3 * Copyright (c) 1998-1999, 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)
37 *d++ = (( uv >> 6) | 0xc0);
38 *d++ = (( uv & 0x3f) | 0x80);
42 *d++ = (( uv >> 12) | 0xe0);
43 *d++ = (((uv >> 6) & 0x3f) | 0x80);
44 *d++ = (( uv & 0x3f) | 0x80);
48 *d++ = (( uv >> 18) | 0xf0);
49 *d++ = (((uv >> 12) & 0x3f) | 0x80);
50 *d++ = (((uv >> 6) & 0x3f) | 0x80);
51 *d++ = (( uv & 0x3f) | 0x80);
55 *d++ = (( uv >> 24) | 0xf8);
56 *d++ = (((uv >> 18) & 0x3f) | 0x80);
57 *d++ = (((uv >> 12) & 0x3f) | 0x80);
58 *d++ = (((uv >> 6) & 0x3f) | 0x80);
59 *d++ = (( uv & 0x3f) | 0x80);
62 if (uv < 0x80000000) {
63 *d++ = (( uv >> 30) | 0xfc);
64 *d++ = (((uv >> 24) & 0x3f) | 0x80);
65 *d++ = (((uv >> 18) & 0x3f) | 0x80);
66 *d++ = (((uv >> 12) & 0x3f) | 0x80);
67 *d++ = (((uv >> 6) & 0x3f) | 0x80);
68 *d++ = (( uv & 0x3f) | 0x80);
72 if (uv < 0x2000000000)
75 *d++ = 0xfe; /* Can't match U+FEFF! */
76 *d++ = (((uv >> 30) & 0x3f) | 0x80);
77 *d++ = (((uv >> 24) & 0x3f) | 0x80);
78 *d++ = (((uv >> 18) & 0x3f) | 0x80);
79 *d++ = (((uv >> 12) & 0x3f) | 0x80);
80 *d++ = (((uv >> 6) & 0x3f) | 0x80);
81 *d++ = (( uv & 0x3f) | 0x80);
86 *d++ = 0xff; /* Can't match U+FFFE! */
87 *d++ = (((uv >> 36) & 0x3f) | 0x80);
88 *d++ = (((uv >> 30) & 0x3f) | 0x80);
89 *d++ = (((uv >> 24) & 0x3f) | 0x80);
90 *d++ = (((uv >> 18) & 0x3f) | 0x80);
91 *d++ = (((uv >> 12) & 0x3f) | 0x80);
92 *d++ = (((uv >> 6) & 0x3f) | 0x80);
93 *d++ = (( uv & 0x3f) | 0x80);
100 Perl_utf8_to_uv(pTHX_ U8* s, I32* retlen)
110 Perl_warn(aTHX_ "Malformed UTF-8 character");
116 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
117 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
118 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
119 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
120 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
121 else if (!(uv & 0x01)) { len = 7; uv &= 0x00; }
122 else len = 8; /* whoa! */
129 if ((*s & 0xc0) != 0x80) {
130 Perl_warn(aTHX_ "Malformed UTF-8 character");
136 uv = (uv << 6) | (*s++ & 0x3f);
141 /* utf8_distance(a,b) is intended to be a - b in pointer arithmetic */
144 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
162 /* WARNING: do not use the following unless you *know* off is within bounds */
165 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
175 while ((*s & 0xc0) == 0x80)
183 /* XXX NOTHING CALLS THE FOLLOWING TWO ROUTINES YET!!! */
185 * Convert native or reversed UTF-16 to UTF-8.
187 * Destination must be pre-extended to 3/2 source. Do not use in-place.
188 * We optimize for native, for obvious reasons. */
191 Perl_utf16_to_utf8(pTHX_ U16* p, U8* d, I32 bytelen)
193 U16* pend = p + bytelen / 2;
201 *d++ = (( uv >> 6) | 0xc0);
202 *d++ = (( uv & 0x3f) | 0x80);
205 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
207 if (low < 0xdc00 || low >= 0xdfff) {
208 Perl_warn(aTHX_ "Malformed UTF-16 surrogate");
212 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
215 *d++ = (( uv >> 12) | 0xe0);
216 *d++ = (((uv >> 6) & 0x3f) | 0x80);
217 *d++ = (( uv & 0x3f) | 0x80);
221 *d++ = (( uv >> 18) | 0xf0);
222 *d++ = (((uv >> 12) & 0x3f) | 0x80);
223 *d++ = (((uv >> 6) & 0x3f) | 0x80);
224 *d++ = (( uv & 0x3f) | 0x80);
231 /* Note: this one is slightly destructive of the source. */
234 Perl_utf16_to_utf8_reversed(pTHX_ U16* p, U8* d, I32 bytelen)
237 U8* send = s + bytelen;
244 return utf16_to_utf8(p, d, bytelen);
247 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
250 Perl_is_uni_alnum(pTHX_ U32 c)
253 uv_to_utf8(tmpbuf, (UV)c);
254 return is_utf8_alnum(tmpbuf);
258 Perl_is_uni_alnumc(pTHX_ U32 c)
261 uv_to_utf8(tmpbuf, (UV)c);
262 return is_utf8_alnumc(tmpbuf);
266 Perl_is_uni_idfirst(pTHX_ U32 c)
269 uv_to_utf8(tmpbuf, (UV)c);
270 return is_utf8_idfirst(tmpbuf);
274 Perl_is_uni_alpha(pTHX_ U32 c)
277 uv_to_utf8(tmpbuf, (UV)c);
278 return is_utf8_alpha(tmpbuf);
282 Perl_is_uni_space(pTHX_ U32 c)
285 uv_to_utf8(tmpbuf, (UV)c);
286 return is_utf8_space(tmpbuf);
290 Perl_is_uni_digit(pTHX_ U32 c)
293 uv_to_utf8(tmpbuf, (UV)c);
294 return is_utf8_digit(tmpbuf);
298 Perl_is_uni_upper(pTHX_ U32 c)
301 uv_to_utf8(tmpbuf, (UV)c);
302 return is_utf8_upper(tmpbuf);
306 Perl_is_uni_lower(pTHX_ U32 c)
309 uv_to_utf8(tmpbuf, (UV)c);
310 return is_utf8_lower(tmpbuf);
314 Perl_is_uni_cntrl(pTHX_ U32 c)
317 uv_to_utf8(tmpbuf, (UV)c);
318 return is_utf8_cntrl(tmpbuf);
322 Perl_is_uni_graph(pTHX_ U32 c)
325 uv_to_utf8(tmpbuf, (UV)c);
326 return is_utf8_graph(tmpbuf);
330 Perl_is_uni_print(pTHX_ U32 c)
333 uv_to_utf8(tmpbuf, (UV)c);
334 return is_utf8_print(tmpbuf);
341 uv_to_utf8(tmpbuf, (UV)c);
342 return is_utf8_punct(tmpbuf);
346 Perl_to_uni_upper(pTHX_ U32 c)
349 uv_to_utf8(tmpbuf, (UV)c);
350 return to_utf8_upper(tmpbuf);
354 Perl_to_uni_title(pTHX_ U32 c)
357 uv_to_utf8(tmpbuf, (UV)c);
358 return to_utf8_title(tmpbuf);
362 Perl_to_uni_lower(pTHX_ U32 c)
365 uv_to_utf8(tmpbuf, (UV)c);
366 return to_utf8_lower(tmpbuf);
369 /* for now these all assume no locale info available for Unicode > 255 */
372 Perl_is_uni_alnum_lc(pTHX_ U32 c)
374 return is_uni_alnum(c); /* XXX no locale support yet */
378 Perl_is_uni_alnumc_lc(pTHX_ U32 c)
380 return is_uni_alnumc(c); /* XXX no locale support yet */
384 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
386 return is_uni_idfirst(c); /* XXX no locale support yet */
390 Perl_is_uni_alpha_lc(pTHX_ U32 c)
392 return is_uni_alpha(c); /* XXX no locale support yet */
396 Perl_is_uni_space_lc(pTHX_ U32 c)
398 return is_uni_space(c); /* XXX no locale support yet */
402 Perl_is_uni_digit_lc(pTHX_ U32 c)
404 return is_uni_digit(c); /* XXX no locale support yet */
408 Perl_is_uni_upper_lc(pTHX_ U32 c)
410 return is_uni_upper(c); /* XXX no locale support yet */
414 Perl_is_uni_lower_lc(pTHX_ U32 c)
416 return is_uni_lower(c); /* XXX no locale support yet */
420 Perl_is_uni_cntrl_lc(pTHX_ U32 c)
422 return is_uni_cntrl(c); /* XXX no locale support yet */
426 Perl_is_uni_graph_lc(pTHX_ U32 c)
428 return is_uni_graph(c); /* XXX no locale support yet */
432 Perl_is_uni_print_lc(pTHX_ U32 c)
434 return is_uni_print(c); /* XXX no locale support yet */
438 Perl_is_uni_punct_lc(pTHX_ U32 c)
440 return is_uni_punct(c); /* XXX no locale support yet */
444 Perl_to_uni_upper_lc(pTHX_ U32 c)
446 return to_uni_upper(c); /* XXX no locale support yet */
450 Perl_to_uni_title_lc(pTHX_ U32 c)
452 return to_uni_title(c); /* XXX no locale support yet */
456 Perl_to_uni_lower_lc(pTHX_ U32 c)
458 return to_uni_lower(c); /* XXX no locale support yet */
462 Perl_is_utf8_alnum(pTHX_ U8 *p)
465 PL_utf8_alnum = swash_init("utf8", "IsAlnum", &PL_sv_undef, 0, 0);
466 return swash_fetch(PL_utf8_alnum, p);
467 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
468 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
470 PL_utf8_alnum = swash_init("utf8", "",
471 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
472 return swash_fetch(PL_utf8_alnum, p);
477 Perl_is_utf8_alnumc(pTHX_ U8 *p)
480 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
481 return swash_fetch(PL_utf8_alnum, p);
482 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
483 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
485 PL_utf8_alnum = swash_init("utf8", "",
486 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
487 return swash_fetch(PL_utf8_alnum, p);
492 Perl_is_utf8_idfirst(pTHX_ U8 *p)
494 return *p == '_' || is_utf8_alpha(p);
498 Perl_is_utf8_alpha(pTHX_ U8 *p)
501 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
502 return swash_fetch(PL_utf8_alpha, p);
506 Perl_is_utf8_ascii(pTHX_ U8 *p)
509 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
510 return swash_fetch(PL_utf8_ascii, p);
514 Perl_is_utf8_space(pTHX_ U8 *p)
517 PL_utf8_space = swash_init("utf8", "IsSpace", &PL_sv_undef, 0, 0);
518 return swash_fetch(PL_utf8_space, p);
522 Perl_is_utf8_digit(pTHX_ U8 *p)
525 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
526 return swash_fetch(PL_utf8_digit, p);
530 Perl_is_utf8_upper(pTHX_ U8 *p)
533 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
534 return swash_fetch(PL_utf8_upper, p);
538 Perl_is_utf8_lower(pTHX_ U8 *p)
541 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
542 return swash_fetch(PL_utf8_lower, p);
546 Perl_is_utf8_cntrl(pTHX_ U8 *p)
549 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
550 return swash_fetch(PL_utf8_cntrl, p);
554 Perl_is_utf8_graph(pTHX_ U8 *p)
557 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
558 return swash_fetch(PL_utf8_graph, p);
562 Perl_is_utf8_print(pTHX_ U8 *p)
565 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
566 return swash_fetch(PL_utf8_print, p);
570 Perl_is_utf8_punct(pTHX_ U8 *p)
573 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
574 return swash_fetch(PL_utf8_punct, p);
578 Perl_is_utf8_xdigit(pTHX_ U8 *p)
581 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
582 return swash_fetch(PL_utf8_xdigit, p);
586 Perl_is_utf8_mark(pTHX_ U8 *p)
589 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
590 return swash_fetch(PL_utf8_mark, p);
594 Perl_to_utf8_upper(pTHX_ U8 *p)
598 if (!PL_utf8_toupper)
599 PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0);
600 uv = swash_fetch(PL_utf8_toupper, p);
601 return uv ? uv : utf8_to_uv(p,0);
605 Perl_to_utf8_title(pTHX_ U8 *p)
609 if (!PL_utf8_totitle)
610 PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0);
611 uv = swash_fetch(PL_utf8_totitle, p);
612 return uv ? uv : utf8_to_uv(p,0);
616 Perl_to_utf8_lower(pTHX_ U8 *p)
620 if (!PL_utf8_tolower)
621 PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0);
622 uv = swash_fetch(PL_utf8_tolower, p);
623 return uv ? uv : utf8_to_uv(p,0);
626 /* a "swash" is a swatch hash */
629 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
634 PUSHSTACKi(PERLSI_MAGIC);
637 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
638 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
640 PUSHs(sv_2mortal(newSViv(minbits)));
641 PUSHs(sv_2mortal(newSViv(none)));
647 if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */
648 strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf);
649 if (call_method("SWASHNEW", G_SCALAR))
650 retval = newSVsv(*PL_stack_sp--);
652 retval = &PL_sv_undef;
655 if (PL_curcop == &PL_compiling) {
656 strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf);
657 PL_curcop->op_private = PL_hints;
659 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
660 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
665 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
667 HV* hv = (HV*)SvRV(sv);
668 U32 klen = UTF8SKIP(ptr) - 1;
669 U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */
671 STRLEN needents = (klen ? 64 : 128);
677 * This single-entry cache saves about 1/3 of the utf8 overhead in test
678 * suite. (That is, only 7-8% overall over just a hash cache. Still,
679 * it's nothing to sniff at.) Pity we usually come through at least
680 * two function calls to get here...
682 * NB: this code assumes that swatches are never modified, once generated!
685 if (hv == PL_last_swash_hv &&
686 klen == PL_last_swash_klen &&
687 (!klen || memEQ(ptr,PL_last_swash_key,klen)) )
689 tmps = PL_last_swash_tmps;
690 slen = PL_last_swash_slen;
693 /* Try our second-level swatch cache, kept in a hash. */
694 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
696 /* If not cached, generate it via utf8::SWASHGET */
697 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
702 PUSHSTACKi(PERLSI_MAGIC);
706 PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, 0) & ~(needents - 1))));
707 PUSHs(sv_2mortal(newSViv(needents)));
709 if (call_method("SWASHGET", G_SCALAR))
710 retval = newSVsv(*PL_stack_sp--);
712 retval = &PL_sv_undef;
716 if (PL_curcop == &PL_compiling)
717 PL_curcop->op_private = PL_hints;
719 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
721 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8)
722 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
725 PL_last_swash_hv = hv;
726 PL_last_swash_klen = klen;
727 PL_last_swash_tmps = tmps;
728 PL_last_swash_slen = slen;
730 Copy(ptr, PL_last_swash_key, klen, U8);
733 switch ((slen << 3) / needents) {
735 bit = 1 << (off & 7);
737 return (tmps[off] & bit) != 0;
742 return (tmps[off] << 8) + tmps[off + 1] ;
745 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
747 Perl_croak(aTHX_ "panic: swash_fetch");