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*|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)
50 uv = NATIVE_TO_ASCII(uv);
56 *d++ = (( uv >> 6) | 0xc0);
57 *d++ = (( uv & 0x3f) | 0x80);
61 *d++ = (( uv >> 12) | 0xe0);
62 *d++ = (((uv >> 6) & 0x3f) | 0x80);
63 *d++ = (( uv & 0x3f) | 0x80);
67 *d++ = (( uv >> 18) | 0xf0);
68 *d++ = (((uv >> 12) & 0x3f) | 0x80);
69 *d++ = (((uv >> 6) & 0x3f) | 0x80);
70 *d++ = (( uv & 0x3f) | 0x80);
74 *d++ = (( uv >> 24) | 0xf8);
75 *d++ = (((uv >> 18) & 0x3f) | 0x80);
76 *d++ = (((uv >> 12) & 0x3f) | 0x80);
77 *d++ = (((uv >> 6) & 0x3f) | 0x80);
78 *d++ = (( uv & 0x3f) | 0x80);
81 if (uv < 0x80000000) {
82 *d++ = (( uv >> 30) | 0xfc);
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);
91 if (uv < UTF8_QUAD_MAX)
94 *d++ = 0xfe; /* Can't match U+FEFF! */
95 *d++ = (((uv >> 30) & 0x3f) | 0x80);
96 *d++ = (((uv >> 24) & 0x3f) | 0x80);
97 *d++ = (((uv >> 18) & 0x3f) | 0x80);
98 *d++ = (((uv >> 12) & 0x3f) | 0x80);
99 *d++ = (((uv >> 6) & 0x3f) | 0x80);
100 *d++ = (( uv & 0x3f) | 0x80);
105 *d++ = 0xff; /* Can't match U+FFFE! */
106 *d++ = 0x80; /* 6 Reserved bits */
107 *d++ = (((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
108 *d++ = (((uv >> 54) & 0x3f) | 0x80);
109 *d++ = (((uv >> 48) & 0x3f) | 0x80);
110 *d++ = (((uv >> 42) & 0x3f) | 0x80);
111 *d++ = (((uv >> 36) & 0x3f) | 0x80);
112 *d++ = (((uv >> 30) & 0x3f) | 0x80);
113 *d++ = (((uv >> 24) & 0x3f) | 0x80);
114 *d++ = (((uv >> 18) & 0x3f) | 0x80);
115 *d++ = (((uv >> 12) & 0x3f) | 0x80);
116 *d++ = (((uv >> 6) & 0x3f) | 0x80);
117 *d++ = (( uv & 0x3f) | 0x80);
124 =for apidoc A|STRLEN|is_utf8_char|U8 *s
126 Tests if some arbitrary number of bytes begins in a valid UTF-8
127 character. Note that an ASCII character is a valid UTF-8 character.
128 The actual number of bytes in the UTF-8 character will be returned if
129 it is valid, otherwise 0.
133 Perl_is_utf8_char(pTHX_ U8 *s)
139 if (UTF8_IS_ASCII(u))
142 if (!UTF8_IS_START(u))
147 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
155 if (!UTF8_IS_CONTINUATION(*s))
157 uv = UTF8_ACCUMULATE(uv, *s);
164 if (UNISKIP(uv) < len)
171 =for apidoc A|bool|is_utf8_string|U8 *s|STRLEN len
173 Returns true if first C<len> bytes of the given string form a valid UTF8
174 string, false otherwise. Note that 'a valid UTF8 string' does not mean
175 'a string that contains UTF8' because a valid ASCII string is a valid
182 Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len)
189 len = strlen((char *)s);
205 =for apidoc A|UV|utf8_to_uv|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
207 Returns the character value of the first character in the string C<s>
208 which is assumed to be in UTF8 encoding and no longer than C<curlen>;
209 C<retlen> will be set to the length, in bytes, of that character.
211 If C<s> does not point to a well-formed UTF8 character, the behaviour
212 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
213 it is assumed that the caller will raise a warning, and this function
214 will silently just set C<retlen> to C<-1> and return zero. If the
215 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
216 malformations will be given, C<retlen> will be set to the expected
217 length of the UTF-8 character in bytes, and zero will be returned.
219 The C<flags> can also contain various flags to allow deviations from
220 the strict UTF-8 encoding (see F<utf8.h>).
225 Perl_utf8_to_uv(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags)
232 bool dowarn = ckWARN_d(WARN_UTF8);
234 STRLEN expectlen = 0;
237 /* This list is a superset of the UTF8_ALLOW_XXX. */
239 #define UTF8_WARN_EMPTY 1
240 #define UTF8_WARN_CONTINUATION 2
241 #define UTF8_WARN_NON_CONTINUATION 3
242 #define UTF8_WARN_FE_FF 4
243 #define UTF8_WARN_SHORT 5
244 #define UTF8_WARN_OVERFLOW 6
245 #define UTF8_WARN_SURROGATE 7
246 #define UTF8_WARN_BOM 8
247 #define UTF8_WARN_LONG 9
248 #define UTF8_WARN_FFFF 10
251 !(flags & UTF8_ALLOW_EMPTY)) {
252 warning = UTF8_WARN_EMPTY;
256 if (UTF8_IS_ASCII(uv)) {
259 return ASCII_TO_NATIVE(*s);
262 if (UTF8_IS_CONTINUATION(uv) &&
263 !(flags & UTF8_ALLOW_CONTINUATION)) {
264 warning = UTF8_WARN_CONTINUATION;
268 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
269 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
270 warning = UTF8_WARN_NON_CONTINUATION;
274 if ((uv == 0xfe || uv == 0xff) &&
275 !(flags & UTF8_ALLOW_FE_FF)) {
276 warning = UTF8_WARN_FE_FF;
280 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
281 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
282 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
283 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
284 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
285 else if (!(uv & 0x01)) { len = 7; uv = 0; }
286 else { len = 13; uv = 0; } /* whoa! */
293 if ((curlen < expectlen) &&
294 !(flags & UTF8_ALLOW_SHORT)) {
295 warning = UTF8_WARN_SHORT;
304 if (!UTF8_IS_CONTINUATION(*s) &&
305 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
307 warning = UTF8_WARN_NON_CONTINUATION;
311 uv = UTF8_ACCUMULATE(uv, *s);
313 /* These cannot be allowed. */
315 if (!(flags & UTF8_ALLOW_LONG)) {
316 warning = UTF8_WARN_LONG;
320 else { /* uv < ouv */
321 /* This cannot be allowed. */
322 warning = UTF8_WARN_OVERFLOW;
330 if (UNICODE_IS_SURROGATE(uv) &&
331 !(flags & UTF8_ALLOW_SURROGATE)) {
332 warning = UTF8_WARN_SURROGATE;
334 } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) &&
335 !(flags & UTF8_ALLOW_BOM)) {
336 warning = UTF8_WARN_BOM;
338 } else if ((expectlen > UNISKIP(uv)) &&
339 !(flags & UTF8_ALLOW_LONG)) {
340 warning = UTF8_WARN_LONG;
342 } else if (UNICODE_IS_ILLEGAL(uv) &&
343 !(flags & UTF8_ALLOW_FFFF)) {
344 warning = UTF8_WARN_FFFF;
352 if (flags & UTF8_CHECK_ONLY) {
359 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
362 case 0: /* Intentionally empty. */ break;
363 case UTF8_WARN_EMPTY:
364 Perl_sv_catpvf(aTHX_ sv, "(empty string)");
366 case UTF8_WARN_CONTINUATION:
367 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf")", uv);
369 case UTF8_WARN_NON_CONTINUATION:
370 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")",
373 case UTF8_WARN_FE_FF:
374 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
376 case UTF8_WARN_SHORT:
377 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
378 curlen, curlen == 1 ? "" : "s", expectlen);
380 case UTF8_WARN_OVERFLOW:
381 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x)",
384 case UTF8_WARN_SURROGATE:
385 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
388 Perl_sv_catpvf(aTHX_ sv, "(byte order mark 0x%04"UVxf")", uv);
391 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
392 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
395 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
398 Perl_sv_catpvf(aTHX_ sv, "(unknown reason)");
406 Perl_warner(aTHX_ WARN_UTF8,
407 "%s in %s", s, PL_op_desc[PL_op->op_type]);
409 Perl_warner(aTHX_ WARN_UTF8, "%s", s);
414 *retlen = expectlen ? expectlen : len;
420 =for apidoc A|U8* s|utf8_to_uv_simple|STRLEN *retlen
422 Returns the character value of the first character in the string C<s>
423 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
424 length, in bytes, of that character.
426 If C<s> does not point to a well-formed UTF8 character, zero is
427 returned and retlen is set, if possible, to -1.
433 Perl_utf8_to_uv_simple(pTHX_ U8* s, STRLEN* retlen)
435 return Perl_utf8_to_uv(aTHX_ s, UTF8_MAXLEN, retlen, 0);
439 =for apidoc A|STRLEN|utf8_length|U8* s|U8 *e
441 Return the length of the UTF-8 char encoded string C<s> in characters.
442 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
443 up past C<e>, croaks.
449 Perl_utf8_length(pTHX_ U8* s, U8* e)
453 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
454 * the bitops (especially ~) can create illegal UTF-8.
455 * In other words: in Perl UTF-8 is not just for Unicode. */
458 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
463 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
472 =for apidoc A|IV|utf8_distance|U8 *a|U8 *b
474 Returns the number of UTF8 characters between the UTF-8 pointers C<a>
477 WARNING: use only if you *know* that the pointers point inside the
483 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
487 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
488 * the bitops (especially ~) can create illegal UTF-8.
489 * In other words: in Perl UTF-8 is not just for Unicode. */
496 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
506 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
516 =for apidoc A|U8*|utf8_hop|U8 *s|I32 off
518 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
521 WARNING: do not use the following unless you *know* C<off> is within
522 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
523 on the first byte of character or just after the last byte of a character.
528 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
530 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
531 * the bitops (especially ~) can create illegal UTF-8.
532 * In other words: in Perl UTF-8 is not just for Unicode. */
541 while (UTF8_IS_CONTINUATION(*s))
549 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
551 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
552 Unlike C<bytes_to_utf8>, this over-writes the original string, and
553 updates len to contain the new length.
554 Returns zero on failure, setting C<len> to -1.
560 Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len)
566 /* ensure valid UTF8 and chars < 256 before updating string */
567 for (send = s + *len; s < send; ) {
572 ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
581 *d++ = (U8)utf8_to_uv_simple(s, &ulen);
590 =for apidoc A|U8 *|bytes_from_utf8|U8 *s|STRLEN *len|bool *is_utf8
592 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
593 Unlike <utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
594 the newly-created string, and updates C<len> to contain the new
595 length. Returns the original string if no conversion occurs, C<len>
596 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
597 0 if C<s> is converted or contains all 7bit characters.
602 Perl_bytes_from_utf8(pTHX_ U8* s, STRLEN *len, bool *is_utf8)
612 /* ensure valid UTF8 and chars < 256 before converting string */
613 for (send = s + *len; s < send;) {
615 if (!UTF8_IS_ASCII(c)) {
616 if (UTF8_IS_CONTINUATION(c) || s >= send ||
617 !UTF8_IS_CONTINUATION(*s) || UTF8_IS_DOWNGRADEABLE_START(c))
628 Newz(801, d, (*len) - count + 1, U8);
629 s = start; start = d;
633 if (UTF8_IS_ASCII(c))
636 *d++ = UTF8_ACCUMULATE(c, *s++);
644 =for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
646 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
647 Returns a pointer to the newly-created string, and sets C<len> to
648 reflect the new length.
654 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
661 Newz(801, d, (*len) * 2 + 1, U8);
665 if (UTF8_IS_ASCII(*s))
670 *d++ = UTF8_EIGHT_BIT_HI(uv);
671 *d++ = UTF8_EIGHT_BIT_LO(uv);
680 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
682 * Destination must be pre-extended to 3/2 source. Do not use in-place.
683 * We optimize for native, for obvious reasons. */
686 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
692 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
697 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
704 *d++ = (( uv >> 6) | 0xc0);
705 *d++ = (( uv & 0x3f) | 0x80);
708 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
710 if (low < 0xdc00 || low >= 0xdfff)
711 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
712 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
715 *d++ = (( uv >> 12) | 0xe0);
716 *d++ = (((uv >> 6) & 0x3f) | 0x80);
717 *d++ = (( uv & 0x3f) | 0x80);
721 *d++ = (( uv >> 18) | 0xf0);
722 *d++ = (((uv >> 12) & 0x3f) | 0x80);
723 *d++ = (((uv >> 6) & 0x3f) | 0x80);
724 *d++ = (( uv & 0x3f) | 0x80);
728 *newlen = d - dstart;
732 /* Note: this one is slightly destructive of the source. */
735 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
738 U8* send = s + bytelen;
745 return utf16_to_utf8(p, d, bytelen, newlen);
748 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
751 Perl_is_uni_alnum(pTHX_ U32 c)
753 U8 tmpbuf[UTF8_MAXLEN+1];
754 uv_to_utf8(tmpbuf, (UV)c);
755 return is_utf8_alnum(tmpbuf);
759 Perl_is_uni_alnumc(pTHX_ U32 c)
761 U8 tmpbuf[UTF8_MAXLEN+1];
762 uv_to_utf8(tmpbuf, (UV)c);
763 return is_utf8_alnumc(tmpbuf);
767 Perl_is_uni_idfirst(pTHX_ U32 c)
769 U8 tmpbuf[UTF8_MAXLEN+1];
770 uv_to_utf8(tmpbuf, (UV)c);
771 return is_utf8_idfirst(tmpbuf);
775 Perl_is_uni_alpha(pTHX_ U32 c)
777 U8 tmpbuf[UTF8_MAXLEN+1];
778 uv_to_utf8(tmpbuf, (UV)c);
779 return is_utf8_alpha(tmpbuf);
783 Perl_is_uni_ascii(pTHX_ U32 c)
785 U8 tmpbuf[UTF8_MAXLEN+1];
786 uv_to_utf8(tmpbuf, (UV)c);
787 return is_utf8_ascii(tmpbuf);
791 Perl_is_uni_space(pTHX_ U32 c)
793 U8 tmpbuf[UTF8_MAXLEN+1];
794 uv_to_utf8(tmpbuf, (UV)c);
795 return is_utf8_space(tmpbuf);
799 Perl_is_uni_digit(pTHX_ U32 c)
801 U8 tmpbuf[UTF8_MAXLEN+1];
802 uv_to_utf8(tmpbuf, (UV)c);
803 return is_utf8_digit(tmpbuf);
807 Perl_is_uni_upper(pTHX_ U32 c)
809 U8 tmpbuf[UTF8_MAXLEN+1];
810 uv_to_utf8(tmpbuf, (UV)c);
811 return is_utf8_upper(tmpbuf);
815 Perl_is_uni_lower(pTHX_ U32 c)
817 U8 tmpbuf[UTF8_MAXLEN+1];
818 uv_to_utf8(tmpbuf, (UV)c);
819 return is_utf8_lower(tmpbuf);
823 Perl_is_uni_cntrl(pTHX_ U32 c)
825 U8 tmpbuf[UTF8_MAXLEN+1];
826 uv_to_utf8(tmpbuf, (UV)c);
827 return is_utf8_cntrl(tmpbuf);
831 Perl_is_uni_graph(pTHX_ U32 c)
833 U8 tmpbuf[UTF8_MAXLEN+1];
834 uv_to_utf8(tmpbuf, (UV)c);
835 return is_utf8_graph(tmpbuf);
839 Perl_is_uni_print(pTHX_ U32 c)
841 U8 tmpbuf[UTF8_MAXLEN+1];
842 uv_to_utf8(tmpbuf, (UV)c);
843 return is_utf8_print(tmpbuf);
847 Perl_is_uni_punct(pTHX_ U32 c)
849 U8 tmpbuf[UTF8_MAXLEN+1];
850 uv_to_utf8(tmpbuf, (UV)c);
851 return is_utf8_punct(tmpbuf);
855 Perl_is_uni_xdigit(pTHX_ U32 c)
857 U8 tmpbuf[UTF8_MAXLEN+1];
858 uv_to_utf8(tmpbuf, (UV)c);
859 return is_utf8_xdigit(tmpbuf);
863 Perl_to_uni_upper(pTHX_ U32 c)
865 U8 tmpbuf[UTF8_MAXLEN+1];
866 uv_to_utf8(tmpbuf, (UV)c);
867 return to_utf8_upper(tmpbuf);
871 Perl_to_uni_title(pTHX_ U32 c)
873 U8 tmpbuf[UTF8_MAXLEN+1];
874 uv_to_utf8(tmpbuf, (UV)c);
875 return to_utf8_title(tmpbuf);
879 Perl_to_uni_lower(pTHX_ U32 c)
881 U8 tmpbuf[UTF8_MAXLEN+1];
882 uv_to_utf8(tmpbuf, (UV)c);
883 return to_utf8_lower(tmpbuf);
886 /* for now these all assume no locale info available for Unicode > 255 */
889 Perl_is_uni_alnum_lc(pTHX_ U32 c)
891 return is_uni_alnum(c); /* XXX no locale support yet */
895 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
897 return is_uni_alnumc(c); /* XXX no locale support yet */
901 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
903 return is_uni_idfirst(c); /* XXX no locale support yet */
907 Perl_is_uni_alpha_lc(pTHX_ U32 c)
909 return is_uni_alpha(c); /* XXX no locale support yet */
913 Perl_is_uni_ascii_lc(pTHX_ U32 c)
915 return is_uni_ascii(c); /* XXX no locale support yet */
919 Perl_is_uni_space_lc(pTHX_ U32 c)
921 return is_uni_space(c); /* XXX no locale support yet */
925 Perl_is_uni_digit_lc(pTHX_ U32 c)
927 return is_uni_digit(c); /* XXX no locale support yet */
931 Perl_is_uni_upper_lc(pTHX_ U32 c)
933 return is_uni_upper(c); /* XXX no locale support yet */
937 Perl_is_uni_lower_lc(pTHX_ U32 c)
939 return is_uni_lower(c); /* XXX no locale support yet */
943 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
945 return is_uni_cntrl(c); /* XXX no locale support yet */
949 Perl_is_uni_graph_lc(pTHX_ U32 c)
951 return is_uni_graph(c); /* XXX no locale support yet */
955 Perl_is_uni_print_lc(pTHX_ U32 c)
957 return is_uni_print(c); /* XXX no locale support yet */
961 Perl_is_uni_punct_lc(pTHX_ U32 c)
963 return is_uni_punct(c); /* XXX no locale support yet */
967 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
969 return is_uni_xdigit(c); /* XXX no locale support yet */
973 Perl_to_uni_upper_lc(pTHX_ U32 c)
975 return to_uni_upper(c); /* XXX no locale support yet */
979 Perl_to_uni_title_lc(pTHX_ U32 c)
981 return to_uni_title(c); /* XXX no locale support yet */
985 Perl_to_uni_lower_lc(pTHX_ U32 c)
987 return to_uni_lower(c); /* XXX no locale support yet */
991 Perl_is_utf8_alnum(pTHX_ U8 *p)
993 if (!is_utf8_char(p))
996 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
997 * descendant of isalnum(3), in other words, it doesn't
998 * contain the '_'. --jhi */
999 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1000 return swash_fetch(PL_utf8_alnum, p);
1001 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1002 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1004 PL_utf8_alnum = swash_init("utf8", "",
1005 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1006 return swash_fetch(PL_utf8_alnum, p);
1011 Perl_is_utf8_alnumc(pTHX_ U8 *p)
1013 if (!is_utf8_char(p))
1016 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1017 return swash_fetch(PL_utf8_alnum, p);
1018 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
1019 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1021 PL_utf8_alnum = swash_init("utf8", "",
1022 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1023 return swash_fetch(PL_utf8_alnum, p);
1028 Perl_is_utf8_idfirst(pTHX_ U8 *p)
1030 return *p == '_' || is_utf8_alpha(p);
1034 Perl_is_utf8_alpha(pTHX_ U8 *p)
1036 if (!is_utf8_char(p))
1039 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1040 return swash_fetch(PL_utf8_alpha, p);
1044 Perl_is_utf8_ascii(pTHX_ U8 *p)
1046 if (!is_utf8_char(p))
1049 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1050 return swash_fetch(PL_utf8_ascii, p);
1054 Perl_is_utf8_space(pTHX_ U8 *p)
1056 if (!is_utf8_char(p))
1059 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1060 return swash_fetch(PL_utf8_space, p);
1064 Perl_is_utf8_digit(pTHX_ U8 *p)
1066 if (!is_utf8_char(p))
1069 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1070 return swash_fetch(PL_utf8_digit, p);
1074 Perl_is_utf8_upper(pTHX_ U8 *p)
1076 if (!is_utf8_char(p))
1079 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
1080 return swash_fetch(PL_utf8_upper, p);
1084 Perl_is_utf8_lower(pTHX_ U8 *p)
1086 if (!is_utf8_char(p))
1089 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
1090 return swash_fetch(PL_utf8_lower, p);
1094 Perl_is_utf8_cntrl(pTHX_ U8 *p)
1096 if (!is_utf8_char(p))
1099 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1100 return swash_fetch(PL_utf8_cntrl, p);
1104 Perl_is_utf8_graph(pTHX_ U8 *p)
1106 if (!is_utf8_char(p))
1109 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1110 return swash_fetch(PL_utf8_graph, p);
1114 Perl_is_utf8_print(pTHX_ U8 *p)
1116 if (!is_utf8_char(p))
1119 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1120 return swash_fetch(PL_utf8_print, p);
1124 Perl_is_utf8_punct(pTHX_ U8 *p)
1126 if (!is_utf8_char(p))
1129 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1130 return swash_fetch(PL_utf8_punct, p);
1134 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1136 if (!is_utf8_char(p))
1138 if (!PL_utf8_xdigit)
1139 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1140 return swash_fetch(PL_utf8_xdigit, p);
1144 Perl_is_utf8_mark(pTHX_ U8 *p)
1146 if (!is_utf8_char(p))
1149 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1150 return swash_fetch(PL_utf8_mark, p);
1154 Perl_to_utf8_upper(pTHX_ U8 *p)
1158 if (!PL_utf8_toupper)
1159 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
1160 uv = swash_fetch(PL_utf8_toupper, p);
1161 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1165 Perl_to_utf8_title(pTHX_ U8 *p)
1169 if (!PL_utf8_totitle)
1170 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
1171 uv = swash_fetch(PL_utf8_totitle, p);
1172 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1176 Perl_to_utf8_lower(pTHX_ U8 *p)
1180 if (!PL_utf8_tolower)
1181 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
1182 uv = swash_fetch(PL_utf8_tolower, p);
1183 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1186 /* a "swash" is a swatch hash */
1189 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1192 SV* tokenbufsv = sv_2mortal(NEWSV(0,0));
1195 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
1197 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1201 PUSHSTACKi(PERLSI_MAGIC);
1204 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1205 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1207 PUSHs(sv_2mortal(newSViv(minbits)));
1208 PUSHs(sv_2mortal(newSViv(none)));
1214 if (PL_curcop == &PL_compiling)
1215 /* XXX ought to be handled by lex_start */
1216 sv_setpv(tokenbufsv, PL_tokenbuf);
1217 if (call_method("SWASHNEW", G_SCALAR))
1218 retval = newSVsv(*PL_stack_sp--);
1220 retval = &PL_sv_undef;
1223 if (PL_curcop == &PL_compiling) {
1225 char* pv = SvPV(tokenbufsv, len);
1227 Copy(pv, PL_tokenbuf, len+1, char);
1228 PL_curcop->op_private = PL_hints;
1230 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1231 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1236 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1238 HV* hv = (HV*)SvRV(sv);
1239 U32 klen = UTF8SKIP(ptr) - 1;
1240 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1242 STRLEN needents = (klen ? 64 : 128);
1248 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1249 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1250 * it's nothing to sniff at.) Pity we usually come through at least
1251 * two function calls to get here...
1253 * NB: this code assumes that swatches are never modified, once generated!
1256 if (hv == PL_last_swash_hv &&
1257 klen == PL_last_swash_klen &&
1258 (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
1260 tmps = PL_last_swash_tmps;
1261 slen = PL_last_swash_slen;
1264 /* Try our second-level swatch cache, kept in a hash. */
1265 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1267 /* If not cached, generate it via utf8::SWASHGET */
1268 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1273 PUSHSTACKi(PERLSI_MAGIC);
1277 PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, UTF8_MAXLEN, 0, 0) & ~(needents - 1))));
1278 PUSHs(sv_2mortal(newSViv(needents)));
1280 if (call_method("SWASHGET", G_SCALAR))
1281 retval = newSVsv(*PL_stack_sp--);
1283 retval = &PL_sv_undef;
1287 if (PL_curcop == &PL_compiling)
1288 PL_curcop->op_private = PL_hints;
1290 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1292 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1293 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1296 PL_last_swash_hv = hv;
1297 PL_last_swash_klen = klen;
1298 PL_last_swash_tmps = tmps;
1299 PL_last_swash_slen = slen;
1301 Copy(ptr, PL_last_swash_key, klen, U8);
1304 switch ((int)((slen << 3) / needents)) {
1306 bit = 1 << (off & 7);
1308 return (tmps[off] & bit) != 0;
1313 return (tmps[off] << 8) + tmps[off + 1] ;
1316 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1318 Perl_croak(aTHX_ "panic: swash_fetch");