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 = UTF8_ACCUMULATE(uv, *s);
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 silently just set C<retlen> to C<-1> and return zero. If the
193 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
194 malformations will be given, C<retlen> will be set to the expected
195 length of the UTF-8 character in bytes, and zero will be returned.
197 The C<flags> can also contain various flags to allow deviations from
198 the strict UTF-8 encoding (see F<utf8.h>).
203 Perl_utf8_to_uv(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags)
210 bool dowarn = ckWARN_d(WARN_UTF8);
212 STRLEN expectlen = 0;
216 Perl_warner(aTHX_ WARN_UTF8,
217 "Malformed UTF-8 character (an empty string)");
221 if (UTF8_IS_ASCII(uv)) {
227 if (UTF8_IS_CONTINUATION(uv) &&
228 !(flags & UTF8_ALLOW_CONTINUATION)) {
230 Perl_warner(aTHX_ WARN_UTF8,
231 "Malformed UTF-8 character (unexpected continuation byte 0x%02"UVxf")",
236 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
237 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
239 Perl_warner(aTHX_ WARN_UTF8,
240 "Malformed UTF-8 character (unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")",
245 if ((uv == 0xfe || uv == 0xff) &&
246 !(flags & UTF8_ALLOW_FE_FF)) {
248 Perl_warner(aTHX_ WARN_UTF8,
249 "Malformed UTF-8 character (byte 0x%02"UVxf")",
254 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
255 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
256 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
257 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
258 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
259 else if (!(uv & 0x01)) { len = 7; uv = 0; }
260 else { len = 13; uv = 0; } /* whoa! */
267 if ((curlen < expectlen) &&
268 !(flags & UTF8_ALLOW_SHORT)) {
270 Perl_warner(aTHX_ WARN_UTF8,
271 "Malformed UTF-8 character (%d byte%s, need %d)",
272 curlen, curlen == 1 ? "" : "s", expectlen);
281 if (!UTF8_IS_CONTINUATION(*s) &&
282 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
284 Perl_warner(aTHX_ WARN_UTF8,
285 "Malformed UTF-8 character (unexpected non-continuation byte 0x%02x)",
290 uv = UTF8_ACCUMULATE(uv, *s);
292 /* This cannot be allowed. */
294 Perl_warner(aTHX_ WARN_UTF8,
295 "Malformed UTF-8 character (overflow at 0x%"UVxf", byte 0x%02x)",
303 if (UNICODE_IS_SURROGATE(uv) &&
304 !(flags & UTF8_ALLOW_SURROGATE)) {
306 Perl_warner(aTHX_ WARN_UTF8,
307 "Malformed UTF-8 character (UTF-16 surrogate 0x%04"UVxf")",
310 } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) &&
311 !(flags & UTF8_ALLOW_BOM)) {
313 Perl_warner(aTHX_ WARN_UTF8,
314 "Malformed UTF-8 character (byte order mark 0x%04"UVxf")",
317 } else if ((expectlen > UNISKIP(uv)) &&
318 !(flags & UTF8_ALLOW_LONG)) {
320 Perl_warner(aTHX_ WARN_UTF8,
321 "Malformed UTF-8 character (%d byte%s, need %d)",
322 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
324 } else if (UNICODE_IS_ILLEGAL(uv) &&
325 !(flags & UTF8_ALLOW_FFFF)) {
327 Perl_warner(aTHX_ WARN_UTF8,
328 "Malformed UTF-8 character (character 0x%04"UVxf")",
337 if (flags & UTF8_CHECK_ONLY) {
344 *retlen = expectlen ? expectlen : len;
350 =for apidoc Am|U8* s|utf8_to_uv_simple|STRLEN *retlen
352 Returns the character value of the first character in the string C<s>
353 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
354 length, in bytes, of that character, and the pointer C<s> will be
355 advanced to the end of the character.
357 If C<s> does not point to a well-formed UTF8 character, zero is
358 returned and retlen is set, if possible, to -1.
364 Perl_utf8_to_uv_simple(pTHX_ U8* s, STRLEN* retlen)
366 return Perl_utf8_to_uv(aTHX_ s, UTF8_MAXLEN, retlen, 0);
370 =for apidoc Am|STRLEN|utf8_length|U8* s|U8 *e
372 Return the length of the UTF-8 char encoded string C<s> in characters.
373 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
374 up past C<e>, croaks.
380 Perl_utf8_length(pTHX_ U8* s, U8* e)
384 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
385 * the bitops (especially ~) can create illegal UTF-8.
386 * In other words: in Perl UTF-8 is not just for Unicode. */
389 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
394 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
403 =for apidoc Am|IV|utf8_distance|U8 *a|U8 *b
405 Returns the number of UTF8 characters between the UTF-8 pointers C<a>
408 WARNING: use only if you *know* that the pointers point inside the
414 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
418 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
419 * the bitops (especially ~) can create illegal UTF-8.
420 * In other words: in Perl UTF-8 is not just for Unicode. */
427 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
437 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
447 =for apidoc Am|U8*|utf8_hop|U8 *s|I32 off
449 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
452 WARNING: do not use the following unless you *know* C<off> is within
453 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
454 on the first byte of character or just after the last byte of a character.
459 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
461 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
462 * the bitops (especially ~) can create illegal UTF-8.
463 * In other words: in Perl UTF-8 is not just for Unicode. */
472 while (UTF8_IS_CONTINUATION(*s))
480 =for apidoc Am|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
482 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
483 Unlike C<bytes_to_utf8>, this over-writes the original string, and
484 updates len to contain the new length.
485 Returns zero on failure, setting C<len> to -1.
491 Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len)
497 /* ensure valid UTF8 and chars < 256 before updating string */
498 for (send = s + *len; s < send; ) {
503 ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
512 *d++ = (U8)utf8_to_uv_simple(s, &ulen);
521 =for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
523 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
524 Returns a pointer to the newly-created string, and sets C<len> to
525 reflect the new length.
531 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
538 Newz(801, d, (*len) * 2 + 1, U8);
546 *d++ = (( uv >> 6) | 0xc0);
547 *d++ = (( uv & 0x3f) | 0x80);
556 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
558 * Destination must be pre-extended to 3/2 source. Do not use in-place.
559 * We optimize for native, for obvious reasons. */
562 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
568 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
573 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
580 *d++ = (( uv >> 6) | 0xc0);
581 *d++ = (( uv & 0x3f) | 0x80);
584 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
586 if (low < 0xdc00 || low >= 0xdfff)
587 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
588 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
591 *d++ = (( uv >> 12) | 0xe0);
592 *d++ = (((uv >> 6) & 0x3f) | 0x80);
593 *d++ = (( uv & 0x3f) | 0x80);
597 *d++ = (( uv >> 18) | 0xf0);
598 *d++ = (((uv >> 12) & 0x3f) | 0x80);
599 *d++ = (((uv >> 6) & 0x3f) | 0x80);
600 *d++ = (( uv & 0x3f) | 0x80);
604 *newlen = d - dstart;
608 /* Note: this one is slightly destructive of the source. */
611 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
614 U8* send = s + bytelen;
621 return utf16_to_utf8(p, d, bytelen, newlen);
624 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
627 Perl_is_uni_alnum(pTHX_ U32 c)
629 U8 tmpbuf[UTF8_MAXLEN+1];
630 uv_to_utf8(tmpbuf, (UV)c);
631 return is_utf8_alnum(tmpbuf);
635 Perl_is_uni_alnumc(pTHX_ U32 c)
637 U8 tmpbuf[UTF8_MAXLEN+1];
638 uv_to_utf8(tmpbuf, (UV)c);
639 return is_utf8_alnumc(tmpbuf);
643 Perl_is_uni_idfirst(pTHX_ U32 c)
645 U8 tmpbuf[UTF8_MAXLEN+1];
646 uv_to_utf8(tmpbuf, (UV)c);
647 return is_utf8_idfirst(tmpbuf);
651 Perl_is_uni_alpha(pTHX_ U32 c)
653 U8 tmpbuf[UTF8_MAXLEN+1];
654 uv_to_utf8(tmpbuf, (UV)c);
655 return is_utf8_alpha(tmpbuf);
659 Perl_is_uni_ascii(pTHX_ U32 c)
661 U8 tmpbuf[UTF8_MAXLEN+1];
662 uv_to_utf8(tmpbuf, (UV)c);
663 return is_utf8_ascii(tmpbuf);
667 Perl_is_uni_space(pTHX_ U32 c)
669 U8 tmpbuf[UTF8_MAXLEN+1];
670 uv_to_utf8(tmpbuf, (UV)c);
671 return is_utf8_space(tmpbuf);
675 Perl_is_uni_digit(pTHX_ U32 c)
677 U8 tmpbuf[UTF8_MAXLEN+1];
678 uv_to_utf8(tmpbuf, (UV)c);
679 return is_utf8_digit(tmpbuf);
683 Perl_is_uni_upper(pTHX_ U32 c)
685 U8 tmpbuf[UTF8_MAXLEN+1];
686 uv_to_utf8(tmpbuf, (UV)c);
687 return is_utf8_upper(tmpbuf);
691 Perl_is_uni_lower(pTHX_ U32 c)
693 U8 tmpbuf[UTF8_MAXLEN+1];
694 uv_to_utf8(tmpbuf, (UV)c);
695 return is_utf8_lower(tmpbuf);
699 Perl_is_uni_cntrl(pTHX_ U32 c)
701 U8 tmpbuf[UTF8_MAXLEN+1];
702 uv_to_utf8(tmpbuf, (UV)c);
703 return is_utf8_cntrl(tmpbuf);
707 Perl_is_uni_graph(pTHX_ U32 c)
709 U8 tmpbuf[UTF8_MAXLEN+1];
710 uv_to_utf8(tmpbuf, (UV)c);
711 return is_utf8_graph(tmpbuf);
715 Perl_is_uni_print(pTHX_ U32 c)
717 U8 tmpbuf[UTF8_MAXLEN+1];
718 uv_to_utf8(tmpbuf, (UV)c);
719 return is_utf8_print(tmpbuf);
723 Perl_is_uni_punct(pTHX_ U32 c)
725 U8 tmpbuf[UTF8_MAXLEN+1];
726 uv_to_utf8(tmpbuf, (UV)c);
727 return is_utf8_punct(tmpbuf);
731 Perl_is_uni_xdigit(pTHX_ U32 c)
733 U8 tmpbuf[UTF8_MAXLEN+1];
734 uv_to_utf8(tmpbuf, (UV)c);
735 return is_utf8_xdigit(tmpbuf);
739 Perl_to_uni_upper(pTHX_ U32 c)
741 U8 tmpbuf[UTF8_MAXLEN+1];
742 uv_to_utf8(tmpbuf, (UV)c);
743 return to_utf8_upper(tmpbuf);
747 Perl_to_uni_title(pTHX_ U32 c)
749 U8 tmpbuf[UTF8_MAXLEN+1];
750 uv_to_utf8(tmpbuf, (UV)c);
751 return to_utf8_title(tmpbuf);
755 Perl_to_uni_lower(pTHX_ U32 c)
757 U8 tmpbuf[UTF8_MAXLEN+1];
758 uv_to_utf8(tmpbuf, (UV)c);
759 return to_utf8_lower(tmpbuf);
762 /* for now these all assume no locale info available for Unicode > 255 */
765 Perl_is_uni_alnum_lc(pTHX_ U32 c)
767 return is_uni_alnum(c); /* XXX no locale support yet */
771 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
773 return is_uni_alnumc(c); /* XXX no locale support yet */
777 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
779 return is_uni_idfirst(c); /* XXX no locale support yet */
783 Perl_is_uni_alpha_lc(pTHX_ U32 c)
785 return is_uni_alpha(c); /* XXX no locale support yet */
789 Perl_is_uni_ascii_lc(pTHX_ U32 c)
791 return is_uni_ascii(c); /* XXX no locale support yet */
795 Perl_is_uni_space_lc(pTHX_ U32 c)
797 return is_uni_space(c); /* XXX no locale support yet */
801 Perl_is_uni_digit_lc(pTHX_ U32 c)
803 return is_uni_digit(c); /* XXX no locale support yet */
807 Perl_is_uni_upper_lc(pTHX_ U32 c)
809 return is_uni_upper(c); /* XXX no locale support yet */
813 Perl_is_uni_lower_lc(pTHX_ U32 c)
815 return is_uni_lower(c); /* XXX no locale support yet */
819 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
821 return is_uni_cntrl(c); /* XXX no locale support yet */
825 Perl_is_uni_graph_lc(pTHX_ U32 c)
827 return is_uni_graph(c); /* XXX no locale support yet */
831 Perl_is_uni_print_lc(pTHX_ U32 c)
833 return is_uni_print(c); /* XXX no locale support yet */
837 Perl_is_uni_punct_lc(pTHX_ U32 c)
839 return is_uni_punct(c); /* XXX no locale support yet */
843 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
845 return is_uni_xdigit(c); /* XXX no locale support yet */
849 Perl_to_uni_upper_lc(pTHX_ U32 c)
851 return to_uni_upper(c); /* XXX no locale support yet */
855 Perl_to_uni_title_lc(pTHX_ U32 c)
857 return to_uni_title(c); /* XXX no locale support yet */
861 Perl_to_uni_lower_lc(pTHX_ U32 c)
863 return to_uni_lower(c); /* XXX no locale support yet */
867 Perl_is_utf8_alnum(pTHX_ U8 *p)
869 if (!is_utf8_char(p))
872 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
873 * descendant of isalnum(3), in other words, it doesn't
874 * contain the '_'. --jhi */
875 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
876 return swash_fetch(PL_utf8_alnum, p);
877 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
878 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
880 PL_utf8_alnum = swash_init("utf8", "",
881 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
882 return swash_fetch(PL_utf8_alnum, p);
887 Perl_is_utf8_alnumc(pTHX_ U8 *p)
889 if (!is_utf8_char(p))
892 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
893 return swash_fetch(PL_utf8_alnum, p);
894 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
895 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
897 PL_utf8_alnum = swash_init("utf8", "",
898 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
899 return swash_fetch(PL_utf8_alnum, p);
904 Perl_is_utf8_idfirst(pTHX_ U8 *p)
906 return *p == '_' || is_utf8_alpha(p);
910 Perl_is_utf8_alpha(pTHX_ U8 *p)
912 if (!is_utf8_char(p))
915 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
916 return swash_fetch(PL_utf8_alpha, p);
920 Perl_is_utf8_ascii(pTHX_ U8 *p)
922 if (!is_utf8_char(p))
925 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
926 return swash_fetch(PL_utf8_ascii, p);
930 Perl_is_utf8_space(pTHX_ U8 *p)
932 if (!is_utf8_char(p))
935 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
936 return swash_fetch(PL_utf8_space, p);
940 Perl_is_utf8_digit(pTHX_ U8 *p)
942 if (!is_utf8_char(p))
945 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
946 return swash_fetch(PL_utf8_digit, p);
950 Perl_is_utf8_upper(pTHX_ U8 *p)
952 if (!is_utf8_char(p))
955 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
956 return swash_fetch(PL_utf8_upper, p);
960 Perl_is_utf8_lower(pTHX_ U8 *p)
962 if (!is_utf8_char(p))
965 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
966 return swash_fetch(PL_utf8_lower, p);
970 Perl_is_utf8_cntrl(pTHX_ U8 *p)
972 if (!is_utf8_char(p))
975 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
976 return swash_fetch(PL_utf8_cntrl, p);
980 Perl_is_utf8_graph(pTHX_ U8 *p)
982 if (!is_utf8_char(p))
985 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
986 return swash_fetch(PL_utf8_graph, p);
990 Perl_is_utf8_print(pTHX_ U8 *p)
992 if (!is_utf8_char(p))
995 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
996 return swash_fetch(PL_utf8_print, p);
1000 Perl_is_utf8_punct(pTHX_ U8 *p)
1002 if (!is_utf8_char(p))
1005 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1006 return swash_fetch(PL_utf8_punct, p);
1010 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1012 if (!is_utf8_char(p))
1014 if (!PL_utf8_xdigit)
1015 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1016 return swash_fetch(PL_utf8_xdigit, p);
1020 Perl_is_utf8_mark(pTHX_ U8 *p)
1022 if (!is_utf8_char(p))
1025 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1026 return swash_fetch(PL_utf8_mark, p);
1030 Perl_to_utf8_upper(pTHX_ U8 *p)
1034 if (!PL_utf8_toupper)
1035 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
1036 uv = swash_fetch(PL_utf8_toupper, p);
1037 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1041 Perl_to_utf8_title(pTHX_ U8 *p)
1045 if (!PL_utf8_totitle)
1046 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
1047 uv = swash_fetch(PL_utf8_totitle, p);
1048 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1052 Perl_to_utf8_lower(pTHX_ U8 *p)
1056 if (!PL_utf8_tolower)
1057 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
1058 uv = swash_fetch(PL_utf8_tolower, p);
1059 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1062 /* a "swash" is a swatch hash */
1065 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1071 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
1073 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1077 PUSHSTACKi(PERLSI_MAGIC);
1080 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1081 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1083 PUSHs(sv_2mortal(newSViv(minbits)));
1084 PUSHs(sv_2mortal(newSViv(none)));
1090 if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */
1091 strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf);
1092 if (call_method("SWASHNEW", G_SCALAR))
1093 retval = newSVsv(*PL_stack_sp--);
1095 retval = &PL_sv_undef;
1098 if (PL_curcop == &PL_compiling) {
1099 strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf);
1100 PL_curcop->op_private = PL_hints;
1102 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1103 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1108 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1110 HV* hv = (HV*)SvRV(sv);
1111 U32 klen = UTF8SKIP(ptr) - 1;
1112 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1114 STRLEN needents = (klen ? 64 : 128);
1120 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1121 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1122 * it's nothing to sniff at.) Pity we usually come through at least
1123 * two function calls to get here...
1125 * NB: this code assumes that swatches are never modified, once generated!
1128 if (hv == PL_last_swash_hv &&
1129 klen == PL_last_swash_klen &&
1130 (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
1132 tmps = PL_last_swash_tmps;
1133 slen = PL_last_swash_slen;
1136 /* Try our second-level swatch cache, kept in a hash. */
1137 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1139 /* If not cached, generate it via utf8::SWASHGET */
1140 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1145 PUSHSTACKi(PERLSI_MAGIC);
1149 PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, UTF8_MAXLEN, 0, 0) & ~(needents - 1))));
1150 PUSHs(sv_2mortal(newSViv(needents)));
1152 if (call_method("SWASHGET", G_SCALAR))
1153 retval = newSVsv(*PL_stack_sp--);
1155 retval = &PL_sv_undef;
1159 if (PL_curcop == &PL_compiling)
1160 PL_curcop->op_private = PL_hints;
1162 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1164 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1165 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1168 PL_last_swash_hv = hv;
1169 PL_last_swash_klen = klen;
1170 PL_last_swash_tmps = tmps;
1171 PL_last_swash_slen = slen;
1173 Copy(ptr, PL_last_swash_key, klen, U8);
1176 switch ((int)((slen << 3) / needents)) {
1178 bit = 1 << (off & 7);
1180 return (tmps[off] & bit) != 0;
1185 return (tmps[off] << 8) + tmps[off + 1] ;
1188 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1190 Perl_croak(aTHX_ "panic: swash_fetch");