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)
49 if (UNICODE_IS_SURROGATE(uv))
50 Perl_croak(aTHX_ "UTF-16 surrogate 0x%04"UVxf, uv);
51 else if ((uv >= 0xFDD0 && uv <= 0xFDEF) ||
52 (uv == 0xFFFE || uv == 0xFFFF))
53 Perl_croak(aTHX_ "Unicode character 0x%04"UVxf" is illegal", uv);
54 if (UNI_IS_INVARIANT(uv)) {
55 *d++ = UTF_TO_NATIVE(uv);
60 STRLEN len = UNISKIP(uv);
63 *p-- = UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
64 uv >>= UTF_ACCUMULATION_SHIFT;
66 *p = UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
69 #else /* Non loop style */
71 *d++ = (( uv >> 6) | 0xc0);
72 *d++ = (( uv & 0x3f) | 0x80);
76 *d++ = (( uv >> 12) | 0xe0);
77 *d++ = (((uv >> 6) & 0x3f) | 0x80);
78 *d++ = (( uv & 0x3f) | 0x80);
82 *d++ = (( uv >> 18) | 0xf0);
83 *d++ = (((uv >> 12) & 0x3f) | 0x80);
84 *d++ = (((uv >> 6) & 0x3f) | 0x80);
85 *d++ = (( uv & 0x3f) | 0x80);
89 *d++ = (( uv >> 24) | 0xf8);
90 *d++ = (((uv >> 18) & 0x3f) | 0x80);
91 *d++ = (((uv >> 12) & 0x3f) | 0x80);
92 *d++ = (((uv >> 6) & 0x3f) | 0x80);
93 *d++ = (( uv & 0x3f) | 0x80);
96 if (uv < 0x80000000) {
97 *d++ = (( uv >> 30) | 0xfc);
98 *d++ = (((uv >> 24) & 0x3f) | 0x80);
99 *d++ = (((uv >> 18) & 0x3f) | 0x80);
100 *d++ = (((uv >> 12) & 0x3f) | 0x80);
101 *d++ = (((uv >> 6) & 0x3f) | 0x80);
102 *d++ = (( uv & 0x3f) | 0x80);
106 if (uv < UTF8_QUAD_MAX)
109 *d++ = 0xfe; /* Can't match U+FEFF! */
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);
120 *d++ = 0xff; /* Can't match U+FFFE! */
121 *d++ = 0x80; /* 6 Reserved bits */
122 *d++ = (((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
123 *d++ = (((uv >> 54) & 0x3f) | 0x80);
124 *d++ = (((uv >> 48) & 0x3f) | 0x80);
125 *d++ = (((uv >> 42) & 0x3f) | 0x80);
126 *d++ = (((uv >> 36) & 0x3f) | 0x80);
127 *d++ = (((uv >> 30) & 0x3f) | 0x80);
128 *d++ = (((uv >> 24) & 0x3f) | 0x80);
129 *d++ = (((uv >> 18) & 0x3f) | 0x80);
130 *d++ = (((uv >> 12) & 0x3f) | 0x80);
131 *d++ = (((uv >> 6) & 0x3f) | 0x80);
132 *d++ = (( uv & 0x3f) | 0x80);
136 #endif /* Loop style */
142 =for apidoc A|STRLEN|is_utf8_char|U8 *s
144 Tests if some arbitrary number of bytes begins in a valid UTF-8
145 character. Note that an INVARIANT (i.e. ASCII) character is a valid UTF-8 character.
146 The actual number of bytes in the UTF-8 character will be returned if
147 it is valid, otherwise 0.
152 Perl_is_utf8_char(pTHX_ U8 *s)
158 if (UTF8_IS_INVARIANT(u))
161 if (!UTF8_IS_START(u))
166 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
171 u &= UTF_START_MASK(len);
175 if (!UTF8_IS_CONTINUATION(*s))
177 uv = UTF8_ACCUMULATE(uv, *s);
184 if (UNISKIP(uv) < len)
191 =for apidoc A|bool|is_utf8_string|U8 *s|STRLEN len
193 Returns true if first C<len> bytes of the given string form a valid UTF8
194 string, false otherwise. Note that 'a valid UTF8 string' does not mean
195 'a string that contains UTF8' because a valid ASCII string is a valid
202 Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len)
209 len = strlen((char *)s);
225 =for apidoc A|UV|utf8n_to_uvuni|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
227 Bottom level UTF-8 decode routine.
228 Returns the unicode code point value of the first character in the string C<s>
229 which is assumed to be in UTF8 encoding and no longer than C<curlen>;
230 C<retlen> will be set to the length, in bytes, of that character.
232 If C<s> does not point to a well-formed UTF8 character, the behaviour
233 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
234 it is assumed that the caller will raise a warning, and this function
235 will silently just set C<retlen> to C<-1> and return zero. If the
236 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
237 malformations will be given, C<retlen> will be set to the expected
238 length of the UTF-8 character in bytes, and zero will be returned.
240 The C<flags> can also contain various flags to allow deviations from
241 the strict UTF-8 encoding (see F<utf8.h>).
243 Most code should use utf8_to_uvchr() rather than call this directly.
249 Perl_utf8n_to_uvuni(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
253 bool dowarn = ckWARN_d(WARN_UTF8);
254 STRLEN expectlen = 0;
257 /* This list is a superset of the UTF8_ALLOW_XXX. */
259 #define UTF8_WARN_EMPTY 1
260 #define UTF8_WARN_CONTINUATION 2
261 #define UTF8_WARN_NON_CONTINUATION 3
262 #define UTF8_WARN_FE_FF 4
263 #define UTF8_WARN_SHORT 5
264 #define UTF8_WARN_OVERFLOW 6
265 #define UTF8_WARN_SURROGATE 7
266 #define UTF8_WARN_BOM 8
267 #define UTF8_WARN_LONG 9
268 #define UTF8_WARN_FFFF 10
271 !(flags & UTF8_ALLOW_EMPTY)) {
272 warning = UTF8_WARN_EMPTY;
276 if (UTF8_IS_INVARIANT(uv)) {
279 return (UV) (NATIVE_TO_UTF(*s));
282 if (UTF8_IS_CONTINUATION(uv) &&
283 !(flags & UTF8_ALLOW_CONTINUATION)) {
284 warning = UTF8_WARN_CONTINUATION;
288 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
289 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
290 warning = UTF8_WARN_NON_CONTINUATION;
295 uv = NATIVE_TO_UTF(uv);
297 if ((uv == 0xfe || uv == 0xff) &&
298 !(flags & UTF8_ALLOW_FE_FF)) {
299 warning = UTF8_WARN_FE_FF;
304 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
305 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
306 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
307 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
309 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
310 else { len = 7; uv &= 0x01; }
312 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
313 else if (!(uv & 0x01)) { len = 7; uv = 0; }
314 else { len = 13; uv = 0; } /* whoa! */
322 if ((curlen < expectlen) &&
323 !(flags & UTF8_ALLOW_SHORT)) {
324 warning = UTF8_WARN_SHORT;
333 if (!UTF8_IS_CONTINUATION(*s) &&
334 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
336 warning = UTF8_WARN_NON_CONTINUATION;
340 uv = UTF8_ACCUMULATE(uv, *s);
342 /* These cannot be allowed. */
344 if (!(flags & UTF8_ALLOW_LONG)) {
345 warning = UTF8_WARN_LONG;
349 else { /* uv < ouv */
350 /* This cannot be allowed. */
351 warning = UTF8_WARN_OVERFLOW;
359 if (UNICODE_IS_SURROGATE(uv) &&
360 !(flags & UTF8_ALLOW_SURROGATE)) {
361 warning = UTF8_WARN_SURROGATE;
363 } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) &&
364 !(flags & UTF8_ALLOW_BOM)) {
365 warning = UTF8_WARN_BOM;
367 } else if ((expectlen > UNISKIP(uv)) &&
368 !(flags & UTF8_ALLOW_LONG)) {
369 warning = UTF8_WARN_LONG;
371 } else if (UNICODE_IS_ILLEGAL(uv) &&
372 !(flags & UTF8_ALLOW_FFFF)) {
373 warning = UTF8_WARN_FFFF;
381 if (flags & UTF8_CHECK_ONLY) {
388 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
391 case 0: /* Intentionally empty. */ break;
392 case UTF8_WARN_EMPTY:
393 Perl_sv_catpvf(aTHX_ sv, "(empty string)");
395 case UTF8_WARN_CONTINUATION:
396 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf")", uv);
398 case UTF8_WARN_NON_CONTINUATION:
399 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")",
402 case UTF8_WARN_FE_FF:
403 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
405 case UTF8_WARN_SHORT:
406 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
407 curlen, curlen == 1 ? "" : "s", expectlen);
408 expectlen = curlen; /* distance for caller to skip */
410 case UTF8_WARN_OVERFLOW:
411 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x)",
414 case UTF8_WARN_SURROGATE:
415 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
418 Perl_sv_catpvf(aTHX_ sv, "(byte order mark 0x%04"UVxf")", uv);
421 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
422 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
425 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
428 Perl_sv_catpvf(aTHX_ sv, "(unknown reason)");
436 Perl_warner(aTHX_ WARN_UTF8,
437 "%s in %s", s, OP_DESC(PL_op));
439 Perl_warner(aTHX_ WARN_UTF8, "%s", s);
444 *retlen = expectlen ? expectlen : len;
450 =for apidoc A|UV|utf8_to_uvchr|U8 *s|STRLEN *retlen
452 Returns the native character value of the first character in the string C<s>
453 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
454 length, in bytes, of that character.
456 If C<s> does not point to a well-formed UTF8 character, zero is
457 returned and retlen is set, if possible, to -1.
463 Perl_utf8_to_uvchr(pTHX_ U8 *s, STRLEN *retlen)
465 return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXLEN, retlen, 0);
469 =for apidoc A|UV|utf8_to_uvuni|U8 *s|STRLEN *retlen
471 Returns the Unicode code point of the first character in the string C<s>
472 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
473 length, in bytes, of that character.
475 This function should only be used when returned UV is considered
476 an index into the Unicode semantic tables (e.g. swashes).
478 If C<s> does not point to a well-formed UTF8 character, zero is
479 returned and retlen is set, if possible, to -1.
485 Perl_utf8_to_uvuni(pTHX_ U8 *s, STRLEN *retlen)
487 /* Call the low level routine asking for checks */
488 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXLEN, retlen, 0);
492 =for apidoc A|STRLEN|utf8_length|U8 *s|U8 *e
494 Return the length of the UTF-8 char encoded string C<s> in characters.
495 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
496 up past C<e>, croaks.
502 Perl_utf8_length(pTHX_ U8 *s, U8 *e)
506 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
507 * the bitops (especially ~) can create illegal UTF-8.
508 * In other words: in Perl UTF-8 is not just for Unicode. */
511 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
516 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
525 =for apidoc A|IV|utf8_distance|U8 *a|U8 *b
527 Returns the number of UTF8 characters between the UTF-8 pointers C<a>
530 WARNING: use only if you *know* that the pointers point inside the
537 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
541 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
542 * the bitops (especially ~) can create illegal UTF-8.
543 * In other words: in Perl UTF-8 is not just for Unicode. */
550 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
560 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
570 =for apidoc A|U8 *|utf8_hop|U8 *s|I32 off
572 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
575 WARNING: do not use the following unless you *know* C<off> is within
576 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
577 on the first byte of character or just after the last byte of a character.
583 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
585 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
586 * the bitops (especially ~) can create illegal UTF-8.
587 * In other words: in Perl UTF-8 is not just for Unicode. */
596 while (UTF8_IS_CONTINUATION(*s))
604 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
606 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
607 Unlike C<bytes_to_utf8>, this over-writes the original string, and
608 updates len to contain the new length.
609 Returns zero on failure, setting C<len> to -1.
615 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
621 /* ensure valid UTF8 and chars < 256 before updating string */
622 for (send = s + *len; s < send; ) {
625 if (!UTF8_IS_INVARIANT(c) &&
626 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
627 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
636 *d++ = (U8)utf8_to_uvchr(s, &ulen);
645 =for apidoc A|U8 *|bytes_from_utf8|U8 *s|STRLEN *len|bool *is_utf8
647 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
648 Unlike <utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
649 the newly-created string, and updates C<len> to contain the new
650 length. Returns the original string if no conversion occurs, C<len>
651 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
652 0 if C<s> is converted or contains all 7bit characters.
658 Perl_bytes_from_utf8(pTHX_ U8 *s, STRLEN *len, bool *is_utf8)
668 /* ensure valid UTF8 and chars < 256 before converting string */
669 for (send = s + *len; s < send;) {
671 if (!UTF8_IS_INVARIANT(c)) {
672 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
673 (c = *s++) && UTF8_IS_CONTINUATION(c))
682 Newz(801, d, (*len) - count + 1, U8);
683 s = start; start = d;
686 if (!UTF8_IS_INVARIANT(c)) {
687 /* Then it is two-byte encoded */
688 c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++);
689 c = ASCII_TO_NATIVE(c);
699 =for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
701 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
702 Returns a pointer to the newly-created string, and sets C<len> to
703 reflect the new length.
709 Perl_bytes_to_utf8(pTHX_ U8 *s, STRLEN *len)
716 Newz(801, d, (*len) * 2 + 1, U8);
720 UV uv = NATIVE_TO_ASCII(*s++);
721 if (UNI_IS_INVARIANT(uv))
722 *d++ = UTF_TO_NATIVE(uv);
724 *d++ = UTF8_EIGHT_BIT_HI(uv);
725 *d++ = UTF8_EIGHT_BIT_LO(uv);
734 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
736 * Destination must be pre-extended to 3/2 source. Do not use in-place.
737 * We optimize for native, for obvious reasons. */
740 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
746 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
751 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
758 *d++ = (( uv >> 6) | 0xc0);
759 *d++ = (( uv & 0x3f) | 0x80);
762 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
764 if (low < 0xdc00 || low >= 0xdfff)
765 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
766 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
769 *d++ = (( uv >> 12) | 0xe0);
770 *d++ = (((uv >> 6) & 0x3f) | 0x80);
771 *d++ = (( uv & 0x3f) | 0x80);
775 *d++ = (( uv >> 18) | 0xf0);
776 *d++ = (((uv >> 12) & 0x3f) | 0x80);
777 *d++ = (((uv >> 6) & 0x3f) | 0x80);
778 *d++ = (( uv & 0x3f) | 0x80);
782 *newlen = d - dstart;
786 /* Note: this one is slightly destructive of the source. */
789 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
792 U8* send = s + bytelen;
799 return utf16_to_utf8(p, d, bytelen, newlen);
802 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
805 Perl_is_uni_alnum(pTHX_ UV c)
807 U8 tmpbuf[UTF8_MAXLEN+1];
808 uvchr_to_utf8(tmpbuf, (UV)c);
809 return is_utf8_alnum(tmpbuf);
813 Perl_is_uni_alnumc(pTHX_ UV c)
815 U8 tmpbuf[UTF8_MAXLEN+1];
816 uvchr_to_utf8(tmpbuf, (UV)c);
817 return is_utf8_alnumc(tmpbuf);
821 Perl_is_uni_idfirst(pTHX_ UV c)
823 U8 tmpbuf[UTF8_MAXLEN+1];
824 uvchr_to_utf8(tmpbuf, (UV)c);
825 return is_utf8_idfirst(tmpbuf);
829 Perl_is_uni_alpha(pTHX_ UV c)
831 U8 tmpbuf[UTF8_MAXLEN+1];
832 uvchr_to_utf8(tmpbuf, (UV)c);
833 return is_utf8_alpha(tmpbuf);
837 Perl_is_uni_ascii(pTHX_ UV c)
839 U8 tmpbuf[UTF8_MAXLEN+1];
840 uvchr_to_utf8(tmpbuf, (UV)c);
841 return is_utf8_ascii(tmpbuf);
845 Perl_is_uni_space(pTHX_ UV c)
847 U8 tmpbuf[UTF8_MAXLEN+1];
848 uvchr_to_utf8(tmpbuf, (UV)c);
849 return is_utf8_space(tmpbuf);
853 Perl_is_uni_digit(pTHX_ UV c)
855 U8 tmpbuf[UTF8_MAXLEN+1];
856 uvchr_to_utf8(tmpbuf, (UV)c);
857 return is_utf8_digit(tmpbuf);
861 Perl_is_uni_upper(pTHX_ UV c)
863 U8 tmpbuf[UTF8_MAXLEN+1];
864 uvchr_to_utf8(tmpbuf, (UV)c);
865 return is_utf8_upper(tmpbuf);
869 Perl_is_uni_lower(pTHX_ UV c)
871 U8 tmpbuf[UTF8_MAXLEN+1];
872 uvchr_to_utf8(tmpbuf, (UV)c);
873 return is_utf8_lower(tmpbuf);
877 Perl_is_uni_cntrl(pTHX_ UV c)
879 U8 tmpbuf[UTF8_MAXLEN+1];
880 uvchr_to_utf8(tmpbuf, (UV)c);
881 return is_utf8_cntrl(tmpbuf);
885 Perl_is_uni_graph(pTHX_ UV c)
887 U8 tmpbuf[UTF8_MAXLEN+1];
888 uvchr_to_utf8(tmpbuf, (UV)c);
889 return is_utf8_graph(tmpbuf);
893 Perl_is_uni_print(pTHX_ UV c)
895 U8 tmpbuf[UTF8_MAXLEN+1];
896 uvchr_to_utf8(tmpbuf, (UV)c);
897 return is_utf8_print(tmpbuf);
901 Perl_is_uni_punct(pTHX_ UV c)
903 U8 tmpbuf[UTF8_MAXLEN+1];
904 uvchr_to_utf8(tmpbuf, (UV)c);
905 return is_utf8_punct(tmpbuf);
909 Perl_is_uni_xdigit(pTHX_ UV c)
911 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
912 uvchr_to_utf8(tmpbuf, (UV)c);
913 return is_utf8_xdigit(tmpbuf);
917 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
919 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
920 uvchr_to_utf8(tmpbuf, (UV)c);
921 return to_utf8_upper(tmpbuf, p, lenp);
925 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
927 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
928 uvchr_to_utf8(tmpbuf, (UV)c);
929 return to_utf8_title(tmpbuf, p, lenp);
933 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
935 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
936 uvchr_to_utf8(tmpbuf, (UV)c);
937 return to_utf8_lower(tmpbuf, p, lenp);
941 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
943 U8 tmpbuf[UTF8_MAXLEN_FOLD+1];
944 uvchr_to_utf8(tmpbuf, (UV)c);
945 return to_utf8_fold(tmpbuf, p, lenp);
948 /* for now these all assume no locale info available for Unicode > 255 */
951 Perl_is_uni_alnum_lc(pTHX_ UV c)
953 return is_uni_alnum(c); /* XXX no locale support yet */
957 Perl_is_uni_alnumc_lc(pTHX_ UV c)
959 return is_uni_alnumc(c); /* XXX no locale support yet */
963 Perl_is_uni_idfirst_lc(pTHX_ UV c)
965 return is_uni_idfirst(c); /* XXX no locale support yet */
969 Perl_is_uni_alpha_lc(pTHX_ UV c)
971 return is_uni_alpha(c); /* XXX no locale support yet */
975 Perl_is_uni_ascii_lc(pTHX_ UV c)
977 return is_uni_ascii(c); /* XXX no locale support yet */
981 Perl_is_uni_space_lc(pTHX_ UV c)
983 return is_uni_space(c); /* XXX no locale support yet */
987 Perl_is_uni_digit_lc(pTHX_ UV c)
989 return is_uni_digit(c); /* XXX no locale support yet */
993 Perl_is_uni_upper_lc(pTHX_ UV c)
995 return is_uni_upper(c); /* XXX no locale support yet */
999 Perl_is_uni_lower_lc(pTHX_ UV c)
1001 return is_uni_lower(c); /* XXX no locale support yet */
1005 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1007 return is_uni_cntrl(c); /* XXX no locale support yet */
1011 Perl_is_uni_graph_lc(pTHX_ UV c)
1013 return is_uni_graph(c); /* XXX no locale support yet */
1017 Perl_is_uni_print_lc(pTHX_ UV c)
1019 return is_uni_print(c); /* XXX no locale support yet */
1023 Perl_is_uni_punct_lc(pTHX_ UV c)
1025 return is_uni_punct(c); /* XXX no locale support yet */
1029 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1031 return is_uni_xdigit(c); /* XXX no locale support yet */
1035 Perl_is_utf8_alnum(pTHX_ U8 *p)
1037 if (!is_utf8_char(p))
1040 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1041 * descendant of isalnum(3), in other words, it doesn't
1042 * contain the '_'. --jhi */
1043 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1044 return swash_fetch(PL_utf8_alnum, p, TRUE);
1045 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1046 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1048 PL_utf8_alnum = swash_init("utf8", "",
1049 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1050 return swash_fetch(PL_utf8_alnum, p, TRUE);
1055 Perl_is_utf8_alnumc(pTHX_ U8 *p)
1057 if (!is_utf8_char(p))
1060 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1061 return swash_fetch(PL_utf8_alnum, p, TRUE);
1062 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
1063 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1065 PL_utf8_alnum = swash_init("utf8", "",
1066 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1067 return swash_fetch(PL_utf8_alnum, p, TRUE);
1072 Perl_is_utf8_idfirst(pTHX_ U8 *p)
1074 return *p == '_' || is_utf8_alpha(p);
1078 Perl_is_utf8_alpha(pTHX_ U8 *p)
1080 if (!is_utf8_char(p))
1083 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1084 return swash_fetch(PL_utf8_alpha, p, TRUE);
1088 Perl_is_utf8_ascii(pTHX_ U8 *p)
1090 if (!is_utf8_char(p))
1093 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1094 return swash_fetch(PL_utf8_ascii, p, TRUE);
1098 Perl_is_utf8_space(pTHX_ U8 *p)
1100 if (!is_utf8_char(p))
1103 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1104 return swash_fetch(PL_utf8_space, p, TRUE);
1108 Perl_is_utf8_digit(pTHX_ U8 *p)
1110 if (!is_utf8_char(p))
1113 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1114 return swash_fetch(PL_utf8_digit, p, TRUE);
1118 Perl_is_utf8_upper(pTHX_ U8 *p)
1120 if (!is_utf8_char(p))
1123 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
1124 return swash_fetch(PL_utf8_upper, p, TRUE);
1128 Perl_is_utf8_lower(pTHX_ U8 *p)
1130 if (!is_utf8_char(p))
1133 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
1134 return swash_fetch(PL_utf8_lower, p, TRUE);
1138 Perl_is_utf8_cntrl(pTHX_ U8 *p)
1140 if (!is_utf8_char(p))
1143 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1144 return swash_fetch(PL_utf8_cntrl, p, TRUE);
1148 Perl_is_utf8_graph(pTHX_ U8 *p)
1150 if (!is_utf8_char(p))
1153 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1154 return swash_fetch(PL_utf8_graph, p, TRUE);
1158 Perl_is_utf8_print(pTHX_ U8 *p)
1160 if (!is_utf8_char(p))
1163 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1164 return swash_fetch(PL_utf8_print, p, TRUE);
1168 Perl_is_utf8_punct(pTHX_ U8 *p)
1170 if (!is_utf8_char(p))
1173 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1174 return swash_fetch(PL_utf8_punct, p, TRUE);
1178 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1180 if (!is_utf8_char(p))
1182 if (!PL_utf8_xdigit)
1183 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1184 return swash_fetch(PL_utf8_xdigit, p, TRUE);
1188 Perl_is_utf8_mark(pTHX_ U8 *p)
1190 if (!is_utf8_char(p))
1193 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1194 return swash_fetch(PL_utf8_mark, p, TRUE);
1198 =for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special
1200 The "p" contains the pointer to the UTF-8 string encoding
1201 the character that is being converted.
1203 The "ustrp" is a pointer to the character buffer to put the
1204 conversion result to. The "lenp" is a pointer to the length
1207 The "swash" is a pointer to the swash to use.
1209 The "normal" is a string like "ToLower" which means the swash
1210 $utf8::ToLower, which is stored in lib/unicore/To/Lower.pl,
1211 and loaded by SWASHGET, using lib/utf8_heavy.pl.
1213 The "special" is a string like "utf8::ToSpecLower", which means
1214 the hash %utf8::ToSpecLower, which is stored in the same file,
1215 lib/unicore/To/Lower.pl, and also loaded by SWASHGET. The access
1216 to the hash is by Perl_to_utf8_case().
1222 Perl_to_utf8_case(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp,char *normal, char *special)
1227 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1228 uv = swash_fetch(*swashp, p, TRUE);
1230 uv = UNI_TO_NATIVE(uv);
1236 uv = utf8_to_uvchr(p, 0);
1238 if ((hv = get_hv(special, FALSE)) &&
1239 (keysv = sv_2mortal(Perl_newSVpvf(aTHX_ "%04"UVXf, uv))) &&
1240 (he = hv_fetch_ent(hv, keysv, FALSE, 0))) {
1241 SV *val = HeVAL(he);
1242 char *s = SvPV(val, *lenp);
1244 if (*lenp > 1 || UNI_IS_INVARIANT(c))
1245 Copy(s, ustrp, *lenp, U8);
1247 /* something in the 0x80..0xFF range */
1248 ustrp[0] = UTF8_EIGHT_BIT_HI(c);
1249 ustrp[1] = UTF8_EIGHT_BIT_LO(c);
1256 *lenp = UNISKIP(uv);
1257 uvuni_to_utf8(ustrp, uv);
1262 Perl_to_utf8_upper(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1264 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1265 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1269 Perl_to_utf8_title(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1271 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1272 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1276 Perl_to_utf8_lower(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1278 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1279 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1283 Perl_to_utf8_fold(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1285 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1286 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1289 /* a "swash" is a swatch hash */
1292 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1295 SV* tokenbufsv = sv_2mortal(NEWSV(0,0));
1297 HV *stash = gv_stashpvn(pkg, strlen(pkg), FALSE);
1300 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
1302 errsv_save = newSVsv(ERRSV);
1303 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1305 sv_setsv(ERRSV, errsv_save);
1306 SvREFCNT_dec(errsv_save);
1310 PUSHSTACKi(PERLSI_MAGIC);
1313 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1314 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1316 PUSHs(sv_2mortal(newSViv(minbits)));
1317 PUSHs(sv_2mortal(newSViv(none)));
1323 if (PL_curcop == &PL_compiling)
1324 /* XXX ought to be handled by lex_start */
1325 sv_setpv(tokenbufsv, PL_tokenbuf);
1326 errsv_save = newSVsv(ERRSV);
1327 if (call_method("SWASHNEW", G_SCALAR))
1328 retval = newSVsv(*PL_stack_sp--);
1330 retval = &PL_sv_undef;
1332 sv_setsv(ERRSV, errsv_save);
1333 SvREFCNT_dec(errsv_save);
1336 if (PL_curcop == &PL_compiling) {
1338 char* pv = SvPV(tokenbufsv, len);
1340 Copy(pv, PL_tokenbuf, len+1, char);
1341 PL_curcop->op_private = PL_hints;
1343 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1344 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1349 /* This API is wrong for special case conversions since we may need to
1350 * return several Unicode characters for a single Unicode character
1351 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1352 * the lower-level routine, and it is similarly broken for returning
1353 * multiple values. --jhi */
1355 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr, bool do_utf8)
1357 HV* hv = (HV*)SvRV(sv);
1366 UV c = NATIVE_TO_ASCII(*ptr);
1368 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1369 tmputf8[0] = UTF8_EIGHT_BIT_HI(c);
1370 tmputf8[1] = UTF8_EIGHT_BIT_LO(c);
1373 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
1374 * then the "swatch" is a vec() for al the chars which start
1376 * So the key in the hash (klen) is length of encoded char -1
1378 klen = UTF8SKIP(ptr) - 1;
1383 /* If char in invariant then swatch is for all the invariant chars
1384 * In both UTF-8 and UTF8-MOD that happens to be UTF_CONTINUATION_MARK
1386 needents = UTF_CONTINUATION_MARK;
1387 off = NATIVE_TO_UTF(ptr[klen]);
1391 /* If char is encoded then swatch is for the prefix */
1392 needents = (1 << UTF_ACCUMULATION_SHIFT);
1393 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
1397 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1398 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1399 * it's nothing to sniff at.) Pity we usually come through at least
1400 * two function calls to get here...
1402 * NB: this code assumes that swatches are never modified, once generated!
1405 if (hv == PL_last_swash_hv &&
1406 klen == PL_last_swash_klen &&
1407 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
1409 tmps = PL_last_swash_tmps;
1410 slen = PL_last_swash_slen;
1413 /* Try our second-level swatch cache, kept in a hash. */
1414 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1416 /* If not cached, generate it via utf8::SWASHGET */
1417 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1419 /* We use utf8n_to_uvuni() as we want an index into
1420 Unicode tables, not a native character number.
1422 UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXLEN, NULL, 0);
1427 PUSHSTACKi(PERLSI_MAGIC);
1431 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
1432 PUSHs(sv_2mortal(newSViv((klen) ?
1433 (code_point & ~(needents - 1)) : 0)));
1434 PUSHs(sv_2mortal(newSViv(needents)));
1436 errsv_save = newSVsv(ERRSV);
1437 if (call_method("SWASHGET", G_SCALAR))
1438 retval = newSVsv(*PL_stack_sp--);
1440 retval = &PL_sv_undef;
1442 sv_setsv(ERRSV, errsv_save);
1443 SvREFCNT_dec(errsv_save);
1447 if (PL_curcop == &PL_compiling)
1448 PL_curcop->op_private = PL_hints;
1450 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1452 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || (slen << 3) < needents)
1453 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1456 PL_last_swash_hv = hv;
1457 PL_last_swash_klen = klen;
1458 PL_last_swash_tmps = tmps;
1459 PL_last_swash_slen = slen;
1461 Copy(ptr, PL_last_swash_key, klen, U8);
1464 switch ((int)((slen << 3) / needents)) {
1466 bit = 1 << (off & 7);
1468 return (tmps[off] & bit) != 0;
1473 return (tmps[off] << 8) + tmps[off + 1] ;
1476 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1478 Perl_croak(aTHX_ "panic: swash_fetch");
1484 =for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv
1486 Adds the UTF8 representation of the Native codepoint C<uv> to the end
1487 of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free
1488 bytes available. The return value is the pointer to the byte after the
1489 end of the new character. In other words,
1491 d = uvchr_to_utf8(d, uv);
1493 is the recommended wide native character-aware way of saying
1500 /* On ASCII machines this is normally a macro but we want a
1501 real function in case XS code wants it
1503 #undef Perl_uvchr_to_utf8
1505 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
1507 return Perl_uvuni_to_utf8(aTHX_ d, NATIVE_TO_UNI(uv));
1512 =for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
1514 Returns the native character value of the first character in the string C<s>
1515 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
1516 length, in bytes, of that character.
1518 Allows length and flags to be passed to low level routine.
1522 /* On ASCII machines this is normally a macro but we want a
1523 real function in case XS code wants it
1525 #undef Perl_utf8n_to_uvchr
1527 Perl_utf8n_to_uvchr(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
1529 UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
1530 return UNI_TO_NATIVE(uv);
1534 =for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags
1536 Build to the scalar dsv a displayable version of the string spv,
1537 length len, the displayable version being at most pvlim bytes long
1538 (if longer, the rest is truncated and "..." will be appended).
1539 The flags argument is currently unused but available for future extensions.
1540 The pointer to the PV of the dsv is returned.
1544 Perl_pv_uni_display(pTHX_ SV *dsv, U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
1549 sv_setpvn(dsv, "", 0);
1550 for (s = (char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
1552 if (pvlim && SvCUR(dsv) >= pvlim) {
1556 u = utf8_to_uvchr((U8*)s, 0);
1557 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
1560 sv_catpvn(dsv, "...", 3);
1566 =for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags
1568 Build to the scalar dsv a displayable version of the scalar sv,
1569 he displayable version being at most pvlim bytes long
1570 (if longer, the rest is truncated and "..." will be appended).
1571 The flags argument is currently unused but available for future extensions.
1572 The pointer to the PV of the dsv is returned.
1576 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
1578 return Perl_pv_uni_display(aTHX_ dsv, (U8*)SvPVX(ssv), SvCUR(ssv),
1583 =for apidoc A|I32|ibcmp_utf8|const char *s1|bool u1|const char *s2|bool u2|register I32 len
1585 Return true if the strings s1 and s2 differ case-insensitively, false
1586 if not (if they are equal case-insensitively). If u1 is true, the
1587 string s1 is assumed to be in UTF-8-encoded Unicode. If u2 is true,
1588 the string s2 is assumed to be in UTF-8-encoded Unicode.
1590 For case-insensitiveness, the "casefolding" of Unicode is used
1591 instead of upper/lowercasing both the characters, see
1592 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
1596 Perl_ibcmp_utf8(pTHX_ const char *s1, bool u1, register I32 len1, const char *s2, bool u2, register I32 len2)
1598 register U8 *a = (U8*)s1;
1599 register U8 *b = (U8*)s2;
1600 register U8 *ae = b + len1;
1601 register U8 *be = b + len2;
1604 STRLEN ulen1, ulen2;
1605 U8 tmpbuf1[UTF8_MAXLEN_FOLD+1];
1606 U8 tmpbuf2[UTF8_MAXLEN_FOLD+1];
1608 while (a < ae && b < be) {
1610 if (a + UTF8SKIP(a) > ae)
1612 ca = utf8_to_uvchr((U8*)a, &la);
1618 if (b + UTF8SKIP(b) > be)
1620 cb = utf8_to_uvchr((U8*)b, &lb);
1627 to_uni_fold(NATIVE_TO_UNI(ca), tmpbuf1, &ulen1);
1631 to_uni_fold(NATIVE_TO_UNI(cb), tmpbuf2, &ulen2);
1635 || (ulen1 == 1 && PL_fold[ca] != PL_fold[cb])
1636 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1))
1637 return 1; /* mismatch */
1642 return a == ae && b == be ? 0 : 1; /* 0 match, 1 mismatch */