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 =for apidoc A|U8*|uv_to_utf8|U8 *d|UV uv
32 Adds the UTF8 representation of the Unicode codepoint C<uv> to the end
33 of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free
34 bytes available. The return value is the pointer to the byte after the
35 end of the new character. In other words,
37 d = uv_to_utf8(d, uv);
39 is the recommended Unicode-aware way of saying
47 Perl_uv_to_utf8(pTHX_ U8 *d, UV uv)
54 *d++ = (( uv >> 6) | 0xc0);
55 *d++ = (( uv & 0x3f) | 0x80);
59 *d++ = (( uv >> 12) | 0xe0);
60 *d++ = (((uv >> 6) & 0x3f) | 0x80);
61 *d++ = (( uv & 0x3f) | 0x80);
65 *d++ = (( uv >> 18) | 0xf0);
66 *d++ = (((uv >> 12) & 0x3f) | 0x80);
67 *d++ = (((uv >> 6) & 0x3f) | 0x80);
68 *d++ = (( uv & 0x3f) | 0x80);
72 *d++ = (( uv >> 24) | 0xf8);
73 *d++ = (((uv >> 18) & 0x3f) | 0x80);
74 *d++ = (((uv >> 12) & 0x3f) | 0x80);
75 *d++ = (((uv >> 6) & 0x3f) | 0x80);
76 *d++ = (( uv & 0x3f) | 0x80);
79 if (uv < 0x80000000) {
80 *d++ = (( uv >> 30) | 0xfc);
81 *d++ = (((uv >> 24) & 0x3f) | 0x80);
82 *d++ = (((uv >> 18) & 0x3f) | 0x80);
83 *d++ = (((uv >> 12) & 0x3f) | 0x80);
84 *d++ = (((uv >> 6) & 0x3f) | 0x80);
85 *d++ = (( uv & 0x3f) | 0x80);
89 if (uv < UTF8_QUAD_MAX)
92 *d++ = 0xfe; /* Can't match U+FEFF! */
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);
103 *d++ = 0xff; /* Can't match U+FFFE! */
104 *d++ = 0x80; /* 6 Reserved bits */
105 *d++ = (((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
106 *d++ = (((uv >> 54) & 0x3f) | 0x80);
107 *d++ = (((uv >> 48) & 0x3f) | 0x80);
108 *d++ = (((uv >> 42) & 0x3f) | 0x80);
109 *d++ = (((uv >> 36) & 0x3f) | 0x80);
110 *d++ = (((uv >> 30) & 0x3f) | 0x80);
111 *d++ = (((uv >> 24) & 0x3f) | 0x80);
112 *d++ = (((uv >> 18) & 0x3f) | 0x80);
113 *d++ = (((uv >> 12) & 0x3f) | 0x80);
114 *d++ = (((uv >> 6) & 0x3f) | 0x80);
115 *d++ = (( uv & 0x3f) | 0x80);
122 =for apidoc A|STRLEN|is_utf8_char|U8 *s
124 Tests if some arbitrary number of bytes begins in a valid UTF-8 character.
125 The actual number of bytes in the UTF-8 character will be returned if it
126 is valid, otherwise 0.
131 Perl_is_utf8_char(pTHX_ U8 *s)
137 if (UTF8_IS_ASCII(u))
140 if (!UTF8_IS_START(u))
145 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
153 if (!UTF8_IS_CONTINUATION(*s))
155 uv = UTF8_ACCUMULATE(uv, *s);
162 if (UNISKIP(uv) < len)
169 =for apidoc A|bool|is_utf8_string|U8 *s|STRLEN len
171 Returns true if first C<len> bytes of the given string form valid a UTF8
172 string, false otherwise.
178 Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len)
185 len = strlen((char *)s);
201 =for apidoc A|U8* s|utf8_to_uv|STRLEN curlen|STRLEN *retlen|U32 flags
203 Returns the character value of the first character in the string C<s>
204 which is assumed to be in UTF8 encoding and no longer than C<curlen>;
205 C<retlen> will be set to the length, in bytes, of that character.
207 If C<s> does not point to a well-formed UTF8 character, the behaviour
208 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
209 it is assumed that the caller will raise a warning, and this function
210 will silently just set C<retlen> to C<-1> and return zero. If the
211 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
212 malformations will be given, C<retlen> will be set to the expected
213 length of the UTF-8 character in bytes, and zero will be returned.
215 The C<flags> can also contain various flags to allow deviations from
216 the strict UTF-8 encoding (see F<utf8.h>).
221 Perl_utf8_to_uv(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags)
228 bool dowarn = ckWARN_d(WARN_UTF8);
230 STRLEN expectlen = 0;
233 /* This list is a superset of the UTF8_ALLOW_XXX. */
235 #define UTF8_WARN_EMPTY 1
236 #define UTF8_WARN_CONTINUATION 2
237 #define UTF8_WARN_NON_CONTINUATION 3
238 #define UTF8_WARN_FE_FF 4
239 #define UTF8_WARN_SHORT 5
240 #define UTF8_WARN_OVERFLOW 6
241 #define UTF8_WARN_SURROGATE 7
242 #define UTF8_WARN_BOM 8
243 #define UTF8_WARN_LONG 9
244 #define UTF8_WARN_FFFF 10
247 !(flags & UTF8_ALLOW_EMPTY)) {
248 warning = UTF8_WARN_EMPTY;
252 if (UTF8_IS_ASCII(uv)) {
258 if (UTF8_IS_CONTINUATION(uv) &&
259 !(flags & UTF8_ALLOW_CONTINUATION)) {
260 warning = UTF8_WARN_CONTINUATION;
264 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
265 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
266 warning = UTF8_WARN_NON_CONTINUATION;
270 if ((uv == 0xfe || uv == 0xff) &&
271 !(flags & UTF8_ALLOW_FE_FF)) {
272 warning = UTF8_WARN_FE_FF;
276 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
277 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
278 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
279 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
280 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
281 else if (!(uv & 0x01)) { len = 7; uv = 0; }
282 else { len = 13; uv = 0; } /* whoa! */
289 if ((curlen < expectlen) &&
290 !(flags & UTF8_ALLOW_SHORT)) {
291 warning = UTF8_WARN_SHORT;
300 if (!UTF8_IS_CONTINUATION(*s) &&
301 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
303 warning = UTF8_WARN_NON_CONTINUATION;
307 uv = UTF8_ACCUMULATE(uv, *s);
309 /* These cannot be allowed. */
311 if (!(flags & UTF8_ALLOW_LONG)) {
312 warning = UTF8_WARN_LONG;
316 else { /* uv < ouv */
317 /* This cannot be allowed. */
318 warning = UTF8_WARN_OVERFLOW;
326 if (UNICODE_IS_SURROGATE(uv) &&
327 !(flags & UTF8_ALLOW_SURROGATE)) {
328 warning = UTF8_WARN_SURROGATE;
330 } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) &&
331 !(flags & UTF8_ALLOW_BOM)) {
332 warning = UTF8_WARN_BOM;
334 } else if ((expectlen > UNISKIP(uv)) &&
335 !(flags & UTF8_ALLOW_LONG)) {
336 warning = UTF8_WARN_LONG;
338 } else if (UNICODE_IS_ILLEGAL(uv) &&
339 !(flags & UTF8_ALLOW_FFFF)) {
340 warning = UTF8_WARN_FFFF;
348 if (flags & UTF8_CHECK_ONLY) {
355 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
358 case 0: /* Intentionally empty. */ break;
359 case UTF8_WARN_EMPTY:
360 Perl_sv_catpvf(aTHX_ sv, "(empty string)");
362 case UTF8_WARN_CONTINUATION:
363 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf")", uv);
365 case UTF8_WARN_NON_CONTINUATION:
366 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")",
369 case UTF8_WARN_FE_FF:
370 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
372 case UTF8_WARN_SHORT:
373 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
374 curlen, curlen == 1 ? "" : "s", expectlen);
376 case UTF8_WARN_OVERFLOW:
377 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x)",
380 case UTF8_WARN_SURROGATE:
381 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
384 Perl_sv_catpvf(aTHX_ sv, "(byte order mark 0x%04"UVxf")", uv);
387 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
388 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
391 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
394 Perl_sv_catpvf(aTHX_ sv, "(unknown reason)");
402 Perl_warner(aTHX_ WARN_UTF8,
403 "%s in %s", s, PL_op_desc[PL_op->op_type]);
405 Perl_warner(aTHX_ WARN_UTF8, "%s", s);
410 *retlen = expectlen ? expectlen : len;
416 =for apidoc A|U8* s|utf8_to_uv_simple|STRLEN *retlen
418 Returns the character value of the first character in the string C<s>
419 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
420 length, in bytes, of that character.
422 If C<s> does not point to a well-formed UTF8 character, zero is
423 returned and retlen is set, if possible, to -1.
429 Perl_utf8_to_uv_simple(pTHX_ U8* s, STRLEN* retlen)
431 return Perl_utf8_to_uv(aTHX_ s, UTF8_MAXLEN, retlen, 0);
435 =for apidoc A|STRLEN|utf8_length|U8* s|U8 *e
437 Return the length of the UTF-8 char encoded string C<s> in characters.
438 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
439 up past C<e>, croaks.
445 Perl_utf8_length(pTHX_ U8* s, U8* e)
449 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
450 * the bitops (especially ~) can create illegal UTF-8.
451 * In other words: in Perl UTF-8 is not just for Unicode. */
454 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
459 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
468 =for apidoc A|IV|utf8_distance|U8 *a|U8 *b
470 Returns the number of UTF8 characters between the UTF-8 pointers C<a>
473 WARNING: use only if you *know* that the pointers point inside the
479 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
483 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
484 * the bitops (especially ~) can create illegal UTF-8.
485 * In other words: in Perl UTF-8 is not just for Unicode. */
492 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
502 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
512 =for apidoc A|U8*|utf8_hop|U8 *s|I32 off
514 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
517 WARNING: do not use the following unless you *know* C<off> is within
518 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
519 on the first byte of character or just after the last byte of a character.
524 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
526 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
527 * the bitops (especially ~) can create illegal UTF-8.
528 * In other words: in Perl UTF-8 is not just for Unicode. */
537 while (UTF8_IS_CONTINUATION(*s))
545 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
547 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
548 Unlike C<bytes_to_utf8>, this over-writes the original string, and
549 updates len to contain the new length.
550 Returns zero on failure, setting C<len> to -1.
556 Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len)
562 /* ensure valid UTF8 and chars < 256 before updating string */
563 for (send = s + *len; s < send; ) {
568 ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
576 if (UTF8_IS_ASCII(*s)) {
581 *d++ = (U8)utf8_to_uv_simple(s, &ulen);
591 =for apidoc A|U8 *|bytes_from_utf8|U8 *s|STRLEN *len|bool *is_utf8
593 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
594 Unlike <utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
595 the newly-created string, and updates C<len> to contain the new
596 length. Returns the original string if no conversion occurs, C<len>
597 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
598 0 if C<s> is converted or contains all 7bit characters.
603 Perl_bytes_from_utf8(pTHX_ U8* s, STRLEN *len, bool *is_utf8)
613 /* ensure valid UTF8 and chars < 256 before converting string */
614 for (send = s + *len; s < send;) {
616 if (!UTF8_IS_ASCII(c)) {
617 if (UTF8_IS_CONTINUATION(c) || s >= send ||
618 !UTF8_IS_CONTINUATION(*s) || UTF8_IS_DOWNGRADEABLE_START(c))
629 Newz(801, d, (*len) - count + 1, U8);
630 s = start; start = d;
634 if (UTF8_IS_ASCII(c))
637 *d++ = UTF8_ACCUMULATE(c, *s++);
645 =for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
647 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
648 Returns a pointer to the newly-created string, and sets C<len> to
649 reflect the new length.
655 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
662 Newz(801, d, (*len) * 2 + 1, U8);
666 if (UTF8_IS_ASCII(*s))
671 *d++ = UTF8_EIGHT_BIT_HI(uv);
672 *d++ = UTF8_EIGHT_BIT_LO(uv);
681 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
683 * Destination must be pre-extended to 3/2 source. Do not use in-place.
684 * We optimize for native, for obvious reasons. */
687 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
693 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
698 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
705 *d++ = (( uv >> 6) | 0xc0);
706 *d++ = (( uv & 0x3f) | 0x80);
709 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
711 if (low < 0xdc00 || low >= 0xdfff)
712 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
713 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
716 *d++ = (( uv >> 12) | 0xe0);
717 *d++ = (((uv >> 6) & 0x3f) | 0x80);
718 *d++ = (( uv & 0x3f) | 0x80);
722 *d++ = (( uv >> 18) | 0xf0);
723 *d++ = (((uv >> 12) & 0x3f) | 0x80);
724 *d++ = (((uv >> 6) & 0x3f) | 0x80);
725 *d++ = (( uv & 0x3f) | 0x80);
729 *newlen = d - dstart;
733 /* Note: this one is slightly destructive of the source. */
736 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
739 U8* send = s + bytelen;
746 return utf16_to_utf8(p, d, bytelen, newlen);
749 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
752 Perl_is_uni_alnum(pTHX_ U32 c)
754 U8 tmpbuf[UTF8_MAXLEN+1];
755 uv_to_utf8(tmpbuf, (UV)c);
756 return is_utf8_alnum(tmpbuf);
760 Perl_is_uni_alnumc(pTHX_ U32 c)
762 U8 tmpbuf[UTF8_MAXLEN+1];
763 uv_to_utf8(tmpbuf, (UV)c);
764 return is_utf8_alnumc(tmpbuf);
768 Perl_is_uni_idfirst(pTHX_ U32 c)
770 U8 tmpbuf[UTF8_MAXLEN+1];
771 uv_to_utf8(tmpbuf, (UV)c);
772 return is_utf8_idfirst(tmpbuf);
776 Perl_is_uni_alpha(pTHX_ U32 c)
778 U8 tmpbuf[UTF8_MAXLEN+1];
779 uv_to_utf8(tmpbuf, (UV)c);
780 return is_utf8_alpha(tmpbuf);
784 Perl_is_uni_ascii(pTHX_ U32 c)
786 U8 tmpbuf[UTF8_MAXLEN+1];
787 uv_to_utf8(tmpbuf, (UV)c);
788 return is_utf8_ascii(tmpbuf);
792 Perl_is_uni_space(pTHX_ U32 c)
794 U8 tmpbuf[UTF8_MAXLEN+1];
795 uv_to_utf8(tmpbuf, (UV)c);
796 return is_utf8_space(tmpbuf);
800 Perl_is_uni_digit(pTHX_ U32 c)
802 U8 tmpbuf[UTF8_MAXLEN+1];
803 uv_to_utf8(tmpbuf, (UV)c);
804 return is_utf8_digit(tmpbuf);
808 Perl_is_uni_upper(pTHX_ U32 c)
810 U8 tmpbuf[UTF8_MAXLEN+1];
811 uv_to_utf8(tmpbuf, (UV)c);
812 return is_utf8_upper(tmpbuf);
816 Perl_is_uni_lower(pTHX_ U32 c)
818 U8 tmpbuf[UTF8_MAXLEN+1];
819 uv_to_utf8(tmpbuf, (UV)c);
820 return is_utf8_lower(tmpbuf);
824 Perl_is_uni_cntrl(pTHX_ U32 c)
826 U8 tmpbuf[UTF8_MAXLEN+1];
827 uv_to_utf8(tmpbuf, (UV)c);
828 return is_utf8_cntrl(tmpbuf);
832 Perl_is_uni_graph(pTHX_ U32 c)
834 U8 tmpbuf[UTF8_MAXLEN+1];
835 uv_to_utf8(tmpbuf, (UV)c);
836 return is_utf8_graph(tmpbuf);
840 Perl_is_uni_print(pTHX_ U32 c)
842 U8 tmpbuf[UTF8_MAXLEN+1];
843 uv_to_utf8(tmpbuf, (UV)c);
844 return is_utf8_print(tmpbuf);
848 Perl_is_uni_punct(pTHX_ U32 c)
850 U8 tmpbuf[UTF8_MAXLEN+1];
851 uv_to_utf8(tmpbuf, (UV)c);
852 return is_utf8_punct(tmpbuf);
856 Perl_is_uni_xdigit(pTHX_ U32 c)
858 U8 tmpbuf[UTF8_MAXLEN+1];
859 uv_to_utf8(tmpbuf, (UV)c);
860 return is_utf8_xdigit(tmpbuf);
864 Perl_to_uni_upper(pTHX_ U32 c)
866 U8 tmpbuf[UTF8_MAXLEN+1];
867 uv_to_utf8(tmpbuf, (UV)c);
868 return to_utf8_upper(tmpbuf);
872 Perl_to_uni_title(pTHX_ U32 c)
874 U8 tmpbuf[UTF8_MAXLEN+1];
875 uv_to_utf8(tmpbuf, (UV)c);
876 return to_utf8_title(tmpbuf);
880 Perl_to_uni_lower(pTHX_ U32 c)
882 U8 tmpbuf[UTF8_MAXLEN+1];
883 uv_to_utf8(tmpbuf, (UV)c);
884 return to_utf8_lower(tmpbuf);
887 /* for now these all assume no locale info available for Unicode > 255 */
890 Perl_is_uni_alnum_lc(pTHX_ U32 c)
892 return is_uni_alnum(c); /* XXX no locale support yet */
896 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
898 return is_uni_alnumc(c); /* XXX no locale support yet */
902 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
904 return is_uni_idfirst(c); /* XXX no locale support yet */
908 Perl_is_uni_alpha_lc(pTHX_ U32 c)
910 return is_uni_alpha(c); /* XXX no locale support yet */
914 Perl_is_uni_ascii_lc(pTHX_ U32 c)
916 return is_uni_ascii(c); /* XXX no locale support yet */
920 Perl_is_uni_space_lc(pTHX_ U32 c)
922 return is_uni_space(c); /* XXX no locale support yet */
926 Perl_is_uni_digit_lc(pTHX_ U32 c)
928 return is_uni_digit(c); /* XXX no locale support yet */
932 Perl_is_uni_upper_lc(pTHX_ U32 c)
934 return is_uni_upper(c); /* XXX no locale support yet */
938 Perl_is_uni_lower_lc(pTHX_ U32 c)
940 return is_uni_lower(c); /* XXX no locale support yet */
944 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
946 return is_uni_cntrl(c); /* XXX no locale support yet */
950 Perl_is_uni_graph_lc(pTHX_ U32 c)
952 return is_uni_graph(c); /* XXX no locale support yet */
956 Perl_is_uni_print_lc(pTHX_ U32 c)
958 return is_uni_print(c); /* XXX no locale support yet */
962 Perl_is_uni_punct_lc(pTHX_ U32 c)
964 return is_uni_punct(c); /* XXX no locale support yet */
968 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
970 return is_uni_xdigit(c); /* XXX no locale support yet */
974 Perl_to_uni_upper_lc(pTHX_ U32 c)
976 return to_uni_upper(c); /* XXX no locale support yet */
980 Perl_to_uni_title_lc(pTHX_ U32 c)
982 return to_uni_title(c); /* XXX no locale support yet */
986 Perl_to_uni_lower_lc(pTHX_ U32 c)
988 return to_uni_lower(c); /* XXX no locale support yet */
992 Perl_is_utf8_alnum(pTHX_ U8 *p)
994 if (!is_utf8_char(p))
997 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
998 * descendant of isalnum(3), in other words, it doesn't
999 * contain the '_'. --jhi */
1000 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1001 return swash_fetch(PL_utf8_alnum, p);
1002 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1003 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1005 PL_utf8_alnum = swash_init("utf8", "",
1006 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1007 return swash_fetch(PL_utf8_alnum, p);
1012 Perl_is_utf8_alnumc(pTHX_ U8 *p)
1014 if (!is_utf8_char(p))
1017 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1018 return swash_fetch(PL_utf8_alnum, p);
1019 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
1020 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1022 PL_utf8_alnum = swash_init("utf8", "",
1023 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1024 return swash_fetch(PL_utf8_alnum, p);
1029 Perl_is_utf8_idfirst(pTHX_ U8 *p)
1031 return *p == '_' || is_utf8_alpha(p);
1035 Perl_is_utf8_alpha(pTHX_ U8 *p)
1037 if (!is_utf8_char(p))
1040 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1041 return swash_fetch(PL_utf8_alpha, p);
1045 Perl_is_utf8_ascii(pTHX_ U8 *p)
1047 if (!is_utf8_char(p))
1050 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1051 return swash_fetch(PL_utf8_ascii, p);
1055 Perl_is_utf8_space(pTHX_ U8 *p)
1057 if (!is_utf8_char(p))
1060 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1061 return swash_fetch(PL_utf8_space, p);
1065 Perl_is_utf8_digit(pTHX_ U8 *p)
1067 if (!is_utf8_char(p))
1070 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1071 return swash_fetch(PL_utf8_digit, p);
1075 Perl_is_utf8_upper(pTHX_ U8 *p)
1077 if (!is_utf8_char(p))
1080 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
1081 return swash_fetch(PL_utf8_upper, p);
1085 Perl_is_utf8_lower(pTHX_ U8 *p)
1087 if (!is_utf8_char(p))
1090 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
1091 return swash_fetch(PL_utf8_lower, p);
1095 Perl_is_utf8_cntrl(pTHX_ U8 *p)
1097 if (!is_utf8_char(p))
1100 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1101 return swash_fetch(PL_utf8_cntrl, p);
1105 Perl_is_utf8_graph(pTHX_ U8 *p)
1107 if (!is_utf8_char(p))
1110 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1111 return swash_fetch(PL_utf8_graph, p);
1115 Perl_is_utf8_print(pTHX_ U8 *p)
1117 if (!is_utf8_char(p))
1120 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1121 return swash_fetch(PL_utf8_print, p);
1125 Perl_is_utf8_punct(pTHX_ U8 *p)
1127 if (!is_utf8_char(p))
1130 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1131 return swash_fetch(PL_utf8_punct, p);
1135 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1137 if (!is_utf8_char(p))
1139 if (!PL_utf8_xdigit)
1140 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1141 return swash_fetch(PL_utf8_xdigit, p);
1145 Perl_is_utf8_mark(pTHX_ U8 *p)
1147 if (!is_utf8_char(p))
1150 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1151 return swash_fetch(PL_utf8_mark, p);
1155 Perl_to_utf8_upper(pTHX_ U8 *p)
1159 if (!PL_utf8_toupper)
1160 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
1161 uv = swash_fetch(PL_utf8_toupper, p);
1162 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1166 Perl_to_utf8_title(pTHX_ U8 *p)
1170 if (!PL_utf8_totitle)
1171 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
1172 uv = swash_fetch(PL_utf8_totitle, p);
1173 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1177 Perl_to_utf8_lower(pTHX_ U8 *p)
1181 if (!PL_utf8_tolower)
1182 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
1183 uv = swash_fetch(PL_utf8_tolower, p);
1184 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1187 /* a "swash" is a swatch hash */
1190 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1193 SV* tokenbufsv = sv_2mortal(NEWSV(0,0));
1196 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
1198 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1202 PUSHSTACKi(PERLSI_MAGIC);
1205 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1206 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1208 PUSHs(sv_2mortal(newSViv(minbits)));
1209 PUSHs(sv_2mortal(newSViv(none)));
1215 if (PL_curcop == &PL_compiling)
1216 /* XXX ought to be handled by lex_start */
1217 sv_setpv(tokenbufsv, PL_tokenbuf);
1218 if (call_method("SWASHNEW", G_SCALAR))
1219 retval = newSVsv(*PL_stack_sp--);
1221 retval = &PL_sv_undef;
1224 if (PL_curcop == &PL_compiling) {
1226 char* pv = SvPV(tokenbufsv, len);
1228 Copy(pv, PL_tokenbuf, len+1, char);
1229 PL_curcop->op_private = PL_hints;
1231 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1232 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1237 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1239 HV* hv = (HV*)SvRV(sv);
1240 U32 klen = UTF8SKIP(ptr) - 1;
1241 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1243 STRLEN needents = (klen ? 64 : 128);
1249 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1250 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1251 * it's nothing to sniff at.) Pity we usually come through at least
1252 * two function calls to get here...
1254 * NB: this code assumes that swatches are never modified, once generated!
1257 if (hv == PL_last_swash_hv &&
1258 klen == PL_last_swash_klen &&
1259 (!klen || memEQ(ptr,PL_last_swash_key,klen)) )
1261 tmps = PL_last_swash_tmps;
1262 slen = PL_last_swash_slen;
1265 /* Try our second-level swatch cache, kept in a hash. */
1266 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1268 /* If not cached, generate it via utf8::SWASHGET */
1269 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1274 PUSHSTACKi(PERLSI_MAGIC);
1278 PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, UTF8_MAXLEN, 0, 0) & ~(needents - 1))));
1279 PUSHs(sv_2mortal(newSViv(needents)));
1281 if (call_method("SWASHGET", G_SCALAR))
1282 retval = newSVsv(*PL_stack_sp--);
1284 retval = &PL_sv_undef;
1288 if (PL_curcop == &PL_compiling)
1289 PL_curcop->op_private = PL_hints;
1291 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1293 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1294 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1297 PL_last_swash_hv = hv;
1298 PL_last_swash_klen = klen;
1299 PL_last_swash_tmps = tmps;
1300 PL_last_swash_slen = slen;
1302 Copy(ptr, PL_last_swash_key, klen, U8);
1305 switch ((int)((slen << 3) / needents)) {
1307 bit = 1 << (off & 7);
1309 return (tmps[off] & bit) != 0;
1314 return (tmps[off] << 8) + tmps[off + 1] ;
1317 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1319 Perl_croak(aTHX_ "panic: swash_fetch");