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)
37 *d++ = (( uv >> 6) | 0xc0);
38 *d++ = (( uv & 0x3f) | 0x80);
42 *d++ = (( uv >> 12) | 0xe0);
43 *d++ = (((uv >> 6) & 0x3f) | 0x80);
44 *d++ = (( uv & 0x3f) | 0x80);
48 *d++ = (( uv >> 18) | 0xf0);
49 *d++ = (((uv >> 12) & 0x3f) | 0x80);
50 *d++ = (((uv >> 6) & 0x3f) | 0x80);
51 *d++ = (( uv & 0x3f) | 0x80);
55 *d++ = (( uv >> 24) | 0xf8);
56 *d++ = (((uv >> 18) & 0x3f) | 0x80);
57 *d++ = (((uv >> 12) & 0x3f) | 0x80);
58 *d++ = (((uv >> 6) & 0x3f) | 0x80);
59 *d++ = (( uv & 0x3f) | 0x80);
62 if (uv < 0x80000000) {
63 *d++ = (( uv >> 30) | 0xfc);
64 *d++ = (((uv >> 24) & 0x3f) | 0x80);
65 *d++ = (((uv >> 18) & 0x3f) | 0x80);
66 *d++ = (((uv >> 12) & 0x3f) | 0x80);
67 *d++ = (((uv >> 6) & 0x3f) | 0x80);
68 *d++ = (( uv & 0x3f) | 0x80);
72 if (uv < UTF8_QUAD_MAX)
75 *d++ = 0xfe; /* Can't match U+FEFF! */
76 *d++ = (((uv >> 30) & 0x3f) | 0x80);
77 *d++ = (((uv >> 24) & 0x3f) | 0x80);
78 *d++ = (((uv >> 18) & 0x3f) | 0x80);
79 *d++ = (((uv >> 12) & 0x3f) | 0x80);
80 *d++ = (((uv >> 6) & 0x3f) | 0x80);
81 *d++ = (( uv & 0x3f) | 0x80);
86 *d++ = 0xff; /* Can't match U+FFFE! */
87 *d++ = 0x80; /* 6 Reserved bits */
88 *d++ = (((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
89 *d++ = (((uv >> 54) & 0x3f) | 0x80);
90 *d++ = (((uv >> 48) & 0x3f) | 0x80);
91 *d++ = (((uv >> 42) & 0x3f) | 0x80);
92 *d++ = (((uv >> 36) & 0x3f) | 0x80);
93 *d++ = (((uv >> 30) & 0x3f) | 0x80);
94 *d++ = (((uv >> 24) & 0x3f) | 0x80);
95 *d++ = (((uv >> 18) & 0x3f) | 0x80);
96 *d++ = (((uv >> 12) & 0x3f) | 0x80);
97 *d++ = (((uv >> 6) & 0x3f) | 0x80);
98 *d++ = (( uv & 0x3f) | 0x80);
104 /* Tests if some arbitrary number of bytes begins in a valid UTF-8 character.
105 * The actual number of bytes in the UTF-8 character will be returned if it
106 * is valid, otherwise 0. */
108 Perl_is_utf8_char(pTHX_ U8 *s)
117 if (u >= 0x80 && u <= 0xbf)
122 if (len < 2 || (u >= 0xc0 && u <= 0xfd && s[1] < 0x80))
130 if ((*s & 0xc0) != 0x80)
132 uv = (uv << 6) | (*s & 0x3f);
139 if (UNISKIP(uv) < len)
146 =for apidoc Am|is_utf8_string|U8 *s|STRLEN len
148 Returns true if first C<len> bytes of the given string form valid a UTF8
149 string, false otherwise.
155 Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len)
174 =for apidoc Am|U8* s|utf8_to_uv|STRLEN curlen|I32 *retlen|U32 flags
176 Returns the character value of the first character in the string C<s>
177 which is assumed to be in UTF8 encoding and no longer than C<curlen>;
178 C<retlen> will be set to the length, in bytes, of that character,
179 and the pointer C<s> will be advanced to the end of the character.
181 If C<s> does not point to a well-formed UTF8 character, the behaviour
182 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
183 it is assumed that the caller will raise a warning, and this function
184 will set C<retlen> to C<-1> and return. The C<flags> can also contain
185 various flags to allow deviations from the strict UTF-8 encoding.
190 Perl_utf8_to_uv(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags)
198 bool dowarn = ckWARN_d(WARN_UTF8);
200 STRLEN expectlen = 0;
204 Perl_warner(aTHX_ WARN_UTF8,
205 "Malformed UTF-8 character (an empty string)");
209 if (uv <= 0x7f) { /* Pure ASCII. */
215 if ((uv >= 0x80 && uv <= 0xbf) &&
216 !(flags & UTF8_ALLOW_CONTINUATION)) {
218 Perl_warner(aTHX_ WARN_UTF8,
219 "Malformed UTF-8 character (unexpected continuation byte 0x%02"UVxf")",
224 if ((uv >= 0xc0 && uv <= 0xfd && curlen > 1 && s[1] < 0x80) &&
225 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
227 Perl_warner(aTHX_ WARN_UTF8,
228 "Malformed UTF-8 character (unexpected non-continuation byte 0x%02"UVxf" after byte 0x%02"UVxf")",
233 if ((uv == 0xfe || uv == 0xff) &&
234 !(flags & UTF8_ALLOW_FE_FF)) {
236 Perl_warner(aTHX_ WARN_UTF8,
237 "Malformed UTF-8 character (byte 0x%02"UVxf")",
242 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
243 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
244 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
245 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
246 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
247 else if (!(uv & 0x01)) { len = 7; uv = 0; }
248 else { len = 13; uv = 0; } /* whoa! */
255 if ((curlen < expectlen) &&
256 !(flags & UTF8_ALLOW_SHORT)) {
258 Perl_warner(aTHX_ WARN_UTF8,
259 "Malformed UTF-8 character (%d byte%s, need %d)",
260 curlen, curlen == 1 ? "" : "s", expectlen);
269 if ((*s & 0xc0) != 0x80) {
271 Perl_warner(aTHX_ WARN_UTF8,
272 "Malformed UTF-8 character (unexpected continuation byte 0x%02x)",
277 uv = (uv << 6) | (*s & 0x3f);
279 /* This cannot be allowed. */
281 Perl_warner(aTHX_ WARN_UTF8,
282 "Malformed UTF-8 character (overflow at 0x%"UVxf", byte 0x%02x)",
290 if ((uv >= 0xd800 && uv <= 0xdfff) &&
291 !(flags & UTF8_ALLOW_SURROGATE)) {
293 Perl_warner(aTHX_ WARN_UTF8,
294 "Malformed UTF-8 character (UTF-16 surrogate 0x%04"UVxf")",
297 } else if ((uv == 0xfffe) &&
298 !(flags & UTF8_ALLOW_BOM)) {
300 Perl_warner(aTHX_ WARN_UTF8,
301 "Malformed UTF-8 character (byte order mark 0x%04"UVxf")",
304 } else if ((uv == 0xffff) &&
305 !(flags & UTF8_ALLOW_FFFF)) {
307 Perl_warner(aTHX_ WARN_UTF8,
308 "Malformed UTF-8 character (character 0x%04"UVxf")",
311 } else if ((expectlen > UNISKIP(uv)) &&
312 !(flags & UTF8_ALLOW_LONG)) {
314 Perl_warner(aTHX_ WARN_UTF8,
315 "Malformed UTF-8 character (%d byte%s, need %d)",
316 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
324 if (flags & UTF8_CHECK_ONLY) {
331 *retlen = expectlen ? expectlen : len;
333 return UNICODE_REPLACEMENT_CHARACTER;
337 =for apidoc Am|U8* s|utf8_to_uv_simple|STRLEN *retlen
339 Returns the character value of the first character in the string C<s>
340 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
341 length, in bytes, of that character, and the pointer C<s> will be
342 advanced to the end of the character.
344 If C<s> does not point to a well-formed UTF8 character, zero is
345 returned and retlen is set, if possible, to -1.
351 Perl_utf8_to_uv_simple(pTHX_ U8* s, STRLEN* retlen)
353 return Perl_utf8_to_uv(aTHX_ s, UTF8_MAXLEN, retlen, 0);
357 =for apidoc|utf8_length|U8 *s|U8 *e
359 Return the length of the UTF-8 char encoded string C<s> in characters.
360 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
361 up past C<e>, croaks.
367 Perl_utf8_length(pTHX_ U8* s, U8* e)
372 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
377 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
385 /* utf8_distance(a,b) returns the number of UTF8 characters between
386 the pointers a and b */
389 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
398 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
408 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
417 /* WARNING: do not use the following unless you *know* off is within bounds */
420 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
430 while ((*s & 0xc0) == 0x80)
439 =for apidoc Am|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
441 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
442 Unlike C<bytes_to_utf8>, this over-writes the original string, and
443 updates len to contain the new length.
444 Returns zero on failure, setting C<len> to -1.
450 Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len)
456 /* ensure valid UTF8 and chars < 256 before updating string */
457 for (send = s + *len; s < send; ) {
462 ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
475 *d++ = (U8)utf8_to_uv_simple(s, &ulen);
485 =for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
487 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
488 Returns a pointer to the newly-created string, and sets C<len> to
489 reflect the new length.
495 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
503 Newz(801, d, (*len) * 2 + 1, U8);
511 *d++ = (( uv >> 6) | 0xc0);
512 *d++ = (( uv & 0x3f) | 0x80);
521 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
523 * Destination must be pre-extended to 3/2 source. Do not use in-place.
524 * We optimize for native, for obvious reasons. */
527 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
533 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
538 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
545 *d++ = (( uv >> 6) | 0xc0);
546 *d++ = (( uv & 0x3f) | 0x80);
549 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
552 if (low < 0xdc00 || low >= 0xdfff)
553 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
554 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
557 *d++ = (( uv >> 12) | 0xe0);
558 *d++ = (((uv >> 6) & 0x3f) | 0x80);
559 *d++ = (( uv & 0x3f) | 0x80);
563 *d++ = (( uv >> 18) | 0xf0);
564 *d++ = (((uv >> 12) & 0x3f) | 0x80);
565 *d++ = (((uv >> 6) & 0x3f) | 0x80);
566 *d++ = (( uv & 0x3f) | 0x80);
570 *newlen = d - dstart;
574 /* Note: this one is slightly destructive of the source. */
577 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
580 U8* send = s + bytelen;
587 return utf16_to_utf8(p, d, bytelen, newlen);
590 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
593 Perl_is_uni_alnum(pTHX_ U32 c)
595 U8 tmpbuf[UTF8_MAXLEN];
596 uv_to_utf8(tmpbuf, (UV)c);
597 return is_utf8_alnum(tmpbuf);
601 Perl_is_uni_alnumc(pTHX_ U32 c)
603 U8 tmpbuf[UTF8_MAXLEN];
604 uv_to_utf8(tmpbuf, (UV)c);
605 return is_utf8_alnumc(tmpbuf);
609 Perl_is_uni_idfirst(pTHX_ U32 c)
611 U8 tmpbuf[UTF8_MAXLEN];
612 uv_to_utf8(tmpbuf, (UV)c);
613 return is_utf8_idfirst(tmpbuf);
617 Perl_is_uni_alpha(pTHX_ U32 c)
619 U8 tmpbuf[UTF8_MAXLEN];
620 uv_to_utf8(tmpbuf, (UV)c);
621 return is_utf8_alpha(tmpbuf);
625 Perl_is_uni_ascii(pTHX_ U32 c)
627 U8 tmpbuf[UTF8_MAXLEN];
628 uv_to_utf8(tmpbuf, (UV)c);
629 return is_utf8_ascii(tmpbuf);
633 Perl_is_uni_space(pTHX_ U32 c)
635 U8 tmpbuf[UTF8_MAXLEN];
636 uv_to_utf8(tmpbuf, (UV)c);
637 return is_utf8_space(tmpbuf);
641 Perl_is_uni_digit(pTHX_ U32 c)
643 U8 tmpbuf[UTF8_MAXLEN];
644 uv_to_utf8(tmpbuf, (UV)c);
645 return is_utf8_digit(tmpbuf);
649 Perl_is_uni_upper(pTHX_ U32 c)
651 U8 tmpbuf[UTF8_MAXLEN];
652 uv_to_utf8(tmpbuf, (UV)c);
653 return is_utf8_upper(tmpbuf);
657 Perl_is_uni_lower(pTHX_ U32 c)
659 U8 tmpbuf[UTF8_MAXLEN];
660 uv_to_utf8(tmpbuf, (UV)c);
661 return is_utf8_lower(tmpbuf);
665 Perl_is_uni_cntrl(pTHX_ U32 c)
667 U8 tmpbuf[UTF8_MAXLEN];
668 uv_to_utf8(tmpbuf, (UV)c);
669 return is_utf8_cntrl(tmpbuf);
673 Perl_is_uni_graph(pTHX_ U32 c)
675 U8 tmpbuf[UTF8_MAXLEN];
676 uv_to_utf8(tmpbuf, (UV)c);
677 return is_utf8_graph(tmpbuf);
681 Perl_is_uni_print(pTHX_ U32 c)
683 U8 tmpbuf[UTF8_MAXLEN];
684 uv_to_utf8(tmpbuf, (UV)c);
685 return is_utf8_print(tmpbuf);
689 Perl_is_uni_punct(pTHX_ U32 c)
691 U8 tmpbuf[UTF8_MAXLEN];
692 uv_to_utf8(tmpbuf, (UV)c);
693 return is_utf8_punct(tmpbuf);
697 Perl_is_uni_xdigit(pTHX_ U32 c)
699 U8 tmpbuf[UTF8_MAXLEN];
700 uv_to_utf8(tmpbuf, (UV)c);
701 return is_utf8_xdigit(tmpbuf);
705 Perl_to_uni_upper(pTHX_ U32 c)
707 U8 tmpbuf[UTF8_MAXLEN];
708 uv_to_utf8(tmpbuf, (UV)c);
709 return to_utf8_upper(tmpbuf);
713 Perl_to_uni_title(pTHX_ U32 c)
715 U8 tmpbuf[UTF8_MAXLEN];
716 uv_to_utf8(tmpbuf, (UV)c);
717 return to_utf8_title(tmpbuf);
721 Perl_to_uni_lower(pTHX_ U32 c)
723 U8 tmpbuf[UTF8_MAXLEN];
724 uv_to_utf8(tmpbuf, (UV)c);
725 return to_utf8_lower(tmpbuf);
728 /* for now these all assume no locale info available for Unicode > 255 */
731 Perl_is_uni_alnum_lc(pTHX_ U32 c)
733 return is_uni_alnum(c); /* XXX no locale support yet */
737 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
739 return is_uni_alnumc(c); /* XXX no locale support yet */
743 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
745 return is_uni_idfirst(c); /* XXX no locale support yet */
749 Perl_is_uni_alpha_lc(pTHX_ U32 c)
751 return is_uni_alpha(c); /* XXX no locale support yet */
755 Perl_is_uni_ascii_lc(pTHX_ U32 c)
757 return is_uni_ascii(c); /* XXX no locale support yet */
761 Perl_is_uni_space_lc(pTHX_ U32 c)
763 return is_uni_space(c); /* XXX no locale support yet */
767 Perl_is_uni_digit_lc(pTHX_ U32 c)
769 return is_uni_digit(c); /* XXX no locale support yet */
773 Perl_is_uni_upper_lc(pTHX_ U32 c)
775 return is_uni_upper(c); /* XXX no locale support yet */
779 Perl_is_uni_lower_lc(pTHX_ U32 c)
781 return is_uni_lower(c); /* XXX no locale support yet */
785 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
787 return is_uni_cntrl(c); /* XXX no locale support yet */
791 Perl_is_uni_graph_lc(pTHX_ U32 c)
793 return is_uni_graph(c); /* XXX no locale support yet */
797 Perl_is_uni_print_lc(pTHX_ U32 c)
799 return is_uni_print(c); /* XXX no locale support yet */
803 Perl_is_uni_punct_lc(pTHX_ U32 c)
805 return is_uni_punct(c); /* XXX no locale support yet */
809 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
811 return is_uni_xdigit(c); /* XXX no locale support yet */
815 Perl_to_uni_upper_lc(pTHX_ U32 c)
817 return to_uni_upper(c); /* XXX no locale support yet */
821 Perl_to_uni_title_lc(pTHX_ U32 c)
823 return to_uni_title(c); /* XXX no locale support yet */
827 Perl_to_uni_lower_lc(pTHX_ U32 c)
829 return to_uni_lower(c); /* XXX no locale support yet */
833 Perl_is_utf8_alnum(pTHX_ U8 *p)
835 if (!is_utf8_char(p))
838 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
839 * descendant of isalnum(3), in other words, it doesn't
840 * contain the '_'. --jhi */
841 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
842 return swash_fetch(PL_utf8_alnum, p);
843 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
844 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
846 PL_utf8_alnum = swash_init("utf8", "",
847 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
848 return swash_fetch(PL_utf8_alnum, p);
853 Perl_is_utf8_alnumc(pTHX_ U8 *p)
855 if (!is_utf8_char(p))
858 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
859 return swash_fetch(PL_utf8_alnum, p);
860 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
861 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
863 PL_utf8_alnum = swash_init("utf8", "",
864 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
865 return swash_fetch(PL_utf8_alnum, p);
870 Perl_is_utf8_idfirst(pTHX_ U8 *p)
872 return *p == '_' || is_utf8_alpha(p);
876 Perl_is_utf8_alpha(pTHX_ U8 *p)
878 if (!is_utf8_char(p))
881 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
882 return swash_fetch(PL_utf8_alpha, p);
886 Perl_is_utf8_ascii(pTHX_ U8 *p)
888 if (!is_utf8_char(p))
891 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
892 return swash_fetch(PL_utf8_ascii, p);
896 Perl_is_utf8_space(pTHX_ U8 *p)
898 if (!is_utf8_char(p))
901 PL_utf8_space = swash_init("utf8", "IsSpace", &PL_sv_undef, 0, 0);
902 return swash_fetch(PL_utf8_space, p);
906 Perl_is_utf8_digit(pTHX_ U8 *p)
908 if (!is_utf8_char(p))
911 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
912 return swash_fetch(PL_utf8_digit, p);
916 Perl_is_utf8_upper(pTHX_ U8 *p)
918 if (!is_utf8_char(p))
921 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
922 return swash_fetch(PL_utf8_upper, p);
926 Perl_is_utf8_lower(pTHX_ U8 *p)
928 if (!is_utf8_char(p))
931 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
932 return swash_fetch(PL_utf8_lower, p);
936 Perl_is_utf8_cntrl(pTHX_ U8 *p)
938 if (!is_utf8_char(p))
941 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
942 return swash_fetch(PL_utf8_cntrl, p);
946 Perl_is_utf8_graph(pTHX_ U8 *p)
948 if (!is_utf8_char(p))
951 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
952 return swash_fetch(PL_utf8_graph, p);
956 Perl_is_utf8_print(pTHX_ U8 *p)
958 if (!is_utf8_char(p))
961 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
962 return swash_fetch(PL_utf8_print, p);
966 Perl_is_utf8_punct(pTHX_ U8 *p)
968 if (!is_utf8_char(p))
971 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
972 return swash_fetch(PL_utf8_punct, p);
976 Perl_is_utf8_xdigit(pTHX_ U8 *p)
978 if (!is_utf8_char(p))
981 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
982 return swash_fetch(PL_utf8_xdigit, p);
986 Perl_is_utf8_mark(pTHX_ U8 *p)
988 if (!is_utf8_char(p))
991 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
992 return swash_fetch(PL_utf8_mark, p);
996 Perl_to_utf8_upper(pTHX_ U8 *p)
1000 if (!PL_utf8_toupper)
1001 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
1002 uv = swash_fetch(PL_utf8_toupper, p);
1003 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1007 Perl_to_utf8_title(pTHX_ U8 *p)
1011 if (!PL_utf8_totitle)
1012 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
1013 uv = swash_fetch(PL_utf8_totitle, p);
1014 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1018 Perl_to_utf8_lower(pTHX_ U8 *p)
1022 if (!PL_utf8_tolower)
1023 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
1024 uv = swash_fetch(PL_utf8_tolower, p);
1025 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1028 /* a "swash" is a swatch hash */
1031 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1037 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
1039 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1043 PUSHSTACKi(PERLSI_MAGIC);
1046 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1047 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1049 PUSHs(sv_2mortal(newSViv(minbits)));
1050 PUSHs(sv_2mortal(newSViv(none)));
1056 if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */
1057 strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf);
1058 if (call_method("SWASHNEW", G_SCALAR))
1059 retval = newSVsv(*PL_stack_sp--);
1061 retval = &PL_sv_undef;
1064 if (PL_curcop == &PL_compiling) {
1065 strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf);
1066 PL_curcop->op_private = PL_hints;
1068 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1069 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1074 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1076 HV* hv = (HV*)SvRV(sv);
1077 U32 klen = UTF8SKIP(ptr) - 1;
1078 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1080 STRLEN needents = (klen ? 64 : 128);
1086 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1087 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1088 * it's nothing to sniff at.) Pity we usually come through at least
1089 * two function calls to get here...
1091 * NB: this code assumes that swatches are never modified, once generated!
1094 if (hv == PL_last_swash_hv &&
1095 klen == PL_last_swash_klen &&
1096 (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
1098 tmps = PL_last_swash_tmps;
1099 slen = PL_last_swash_slen;
1102 /* Try our second-level swatch cache, kept in a hash. */
1103 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1105 /* If not cached, generate it via utf8::SWASHGET */
1106 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1111 PUSHSTACKi(PERLSI_MAGIC);
1115 PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, UTF8_MAXLEN, 0, 0) & ~(needents - 1))));
1116 PUSHs(sv_2mortal(newSViv(needents)));
1118 if (call_method("SWASHGET", G_SCALAR))
1119 retval = newSVsv(*PL_stack_sp--);
1121 retval = &PL_sv_undef;
1125 if (PL_curcop == &PL_compiling)
1126 PL_curcop->op_private = PL_hints;
1128 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1130 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1131 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1134 PL_last_swash_hv = hv;
1135 PL_last_swash_klen = klen;
1136 PL_last_swash_tmps = tmps;
1137 PL_last_swash_slen = slen;
1139 Copy(ptr, PL_last_swash_key, klen, U8);
1142 switch ((int)((slen << 3) / needents)) {
1144 bit = 1 << (off & 7);
1146 return (tmps[off] & bit) != 0;
1151 return (tmps[off] << 8) + tmps[off + 1] ;
1154 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1156 Perl_croak(aTHX_ "panic: swash_fetch");