locale support yet */
772 Perl_is_uni_alpha_lc(pTHX_ U32 c)
774 return is_uni_alpha(c); /* XXX no locale support yet */
778 Perl_is_uni_ascii_lc(pTHX_ U32 c)
780 return is_uni_ascii(c); /* XXX no locale support yet */
784 Perl_is_uni_space_lc(pTHX_ U32 c)
786 return is_uni_space(c); /* XXX no locale support yet */
790 Perl_is_uni_digit_lc(pTHX_ U32 c)
792 return is_uni_digit(c); /* XXX no locale support yet */
796 Perl_is_uni_upper_lc(pTHX_ U32 c)
798 return is_uni_upper(c); /* XXX no locale support yet */
802 Perl_is_uni_lower_lc(pTHX_ U32 c)
804 return is_uni_lower(c); /* XXX no locale support yet */
808 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
810 return is_uni_cntrl(c); /* XXX no locale support yet */
814 Perl_is_uni_graph_lc(pTHX_ U32 c)
816 return is_uni_graph(c); /* XXX no locale support yet */
820 Perl_is_uni_print_lc(pTHX_ U32 c)
822 return is_uni_print(c); /* XXX no locale support yet */
826 Perl_is_uni_punct_lc(pTHX_ U32 c)
828 return is_uni_punct(c); /* XXX no locale support yet */
832 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
834 return is_uni_xdigit(c); /* XXX no locale support yet */
838 Perl_to_uni_upper_lc(pTHX_ U32 c)
840 return to_uni_upper(c); /* XXX no locale support yet */
844 Perl_to_uni_title_lc(pTHX_ U32 c)
846 return to_uni_title(c); /* XXX no locale support yet */
850 Perl_to_uni_lower_lc(pTHX_ U32 c)
852 return to_uni_lower(c); /* XXX no locale support yet */
856 Perl_is_utf8_alnum(pTHX_ U8 *p)
858 if (!is_utf8_char(p))
861 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
862 * descendant of isalnum(3), in other words, it doesn't
863 * contain the '_'. --jhi */
864 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
865 return swash_fetch(PL_utf8_alnum, p);
866 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
867 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
869 PL_utf8_alnum = swash_init("utf8", "",
870 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
871 return swash_fetch(PL_utf8_alnum, p);
876 Perl_is_utf8_alnumc(pTHX_ U8 *p)
878 if (!is_utf8_char(p))
881 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
882 return swash_fetch(PL_utf8_alnum, p);
883 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
884 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
886 PL_utf8_alnum = swash_init("utf8", "",
887 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
888 return swash_fetch(PL_utf8_alnum, p);
893 Perl_is_utf8_idfirst(pTHX_ U8 *p)
895 return *p == '_' || is_utf8_alpha(p);
899 Perl_is_utf8_alpha(pTHX_ U8 *p)
901 if (!is_utf8_char(p))
904 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
905 return swash_fetch(PL_utf8_alpha, p);
909 Perl_is_utf8_ascii(pTHX_ U8 *p)
911 if (!is_utf8_char(p))
914 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
915 return swash_fetch(PL_utf8_ascii, p);
919 Perl_is_utf8_space(pTHX_ U8 *p)
921 if (!is_utf8_char(p))
924 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
925 return swash_fetch(PL_utf8_space, p);
929 Perl_is_utf8_digit(pTHX_ U8 *p)
931 if (!is_utf8_char(p))
934 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
935 return swash_fetch(PL_utf8_digit, p);
939 Perl_is_utf8_upper(pTHX_ U8 *p)
941 if (!is_utf8_char(p))
944 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
945 return swash_fetch(PL_utf8_upper, p);
949 Perl_is_utf8_lower(pTHX_ U8 *p)
951 if (!is_utf8_char(p))
954 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
955 return swash_fetch(PL_utf8_lower, p);
959 Perl_is_utf8_cntrl(pTHX_ U8 *p)
961 if (!is_utf8_char(p))
964 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
965 return swash_fetch(PL_utf8_cntrl, p);
969 Perl_is_utf8_graph(pTHX_ U8 *p)
971 if (!is_utf8_char(p))
974 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
975 return swash_fetch(PL_utf8_graph, p);
979 Perl_is_utf8_print(pTHX_ U8 *p)
981 if (!is_utf8_char(p))
984 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
985 return swash_fetch(PL_utf8_print, p);
989 Perl_is_utf8_punct(pTHX_ U8 *p)
991 if (!is_utf8_char(p))
994 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
995 return swash_fetch(PL_utf8_punct, p);
999 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1001 if (!is_utf8_char(p))
1003 if (!PL_utf8_xdigit)
1004 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1005 return swash_fetch(PL_utf8_xdigit, p);
1009 Perl_is_utf8_mark(pTHX_ U8 *p)
1011 if (!is_utf8_char(p))
1014 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1015 return swash_fetch(PL_utf8_mark, p);
1019 Perl_to_utf8_upper(pTHX_ U8 *p)
1023 if (!PL_utf8_toupper)
1024 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
1025 uv = swash_fetch(PL_utf8_toupper, p);
1026 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1030 Perl_to_utf8_title(pTHX_ U8 *p)
1034 if (!PL_utf8_totitle)
1035 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
1036 uv = swash_fetch(PL_utf8_totitle, p);
1037 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1041 Perl_to_utf8_lower(pTHX_ U8 *p)
1045 if (!PL_utf8_tolower)
1046 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
1047 uv = swash_fetch(PL_utf8_tolower, p);
1048 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1051 /* a "swash" is a swatch hash */
1054 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1060 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
1062 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1066 PUSHSTACKi(PERLSI_MAGIC);
1069 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1070 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1072 PUSHs(sv_2mortal(newSViv(minbits)));
1073 PUSHs(sv_2mortal(newSViv(none)));
1079 if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */
1080 strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf);
1081 if (call_method("SWASHNEW", G_SCALAR))
1082 retval = newSVsv(*PL_stack_sp--);
1084 retval = &PL_sv_undef;
1087 if (PL_curcop == &PL_compiling) {
1088 strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf);
1089 PL_curcop->op_private = PL_hints;
1091 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1092 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1097 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1099 HV* hv = (HV*)SvRV(sv);
1100 U32 klen = UTF8SKIP(ptr) - 1;
1101 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1103 STRLEN needents = (klen ? 64 : 128);
1109 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1110 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1111 * it's nothing to sniff at.) Pity we usually come through at least
1112 * two function calls to get here...
1114 * NB: this code assumes that swatches are never modified, once generated!
1117 if (hv == PL_last_swash_hv &&
1118 klen == PL_last_swash_klen &&
1119 (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
1121 tmps = PL_last_swash_tmps;
1122 slen = PL_last_swash_slen;
1125 /* Try our second-level swatch cache, kept in a hash. */
1126 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1128 /* If not cached, generate it via utf8::SWASHGET */
1129 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1134 PUSHSTACKi(PERLSI_MAGIC);
1138 PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, UTF8_MAXLEN, 0, 0) & ~(needents - 1))));
1139 PUSHs(sv_2mortal(newSViv(needents)));
1141 if (call_method("SWASHGET", G_SCALAR))
1142 retval = newSVsv(*PL_stack_sp--);
1144 retval = &PL_sv_undef;
1148 if (PL_curcop == &PL_compiling)
1149 PL_curcop->op_private = PL_hints;
1151 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1153 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1154 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1157 PL_last_swash_hv = hv;
1158 PL_last_swash_klen = klen;
1159 PL_last_swash_tmps = tmps;
1160 PL_last_swash_slen = slen;
1162 Copy(ptr, PL_last_swash_key, klen, U8);
1165 switch ((int)((slen << 3) / needents)) {
1167 bit = 1 << (off & 7);
1169 return (tmps[off] & bit) != 0;
1174 return (tmps[off] << 8) + tmps[off + 1] ;
1177 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1179 Perl_croak(aTHX_ "panic: swash_fetch");