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_CONTINUATION(c) || s >= send ||
684 !UTF8_IS_CONTINUATION(*s) || UTF8_IS_DOWNGRADEABLE_START(c))
695 Newz(801, d, (*len) - count + 1, U8);
696 s = start; start = d;
700 if (UTF8_IS_ASCII(c))
703 *d++ = UTF8_ACCUMULATE(c, *s++);
711 =for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
713 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
714 Returns a pointer to the newly-created string, and sets C<len> to
715 reflect the new length.
721 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
728 Newz(801, d, (*len) * 2 + 1, U8);
732 if (UTF8_IS_ASCII(*s))
737 *d++ = UTF8_EIGHT_BIT_HI(uv);
738 *d++ = UTF8_EIGHT_BIT_LO(uv);
747 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
749 * Destination must be pre-extended to 3/2 source. Do not use in-place.
750 * We optimize for native, for obvious reasons. */
753 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
759 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
764 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
771 *d++ = (( uv >> 6) | 0xc0);
772 *d++ = (( uv & 0x3f) | 0x80);
775 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
777 if (low < 0xdc00 || low >= 0xdfff)
778 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
779 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
782 *d++ = (( uv >> 12) | 0xe0);
783 *d++ = (((uv >> 6) & 0x3f) | 0x80);
784 *d++ = (( uv & 0x3f) | 0x80);
788 *d++ = (( uv >> 18) | 0xf0);
789 *d++ = (((uv >> 12) & 0x3f) | 0x80);
790 *d++ = (((uv >> 6) & 0x3f) | 0x80);
791 *d++ = (( uv & 0x3f) | 0x80);
795 *newlen = d - dstart;
799 /* Note: this one is slightly destructive of the source. */
802 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
805 U8* send = s + bytelen;
812 return utf16_to_utf8(p, d, bytelen, newlen);
815 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
818 Perl_is_uni_alnum(pTHX_ U32 c)
820 U8 tmpbuf[UTF8_MAXLEN+1];
821 uvuni_to_utf8(tmpbuf, (UV)c);
822 return is_utf8_alnum(tmpbuf);
826 Perl_is_uni_alnumc(pTHX_ U32 c)
828 U8 tmpbuf[UTF8_MAXLEN+1];
829 uvuni_to_utf8(tmpbuf, (UV)c);
830 return is_utf8_alnumc(tmpbuf);
834 Perl_is_uni_idfirst(pTHX_ U32 c)
836 U8 tmpbuf[UTF8_MAXLEN+1];
837 uvuni_to_utf8(tmpbuf, (UV)c);
838 return is_utf8_idfirst(tmpbuf);
842 Perl_is_uni_alpha(pTHX_ U32 c)
844 U8 tmpbuf[UTF8_MAXLEN+1];
845 uvuni_to_utf8(tmpbuf, (UV)c);
846 return is_utf8_alpha(tmpbuf);
850 Perl_is_uni_ascii(pTHX_ U32 c)
852 U8 tmpbuf[UTF8_MAXLEN+1];
853 uvuni_to_utf8(tmpbuf, (UV)c);
854 return is_utf8_ascii(tmpbuf);
858 Perl_is_uni_space(pTHX_ U32 c)
860 U8 tmpbuf[UTF8_MAXLEN+1];
861 uvuni_to_utf8(tmpbuf, (UV)c);
862 return is_utf8_space(tmpbuf);
866 Perl_is_uni_digit(pTHX_ U32 c)
868 U8 tmpbuf[UTF8_MAXLEN+1];
869 uvuni_to_utf8(tmpbuf, (UV)c);
870 return is_utf8_digit(tmpbuf);
874 Perl_is_uni_upper(pTHX_ U32 c)
876 U8 tmpbuf[UTF8_MAXLEN+1];
877 uvuni_to_utf8(tmpbuf, (UV)c);
878 return is_utf8_upper(tmpbuf);
882 Perl_is_uni_lower(pTHX_ U32 c)
884 U8 tmpbuf[UTF8_MAXLEN+1];
885 uvuni_to_utf8(tmpbuf, (UV)c);
886 return is_utf8_lower(tmpbuf);
890 Perl_is_uni_cntrl(pTHX_ U32 c)
892 U8 tmpbuf[UTF8_MAXLEN+1];
893 uvuni_to_utf8(tmpbuf, (UV)c);
894 return is_utf8_cntrl(tmpbuf);
898 Perl_is_uni_graph(pTHX_ U32 c)
900 U8 tmpbuf[UTF8_MAXLEN+1];
901 uvuni_to_utf8(tmpbuf, (UV)c);
902 return is_utf8_graph(tmpbuf);
906 Perl_is_uni_print(pTHX_ U32 c)
908 U8 tmpbuf[UTF8_MAXLEN+1];
909 uvuni_to_utf8(tmpbuf, (UV)c);
910 return is_utf8_print(tmpbuf);
914 Perl_is_uni_punct(pTHX_ U32 c)
916 U8 tmpbuf[UTF8_MAXLEN+1];
917 uvuni_to_utf8(tmpbuf, (UV)c);
918 return is_utf8_punct(tmpbuf);
922 Perl_is_uni_xdigit(pTHX_ U32 c)
924 U8 tmpbuf[UTF8_MAXLEN+1];
925 uvuni_to_utf8(tmpbuf, (UV)c);
926 return is_utf8_xdigit(tmpbuf);
930 Perl_to_uni_upper(pTHX_ U32 c)
932 U8 tmpbuf[UTF8_MAXLEN+1];
933 uvuni_to_utf8(tmpbuf, (UV)c);
934 return to_utf8_upper(tmpbuf);
938 Perl_to_uni_title(pTHX_ U32 c)
940 U8 tmpbuf[UTF8_MAXLEN+1];
941 uvuni_to_utf8(tmpbuf, (UV)c);
942 return to_utf8_title(tmpbuf);
946 Perl_to_uni_lower(pTHX_ U32 c)
948 U8 tmpbuf[UTF8_MAXLEN+1];
949 uvuni_to_utf8(tmpbuf, (UV)c);
950 return to_utf8_lower(tmpbuf);
953 /* for now these all assume no locale info available for Unicode > 255 */
956 Perl_is_uni_alnum_lc(pTHX_ U32 c)
958 return is_uni_alnum(c); /* XXX no locale support yet */
962 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
964 return is_uni_alnumc(c); /* XXX no locale support yet */
968 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
970 return is_uni_idfirst(c); /* XXX no locale support yet */
974 Perl_is_uni_alpha_lc(pTHX_ U32 c)
976 return is_uni_alpha(c); /* XXX no locale support yet */
980 Perl_is_uni_ascii_lc(pTHX_ U32 c)
982 return is_uni_ascii(c); /* XXX no locale support yet */
986 Perl_is_uni_space_lc(pTHX_ U32 c)
988 return is_uni_space(c); /* XXX no locale support yet */
992 Perl_is_uni_digit_lc(pTHX_ U32 c)
994 return is_uni_digit(c); /* XXX no locale support yet */
998 Perl_is_uni_upper_lc(pTHX_ U32 c)
1000 return is_uni_upper(c); /* XXX no locale support yet */
1004 Perl_is_uni_lower_lc(pTHX_ U32 c)
1006 return is_uni_lower(c); /* XXX no locale support yet */
1010 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
1012 return is_uni_cntrl(c); /* XXX no locale support yet */
1016 Perl_is_uni_graph_lc(pTHX_ U32 c)
1018 return is_uni_graph(c); /* XXX no locale support yet */
1022 Perl_is_uni_print_lc(pTHX_ U32 c)
1024 return is_uni_print(c); /* XXX no locale support yet */
1028 Perl_is_uni_punct_lc(pTHX_ U32 c)
1030 return is_uni_punct(c); /* XXX no locale support yet */
1034 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
1036 return is_uni_xdigit(c); /* XXX no locale support yet */
1040 Perl_to_uni_upper_lc(pTHX_ U32 c)
1042 return to_uni_upper(c); /* XXX no locale support yet */
1046 Perl_to_uni_title_lc(pTHX_ U32 c)
1048 return to_uni_title(c); /* XXX no locale support yet */
1052 Perl_to_uni_lower_lc(pTHX_ U32 c)
1054 return to_uni_lower(c); /* XXX no locale support yet */
1058 Perl_is_utf8_alnum(pTHX_ U8 *p)
1060 if (!is_utf8_char(p))
1063 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1064 * descendant of isalnum(3), in other words, it doesn't
1065 * contain the '_'. --jhi */
1066 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1067 return swash_fetch(PL_utf8_alnum, p);
1068 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1069 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1071 PL_utf8_alnum = swash_init("utf8", "",
1072 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1073 return swash_fetch(PL_utf8_alnum, p);
1078 Perl_is_utf8_alnumc(pTHX_ U8 *p)
1080 if (!is_utf8_char(p))
1083 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1084 return swash_fetch(PL_utf8_alnum, p);
1085 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
1086 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1088 PL_utf8_alnum = swash_init("utf8", "",
1089 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1090 return swash_fetch(PL_utf8_alnum, p);
1095 Perl_is_utf8_idfirst(pTHX_ U8 *p)
1097 return *p == '_' || is_utf8_alpha(p);
1101 Perl_is_utf8_alpha(pTHX_ U8 *p)
1103 if (!is_utf8_char(p))
1106 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1107 return swash_fetch(PL_utf8_alpha, p);
1111 Perl_is_utf8_ascii(pTHX_ U8 *p)
1113 if (!is_utf8_char(p))
1116 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1117 return swash_fetch(PL_utf8_ascii, p);
1121 Perl_is_utf8_space(pTHX_ U8 *p)
1123 if (!is_utf8_char(p))
1126 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1127 return swash_fetch(PL_utf8_space, p);
1131 Perl_is_utf8_digit(pTHX_ U8 *p)
1133 if (!is_utf8_char(p))
1136 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1137 return swash_fetch(PL_utf8_digit, p);
1141 Perl_is_utf8_upper(pTHX_ U8 *p)
1143 if (!is_utf8_char(p))
1146 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
1147 return swash_fetch(PL_utf8_upper, p);
1151 Perl_is_utf8_lower(pTHX_ U8 *p)
1153 if (!is_utf8_char(p))
1156 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
1157 return swash_fetch(PL_utf8_lower, p);
1161 Perl_is_utf8_cntrl(pTHX_ U8 *p)
1163 if (!is_utf8_char(p))
1166 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1167 return swash_fetch(PL_utf8_cntrl, p);
1171 Perl_is_utf8_graph(pTHX_ U8 *p)
1173 if (!is_utf8_char(p))
1176 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1177 return swash_fetch(PL_utf8_graph, p);
1181 Perl_is_utf8_print(pTHX_ U8 *p)
1183 if (!is_utf8_char(p))
1186 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1187 return swash_fetch(PL_utf8_print, p);
1191 Perl_is_utf8_punct(pTHX_ U8 *p)
1193 if (!is_utf8_char(p))
1196 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1197 return swash_fetch(PL_utf8_punct, p);
1201 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1203 if (!is_utf8_char(p))
1205 if (!PL_utf8_xdigit)
1206 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1207 return swash_fetch(PL_utf8_xdigit, p);
1211 Perl_is_utf8_mark(pTHX_ U8 *p)
1213 if (!is_utf8_char(p))
1216 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1217 return swash_fetch(PL_utf8_mark, p);
1221 Perl_to_utf8_upper(pTHX_ U8 *p)
1225 if (!PL_utf8_toupper)
1226 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
1227 uv = swash_fetch(PL_utf8_toupper, p);
1228 return uv ? UNI_TO_NATIVE(uv) : utf8_to_uvchr(p,0);
1232 Perl_to_utf8_title(pTHX_ U8 *p)
1236 if (!PL_utf8_totitle)
1237 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
1238 uv = swash_fetch(PL_utf8_totitle, p);
1239 return uv ? UNI_TO_NATIVE(uv) : utf8_to_uvchr(p,0);
1243 Perl_to_utf8_lower(pTHX_ U8 *p)
1247 if (!PL_utf8_tolower)
1248 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
1249 uv = swash_fetch(PL_utf8_tolower, p);
1250 return uv ? UNI_TO_NATIVE(uv) : utf8_to_uvchr(p,0);
1253 /* a "swash" is a swatch hash */
1256 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1259 SV* tokenbufsv = sv_2mortal(NEWSV(0,0));
1262 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
1264 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1268 PUSHSTACKi(PERLSI_MAGIC);
1271 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1272 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1274 PUSHs(sv_2mortal(newSViv(minbits)));
1275 PUSHs(sv_2mortal(newSViv(none)));
1281 if (PL_curcop == &PL_compiling)
1282 /* XXX ought to be handled by lex_start */
1283 sv_setpv(tokenbufsv, PL_tokenbuf);
1284 if (call_method("SWASHNEW", G_SCALAR))
1285 retval = newSVsv(*PL_stack_sp--);
1287 retval = &PL_sv_undef;
1290 if (PL_curcop == &PL_compiling) {
1292 char* pv = SvPV(tokenbufsv, len);
1294 Copy(pv, PL_tokenbuf, len+1, char);
1295 PL_curcop->op_private = PL_hints;
1297 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1298 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1303 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1305 HV* hv = (HV*)SvRV(sv);
1306 U32 klen = UTF8SKIP(ptr) - 1;
1307 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1309 STRLEN needents = (klen ? 64 : 128);
1315 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1316 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1317 * it's nothing to sniff at.) Pity we usually come through at least
1318 * two function calls to get here...
1320 * NB: this code assumes that swatches are never modified, once generated!
1323 if (hv == PL_last_swash_hv &&
1324 klen == PL_last_swash_klen &&
1325 (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
1327 tmps = PL_last_swash_tmps;
1328 slen = PL_last_swash_slen;
1331 /* Try our second-level swatch cache, kept in a hash. */
1332 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1334 /* If not cached, generate it via utf8::SWASHGET */
1335 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1340 PUSHSTACKi(PERLSI_MAGIC);
1344 /* We call utf8_to_uni as we want and index into Unicode tables,
1345 not a native character number.
1347 PUSHs(sv_2mortal(newSViv(utf8_to_uvuni(ptr, 0) & ~(needents - 1))));
1348 PUSHs(sv_2mortal(newSViv(needents)));
1350 if (call_method("SWASHGET", G_SCALAR))
1351 retval = newSVsv(*PL_stack_sp--);
1353 retval = &PL_sv_undef;
1357 if (PL_curcop == &PL_compiling)
1358 PL_curcop->op_private = PL_hints;
1360 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1362 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1363 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1366 PL_last_swash_hv = hv;
1367 PL_last_swash_klen = klen;
1368 PL_last_swash_tmps = tmps;
1369 PL_last_swash_slen = slen;
1371 Copy(ptr, PL_last_swash_key, klen, U8);
1374 switch ((int)((slen << 3) / needents)) {
1376 bit = 1 << (off & 7);
1378 return (tmps[off] & bit) != 0;
1383 return (tmps[off] << 8) + tmps[off + 1] ;
1386 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1388 Perl_croak(aTHX_ "panic: swash_fetch");