3 * Copyright (c) 1998-2001, 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)
122 if (UTF8_IS_ASCII(u))
125 if (!UTF8_IS_START(u))
130 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
138 if (!UTF8_IS_CONTINUATION(*s))
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)
170 len = strlen((char *)s);
186 =for apidoc Am|U8* s|utf8_to_uv|STRLEN curlen|STRLEN *retlen|U32 flags
188 Returns the character value of the first character in the string C<s>
189 which is assumed to be in UTF8 encoding and no longer than C<curlen>;
190 C<retlen> will be set to the length, in bytes, of that character.
192 If C<s> does not point to a well-formed UTF8 character, the behaviour
193 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
194 it is assumed that the caller will raise a warning, and this function
195 will silently just set C<retlen> to C<-1> and return zero. If the
196 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
197 malformations will be given, C<retlen> will be set to the expected
198 length of the UTF-8 character in bytes, and zero will be returned.
200 The C<flags> can also contain various flags to allow deviations from
201 the strict UTF-8 encoding (see F<utf8.h>).
206 Perl_utf8_to_uv(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags)
213 bool dowarn = ckWARN_d(WARN_UTF8);
215 STRLEN expectlen = 0;
219 Perl_warner(aTHX_ WARN_UTF8,
220 "Malformed UTF-8 character (an empty string)");
224 if (UTF8_IS_ASCII(uv)) {
230 if (UTF8_IS_CONTINUATION(uv) &&
231 !(flags & UTF8_ALLOW_CONTINUATION)) {
233 Perl_warner(aTHX_ WARN_UTF8,
234 "Malformed UTF-8 character (unexpected continuation byte 0x%02"UVxf")",
239 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
240 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
242 Perl_warner(aTHX_ WARN_UTF8,
243 "Malformed UTF-8 character (unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")",
248 if ((uv == 0xfe || uv == 0xff) &&
249 !(flags & UTF8_ALLOW_FE_FF)) {
251 Perl_warner(aTHX_ WARN_UTF8,
252 "Malformed UTF-8 character (byte 0x%02"UVxf")",
257 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
258 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
259 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
260 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
261 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
262 else if (!(uv & 0x01)) { len = 7; uv = 0; }
263 else { len = 13; uv = 0; } /* whoa! */
270 if ((curlen < expectlen) &&
271 !(flags & UTF8_ALLOW_SHORT)) {
273 Perl_warner(aTHX_ WARN_UTF8,
274 "Malformed UTF-8 character (%d byte%s, need %d)",
275 curlen, curlen == 1 ? "" : "s", expectlen);
284 if (!UTF8_IS_CONTINUATION(*s) &&
285 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
287 Perl_warner(aTHX_ WARN_UTF8,
288 "Malformed UTF-8 character (unexpected non-continuation byte 0x%02x)",
293 uv = UTF8_ACCUMULATE(uv, *s);
295 /* This cannot be allowed. */
297 Perl_warner(aTHX_ WARN_UTF8,
298 "Malformed UTF-8 character (overflow at 0x%"UVxf", byte 0x%02x)",
306 if (UNICODE_IS_SURROGATE(uv) &&
307 !(flags & UTF8_ALLOW_SURROGATE)) {
309 Perl_warner(aTHX_ WARN_UTF8,
310 "Malformed UTF-8 character (UTF-16 surrogate 0x%04"UVxf")",
313 } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) &&
314 !(flags & UTF8_ALLOW_BOM)) {
316 Perl_warner(aTHX_ WARN_UTF8,
317 "Malformed UTF-8 character (byte order mark 0x%04"UVxf")",
320 } else if ((expectlen > UNISKIP(uv)) &&
321 !(flags & UTF8_ALLOW_LONG)) {
323 Perl_warner(aTHX_ WARN_UTF8,
324 "Malformed UTF-8 character (%d byte%s, need %d)",
325 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
327 } else if (UNICODE_IS_ILLEGAL(uv) &&
328 !(flags & UTF8_ALLOW_FFFF)) {
330 Perl_warner(aTHX_ WARN_UTF8,
331 "Malformed UTF-8 character (character 0x%04"UVxf")",
340 if (flags & UTF8_CHECK_ONLY) {
347 *retlen = expectlen ? expectlen : len;
353 =for apidoc Am|U8* s|utf8_to_uv_simple|STRLEN *retlen
355 Returns the character value of the first character in the string C<s>
356 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
357 length, in bytes, of that character.
359 If C<s> does not point to a well-formed UTF8 character, zero is
360 returned and retlen is set, if possible, to -1.
366 Perl_utf8_to_uv_simple(pTHX_ U8* s, STRLEN* retlen)
368 return Perl_utf8_to_uv(aTHX_ s, UTF8_MAXLEN, retlen, 0);
372 =for apidoc Am|STRLEN|utf8_length|U8* s|U8 *e
374 Return the length of the UTF-8 char encoded string C<s> in characters.
375 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
376 up past C<e>, croaks.
382 Perl_utf8_length(pTHX_ U8* s, U8* e)
386 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
387 * the bitops (especially ~) can create illegal UTF-8.
388 * In other words: in Perl UTF-8 is not just for Unicode. */
391 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
396 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
405 =for apidoc Am|IV|utf8_distance|U8 *a|U8 *b
407 Returns the number of UTF8 characters between the UTF-8 pointers C<a>
410 WARNING: use only if you *know* that the pointers point inside the
416 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
420 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
421 * the bitops (especially ~) can create illegal UTF-8.
422 * In other words: in Perl UTF-8 is not just for Unicode. */
429 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
439 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
449 =for apidoc Am|U8*|utf8_hop|U8 *s|I32 off
451 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
454 WARNING: do not use the following unless you *know* C<off> is within
455 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
456 on the first byte of character or just after the last byte of a character.
461 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
463 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
464 * the bitops (especially ~) can create illegal UTF-8.
465 * In other words: in Perl UTF-8 is not just for Unicode. */
474 while (UTF8_IS_CONTINUATION(*s))
482 =for apidoc Am|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
484 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
485 Unlike C<bytes_to_utf8>, this over-writes the original string, and
486 updates len to contain the new length.
487 Returns zero on failure, setting C<len> to -1.
493 Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len)
499 /* ensure valid UTF8 and chars < 256 before updating string */
500 for (send = s + *len; s < send; ) {
505 ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
514 *d++ = (U8)utf8_to_uv_simple(s, &ulen);
523 =for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
525 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
526 Returns a pointer to the newly-created string, and sets C<len> to
527 reflect the new length.
533 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
540 Newz(801, d, (*len) * 2 + 1, U8);
548 *d++ = (( uv >> 6) | 0xc0);
549 *d++ = (( uv & 0x3f) | 0x80);
558 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
560 * Destination must be pre-extended to 3/2 source. Do not use in-place.
561 * We optimize for native, for obvious reasons. */
564 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
570 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
575 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
582 *d++ = (( uv >> 6) | 0xc0);
583 *d++ = (( uv & 0x3f) | 0x80);
586 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
588 if (low < 0xdc00 || low >= 0xdfff)
589 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
590 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
593 *d++ = (( uv >> 12) | 0xe0);
594 *d++ = (((uv >> 6) & 0x3f) | 0x80);
595 *d++ = (( uv & 0x3f) | 0x80);
599 *d++ = (( uv >> 18) | 0xf0);
600 *d++ = (((uv >> 12) & 0x3f) | 0x80);
601 *d++ = (((uv >> 6) & 0x3f) | 0x80);
602 *d++ = (( uv & 0x3f) | 0x80);
606 *newlen = d - dstart;
610 /* Note: this one is slightly destructive of the source. */
613 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
616 U8* send = s + bytelen;
623 return utf16_to_utf8(p, d, bytelen, newlen);
626 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
629 Perl_is_uni_alnum(pTHX_ U32 c)
631 U8 tmpbuf[UTF8_MAXLEN+1];
632 uv_to_utf8(tmpbuf, (UV)c);
633 return is_utf8_alnum(tmpbuf);
637 Perl_is_uni_alnumc(pTHX_ U32 c)
639 U8 tmpbuf[UTF8_MAXLEN+1];
640 uv_to_utf8(tmpbuf, (UV)c);
641 return is_utf8_alnumc(tmpbuf);
645 Perl_is_uni_idfirst(pTHX_ U32 c)
647 U8 tmpbuf[UTF8_MAXLEN+1];
648 uv_to_utf8(tmpbuf, (UV)c);
649 return is_utf8_idfirst(tmpbuf);
653 Perl_is_uni_alpha(pTHX_ U32 c)
655 U8 tmpbuf[UTF8_MAXLEN+1];
656 uv_to_utf8(tmpbuf, (UV)c);
657 return is_utf8_alpha(tmpbuf);
661 Perl_is_uni_ascii(pTHX_ U32 c)
663 U8 tmpbuf[UTF8_MAXLEN+1];
664 uv_to_utf8(tmpbuf, (UV)c);
665 return is_utf8_ascii(tmpbuf);
669 Perl_is_uni_space(pTHX_ U32 c)
671 U8 tmpbuf[UTF8_MAXLEN+1];
672 uv_to_utf8(tmpbuf, (UV)c);
673 return is_utf8_space(tmpbuf);
677 Perl_is_uni_digit(pTHX_ U32 c)
679 U8 tmpbuf[UTF8_MAXLEN+1];
680 uv_to_utf8(tmpbuf, (UV)c);
681 return is_utf8_digit(tmpbuf);
685 Perl_is_uni_upper(pTHX_ U32 c)
687 U8 tmpbuf[UTF8_MAXLEN+1];
688 uv_to_utf8(tmpbuf, (UV)c);
689 return is_utf8_upper(tmpbuf);
693 Perl_is_uni_lower(pTHX_ U32 c)
695 U8 tmpbuf[UTF8_MAXLEN+1];
696 uv_to_utf8(tmpbuf, (UV)c);
697 return is_utf8_lower(tmpbuf);
701 Perl_is_uni_cntrl(pTHX_ U32 c)
703 U8 tmpbuf[UTF8_MAXLEN+1];
704 uv_to_utf8(tmpbuf, (UV)c);
705 return is_utf8_cntrl(tmpbuf);
709 Perl_is_uni_graph(pTHX_ U32 c)
711 U8 tmpbuf[UTF8_MAXLEN+1];
712 uv_to_utf8(tmpbuf, (UV)c);
713 return is_utf8_graph(tmpbuf);
717 Perl_is_uni_print(pTHX_ U32 c)
719 U8 tmpbuf[UTF8_MAXLEN+1];
720 uv_to_utf8(tmpbuf, (UV)c);
721 return is_utf8_print(tmpbuf);
725 Perl_is_uni_punct(pTHX_ U32 c)
727 U8 tmpbuf[UTF8_MAXLEN+1];
728 uv_to_utf8(tmpbuf, (UV)c);
729 return is_utf8_punct(tmpbuf);
733 Perl_is_uni_xdigit(pTHX_ U32 c)
735 U8 tmpbuf[UTF8_MAXLEN+1];
736 uv_to_utf8(tmpbuf, (UV)c);
737 return is_utf8_xdigit(tmpbuf);
741 Perl_to_uni_upper(pTHX_ U32 c)
743 U8 tmpbuf[UTF8_MAXLEN+1];
744 uv_to_utf8(tmpbuf, (UV)c);
745 return to_utf8_upper(tmpbuf);
749 Perl_to_uni_title(pTHX_ U32 c)
751 U8 tmpbuf[UTF8_MAXLEN+1];
752 uv_to_utf8(tmpbuf, (UV)c);
753 return to_utf8_title(tmpbuf);
757 Perl_to_uni_lower(pTHX_ U32 c)
759 U8 tmpbuf[UTF8_MAXLEN+1];
760 uv_to_utf8(tmpbuf, (UV)c);
761 return to_utf8_lower(tmpbuf);
764 /* for now these all assume no locale info available for Unicode > 255 */
767 Perl_is_uni_alnum_lc(pTHX_ U32 c)
769 return is_uni_alnum(c); /* XXX no locale support yet */
773 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
775 return is_uni_alnumc(c); /* XXX no locale support yet */
779 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
781 return is_uni_idfirst(c); /* XXX no locale support yet */
785 Perl_is_uni_alpha_lc(pTHX_ U32 c)
787 return is_uni_alpha(c); /* XXX no locale support yet */
791 Perl_is_uni_ascii_lc(pTHX_ U32 c)
793 return is_uni_ascii(c); /* XXX no locale support yet */
797 Perl_is_uni_space_lc(pTHX_ U32 c)
799 return is_uni_space(c); /* XXX no locale support yet */
803 Perl_is_uni_digit_lc(pTHX_ U32 c)
805 return is_uni_digit(c); /* XXX no locale support yet */
809 Perl_is_uni_upper_lc(pTHX_ U32 c)
811 return is_uni_upper(c); /* XXX no locale support yet */
815 Perl_is_uni_lower_lc(pTHX_ U32 c)
817 return is_uni_lower(c); /* XXX no locale support yet */
821 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
823 return is_uni_cntrl(c); /* XXX no locale support yet */
827 Perl_is_uni_graph_lc(pTHX_ U32 c)
829 return is_uni_graph(c); /* XXX no locale support yet */
833 Perl_is_uni_print_lc(pTHX_ U32 c)
835 return is_uni_print(c); /* XXX no locale support yet */
839 Perl_is_uni_punct_lc(pTHX_ U32 c)
841 return is_uni_punct(c); /* XXX no locale support yet */
845 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
847 return is_uni_xdigit(c); /* XXX no locale support yet */
851 Perl_to_uni_upper_lc(pTHX_ U32 c)
853 return to_uni_upper(c); /* XXX no locale support yet */
857 Perl_to_uni_title_lc(pTHX_ U32 c)
859 return to_uni_title(c); /* XXX no locale support yet */
863 Perl_to_uni_lower_lc(pTHX_ U32 c)
865 return to_uni_lower(c); /* XXX no locale support yet */
869 Perl_is_utf8_alnum(pTHX_ U8 *p)
871 if (!is_utf8_char(p))
874 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
875 * descendant of isalnum(3), in other words, it doesn't
876 * contain the '_'. --jhi */
877 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
878 return swash_fetch(PL_utf8_alnum, p);
879 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
880 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
882 PL_utf8_alnum = swash_init("utf8", "",
883 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
884 return swash_fetch(PL_utf8_alnum, p);
889 Perl_is_utf8_alnumc(pTHX_ U8 *p)
891 if (!is_utf8_char(p))
894 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
895 return swash_fetch(PL_utf8_alnum, p);
896 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
897 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
899 PL_utf8_alnum = swash_init("utf8", "",
900 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
901 return swash_fetch(PL_utf8_alnum, p);
906 Perl_is_utf8_idfirst(pTHX_ U8 *p)
908 return *p == '_' || is_utf8_alpha(p);
912 Perl_is_utf8_alpha(pTHX_ U8 *p)
914 if (!is_utf8_char(p))
917 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
918 return swash_fetch(PL_utf8_alpha, p);
922 Perl_is_utf8_ascii(pTHX_ U8 *p)
924 if (!is_utf8_char(p))
927 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
928 return swash_fetch(PL_utf8_ascii, p);
932 Perl_is_utf8_space(pTHX_ U8 *p)
934 if (!is_utf8_char(p))
937 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
938 return swash_fetch(PL_utf8_space, p);
942 Perl_is_utf8_digit(pTHX_ U8 *p)
944 if (!is_utf8_char(p))
947 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
948 return swash_fetch(PL_utf8_digit, p);
952 Perl_is_utf8_upper(pTHX_ U8 *p)
954 if (!is_utf8_char(p))
957 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
958 return swash_fetch(PL_utf8_upper, p);
962 Perl_is_utf8_lower(pTHX_ U8 *p)
964 if (!is_utf8_char(p))
967 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
968 return swash_fetch(PL_utf8_lower, p);
972 Perl_is_utf8_cntrl(pTHX_ U8 *p)
974 if (!is_utf8_char(p))
977 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
978 return swash_fetch(PL_utf8_cntrl, p);
982 Perl_is_utf8_graph(pTHX_ U8 *p)
984 if (!is_utf8_char(p))
987 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
988 return swash_fetch(PL_utf8_graph, p);
992 Perl_is_utf8_print(pTHX_ U8 *p)
994 if (!is_utf8_char(p))
997 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
998 return swash_fetch(PL_utf8_print, p);
1002 Perl_is_utf8_punct(pTHX_ U8 *p)
1004 if (!is_utf8_char(p))
1007 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1008 return swash_fetch(PL_utf8_punct, p);
1012 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1014 if (!is_utf8_char(p))
1016 if (!PL_utf8_xdigit)
1017 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1018 return swash_fetch(PL_utf8_xdigit, p);
1022 Perl_is_utf8_mark(pTHX_ U8 *p)
1024 if (!is_utf8_char(p))
1027 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1028 return swash_fetch(PL_utf8_mark, p);
1032 Perl_to_utf8_upper(pTHX_ U8 *p)
1036 if (!PL_utf8_toupper)
1037 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
1038 uv = swash_fetch(PL_utf8_toupper, p);
1039 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1043 Perl_to_utf8_title(pTHX_ U8 *p)
1047 if (!PL_utf8_totitle)
1048 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
1049 uv = swash_fetch(PL_utf8_totitle, p);
1050 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1054 Perl_to_utf8_lower(pTHX_ U8 *p)
1058 if (!PL_utf8_tolower)
1059 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
1060 uv = swash_fetch(PL_utf8_tolower, p);
1061 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1064 /* a "swash" is a swatch hash */
1067 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1073 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
1075 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1079 PUSHSTACKi(PERLSI_MAGIC);
1082 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1083 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1085 PUSHs(sv_2mortal(newSViv(minbits)));
1086 PUSHs(sv_2mortal(newSViv(none)));
1092 if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */
1093 strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf);
1094 if (call_method("SWASHNEW", G_SCALAR))
1095 retval = newSVsv(*PL_stack_sp--);
1097 retval = &PL_sv_undef;
1100 if (PL_curcop == &PL_compiling) {
1101 strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf);
1102 PL_curcop->op_private = PL_hints;
1104 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1105 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1110 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1112 HV* hv = (HV*)SvRV(sv);
1113 U32 klen = UTF8SKIP(ptr) - 1;
1114 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1116 STRLEN needents = (klen ? 64 : 128);
1122 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1123 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1124 * it's nothing to sniff at.) Pity we usually come through at least
1125 * two function calls to get here...
1127 * NB: this code assumes that swatches are never modified, once generated!
1130 if (hv == PL_last_swash_hv &&
1131 klen == PL_last_swash_klen &&
1132 (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
1134 tmps = PL_last_swash_tmps;
1135 slen = PL_last_swash_slen;
1138 /* Try our second-level swatch cache, kept in a hash. */
1139 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1141 /* If not cached, generate it via utf8::SWASHGET */
1142 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1147 PUSHSTACKi(PERLSI_MAGIC);
1151 PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, UTF8_MAXLEN, 0, 0) & ~(needents - 1))));
1152 PUSHs(sv_2mortal(newSViv(needents)));
1154 if (call_method("SWASHGET", G_SCALAR))
1155 retval = newSVsv(*PL_stack_sp--);
1157 retval = &PL_sv_undef;
1161 if (PL_curcop == &PL_compiling)
1162 PL_curcop->op_private = PL_hints;
1164 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1166 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1167 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1170 PL_last_swash_hv = hv;
1171 PL_last_swash_klen = klen;
1172 PL_last_swash_tmps = tmps;
1173 PL_last_swash_slen = slen;
1175 Copy(ptr, PL_last_swash_key, klen, U8);
1178 switch ((int)((slen << 3) / needents)) {
1180 bit = 1 << (off & 7);
1182 return (tmps[off] & bit) != 0;
1187 return (tmps[off] << 8) + tmps[off + 1] ;
1190 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1192 Perl_croak(aTHX_ "panic: swash_fetch");