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 < 0x1000000000LL)
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 (UTF8LEN(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_chk|STRLEN curlen|I32 *retlen|I32 checking
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<checking>: if this is true, it is
183 assumed that the caller will raise a warning, and this function will
184 set C<retlen> to C<-1> and return. If C<checking> is not true, an optional UTF8
191 Perl_utf8_to_uv_chk(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, bool checking)
196 bool dowarn = ckWARN_d(WARN_UTF8);
197 STRLEN expectlen = 0;
199 if (uv <= 0x7f) { /* Pure ASCII. */
205 if (uv >= 0x80 && uv <= 0xbf) {
207 Perl_warner(aTHX_ WARN_UTF8,
208 "Malformed UTF-8 character (unexpected continuation byte 0x%02x)",
213 if (uv >= 0xc0 && uv <= 0xfd && curlen > 1 && s[1] < 0x80) {
215 Perl_warner(aTHX_ WARN_UTF8,
216 "Malformed UTF-8 character (unexpected non-continuation byte 0x%02x after byte 0x%02x)",
221 if ((uv == 0xfe || uv == 0xff) && IN_UTF8){
223 Perl_warner(aTHX_ WARN_UTF8,
224 "Malformed UTF-8 character (impossible byte 0x%02x)",
229 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
230 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
231 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
232 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
233 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
234 else if (!(uv & 0x01)) { len = 7; uv = 0; }
235 else { len = 13; uv = 0; } /* whoa! */
242 if (curlen < expectlen) {
244 Perl_warner(aTHX_ WARN_UTF8,
245 "Malformed UTF-8 character (%d byte%s, need %d)",
246 curlen, curlen > 1 ? "s" : "", expectlen);
255 if ((*s & 0xc0) != 0x80) {
257 Perl_warner(aTHX_ WARN_UTF8,
258 "Malformed UTF-8 character (unexpected continuation byte 0x%02x)",
263 uv = (uv << 6) | (*s & 0x3f);
266 Perl_warner(aTHX_ WARN_UTF8,
267 "Malformed UTF-8 character (overflow at 0x%"UVxf", byte 0x%02x)",
275 if (uv >= 0xd800 && uv <= 0xdfff) {
277 Perl_warner(aTHX_ WARN_UTF8,
278 "Malformed UTF-8 character (UTF-16 surrogate 0x%04"UVxf")",
281 } else if (uv == 0xfffe) {
283 Perl_warner(aTHX_ WARN_UTF8,
284 "Malformed UTF-8 character (byte order mark 0x%04"UVxf")",
287 } else if (uv == 0xffff && IN_UTF8) {
289 Perl_warner(aTHX_ WARN_UTF8,
290 "Malformed UTF-8 character (impossible character 0x%04"UVxf")",
293 } else if (expectlen > UTF8LEN(uv)) {
295 Perl_warner(aTHX_ WARN_UTF8,
296 "Malformed UTF-8 character (%d byte%s, need %d)",
297 expectlen, expectlen > 1 ? "s": "", UTF8LEN(uv));
314 return UNICODE_REPLACEMENT_CHARACTER;
318 =for apidoc Am|U8* s|utf8_to_uv|STRLEN *retlen
320 Returns the character value of the first character in the string C<s>
321 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
322 length, in bytes, of that character, and the pointer C<s> will be
323 advanced to the end of the character.
325 If C<s> does not point to a well-formed UTF8 character, zero is
326 returned and retlen is set, if possible, to -1.
332 Perl_utf8_to_uv(pTHX_ U8* s, STRLEN* retlen)
334 return Perl_utf8_to_uv_chk(aTHX_ s, (STRLEN)-1, retlen, 0);
337 /* utf8_distance(a,b) returns the number of UTF8 characters between
338 the pointers a and b */
341 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
359 /* WARNING: do not use the following unless you *know* off is within bounds */
362 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
372 while ((*s & 0xc0) == 0x80)
381 =for apidoc Am|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
383 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
384 Unlike C<bytes_to_utf8>, this over-writes the original string, and
385 updates len to contain the new length.
386 Returns zero on failure, setting C<len> to -1.
392 Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len)
402 /* ensure valid UTF8 and chars < 256 before updating string */
406 ( (s >= send) || ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
417 *d++ = (U8)utf8_to_uv(s, &ulen);
427 =for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
429 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
430 Returns a pointer to the newly-created string, and sets C<len> to
431 reflect the new length.
437 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
445 Newz(801, d, (*len) * 2 + 1, U8);
453 *d++ = (( uv >> 6) | 0xc0);
454 *d++ = (( uv & 0x3f) | 0x80);
463 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
465 * Destination must be pre-extended to 3/2 source. Do not use in-place.
466 * We optimize for native, for obvious reasons. */
469 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
475 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
480 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
487 *d++ = (( uv >> 6) | 0xc0);
488 *d++ = (( uv & 0x3f) | 0x80);
491 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
494 if (low < 0xdc00 || low >= 0xdfff)
495 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
496 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
499 *d++ = (( uv >> 12) | 0xe0);
500 *d++ = (((uv >> 6) & 0x3f) | 0x80);
501 *d++ = (( uv & 0x3f) | 0x80);
505 *d++ = (( uv >> 18) | 0xf0);
506 *d++ = (((uv >> 12) & 0x3f) | 0x80);
507 *d++ = (((uv >> 6) & 0x3f) | 0x80);
508 *d++ = (( uv & 0x3f) | 0x80);
512 *newlen = d - dstart;
516 /* Note: this one is slightly destructive of the source. */
519 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
522 U8* send = s + bytelen;
529 return utf16_to_utf8(p, d, bytelen, newlen);
532 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
535 Perl_is_uni_alnum(pTHX_ U32 c)
537 U8 tmpbuf[UTF8_MAXLEN];
538 uv_to_utf8(tmpbuf, (UV)c);
539 return is_utf8_alnum(tmpbuf);
543 Perl_is_uni_alnumc(pTHX_ U32 c)
545 U8 tmpbuf[UTF8_MAXLEN];
546 uv_to_utf8(tmpbuf, (UV)c);
547 return is_utf8_alnumc(tmpbuf);
551 Perl_is_uni_idfirst(pTHX_ U32 c)
553 U8 tmpbuf[UTF8_MAXLEN];
554 uv_to_utf8(tmpbuf, (UV)c);
555 return is_utf8_idfirst(tmpbuf);
559 Perl_is_uni_alpha(pTHX_ U32 c)
561 U8 tmpbuf[UTF8_MAXLEN];
562 uv_to_utf8(tmpbuf, (UV)c);
563 return is_utf8_alpha(tmpbuf);
567 Perl_is_uni_ascii(pTHX_ U32 c)
569 U8 tmpbuf[UTF8_MAXLEN];
570 uv_to_utf8(tmpbuf, (UV)c);
571 return is_utf8_ascii(tmpbuf);
575 Perl_is_uni_space(pTHX_ U32 c)
577 U8 tmpbuf[UTF8_MAXLEN];
578 uv_to_utf8(tmpbuf, (UV)c);
579 return is_utf8_space(tmpbuf);
583 Perl_is_uni_digit(pTHX_ U32 c)
585 U8 tmpbuf[UTF8_MAXLEN];
586 uv_to_utf8(tmpbuf, (UV)c);
587 return is_utf8_digit(tmpbuf);
591 Perl_is_uni_upper(pTHX_ U32 c)
593 U8 tmpbuf[UTF8_MAXLEN];
594 uv_to_utf8(tmpbuf, (UV)c);
595 return is_utf8_upper(tmpbuf);
599 Perl_is_uni_lower(pTHX_ U32 c)
601 U8 tmpbuf[UTF8_MAXLEN];
602 uv_to_utf8(tmpbuf, (UV)c);
603 return is_utf8_lower(tmpbuf);
607 Perl_is_uni_cntrl(pTHX_ U32 c)
609 U8 tmpbuf[UTF8_MAXLEN];
610 uv_to_utf8(tmpbuf, (UV)c);
611 return is_utf8_cntrl(tmpbuf);
615 Perl_is_uni_graph(pTHX_ U32 c)
617 U8 tmpbuf[UTF8_MAXLEN];
618 uv_to_utf8(tmpbuf, (UV)c);
619 return is_utf8_graph(tmpbuf);
623 Perl_is_uni_print(pTHX_ U32 c)
625 U8 tmpbuf[UTF8_MAXLEN];
626 uv_to_utf8(tmpbuf, (UV)c);
627 return is_utf8_print(tmpbuf);
631 Perl_is_uni_punct(pTHX_ U32 c)
633 U8 tmpbuf[UTF8_MAXLEN];
634 uv_to_utf8(tmpbuf, (UV)c);
635 return is_utf8_punct(tmpbuf);
639 Perl_is_uni_xdigit(pTHX_ U32 c)
641 U8 tmpbuf[UTF8_MAXLEN];
642 uv_to_utf8(tmpbuf, (UV)c);
643 return is_utf8_xdigit(tmpbuf);
647 Perl_to_uni_upper(pTHX_ U32 c)
649 U8 tmpbuf[UTF8_MAXLEN];
650 uv_to_utf8(tmpbuf, (UV)c);
651 return to_utf8_upper(tmpbuf);
655 Perl_to_uni_title(pTHX_ U32 c)
657 U8 tmpbuf[UTF8_MAXLEN];
658 uv_to_utf8(tmpbuf, (UV)c);
659 return to_utf8_title(tmpbuf);
663 Perl_to_uni_lower(pTHX_ U32 c)
665 U8 tmpbuf[UTF8_MAXLEN];
666 uv_to_utf8(tmpbuf, (UV)c);
667 return to_utf8_lower(tmpbuf);
670 /* for now these all assume no locale info available for Unicode > 255 */
673 Perl_is_uni_alnum_lc(pTHX_ U32 c)
675 return is_uni_alnum(c); /* XXX no locale support yet */
679 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
681 return is_uni_alnumc(c); /* XXX no locale support yet */
685 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
687 return is_uni_idfirst(c); /* XXX no locale support yet */
691 Perl_is_uni_alpha_lc(pTHX_ U32 c)
693 return is_uni_alpha(c); /* XXX no locale support yet */
697 Perl_is_uni_ascii_lc(pTHX_ U32 c)
699 return is_uni_ascii(c); /* XXX no locale support yet */
703 Perl_is_uni_space_lc(pTHX_ U32 c)
705 return is_uni_space(c); /* XXX no locale support yet */
709 Perl_is_uni_digit_lc(pTHX_ U32 c)
711 return is_uni_digit(c); /* XXX no locale support yet */
715 Perl_is_uni_upper_lc(pTHX_ U32 c)
717 return is_uni_upper(c); /* XXX no locale support yet */
721 Perl_is_uni_lower_lc(pTHX_ U32 c)
723 return is_uni_lower(c); /* XXX no locale support yet */
727 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
729 return is_uni_cntrl(c); /* XXX no locale support yet */
733 Perl_is_uni_graph_lc(pTHX_ U32 c)
735 return is_uni_graph(c); /* XXX no locale support yet */
739 Perl_is_uni_print_lc(pTHX_ U32 c)
741 return is_uni_print(c); /* XXX no locale support yet */
745 Perl_is_uni_punct_lc(pTHX_ U32 c)
747 return is_uni_punct(c); /* XXX no locale support yet */
751 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
753 return is_uni_xdigit(c); /* XXX no locale support yet */
757 Perl_to_uni_upper_lc(pTHX_ U32 c)
759 return to_uni_upper(c); /* XXX no locale support yet */
763 Perl_to_uni_title_lc(pTHX_ U32 c)
765 return to_uni_title(c); /* XXX no locale support yet */
769 Perl_to_uni_lower_lc(pTHX_ U32 c)
771 return to_uni_lower(c); /* XXX no locale support yet */
775 Perl_is_utf8_alnum(pTHX_ U8 *p)
777 if (!is_utf8_char(p))
780 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
781 * descendant of isalnum(3), in other words, it doesn't
782 * contain the '_'. --jhi */
783 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
784 return swash_fetch(PL_utf8_alnum, p);
785 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
786 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
788 PL_utf8_alnum = swash_init("utf8", "",
789 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
790 return swash_fetch(PL_utf8_alnum, p);
795 Perl_is_utf8_alnumc(pTHX_ U8 *p)
797 if (!is_utf8_char(p))
800 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
801 return swash_fetch(PL_utf8_alnum, p);
802 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
803 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
805 PL_utf8_alnum = swash_init("utf8", "",
806 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
807 return swash_fetch(PL_utf8_alnum, p);
812 Perl_is_utf8_idfirst(pTHX_ U8 *p)
814 return *p == '_' || is_utf8_alpha(p);
818 Perl_is_utf8_alpha(pTHX_ U8 *p)
820 if (!is_utf8_char(p))
823 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
824 return swash_fetch(PL_utf8_alpha, p);
828 Perl_is_utf8_ascii(pTHX_ U8 *p)
830 if (!is_utf8_char(p))
833 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
834 return swash_fetch(PL_utf8_ascii, p);
838 Perl_is_utf8_space(pTHX_ U8 *p)
840 if (!is_utf8_char(p))
843 PL_utf8_space = swash_init("utf8", "IsSpace", &PL_sv_undef, 0, 0);
844 return swash_fetch(PL_utf8_space, p);
848 Perl_is_utf8_digit(pTHX_ U8 *p)
850 if (!is_utf8_char(p))
853 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
854 return swash_fetch(PL_utf8_digit, p);
858 Perl_is_utf8_upper(pTHX_ U8 *p)
860 if (!is_utf8_char(p))
863 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
864 return swash_fetch(PL_utf8_upper, p);
868 Perl_is_utf8_lower(pTHX_ U8 *p)
870 if (!is_utf8_char(p))
873 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
874 return swash_fetch(PL_utf8_lower, p);
878 Perl_is_utf8_cntrl(pTHX_ U8 *p)
880 if (!is_utf8_char(p))
883 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
884 return swash_fetch(PL_utf8_cntrl, p);
888 Perl_is_utf8_graph(pTHX_ U8 *p)
890 if (!is_utf8_char(p))
893 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
894 return swash_fetch(PL_utf8_graph, p);
898 Perl_is_utf8_print(pTHX_ U8 *p)
900 if (!is_utf8_char(p))
903 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
904 return swash_fetch(PL_utf8_print, p);
908 Perl_is_utf8_punct(pTHX_ U8 *p)
910 if (!is_utf8_char(p))
913 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
914 return swash_fetch(PL_utf8_punct, p);
918 Perl_is_utf8_xdigit(pTHX_ U8 *p)
920 if (!is_utf8_char(p))
923 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
924 return swash_fetch(PL_utf8_xdigit, p);
928 Perl_is_utf8_mark(pTHX_ U8 *p)
930 if (!is_utf8_char(p))
933 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
934 return swash_fetch(PL_utf8_mark, p);
938 Perl_to_utf8_upper(pTHX_ U8 *p)
942 if (!PL_utf8_toupper)
943 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
944 uv = swash_fetch(PL_utf8_toupper, p);
945 return uv ? uv : utf8_to_uv_chk(p,STRLEN_MAX,0,0);
949 Perl_to_utf8_title(pTHX_ U8 *p)
953 if (!PL_utf8_totitle)
954 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
955 uv = swash_fetch(PL_utf8_totitle, p);
956 return uv ? uv : utf8_to_uv_chk(p,STRLEN_MAX,0,0);
960 Perl_to_utf8_lower(pTHX_ U8 *p)
964 if (!PL_utf8_tolower)
965 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
966 uv = swash_fetch(PL_utf8_tolower, p);
967 return uv ? uv : utf8_to_uv_chk(p,STRLEN_MAX,0,0);
970 /* a "swash" is a swatch hash */
973 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
979 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
981 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
985 PUSHSTACKi(PERLSI_MAGIC);
988 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
989 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
991 PUSHs(sv_2mortal(newSViv(minbits)));
992 PUSHs(sv_2mortal(newSViv(none)));
998 if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */
999 strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf);
1000 if (call_method("SWASHNEW", G_SCALAR))
1001 retval = newSVsv(*PL_stack_sp--);
1003 retval = &PL_sv_undef;
1006 if (PL_curcop == &PL_compiling) {
1007 strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf);
1008 PL_curcop->op_private = PL_hints;
1010 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1011 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1016 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1018 HV* hv = (HV*)SvRV(sv);
1019 U32 klen = UTF8SKIP(ptr) - 1;
1020 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1022 STRLEN needents = (klen ? 64 : 128);
1028 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1029 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1030 * it's nothing to sniff at.) Pity we usually come through at least
1031 * two function calls to get here...
1033 * NB: this code assumes that swatches are never modified, once generated!
1036 if (hv == PL_last_swash_hv &&
1037 klen == PL_last_swash_klen &&
1038 (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
1040 tmps = PL_last_swash_tmps;
1041 slen = PL_last_swash_slen;
1044 /* Try our second-level swatch cache, kept in a hash. */
1045 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1047 /* If not cached, generate it via utf8::SWASHGET */
1048 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1053 PUSHSTACKi(PERLSI_MAGIC);
1057 PUSHs(sv_2mortal(newSViv(utf8_to_uv_chk(ptr, STRLEN_MAX, 0, 0) & ~(needents - 1))));
1058 PUSHs(sv_2mortal(newSViv(needents)));
1060 if (call_method("SWASHGET", G_SCALAR))
1061 retval = newSVsv(*PL_stack_sp--);
1063 retval = &PL_sv_undef;
1067 if (PL_curcop == &PL_compiling)
1068 PL_curcop->op_private = PL_hints;
1070 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1072 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1073 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1076 PL_last_swash_hv = hv;
1077 PL_last_swash_klen = klen;
1078 PL_last_swash_tmps = tmps;
1079 PL_last_swash_slen = slen;
1081 Copy(ptr, PL_last_swash_key, klen, U8);
1084 switch ((slen << 3) / needents) {
1086 bit = 1 << (off & 7);
1088 return (tmps[off] & bit) != 0;
1093 return (tmps[off] << 8) + tmps[off + 1] ;
1096 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1098 Perl_croak(aTHX_ "panic: swash_fetch");