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|STRLEN *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
191 Perl_utf8_to_uv(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags)
199 bool dowarn = ckWARN_d(WARN_UTF8);
201 STRLEN expectlen = 0;
205 Perl_warner(aTHX_ WARN_UTF8,
206 "Malformed UTF-8 character (an empty string)");
210 if (uv <= 0x7f) { /* Pure ASCII. */
216 if ((uv >= 0x80 && uv <= 0xbf) &&
217 !(flags & UTF8_ALLOW_CONTINUATION)) {
219 Perl_warner(aTHX_ WARN_UTF8,
220 "Malformed UTF-8 character (unexpected continuation byte 0x%02"UVxf")",
225 if ((uv >= 0xc0 && uv <= 0xfd && curlen > 1 && s[1] < 0x80) &&
226 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
228 Perl_warner(aTHX_ WARN_UTF8,
229 "Malformed UTF-8 character (unexpected non-continuation byte 0x%02"UVxf" after byte 0x%02"UVxf")",
234 if ((uv == 0xfe || uv == 0xff) &&
235 !(flags & UTF8_ALLOW_FE_FF)) {
237 Perl_warner(aTHX_ WARN_UTF8,
238 "Malformed UTF-8 character (byte 0x%02"UVxf")",
243 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
244 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
245 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
246 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
247 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
248 else if (!(uv & 0x01)) { len = 7; uv = 0; }
249 else { len = 13; uv = 0; } /* whoa! */
256 if ((curlen < expectlen) &&
257 !(flags & UTF8_ALLOW_SHORT)) {
259 Perl_warner(aTHX_ WARN_UTF8,
260 "Malformed UTF-8 character (%d byte%s, need %d)",
261 curlen, curlen == 1 ? "" : "s", expectlen);
270 if ((*s & 0xc0) != 0x80) {
272 Perl_warner(aTHX_ WARN_UTF8,
273 "Malformed UTF-8 character (unexpected continuation byte 0x%02x)",
278 uv = (uv << 6) | (*s & 0x3f);
280 /* This cannot be allowed. */
282 Perl_warner(aTHX_ WARN_UTF8,
283 "Malformed UTF-8 character (overflow at 0x%"UVxf", byte 0x%02x)",
291 if ((uv >= 0xd800 && uv <= 0xdfff) &&
292 !(flags & UTF8_ALLOW_SURROGATE)) {
294 Perl_warner(aTHX_ WARN_UTF8,
295 "Malformed UTF-8 character (UTF-16 surrogate 0x%04"UVxf")",
298 } else if ((uv == 0xfffe) &&
299 !(flags & UTF8_ALLOW_BOM)) {
301 Perl_warner(aTHX_ WARN_UTF8,
302 "Malformed UTF-8 character (byte order mark 0x%04"UVxf")",
305 } else if ((uv == 0xffff) &&
306 !(flags & UTF8_ALLOW_FFFF)) {
308 Perl_warner(aTHX_ WARN_UTF8,
309 "Malformed UTF-8 character (character 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));
325 if (flags & UTF8_CHECK_ONLY) {
332 *retlen = expectlen ? expectlen : len;
334 return UNICODE_REPLACEMENT_CHARACTER;
338 =for apidoc Am|U8* s|utf8_to_uv_simple|STRLEN *retlen
340 Returns the character value of the first character in the string C<s>
341 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
342 length, in bytes, of that character, and the pointer C<s> will be
343 advanced to the end of the character.
345 If C<s> does not point to a well-formed UTF8 character, zero is
346 returned and retlen is set, if possible, to -1.
352 Perl_utf8_to_uv_simple(pTHX_ U8* s, STRLEN* retlen)
354 return Perl_utf8_to_uv(aTHX_ s, UTF8_MAXLEN, retlen, 0);
358 =for apidoc|utf8_length|U8 *s|U8 *e
360 Return the length of the UTF-8 char encoded string C<s> in characters.
361 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
362 up past C<e>, croaks.
368 Perl_utf8_length(pTHX_ U8* s, U8* e)
373 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
378 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
386 /* utf8_distance(a,b) returns the number of UTF8 characters between
387 the pointers a and b */
390 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
399 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
409 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
418 /* WARNING: do not use the following unless you *know* off is within bounds */
421 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
431 while ((*s & 0xc0) == 0x80)
440 =for apidoc Am|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
442 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
443 Unlike C<bytes_to_utf8>, this over-writes the original string, and
444 updates len to contain the new length.
445 Returns zero on failure, setting C<len> to -1.
451 Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len)
457 /* ensure valid UTF8 and chars < 256 before updating string */
458 for (send = s + *len; s < send; ) {
463 ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
476 *d++ = (U8)utf8_to_uv_simple(s, &ulen);
486 =for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
488 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
489 Returns a pointer to the newly-created string, and sets C<len> to
490 reflect the new length.
496 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
504 Newz(801, d, (*len) * 2 + 1, U8);
512 *d++ = (( uv >> 6) | 0xc0);
513 *d++ = (( uv & 0x3f) | 0x80);
522 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
524 * Destination must be pre-extended to 3/2 source. Do not use in-place.
525 * We optimize for native, for obvious reasons. */
528 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
534 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
539 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
546 *d++ = (( uv >> 6) | 0xc0);
547 *d++ = (( uv & 0x3f) | 0x80);
550 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
553 if (low < 0xdc00 || low >= 0xdfff)
554 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
555 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
558 *d++ = (( uv >> 12) | 0xe0);
559 *d++ = (((uv >> 6) & 0x3f) | 0x80);
560 *d++ = (( uv & 0x3f) | 0x80);
564 *d++ = (( uv >> 18) | 0xf0);
565 *d++ = (((uv >> 12) & 0x3f) | 0x80);
566 *d++ = (((uv >> 6) & 0x3f) | 0x80);
567 *d++ = (( uv & 0x3f) | 0x80);
571 *newlen = d - dstart;
575 /* Note: this one is slightly destructive of the source. */
578 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
581 U8* send = s + bytelen;
588 return utf16_to_utf8(p, d, bytelen, newlen);
591 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
594 Perl_is_uni_alnum(pTHX_ U32 c)
596 U8 tmpbuf[UTF8_MAXLEN];
597 uv_to_utf8(tmpbuf, (UV)c);
598 return is_utf8_alnum(tmpbuf);
602 Perl_is_uni_alnumc(pTHX_ U32 c)
604 U8 tmpbuf[UTF8_MAXLEN];
605 uv_to_utf8(tmpbuf, (UV)c);
606 return is_utf8_alnumc(tmpbuf);
610 Perl_is_uni_idfirst(pTHX_ U32 c)
612 U8 tmpbuf[UTF8_MAXLEN];
613 uv_to_utf8(tmpbuf, (UV)c);
614 return is_utf8_idfirst(tmpbuf);
618 Perl_is_uni_alpha(pTHX_ U32 c)
620 U8 tmpbuf[UTF8_MAXLEN];
621 uv_to_utf8(tmpbuf, (UV)c);
622 return is_utf8_alpha(tmpbuf);
626 Perl_is_uni_ascii(pTHX_ U32 c)
628 U8 tmpbuf[UTF8_MAXLEN];
629 uv_to_utf8(tmpbuf, (UV)c);
630 return is_utf8_ascii(tmpbuf);
634 Perl_is_uni_space(pTHX_ U32 c)
636 U8 tmpbuf[UTF8_MAXLEN];
637 uv_to_utf8(tmpbuf, (UV)c);
638 return is_utf8_space(tmpbuf);
642 Perl_is_uni_digit(pTHX_ U32 c)
644 U8 tmpbuf[UTF8_MAXLEN];
645 uv_to_utf8(tmpbuf, (UV)c);
646 return is_utf8_digit(tmpbuf);
650 Perl_is_uni_upper(pTHX_ U32 c)
652 U8 tmpbuf[UTF8_MAXLEN];
653 uv_to_utf8(tmpbuf, (UV)c);
654 return is_utf8_upper(tmpbuf);
658 Perl_is_uni_lower(pTHX_ U32 c)
660 U8 tmpbuf[UTF8_MAXLEN];
661 uv_to_utf8(tmpbuf, (UV)c);
662 return is_utf8_lower(tmpbuf);
666 Perl_is_uni_cntrl(pTHX_ U32 c)
668 U8 tmpbuf[UTF8_MAXLEN];
669 uv_to_utf8(tmpbuf, (UV)c);
670 return is_utf8_cntrl(tmpbuf);
674 Perl_is_uni_graph(pTHX_ U32 c)
676 U8 tmpbuf[UTF8_MAXLEN];
677 uv_to_utf8(tmpbuf, (UV)c);
678 return is_utf8_graph(tmpbuf);
682 Perl_is_uni_print(pTHX_ U32 c)
684 U8 tmpbuf[UTF8_MAXLEN];
685 uv_to_utf8(tmpbuf, (UV)c);
686 return is_utf8_print(tmpbuf);
690 Perl_is_uni_punct(pTHX_ U32 c)
692 U8 tmpbuf[UTF8_MAXLEN];
693 uv_to_utf8(tmpbuf, (UV)c);
694 return is_utf8_punct(tmpbuf);
698 Perl_is_uni_xdigit(pTHX_ U32 c)
700 U8 tmpbuf[UTF8_MAXLEN];
701 uv_to_utf8(tmpbuf, (UV)c);
702 return is_utf8_xdigit(tmpbuf);
706 Perl_to_uni_upper(pTHX_ U32 c)
708 U8 tmpbuf[UTF8_MAXLEN];
709 uv_to_utf8(tmpbuf, (UV)c);
710 return to_utf8_upper(tmpbuf);
714 Perl_to_uni_title(pTHX_ U32 c)
716 U8 tmpbuf[UTF8_MAXLEN];
717 uv_to_utf8(tmpbuf, (UV)c);
718 return to_utf8_title(tmpbuf);
722 Perl_to_uni_lower(pTHX_ U32 c)
724 U8 tmpbuf[UTF8_MAXLEN];
725 uv_to_utf8(tmpbuf, (UV)c);
726 return to_utf8_lower(tmpbuf);
729 /* for now these all assume no locale info available for Unicode > 255 */
732 Perl_is_uni_alnum_lc(pTHX_ U32 c)
734 return is_uni_alnum(c); /* XXX no locale support yet */
738 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
740 return is_uni_alnumc(c); /* XXX no locale support yet */
744 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
746 return is_uni_idfirst(c); /* XXX no locale support yet */
750 Perl_is_uni_alpha_lc(pTHX_ U32 c)
752 return is_uni_alpha(c); /* XXX no locale support yet */
756 Perl_is_uni_ascii_lc(pTHX_ U32 c)
758 return is_uni_ascii(c); /* XXX no locale support yet */
762 Perl_is_uni_space_lc(pTHX_ U32 c)
764 return is_uni_space(c); /* XXX no locale support yet */
768 Perl_is_uni_digit_lc(pTHX_ U32 c)
770 return is_uni_digit(c); /* XXX no locale support yet */
774 Perl_is_uni_upper_lc(pTHX_ U32 c)
776 return is_uni_upper(c); /* XXX no locale support yet */
780 Perl_is_uni_lower_lc(pTHX_ U32 c)
782 return is_uni_lower(c); /* XXX no locale support yet */
786 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
788 return is_uni_cntrl(c); /* XXX no locale support yet */
792 Perl_is_uni_graph_lc(pTHX_ U32 c)
794 return is_uni_graph(c); /* XXX no locale support yet */
798 Perl_is_uni_print_lc(pTHX_ U32 c)
800 return is_uni_print(c); /* XXX no locale support yet */
804 Perl_is_uni_punct_lc(pTHX_ U32 c)
806 return is_uni_punct(c); /* XXX no locale support yet */
810 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
812 return is_uni_xdigit(c); /* XXX no locale support yet */
816 Perl_to_uni_upper_lc(pTHX_ U32 c)
818 return to_uni_upper(c); /* XXX no locale support yet */
822 Perl_to_uni_title_lc(pTHX_ U32 c)
824 return to_uni_title(c); /* XXX no locale support yet */
828 Perl_to_uni_lower_lc(pTHX_ U32 c)
830 return to_uni_lower(c); /* XXX no locale support yet */
834 Perl_is_utf8_alnum(pTHX_ U8 *p)
836 if (!is_utf8_char(p))
839 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
840 * descendant of isalnum(3), in other words, it doesn't
841 * contain the '_'. --jhi */
842 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
843 return swash_fetch(PL_utf8_alnum, p);
844 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
845 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
847 PL_utf8_alnum = swash_init("utf8", "",
848 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
849 return swash_fetch(PL_utf8_alnum, p);
854 Perl_is_utf8_alnumc(pTHX_ U8 *p)
856 if (!is_utf8_char(p))
859 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
860 return swash_fetch(PL_utf8_alnum, p);
861 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
862 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
864 PL_utf8_alnum = swash_init("utf8", "",
865 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
866 return swash_fetch(PL_utf8_alnum, p);
871 Perl_is_utf8_idfirst(pTHX_ U8 *p)
873 return *p == '_' || is_utf8_alpha(p);
877 Perl_is_utf8_alpha(pTHX_ U8 *p)
879 if (!is_utf8_char(p))
882 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
883 return swash_fetch(PL_utf8_alpha, p);
887 Perl_is_utf8_ascii(pTHX_ U8 *p)
889 if (!is_utf8_char(p))
892 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
893 return swash_fetch(PL_utf8_ascii, p);
897 Perl_is_utf8_space(pTHX_ U8 *p)
899 if (!is_utf8_char(p))
902 PL_utf8_space = swash_init("utf8", "IsSpace", &PL_sv_undef, 0, 0);
903 return swash_fetch(PL_utf8_space, p);
907 Perl_is_utf8_digit(pTHX_ U8 *p)
909 if (!is_utf8_char(p))
912 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
913 return swash_fetch(PL_utf8_digit, p);
917 Perl_is_utf8_upper(pTHX_ U8 *p)
919 if (!is_utf8_char(p))
922 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
923 return swash_fetch(PL_utf8_upper, p);
927 Perl_is_utf8_lower(pTHX_ U8 *p)
929 if (!is_utf8_char(p))
932 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
933 return swash_fetch(PL_utf8_lower, p);
937 Perl_is_utf8_cntrl(pTHX_ U8 *p)
939 if (!is_utf8_char(p))
942 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
943 return swash_fetch(PL_utf8_cntrl, p);
947 Perl_is_utf8_graph(pTHX_ U8 *p)
949 if (!is_utf8_char(p))
952 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
953 return swash_fetch(PL_utf8_graph, p);
957 Perl_is_utf8_print(pTHX_ U8 *p)
959 if (!is_utf8_char(p))
962 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
963 return swash_fetch(PL_utf8_print, p);
967 Perl_is_utf8_punct(pTHX_ U8 *p)
969 if (!is_utf8_char(p))
972 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
973 return swash_fetch(PL_utf8_punct, p);
977 Perl_is_utf8_xdigit(pTHX_ U8 *p)
979 if (!is_utf8_char(p))
982 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
983 return swash_fetch(PL_utf8_xdigit, p);
987 Perl_is_utf8_mark(pTHX_ U8 *p)
989 if (!is_utf8_char(p))
992 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
993 return swash_fetch(PL_utf8_mark, p);
997 Perl_to_utf8_upper(pTHX_ U8 *p)
1001 if (!PL_utf8_toupper)
1002 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
1003 uv = swash_fetch(PL_utf8_toupper, p);
1004 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1008 Perl_to_utf8_title(pTHX_ U8 *p)
1012 if (!PL_utf8_totitle)
1013 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
1014 uv = swash_fetch(PL_utf8_totitle, p);
1015 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1019 Perl_to_utf8_lower(pTHX_ U8 *p)
1023 if (!PL_utf8_tolower)
1024 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
1025 uv = swash_fetch(PL_utf8_tolower, p);
1026 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1029 /* a "swash" is a swatch hash */
1032 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1038 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
1040 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1044 PUSHSTACKi(PERLSI_MAGIC);
1047 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1048 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1050 PUSHs(sv_2mortal(newSViv(minbits)));
1051 PUSHs(sv_2mortal(newSViv(none)));
1057 if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */
1058 strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf);
1059 if (call_method("SWASHNEW", G_SCALAR))
1060 retval = newSVsv(*PL_stack_sp--);
1062 retval = &PL_sv_undef;
1065 if (PL_curcop == &PL_compiling) {
1066 strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf);
1067 PL_curcop->op_private = PL_hints;
1069 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1070 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1075 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1077 HV* hv = (HV*)SvRV(sv);
1078 U32 klen = UTF8SKIP(ptr) - 1;
1079 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1081 STRLEN needents = (klen ? 64 : 128);
1087 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1088 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1089 * it's nothing to sniff at.) Pity we usually come through at least
1090 * two function calls to get here...
1092 * NB: this code assumes that swatches are never modified, once generated!
1095 if (hv == PL_last_swash_hv &&
1096 klen == PL_last_swash_klen &&
1097 (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
1099 tmps = PL_last_swash_tmps;
1100 slen = PL_last_swash_slen;
1103 /* Try our second-level swatch cache, kept in a hash. */
1104 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1106 /* If not cached, generate it via utf8::SWASHGET */
1107 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1112 PUSHSTACKi(PERLSI_MAGIC);
1116 PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, UTF8_MAXLEN, 0, 0) & ~(needents - 1))));
1117 PUSHs(sv_2mortal(newSViv(needents)));
1119 if (call_method("SWASHGET", G_SCALAR))
1120 retval = newSVsv(*PL_stack_sp--);
1122 retval = &PL_sv_undef;
1126 if (PL_curcop == &PL_compiling)
1127 PL_curcop->op_private = PL_hints;
1129 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1131 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1132 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1135 PL_last_swash_hv = hv;
1136 PL_last_swash_klen = klen;
1137 PL_last_swash_tmps = tmps;
1138 PL_last_swash_slen = slen;
1140 Copy(ptr, PL_last_swash_key, klen, U8);
1143 switch ((int)((slen << 3) / needents)) {
1145 bit = 1 << (off & 7);
1147 return (tmps[off] & bit) != 0;
1152 return (tmps[off] << 8) + tmps[off + 1] ;
1155 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1157 Perl_croak(aTHX_ "panic: swash_fetch");