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 =for apidoc A|U8*|uvuni_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 = uvuni_to_utf8(d, uv);
39 is the recommended Unicode-aware way of saying
47 Perl_uvuni_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|U8*|uvchr_to_utf8|U8 *d|UV uv
124 Adds the UTF8 representation of the Native codepoint C<uv> to the end
125 of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free
126 bytes available. The return value is the pointer to the byte after the
127 end of the new character. In other words,
129 d = uvchr_to_utf8(d, uv);
131 is the recommended wide native character-aware way of saying
139 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
142 uv = NATIVE_TO_ASCII(uv);
143 return Perl_uvuni_to_utf8(aTHX_ d, uv);
148 =for apidoc A|STRLEN|is_utf8_char|U8 *s
150 Tests if some arbitrary number of bytes begins in a valid UTF-8
151 character. Note that an ASCII character is a valid UTF-8 character.
152 The actual number of bytes in the UTF-8 character will be returned if
153 it is valid, otherwise 0.
157 Perl_is_utf8_char(pTHX_ U8 *s)
163 if (UTF8_IS_ASCII(u))
166 if (!UTF8_IS_START(u))
171 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
179 if (!UTF8_IS_CONTINUATION(*s))
181 uv = UTF8_ACCUMULATE(uv, *s);
188 if (UNISKIP(uv) < len)
195 =for apidoc A|bool|is_utf8_string|U8 *s|STRLEN len
197 Returns true if first C<len> bytes of the given string form a valid UTF8
198 string, false otherwise. Note that 'a valid UTF8 string' does not mean
199 'a string that contains UTF8' because a valid ASCII string is a valid
206 Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len)
213 len = strlen((char *)s);
229 =for apidoc A|UV|utf8n_to_uvuni|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
231 Bottom level UTF-8 decode routine.
232 Returns the unicode code point value of the first character in the string C<s>
233 which is assumed to be in UTF8 encoding and no longer than C<curlen>;
234 C<retlen> will be set to the length, in bytes, of that character.
236 If C<s> does not point to a well-formed UTF8 character, the behaviour
237 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
238 it is assumed that the caller will raise a warning, and this function
239 will silently just set C<retlen> to C<-1> and return zero. If the
240 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
241 malformations will be given, C<retlen> will be set to the expected
242 length of the UTF-8 character in bytes, and zero will be returned.
244 The C<flags> can also contain various flags to allow deviations from
245 the strict UTF-8 encoding (see F<utf8.h>).
247 Most code should use utf8_to_uvchr() rather than call this directly.
252 Perl_utf8n_to_uvuni(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags)
256 bool dowarn = ckWARN_d(WARN_UTF8);
257 STRLEN expectlen = 0;
260 /* This list is a superset of the UTF8_ALLOW_XXX. */
262 #define UTF8_WARN_EMPTY 1
263 #define UTF8_WARN_CONTINUATION 2
264 #define UTF8_WARN_NON_CONTINUATION 3
265 #define UTF8_WARN_FE_FF 4
266 #define UTF8_WARN_SHORT 5
267 #define UTF8_WARN_OVERFLOW 6
268 #define UTF8_WARN_SURROGATE 7
269 #define UTF8_WARN_BOM 8
270 #define UTF8_WARN_LONG 9
271 #define UTF8_WARN_FFFF 10
274 !(flags & UTF8_ALLOW_EMPTY)) {
275 warning = UTF8_WARN_EMPTY;
279 if (UTF8_IS_ASCII(uv)) {
285 if (UTF8_IS_CONTINUATION(uv) &&
286 !(flags & UTF8_ALLOW_CONTINUATION)) {
287 warning = UTF8_WARN_CONTINUATION;
291 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
292 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
293 warning = UTF8_WARN_NON_CONTINUATION;
297 if ((uv == 0xfe || uv == 0xff) &&
298 !(flags & UTF8_ALLOW_FE_FF)) {
299 warning = UTF8_WARN_FE_FF;
303 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
304 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
305 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
306 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
307 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
308 else if (!(uv & 0x01)) { len = 7; uv = 0; }
309 else { len = 13; uv = 0; } /* whoa! */
316 if ((curlen < expectlen) &&
317 !(flags & UTF8_ALLOW_SHORT)) {
318 warning = UTF8_WARN_SHORT;
327 if (!UTF8_IS_CONTINUATION(*s) &&
328 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
330 warning = UTF8_WARN_NON_CONTINUATION;
334 uv = UTF8_ACCUMULATE(uv, *s);
336 /* These cannot be allowed. */
338 if (!(flags & UTF8_ALLOW_LONG)) {
339 warning = UTF8_WARN_LONG;
343 else { /* uv < ouv */
344 /* This cannot be allowed. */
345 warning = UTF8_WARN_OVERFLOW;
353 if (UNICODE_IS_SURROGATE(uv) &&
354 !(flags & UTF8_ALLOW_SURROGATE)) {
355 warning = UTF8_WARN_SURROGATE;
357 } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) &&
358 !(flags & UTF8_ALLOW_BOM)) {
359 warning = UTF8_WARN_BOM;
361 } else if ((expectlen > UNISKIP(uv)) &&
362 !(flags & UTF8_ALLOW_LONG)) {
363 warning = UTF8_WARN_LONG;
365 } else if (UNICODE_IS_ILLEGAL(uv) &&
366 !(flags & UTF8_ALLOW_FFFF)) {
367 warning = UTF8_WARN_FFFF;
375 if (flags & UTF8_CHECK_ONLY) {
382 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
385 case 0: /* Intentionally empty. */ break;
386 case UTF8_WARN_EMPTY:
387 Perl_sv_catpvf(aTHX_ sv, "(empty string)");
389 case UTF8_WARN_CONTINUATION:
390 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf")", uv);
392 case UTF8_WARN_NON_CONTINUATION:
393 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")",
396 case UTF8_WARN_FE_FF:
397 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
399 case UTF8_WARN_SHORT:
400 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
401 curlen, curlen == 1 ? "" : "s", expectlen);
403 case UTF8_WARN_OVERFLOW:
404 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x)",
407 case UTF8_WARN_SURROGATE:
408 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
411 Perl_sv_catpvf(aTHX_ sv, "(byte order mark 0x%04"UVxf")", uv);
414 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
415 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
418 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
421 Perl_sv_catpvf(aTHX_ sv, "(unknown reason)");
429 Perl_warner(aTHX_ WARN_UTF8,
430 "%s in %s", s, PL_op_desc[PL_op->op_type]);
432 Perl_warner(aTHX_ WARN_UTF8, "%s", s);
437 *retlen = expectlen ? expectlen : len;
443 =for apidoc A|U8* s|utf8n_to_uvchr|STRLEN curlen, STRLEN *retlen, U32 flags
445 Returns the native character value of the first character in the string C<s>
446 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
447 length, in bytes, of that character.
449 Allows length and flags to be passed to low level routine.
455 Perl_utf8n_to_uvchr(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags)
457 UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
459 return (UV) ASCII_TO_NATIVE(uv);
464 =for apidoc A|U8* s|utf8_to_uvchr|STRLEN *retlen
466 Returns the native character value of the first character in the string C<s>
467 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
468 length, in bytes, of that character.
470 If C<s> does not point to a well-formed UTF8 character, zero is
471 returned and retlen is set, if possible, to -1.
477 Perl_utf8_to_uvchr(pTHX_ U8* s, STRLEN* retlen)
479 return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXLEN, retlen, 0);
483 =for apidoc A|U8* s|utf8_to_uvuni|STRLEN *retlen
485 Returns the Unicode code point of the first character in the string C<s>
486 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
487 length, in bytes, of that character.
489 This function should only be used when returned UV is considered
490 an index into the Unicode semantic tables (e.g. swashes).
492 If C<s> does not point to a well-formed UTF8 character, zero is
493 returned and retlen is set, if possible, to -1.
499 Perl_utf8_to_uvuni(pTHX_ U8* s, STRLEN* retlen)
501 /* Call the low level routine asking for checks */
502 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXLEN, retlen, 0);
506 =for apidoc A|STRLEN|utf8_length|U8* s|U8 *e
508 Return the length of the UTF-8 char encoded string C<s> in characters.
509 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
510 up past C<e>, croaks.
516 Perl_utf8_length(pTHX_ U8* s, U8* e)
520 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
521 * the bitops (especially ~) can create illegal UTF-8.
522 * In other words: in Perl UTF-8 is not just for Unicode. */
525 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
530 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
539 =for apidoc A|IV|utf8_distance|U8 *a|U8 *b
541 Returns the number of UTF8 characters between the UTF-8 pointers C<a>
544 WARNING: use only if you *know* that the pointers point inside the
550 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
554 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
555 * the bitops (especially ~) can create illegal UTF-8.
556 * In other words: in Perl UTF-8 is not just for Unicode. */
563 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
573 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
583 =for apidoc A|U8*|utf8_hop|U8 *s|I32 off
585 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
588 WARNING: do not use the following unless you *know* C<off> is within
589 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
590 on the first byte of character or just after the last byte of a character.
595 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
597 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
598 * the bitops (especially ~) can create illegal UTF-8.
599 * In other words: in Perl UTF-8 is not just for Unicode. */
608 while (UTF8_IS_CONTINUATION(*s))
616 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
618 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
619 Unlike C<bytes_to_utf8>, this over-writes the original string, and
620 updates len to contain the new length.
621 Returns zero on failure, setting C<len> to -1.
627 Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len)
633 /* ensure valid UTF8 and chars < 256 before updating string */
634 for (send = s + *len; s < send; ) {
639 ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
648 *d++ = (U8)utf8_to_uvchr(s, &ulen);
657 =for apidoc A|U8 *|bytes_from_utf8|U8 *s|STRLEN *len|bool *is_utf8
659 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
660 Unlike <utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
661 the newly-created string, and updates C<len> to contain the new
662 length. Returns the original string if no conversion occurs, C<len>
663 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
664 0 if C<s> is converted or contains all 7bit characters.
669 Perl_bytes_from_utf8(pTHX_ U8* s, STRLEN *len, bool *is_utf8)
679 /* ensure valid UTF8 and chars < 256 before converting string */
680 for (send = s + *len; s < send;) {
682 if (!UTF8_IS_ASCII(c)) {
683 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
684 (c = *s++) && UTF8_IS_CONTINUATION(c))
694 /* Can use as-is if no high chars */
699 Newz(801, d, (*len) - count + 1, U8);
700 s = start; start = d;
703 if (!UTF8_IS_ASCII(c))
704 c = UTF8_ACCUMULATE(c, *s++);
705 *d++ = ASCII_TO_NATIVE(c);
713 =for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
715 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
716 Returns a pointer to the newly-created string, and sets C<len> to
717 reflect the new length.
723 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
730 Newz(801, d, (*len) * 2 + 1, U8);
734 UV uv = NATIVE_TO_ASCII(*s++);
735 if (UTF8_IS_ASCII(uv))
738 *d++ = UTF8_EIGHT_BIT_HI(uv);
739 *d++ = UTF8_EIGHT_BIT_LO(uv);
748 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
750 * Destination must be pre-extended to 3/2 source. Do not use in-place.
751 * We optimize for native, for obvious reasons. */
754 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
760 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
765 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
772 *d++ = (( uv >> 6) | 0xc0);
773 *d++ = (( uv & 0x3f) | 0x80);
776 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
778 if (low < 0xdc00 || low >= 0xdfff)
779 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
780 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
783 *d++ = (( uv >> 12) | 0xe0);
784 *d++ = (((uv >> 6) & 0x3f) | 0x80);
785 *d++ = (( uv & 0x3f) | 0x80);
789 *d++ = (( uv >> 18) | 0xf0);
790 *d++ = (((uv >> 12) & 0x3f) | 0x80);
791 *d++ = (((uv >> 6) & 0x3f) | 0x80);
792 *d++ = (( uv & 0x3f) | 0x80);
796 *newlen = d - dstart;
800 /* Note: this one is slightly destructive of the source. */
803 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
806 U8* send = s + bytelen;
813 return utf16_to_utf8(p, d, bytelen, newlen);
816 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
819 Perl_is_uni_alnum(pTHX_ U32 c)
821 U8 tmpbuf[UTF8_MAXLEN+1];
822 uvuni_to_utf8(tmpbuf, (UV)c);
823 return is_utf8_alnum(tmpbuf);
827 Perl_is_uni_alnumc(pTHX_ U32 c)
829 U8 tmpbuf[UTF8_MAXLEN+1];
830 uvuni_to_utf8(tmpbuf, (UV)c);
831 return is_utf8_alnumc(tmpbuf);
835 Perl_is_uni_idfirst(pTHX_ U32 c)
837 U8 tmpbuf[UTF8_MAXLEN+1];
838 uvuni_to_utf8(tmpbuf, (UV)c);
839 return is_utf8_idfirst(tmpbuf);
843 Perl_is_uni_alpha(pTHX_ U32 c)
845 U8 tmpbuf[UTF8_MAXLEN+1];
846 uvuni_to_utf8(tmpbuf, (UV)c);
847 return is_utf8_alpha(tmpbuf);
851 Perl_is_uni_ascii(pTHX_ U32 c)
853 U8 tmpbuf[UTF8_MAXLEN+1];
854 uvuni_to_utf8(tmpbuf, (UV)c);
855 return is_utf8_ascii(tmpbuf);
859 Perl_is_uni_space(pTHX_ U32 c)
861 U8 tmpbuf[UTF8_MAXLEN+1];
862 uvuni_to_utf8(tmpbuf, (UV)c);
863 return is_utf8_space(tmpbuf);
867 Perl_is_uni_digit(pTHX_ U32 c)
869 U8 tmpbuf[UTF8_MAXLEN+1];
870 uvuni_to_utf8(tmpbuf, (UV)c);
871 return is_utf8_digit(tmpbuf);
875 Perl_is_uni_upper(pTHX_ U32 c)
877 U8 tmpbuf[UTF8_MAXLEN+1];
878 uvuni_to_utf8(tmpbuf, (UV)c);
879 return is_utf8_upper(tmpbuf);
883 Perl_is_uni_lower(pTHX_ U32 c)
885 U8 tmpbuf[UTF8_MAXLEN+1];
886 uvuni_to_utf8(tmpbuf, (UV)c);
887 return is_utf8_lower(tmpbuf);
891 Perl_is_uni_cntrl(pTHX_ U32 c)
893 U8 tmpbuf[UTF8_MAXLEN+1];
894 uvuni_to_utf8(tmpbuf, (UV)c);
895 return is_utf8_cntrl(tmpbuf);
899 Perl_is_uni_graph(pTHX_ U32 c)
901 U8 tmpbuf[UTF8_MAXLEN+1];
902 uvuni_to_utf8(tmpbuf, (UV)c);
903 return is_utf8_graph(tmpbuf);
907 Perl_is_uni_print(pTHX_ U32 c)
909 U8 tmpbuf[UTF8_MAXLEN+1];
910 uvuni_to_utf8(tmpbuf, (UV)c);
911 return is_utf8_print(tmpbuf);
915 Perl_is_uni_punct(pTHX_ U32 c)
917 U8 tmpbuf[UTF8_MAXLEN+1];
918 uvuni_to_utf8(tmpbuf, (UV)c);
919 return is_utf8_punct(tmpbuf);
923 Perl_is_uni_xdigit(pTHX_ U32 c)
925 U8 tmpbuf[UTF8_MAXLEN+1];
926 uvuni_to_utf8(tmpbuf, (UV)c);
927 return is_utf8_xdigit(tmpbuf);
931 Perl_to_uni_upper(pTHX_ U32 c)
933 U8 tmpbuf[UTF8_MAXLEN+1];
934 uvuni_to_utf8(tmpbuf, (UV)c);
935 return to_utf8_upper(tmpbuf);
939 Perl_to_uni_title(pTHX_ U32 c)
941 U8 tmpbuf[UTF8_MAXLEN+1];
942 uvuni_to_utf8(tmpbuf, (UV)c);
943 return to_utf8_title(tmpbuf);
947 Perl_to_uni_lower(pTHX_ U32 c)
949 U8 tmpbuf[UTF8_MAXLEN+1];
950 uvuni_to_utf8(tmpbuf, (UV)c);
951 return to_utf8_lower(tmpbuf);
954 /* for now these all assume no locale info available for Unicode > 255 */
957 Perl_is_uni_alnum_lc(pTHX_ U32 c)
959 return is_uni_alnum(c); /* XXX no locale support yet */
963 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
965 return is_uni_alnumc(c); /* XXX no locale support yet */
969 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
971 return is_uni_idfirst(c); /* XXX no locale support yet */
975 Perl_is_uni_alpha_lc(pTHX_ U32 c)
977 return is_uni_alpha(c); /* XXX no locale support yet */
981 Perl_is_uni_ascii_lc(pTHX_ U32 c)
983 return is_uni_ascii(c); /* XXX no locale support yet */
987 Perl_is_uni_space_lc(pTHX_ U32 c)
989 return is_uni_space(c); /* XXX no locale support yet */
993 Perl_is_uni_digit_lc(pTHX_ U32 c)
995 return is_uni_digit(c); /* XXX no locale support yet */
999 Perl_is_uni_upper_lc(pTHX_ U32 c)
1001 return is_uni_upper(c); /* XXX no locale support yet */
1005 Perl_is_uni_lower_lc(pTHX_ U32 c)
1007 return is_uni_lower(c); /* XXX no locale support yet */
1011 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
1013 return is_uni_cntrl(c); /* XXX no locale support yet */
1017 Perl_is_uni_graph_lc(pTHX_ U32 c)
1019 return is_uni_graph(c); /* XXX no locale support yet */
1023 Perl_is_uni_print_lc(pTHX_ U32 c)
1025 return is_uni_print(c); /* XXX no locale support yet */
1029 Perl_is_uni_punct_lc(pTHX_ U32 c)
1031 return is_uni_punct(c); /* XXX no locale support yet */
1035 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
1037 return is_uni_xdigit(c); /* XXX no locale support yet */
1041 Perl_to_uni_upper_lc(pTHX_ U32 c)
1043 return to_uni_upper(c); /* XXX no locale support yet */
1047 Perl_to_uni_title_lc(pTHX_ U32 c)
1049 return to_uni_title(c); /* XXX no locale support yet */
1053 Perl_to_uni_lower_lc(pTHX_ U32 c)
1055 return to_uni_lower(c); /* XXX no locale support yet */
1059 Perl_is_utf8_alnum(pTHX_ U8 *p)
1061 if (!is_utf8_char(p))
1064 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1065 * descendant of isalnum(3), in other words, it doesn't
1066 * contain the '_'. --jhi */
1067 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1068 return swash_fetch(PL_utf8_alnum, p);
1069 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1070 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1072 PL_utf8_alnum = swash_init("utf8", "",
1073 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1074 return swash_fetch(PL_utf8_alnum, p);
1079 Perl_is_utf8_alnumc(pTHX_ U8 *p)
1081 if (!is_utf8_char(p))
1084 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1085 return swash_fetch(PL_utf8_alnum, p);
1086 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
1087 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1089 PL_utf8_alnum = swash_init("utf8", "",
1090 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1091 return swash_fetch(PL_utf8_alnum, p);
1096 Perl_is_utf8_idfirst(pTHX_ U8 *p)
1098 return *p == '_' || is_utf8_alpha(p);
1102 Perl_is_utf8_alpha(pTHX_ U8 *p)
1104 if (!is_utf8_char(p))
1107 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1108 return swash_fetch(PL_utf8_alpha, p);
1112 Perl_is_utf8_ascii(pTHX_ U8 *p)
1114 if (!is_utf8_char(p))
1117 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1118 return swash_fetch(PL_utf8_ascii, p);
1122 Perl_is_utf8_space(pTHX_ U8 *p)
1124 if (!is_utf8_char(p))
1127 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1128 return swash_fetch(PL_utf8_space, p);
1132 Perl_is_utf8_digit(pTHX_ U8 *p)
1134 if (!is_utf8_char(p))
1137 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1138 return swash_fetch(PL_utf8_digit, p);
1142 Perl_is_utf8_upper(pTHX_ U8 *p)
1144 if (!is_utf8_char(p))
1147 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
1148 return swash_fetch(PL_utf8_upper, p);
1152 Perl_is_utf8_lower(pTHX_ U8 *p)
1154 if (!is_utf8_char(p))
1157 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
1158 return swash_fetch(PL_utf8_lower, p);
1162 Perl_is_utf8_cntrl(pTHX_ U8 *p)
1164 if (!is_utf8_char(p))
1167 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1168 return swash_fetch(PL_utf8_cntrl, p);
1172 Perl_is_utf8_graph(pTHX_ U8 *p)
1174 if (!is_utf8_char(p))
1177 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1178 return swash_fetch(PL_utf8_graph, p);
1182 Perl_is_utf8_print(pTHX_ U8 *p)
1184 if (!is_utf8_char(p))
1187 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1188 return swash_fetch(PL_utf8_print, p);
1192 Perl_is_utf8_punct(pTHX_ U8 *p)
1194 if (!is_utf8_char(p))
1197 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1198 return swash_fetch(PL_utf8_punct, p);
1202 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1204 if (!is_utf8_char(p))
1206 if (!PL_utf8_xdigit)
1207 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1208 return swash_fetch(PL_utf8_xdigit, p);
1212 Perl_is_utf8_mark(pTHX_ U8 *p)
1214 if (!is_utf8_char(p))
1217 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1218 return swash_fetch(PL_utf8_mark, p);
1222 Perl_to_utf8_upper(pTHX_ U8 *p)
1226 if (!PL_utf8_toupper)
1227 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
1228 uv = swash_fetch(PL_utf8_toupper, p);
1229 return uv ? UNI_TO_NATIVE(uv) : utf8_to_uvchr(p,0);
1233 Perl_to_utf8_title(pTHX_ U8 *p)
1237 if (!PL_utf8_totitle)
1238 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
1239 uv = swash_fetch(PL_utf8_totitle, p);
1240 return uv ? UNI_TO_NATIVE(uv) : utf8_to_uvchr(p,0);
1244 Perl_to_utf8_lower(pTHX_ U8 *p)
1248 if (!PL_utf8_tolower)
1249 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
1250 uv = swash_fetch(PL_utf8_tolower, p);
1251 return uv ? UNI_TO_NATIVE(uv) : utf8_to_uvchr(p,0);
1254 /* a "swash" is a swatch hash */
1257 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1260 SV* tokenbufsv = sv_2mortal(NEWSV(0,0));
1263 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
1265 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1269 PUSHSTACKi(PERLSI_MAGIC);
1272 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1273 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1275 PUSHs(sv_2mortal(newSViv(minbits)));
1276 PUSHs(sv_2mortal(newSViv(none)));
1282 if (PL_curcop == &PL_compiling)
1283 /* XXX ought to be handled by lex_start */
1284 sv_setpv(tokenbufsv, PL_tokenbuf);
1285 if (call_method("SWASHNEW", G_SCALAR))
1286 retval = newSVsv(*PL_stack_sp--);
1288 retval = &PL_sv_undef;
1291 if (PL_curcop == &PL_compiling) {
1293 char* pv = SvPV(tokenbufsv, len);
1295 Copy(pv, PL_tokenbuf, len+1, char);
1296 PL_curcop->op_private = PL_hints;
1298 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1299 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1304 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1306 HV* hv = (HV*)SvRV(sv);
1307 U32 klen = UTF8SKIP(ptr) - 1;
1308 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1310 STRLEN needents = (klen ? 64 : 128);
1316 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1317 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1318 * it's nothing to sniff at.) Pity we usually come through at least
1319 * two function calls to get here...
1321 * NB: this code assumes that swatches are never modified, once generated!
1324 if (hv == PL_last_swash_hv &&
1325 klen == PL_last_swash_klen &&
1326 (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
1328 tmps = PL_last_swash_tmps;
1329 slen = PL_last_swash_slen;
1332 /* Try our second-level swatch cache, kept in a hash. */
1333 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1335 /* If not cached, generate it via utf8::SWASHGET */
1336 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1341 PUSHSTACKi(PERLSI_MAGIC);
1345 /* We call utf8_to_uni as we want and index into Unicode tables,
1346 not a native character number.
1348 PUSHs(sv_2mortal(newSViv(utf8_to_uvuni(ptr, 0) & ~(needents - 1))));
1349 PUSHs(sv_2mortal(newSViv(needents)));
1351 if (call_method("SWASHGET", G_SCALAR))
1352 retval = newSVsv(*PL_stack_sp--);
1354 retval = &PL_sv_undef;
1358 if (PL_curcop == &PL_compiling)
1359 PL_curcop->op_private = PL_hints;
1361 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1363 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1364 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1367 PL_last_swash_hv = hv;
1368 PL_last_swash_klen = klen;
1369 PL_last_swash_tmps = tmps;
1370 PL_last_swash_slen = slen;
1372 Copy(ptr, PL_last_swash_key, klen, U8);
1375 switch ((int)((slen << 3) / needents)) {
1377 bit = 1 << (off & 7);
1379 return (tmps[off] & bit) != 0;
1384 return (tmps[off] << 8) + tmps[off + 1] ;
1387 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1389 Perl_croak(aTHX_ "panic: swash_fetch");