3 * Copyright (c) 1998-2001, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * 'What a fix!' said Sam. 'That's the one place in all the lands we've ever
12 * heard of that we don't want to see any closer; and that's the one place
13 * we're trying to get to! And that's just where we can't get, nohow.'
15 * 'Well do I understand your speech,' he answered in the same language;
16 * 'yet few strangers do so. Why then do you not speak in the Common Tongue,
17 * as is the custom in the West, if you wish to be answered?'
19 * ...the travellers perceived that the floor was paved with stones of many
20 * hues; branching runes and strange devices intertwined beneath their feet.
24 #define PERL_IN_UTF8_C
30 Perl_uv_to_utf8(pTHX_ U8 *d, UV uv) /* the d must be UTF8_MAXLEN+1 deep */
38 *d++ = (( uv >> 6) | 0xc0);
39 *d++ = (( uv & 0x3f) | 0x80);
44 *d++ = (( uv >> 12) | 0xe0);
45 *d++ = (((uv >> 6) & 0x3f) | 0x80);
46 *d++ = (( uv & 0x3f) | 0x80);
51 *d++ = (( uv >> 18) | 0xf0);
52 *d++ = (((uv >> 12) & 0x3f) | 0x80);
53 *d++ = (((uv >> 6) & 0x3f) | 0x80);
54 *d++ = (( uv & 0x3f) | 0x80);
59 *d++ = (( uv >> 24) | 0xf8);
60 *d++ = (((uv >> 18) & 0x3f) | 0x80);
61 *d++ = (((uv >> 12) & 0x3f) | 0x80);
62 *d++ = (((uv >> 6) & 0x3f) | 0x80);
63 *d++ = (( uv & 0x3f) | 0x80);
67 if (uv < 0x80000000) {
68 *d++ = (( uv >> 30) | 0xfc);
69 *d++ = (((uv >> 24) & 0x3f) | 0x80);
70 *d++ = (((uv >> 18) & 0x3f) | 0x80);
71 *d++ = (((uv >> 12) & 0x3f) | 0x80);
72 *d++ = (((uv >> 6) & 0x3f) | 0x80);
73 *d++ = (( uv & 0x3f) | 0x80);
78 if (uv < UTF8_QUAD_MAX)
81 *d++ = 0xfe; /* Can't match U+FEFF! */
82 *d++ = (((uv >> 30) & 0x3f) | 0x80);
83 *d++ = (((uv >> 24) & 0x3f) | 0x80);
84 *d++ = (((uv >> 18) & 0x3f) | 0x80);
85 *d++ = (((uv >> 12) & 0x3f) | 0x80);
86 *d++ = (((uv >> 6) & 0x3f) | 0x80);
87 *d++ = (( uv & 0x3f) | 0x80);
93 *d++ = 0xff; /* Can't match U+FFFE! */
94 *d++ = 0x80; /* 6 Reserved bits */
95 *d++ = (((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
96 *d++ = (((uv >> 54) & 0x3f) | 0x80);
97 *d++ = (((uv >> 48) & 0x3f) | 0x80);
98 *d++ = (((uv >> 42) & 0x3f) | 0x80);
99 *d++ = (((uv >> 36) & 0x3f) | 0x80);
100 *d++ = (((uv >> 30) & 0x3f) | 0x80);
101 *d++ = (((uv >> 24) & 0x3f) | 0x80);
102 *d++ = (((uv >> 18) & 0x3f) | 0x80);
103 *d++ = (((uv >> 12) & 0x3f) | 0x80);
104 *d++ = (((uv >> 6) & 0x3f) | 0x80);
105 *d++ = (( uv & 0x3f) | 0x80);
112 /* Tests if some arbitrary number of bytes begins in a valid UTF-8 character.
113 * The actual number of bytes in the UTF-8 character will be returned if it
114 * is valid, otherwise 0. */
116 Perl_is_utf8_char(pTHX_ U8 *s)
122 if (UTF8_IS_ASCII(u))
125 if (!UTF8_IS_START(u))
130 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
138 if (!UTF8_IS_CONTINUATION(*s))
140 uv = UTF8_ACCUMULATE(uv, *s);
147 if (UNISKIP(uv) < len)
154 =for apidoc Am|is_utf8_string|U8 *s|STRLEN len
156 Returns true if first C<len> bytes of the given string form valid a UTF8
157 string, false otherwise.
163 Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len)
170 len = strlen((char *)s);
186 =for apidoc Am|U8* s|utf8_to_uv|STRLEN curlen|STRLEN *retlen|U32 flags
188 Returns the character value of the first character in the string C<s>
189 which is assumed to be in UTF8 encoding and no longer than C<curlen>;
190 C<retlen> will be set to the length, in bytes, of that character.
192 If C<s> does not point to a well-formed UTF8 character, the behaviour
193 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
194 it is assumed that the caller will raise a warning, and this function
195 will silently just set C<retlen> to C<-1> and return zero. If the
196 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
197 malformations will be given, C<retlen> will be set to the expected
198 length of the UTF-8 character in bytes, and zero will be returned.
200 The C<flags> can also contain various flags to allow deviations from
201 the strict UTF-8 encoding (see F<utf8.h>).
206 Perl_utf8_to_uv(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags)
213 bool dowarn = ckWARN_d(WARN_UTF8);
215 STRLEN expectlen = 0;
218 /* This list is a superset of the UTF8_ALLOW_XXX. */
220 #define UTF8_WARN_EMPTY 1
221 #define UTF8_WARN_CONTINUATION 2
222 #define UTF8_WARN_NON_CONTINUATION 3
223 #define UTF8_WARN_FE_FF 4
224 #define UTF8_WARN_SHORT 5
225 #define UTF8_WARN_OVERFLOW 6
226 #define UTF8_WARN_SURROGATE 7
227 #define UTF8_WARN_BOM 8
228 #define UTF8_WARN_LONG 9
229 #define UTF8_WARN_FFFF 10
232 !(flags & UTF8_ALLOW_EMPTY)) {
233 warning = UTF8_WARN_EMPTY;
237 if (UTF8_IS_ASCII(uv)) {
243 if (UTF8_IS_CONTINUATION(uv) &&
244 !(flags & UTF8_ALLOW_CONTINUATION)) {
245 warning = UTF8_WARN_CONTINUATION;
249 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
250 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
251 warning = UTF8_WARN_NON_CONTINUATION;
255 if ((uv == 0xfe || uv == 0xff) &&
256 !(flags & UTF8_ALLOW_FE_FF)) {
257 warning = UTF8_WARN_FE_FF;
261 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
262 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
263 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
264 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
265 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
266 else if (!(uv & 0x01)) { len = 7; uv = 0; }
267 else { len = 13; uv = 0; } /* whoa! */
274 if ((curlen < expectlen) &&
275 !(flags & UTF8_ALLOW_SHORT)) {
276 warning = UTF8_WARN_SHORT;
285 if (!UTF8_IS_CONTINUATION(*s) &&
286 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
288 warning = UTF8_WARN_NON_CONTINUATION;
292 uv = UTF8_ACCUMULATE(uv, *s);
294 /* These cannot be allowed. */
296 if (!(flags & UTF8_ALLOW_LONG)) {
297 warning = UTF8_WARN_LONG;
301 else { /* uv < ouv */
302 /* This cannot be allowed. */
303 warning = UTF8_WARN_OVERFLOW;
311 if (UNICODE_IS_SURROGATE(uv) &&
312 !(flags & UTF8_ALLOW_SURROGATE)) {
313 warning = UTF8_WARN_SURROGATE;
315 } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) &&
316 !(flags & UTF8_ALLOW_BOM)) {
317 warning = UTF8_WARN_BOM;
319 } else if ((expectlen > UNISKIP(uv)) &&
320 !(flags & UTF8_ALLOW_LONG)) {
321 warning = UTF8_WARN_LONG;
323 } else if (UNICODE_IS_ILLEGAL(uv) &&
324 !(flags & UTF8_ALLOW_FFFF)) {
325 warning = UTF8_WARN_FFFF;
333 if (flags & UTF8_CHECK_ONLY) {
340 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
343 case 0: /* Intentionally empty. */ break;
344 case UTF8_WARN_EMPTY:
345 Perl_sv_catpvf(aTHX_ sv, "(empty string)");
347 case UTF8_WARN_CONTINUATION:
348 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf")", uv);
350 case UTF8_WARN_NON_CONTINUATION:
351 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")",
354 case UTF8_WARN_FE_FF:
355 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
357 case UTF8_WARN_SHORT:
358 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
359 curlen, curlen == 1 ? "" : "s", expectlen);
361 case UTF8_WARN_OVERFLOW:
362 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x)",
365 case UTF8_WARN_SURROGATE:
366 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
369 Perl_sv_catpvf(aTHX_ sv, "(byte order mark 0x%04"UVxf")", uv);
372 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
373 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
376 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
379 Perl_sv_catpvf(aTHX_ sv, "(unknown reason)");
387 Perl_warner(aTHX_ WARN_UTF8,
388 "%s in %s", s, PL_op_desc[PL_op->op_type]);
390 Perl_warner(aTHX_ WARN_UTF8, "%s", s);
395 *retlen = expectlen ? expectlen : len;
401 =for apidoc Am|U8* s|utf8_to_uv_simple|STRLEN *retlen
403 Returns the character value of the first character in the string C<s>
404 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
405 length, in bytes, of that character.
407 If C<s> does not point to a well-formed UTF8 character, zero is
408 returned and retlen is set, if possible, to -1.
414 Perl_utf8_to_uv_simple(pTHX_ U8* s, STRLEN* retlen)
416 return Perl_utf8_to_uv(aTHX_ s, UTF8_MAXLEN, retlen, 0);
420 =for apidoc Am|STRLEN|utf8_length|U8* s|U8 *e
422 Return the length of the UTF-8 char encoded string C<s> in characters.
423 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
424 up past C<e>, croaks.
430 Perl_utf8_length(pTHX_ U8* s, U8* e)
434 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
435 * the bitops (especially ~) can create illegal UTF-8.
436 * In other words: in Perl UTF-8 is not just for Unicode. */
439 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
444 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
453 =for apidoc Am|IV|utf8_distance|U8 *a|U8 *b
455 Returns the number of UTF8 characters between the UTF-8 pointers C<a>
458 WARNING: use only if you *know* that the pointers point inside the
464 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
468 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
469 * the bitops (especially ~) can create illegal UTF-8.
470 * In other words: in Perl UTF-8 is not just for Unicode. */
477 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
487 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
497 =for apidoc Am|U8*|utf8_hop|U8 *s|I32 off
499 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
502 WARNING: do not use the following unless you *know* C<off> is within
503 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
504 on the first byte of character or just after the last byte of a character.
509 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
511 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
512 * the bitops (especially ~) can create illegal UTF-8.
513 * In other words: in Perl UTF-8 is not just for Unicode. */
522 while (UTF8_IS_CONTINUATION(*s))
530 =for apidoc Am|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
532 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
533 Unlike C<bytes_to_utf8>, this over-writes the original string, and
534 updates len to contain the new length.
535 Returns zero on failure, setting C<len> to -1.
541 Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len)
547 /* ensure valid UTF8 and chars < 256 before updating string */
548 for (send = s + *len; s < send; ) {
553 ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) {
562 *d++ = (U8)utf8_to_uv_simple(s, &ulen);
571 =for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
573 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
574 Returns a pointer to the newly-created string, and sets C<len> to
575 reflect the new length.
581 Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
588 Newz(801, d, (*len) * 2 + 1, U8);
596 *d++ = (( uv >> 6) | 0xc0);
597 *d++ = (( uv & 0x3f) | 0x80);
606 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
608 * Destination must be pre-extended to 3/2 source. Do not use in-place.
609 * We optimize for native, for obvious reasons. */
612 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
618 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
623 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
630 *d++ = (( uv >> 6) | 0xc0);
631 *d++ = (( uv & 0x3f) | 0x80);
634 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
636 if (low < 0xdc00 || low >= 0xdfff)
637 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
638 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
641 *d++ = (( uv >> 12) | 0xe0);
642 *d++ = (((uv >> 6) & 0x3f) | 0x80);
643 *d++ = (( uv & 0x3f) | 0x80);
647 *d++ = (( uv >> 18) | 0xf0);
648 *d++ = (((uv >> 12) & 0x3f) | 0x80);
649 *d++ = (((uv >> 6) & 0x3f) | 0x80);
650 *d++ = (( uv & 0x3f) | 0x80);
654 *newlen = d - dstart;
658 /* Note: this one is slightly destructive of the source. */
661 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
664 U8* send = s + bytelen;
671 return utf16_to_utf8(p, d, bytelen, newlen);
674 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
677 Perl_is_uni_alnum(pTHX_ U32 c)
679 U8 tmpbuf[UTF8_MAXLEN+1];
680 uv_to_utf8(tmpbuf, (UV)c);
681 return is_utf8_alnum(tmpbuf);
685 Perl_is_uni_alnumc(pTHX_ U32 c)
687 U8 tmpbuf[UTF8_MAXLEN+1];
688 uv_to_utf8(tmpbuf, (UV)c);
689 return is_utf8_alnumc(tmpbuf);
693 Perl_is_uni_idfirst(pTHX_ U32 c)
695 U8 tmpbuf[UTF8_MAXLEN+1];
696 uv_to_utf8(tmpbuf, (UV)c);
697 return is_utf8_idfirst(tmpbuf);
701 Perl_is_uni_alpha(pTHX_ U32 c)
703 U8 tmpbuf[UTF8_MAXLEN+1];
704 uv_to_utf8(tmpbuf, (UV)c);
705 return is_utf8_alpha(tmpbuf);
709 Perl_is_uni_ascii(pTHX_ U32 c)
711 U8 tmpbuf[UTF8_MAXLEN+1];
712 uv_to_utf8(tmpbuf, (UV)c);
713 return is_utf8_ascii(tmpbuf);
717 Perl_is_uni_space(pTHX_ U32 c)
719 U8 tmpbuf[UTF8_MAXLEN+1];
720 uv_to_utf8(tmpbuf, (UV)c);
721 return is_utf8_space(tmpbuf);
725 Perl_is_uni_digit(pTHX_ U32 c)
727 U8 tmpbuf[UTF8_MAXLEN+1];
728 uv_to_utf8(tmpbuf, (UV)c);
729 return is_utf8_digit(tmpbuf);
733 Perl_is_uni_upper(pTHX_ U32 c)
735 U8 tmpbuf[UTF8_MAXLEN+1];
736 uv_to_utf8(tmpbuf, (UV)c);
737 return is_utf8_upper(tmpbuf);
741 Perl_is_uni_lower(pTHX_ U32 c)
743 U8 tmpbuf[UTF8_MAXLEN+1];
744 uv_to_utf8(tmpbuf, (UV)c);
745 return is_utf8_lower(tmpbuf);
749 Perl_is_uni_cntrl(pTHX_ U32 c)
751 U8 tmpbuf[UTF8_MAXLEN+1];
752 uv_to_utf8(tmpbuf, (UV)c);
753 return is_utf8_cntrl(tmpbuf);
757 Perl_is_uni_graph(pTHX_ U32 c)
759 U8 tmpbuf[UTF8_MAXLEN+1];
760 uv_to_utf8(tmpbuf, (UV)c);
761 return is_utf8_graph(tmpbuf);
765 Perl_is_uni_print(pTHX_ U32 c)
767 U8 tmpbuf[UTF8_MAXLEN+1];
768 uv_to_utf8(tmpbuf, (UV)c);
769 return is_utf8_print(tmpbuf);
773 Perl_is_uni_punct(pTHX_ U32 c)
775 U8 tmpbuf[UTF8_MAXLEN+1];
776 uv_to_utf8(tmpbuf, (UV)c);
777 return is_utf8_punct(tmpbuf);
781 Perl_is_uni_xdigit(pTHX_ U32 c)
783 U8 tmpbuf[UTF8_MAXLEN+1];
784 uv_to_utf8(tmpbuf, (UV)c);
785 return is_utf8_xdigit(tmpbuf);
789 Perl_to_uni_upper(pTHX_ U32 c)
791 U8 tmpbuf[UTF8_MAXLEN+1];
792 uv_to_utf8(tmpbuf, (UV)c);
793 return to_utf8_upper(tmpbuf);
797 Perl_to_uni_title(pTHX_ U32 c)
799 U8 tmpbuf[UTF8_MAXLEN+1];
800 uv_to_utf8(tmpbuf, (UV)c);
801 return to_utf8_title(tmpbuf);
805 Perl_to_uni_lower(pTHX_ U32 c)
807 U8 tmpbuf[UTF8_MAXLEN+1];
808 uv_to_utf8(tmpbuf, (UV)c);
809 return to_utf8_lower(tmpbuf);
812 /* for now these all assume no locale info available for Unicode > 255 */
815 Perl_is_uni_alnum_lc(pTHX_ U32 c)
817 return is_uni_alnum(c); /* XXX no locale support yet */
821 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
823 return is_uni_alnumc(c); /* XXX no locale support yet */
827 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
829 return is_uni_idfirst(c); /* XXX no locale support yet */
833 Perl_is_uni_alpha_lc(pTHX_ U32 c)
835 return is_uni_alpha(c); /* XXX no locale support yet */
839 Perl_is_uni_ascii_lc(pTHX_ U32 c)
841 return is_uni_ascii(c); /* XXX no locale support yet */
845 Perl_is_uni_space_lc(pTHX_ U32 c)
847 return is_uni_space(c); /* XXX no locale support yet */
851 Perl_is_uni_digit_lc(pTHX_ U32 c)
853 return is_uni_digit(c); /* XXX no locale support yet */
857 Perl_is_uni_upper_lc(pTHX_ U32 c)
859 return is_uni_upper(c); /* XXX no locale support yet */
863 Perl_is_uni_lower_lc(pTHX_ U32 c)
865 return is_uni_lower(c); /* XXX no locale support yet */
869 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
871 return is_uni_cntrl(c); /* XXX no locale support yet */
875 Perl_is_uni_graph_lc(pTHX_ U32 c)
877 return is_uni_graph(c); /* XXX no locale support yet */
881 Perl_is_uni_print_lc(pTHX_ U32 c)
883 return is_uni_print(c); /* XXX no locale support yet */
887 Perl_is_uni_punct_lc(pTHX_ U32 c)
889 return is_uni_punct(c); /* XXX no locale support yet */
893 Perl_is_uni_xdigit_lc(pTHX_ U32 c)
895 return is_uni_xdigit(c); /* XXX no locale support yet */
899 Perl_to_uni_upper_lc(pTHX_ U32 c)
901 return to_uni_upper(c); /* XXX no locale support yet */
905 Perl_to_uni_title_lc(pTHX_ U32 c)
907 return to_uni_title(c); /* XXX no locale support yet */
911 Perl_to_uni_lower_lc(pTHX_ U32 c)
913 return to_uni_lower(c); /* XXX no locale support yet */
917 Perl_is_utf8_alnum(pTHX_ U8 *p)
919 if (!is_utf8_char(p))
922 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
923 * descendant of isalnum(3), in other words, it doesn't
924 * contain the '_'. --jhi */
925 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
926 return swash_fetch(PL_utf8_alnum, p);
927 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
928 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
930 PL_utf8_alnum = swash_init("utf8", "",
931 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
932 return swash_fetch(PL_utf8_alnum, p);
937 Perl_is_utf8_alnumc(pTHX_ U8 *p)
939 if (!is_utf8_char(p))
942 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
943 return swash_fetch(PL_utf8_alnum, p);
944 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
945 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
947 PL_utf8_alnum = swash_init("utf8", "",
948 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
949 return swash_fetch(PL_utf8_alnum, p);
954 Perl_is_utf8_idfirst(pTHX_ U8 *p)
956 return *p == '_' || is_utf8_alpha(p);
960 Perl_is_utf8_alpha(pTHX_ U8 *p)
962 if (!is_utf8_char(p))
965 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
966 return swash_fetch(PL_utf8_alpha, p);
970 Perl_is_utf8_ascii(pTHX_ U8 *p)
972 if (!is_utf8_char(p))
975 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
976 return swash_fetch(PL_utf8_ascii, p);
980 Perl_is_utf8_space(pTHX_ U8 *p)
982 if (!is_utf8_char(p))
985 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
986 return swash_fetch(PL_utf8_space, p);
990 Perl_is_utf8_digit(pTHX_ U8 *p)
992 if (!is_utf8_char(p))
995 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
996 return swash_fetch(PL_utf8_digit, p);
1000 Perl_is_utf8_upper(pTHX_ U8 *p)
1002 if (!is_utf8_char(p))
1005 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
1006 return swash_fetch(PL_utf8_upper, p);
1010 Perl_is_utf8_lower(pTHX_ U8 *p)
1012 if (!is_utf8_char(p))
1015 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
1016 return swash_fetch(PL_utf8_lower, p);
1020 Perl_is_utf8_cntrl(pTHX_ U8 *p)
1022 if (!is_utf8_char(p))
1025 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1026 return swash_fetch(PL_utf8_cntrl, p);
1030 Perl_is_utf8_graph(pTHX_ U8 *p)
1032 if (!is_utf8_char(p))
1035 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1036 return swash_fetch(PL_utf8_graph, p);
1040 Perl_is_utf8_print(pTHX_ U8 *p)
1042 if (!is_utf8_char(p))
1045 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1046 return swash_fetch(PL_utf8_print, p);
1050 Perl_is_utf8_punct(pTHX_ U8 *p)
1052 if (!is_utf8_char(p))
1055 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1056 return swash_fetch(PL_utf8_punct, p);
1060 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1062 if (!is_utf8_char(p))
1064 if (!PL_utf8_xdigit)
1065 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1066 return swash_fetch(PL_utf8_xdigit, p);
1070 Perl_is_utf8_mark(pTHX_ U8 *p)
1072 if (!is_utf8_char(p))
1075 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1076 return swash_fetch(PL_utf8_mark, p);
1080 Perl_to_utf8_upper(pTHX_ U8 *p)
1084 if (!PL_utf8_toupper)
1085 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
1086 uv = swash_fetch(PL_utf8_toupper, p);
1087 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1091 Perl_to_utf8_title(pTHX_ U8 *p)
1095 if (!PL_utf8_totitle)
1096 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
1097 uv = swash_fetch(PL_utf8_totitle, p);
1098 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1102 Perl_to_utf8_lower(pTHX_ U8 *p)
1106 if (!PL_utf8_tolower)
1107 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
1108 uv = swash_fetch(PL_utf8_tolower, p);
1109 return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0);
1112 /* a "swash" is a swatch hash */
1115 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1121 if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */
1123 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1127 PUSHSTACKi(PERLSI_MAGIC);
1130 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1131 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1133 PUSHs(sv_2mortal(newSViv(minbits)));
1134 PUSHs(sv_2mortal(newSViv(none)));
1140 if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */
1141 strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf);
1142 if (call_method("SWASHNEW", G_SCALAR))
1143 retval = newSVsv(*PL_stack_sp--);
1145 retval = &PL_sv_undef;
1148 if (PL_curcop == &PL_compiling) {
1149 strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf);
1150 PL_curcop->op_private = PL_hints;
1152 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1153 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1158 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
1160 HV* hv = (HV*)SvRV(sv);
1161 U32 klen = UTF8SKIP(ptr) - 1;
1162 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
1164 STRLEN needents = (klen ? 64 : 128);
1170 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1171 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1172 * it's nothing to sniff at.) Pity we usually come through at least
1173 * two function calls to get here...
1175 * NB: this code assumes that swatches are never modified, once generated!
1178 if (hv == PL_last_swash_hv &&
1179 klen == PL_last_swash_klen &&
1180 (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
1182 tmps = PL_last_swash_tmps;
1183 slen = PL_last_swash_slen;
1186 /* Try our second-level swatch cache, kept in a hash. */
1187 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1189 /* If not cached, generate it via utf8::SWASHGET */
1190 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1195 PUSHSTACKi(PERLSI_MAGIC);
1199 PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, UTF8_MAXLEN, 0, 0) & ~(needents - 1))));
1200 PUSHs(sv_2mortal(newSViv(needents)));
1202 if (call_method("SWASHGET", G_SCALAR))
1203 retval = newSVsv(*PL_stack_sp--);
1205 retval = &PL_sv_undef;
1209 if (PL_curcop == &PL_compiling)
1210 PL_curcop->op_private = PL_hints;
1212 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1214 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
1215 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1218 PL_last_swash_hv = hv;
1219 PL_last_swash_klen = klen;
1220 PL_last_swash_tmps = tmps;
1221 PL_last_swash_slen = slen;
1223 Copy(ptr, PL_last_swash_key, klen, U8);
1226 switch ((int)((slen << 3) / needents)) {
1228 bit = 1 << (off & 7);
1230 return (tmps[off] & bit) != 0;
1235 return (tmps[off] << 8) + tmps[off + 1] ;
1238 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1240 Perl_croak(aTHX_ "panic: swash_fetch");