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 (UNI_IS_INVARIANT(uv)) {
50 *d++ = UTF_TO_NATIVE(uv);
55 STRLEN len = UNISKIP(uv);
58 *p-- = UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
59 uv >>= UTF_ACCUMULATION_SHIFT;
61 *p = UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
64 #else /* Non loop style */
66 *d++ = (( uv >> 6) | 0xc0);
67 *d++ = (( uv & 0x3f) | 0x80);
71 *d++ = (( uv >> 12) | 0xe0);
72 *d++ = (((uv >> 6) & 0x3f) | 0x80);
73 *d++ = (( uv & 0x3f) | 0x80);
77 *d++ = (( uv >> 18) | 0xf0);
78 *d++ = (((uv >> 12) & 0x3f) | 0x80);
79 *d++ = (((uv >> 6) & 0x3f) | 0x80);
80 *d++ = (( uv & 0x3f) | 0x80);
84 *d++ = (( uv >> 24) | 0xf8);
85 *d++ = (((uv >> 18) & 0x3f) | 0x80);
86 *d++ = (((uv >> 12) & 0x3f) | 0x80);
87 *d++ = (((uv >> 6) & 0x3f) | 0x80);
88 *d++ = (( uv & 0x3f) | 0x80);
91 if (uv < 0x80000000) {
92 *d++ = (( uv >> 30) | 0xfc);
93 *d++ = (((uv >> 24) & 0x3f) | 0x80);
94 *d++ = (((uv >> 18) & 0x3f) | 0x80);
95 *d++ = (((uv >> 12) & 0x3f) | 0x80);
96 *d++ = (((uv >> 6) & 0x3f) | 0x80);
97 *d++ = (( uv & 0x3f) | 0x80);
101 if (uv < UTF8_QUAD_MAX)
104 *d++ = 0xfe; /* Can't match U+FEFF! */
105 *d++ = (((uv >> 30) & 0x3f) | 0x80);
106 *d++ = (((uv >> 24) & 0x3f) | 0x80);
107 *d++ = (((uv >> 18) & 0x3f) | 0x80);
108 *d++ = (((uv >> 12) & 0x3f) | 0x80);
109 *d++ = (((uv >> 6) & 0x3f) | 0x80);
110 *d++ = (( uv & 0x3f) | 0x80);
115 *d++ = 0xff; /* Can't match U+FFFE! */
116 *d++ = 0x80; /* 6 Reserved bits */
117 *d++ = (((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
118 *d++ = (((uv >> 54) & 0x3f) | 0x80);
119 *d++ = (((uv >> 48) & 0x3f) | 0x80);
120 *d++ = (((uv >> 42) & 0x3f) | 0x80);
121 *d++ = (((uv >> 36) & 0x3f) | 0x80);
122 *d++ = (((uv >> 30) & 0x3f) | 0x80);
123 *d++ = (((uv >> 24) & 0x3f) | 0x80);
124 *d++ = (((uv >> 18) & 0x3f) | 0x80);
125 *d++ = (((uv >> 12) & 0x3f) | 0x80);
126 *d++ = (((uv >> 6) & 0x3f) | 0x80);
127 *d++ = (( uv & 0x3f) | 0x80);
131 #endif /* Loop style */
137 =for apidoc A|STRLEN|is_utf8_char|U8 *s
139 Tests if some arbitrary number of bytes begins in a valid UTF-8
140 character. Note that an INVARIANT (i.e. ASCII) character is a valid UTF-8 character.
141 The actual number of bytes in the UTF-8 character will be returned if
142 it is valid, otherwise 0.
147 Perl_is_utf8_char(pTHX_ U8 *s)
153 if (UTF8_IS_INVARIANT(u))
156 if (!UTF8_IS_START(u))
161 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
166 u &= UTF_START_MASK(len);
170 if (!UTF8_IS_CONTINUATION(*s))
172 uv = UTF8_ACCUMULATE(uv, *s);
179 if (UNISKIP(uv) < len)
186 =for apidoc A|bool|is_utf8_string|U8 *s|STRLEN len
188 Returns true if first C<len> bytes of the given string form a valid UTF8
189 string, false otherwise. Note that 'a valid UTF8 string' does not mean
190 'a string that contains UTF8' because a valid ASCII string is a valid
197 Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len)
204 len = strlen((char *)s);
220 =for apidoc A|UV|utf8n_to_uvuni|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
222 Bottom level UTF-8 decode routine.
223 Returns the unicode code point value of the first character in the string C<s>
224 which is assumed to be in UTF8 encoding and no longer than C<curlen>;
225 C<retlen> will be set to the length, in bytes, of that character.
227 If C<s> does not point to a well-formed UTF8 character, the behaviour
228 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
229 it is assumed that the caller will raise a warning, and this function
230 will silently just set C<retlen> to C<-1> and return zero. If the
231 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
232 malformations will be given, C<retlen> will be set to the expected
233 length of the UTF-8 character in bytes, and zero will be returned.
235 The C<flags> can also contain various flags to allow deviations from
236 the strict UTF-8 encoding (see F<utf8.h>).
238 Most code should use utf8_to_uvchr() rather than call this directly.
244 Perl_utf8n_to_uvuni(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
248 bool dowarn = ckWARN_d(WARN_UTF8);
249 STRLEN expectlen = 0;
252 /* This list is a superset of the UTF8_ALLOW_XXX. */
254 #define UTF8_WARN_EMPTY 1
255 #define UTF8_WARN_CONTINUATION 2
256 #define UTF8_WARN_NON_CONTINUATION 3
257 #define UTF8_WARN_FE_FF 4
258 #define UTF8_WARN_SHORT 5
259 #define UTF8_WARN_OVERFLOW 6
260 #define UTF8_WARN_SURROGATE 7
261 #define UTF8_WARN_BOM 8
262 #define UTF8_WARN_LONG 9
263 #define UTF8_WARN_FFFF 10
266 !(flags & UTF8_ALLOW_EMPTY)) {
267 warning = UTF8_WARN_EMPTY;
271 if (UTF8_IS_INVARIANT(uv)) {
274 return (UV) (NATIVE_TO_UTF(*s));
277 if (UTF8_IS_CONTINUATION(uv) &&
278 !(flags & UTF8_ALLOW_CONTINUATION)) {
279 warning = UTF8_WARN_CONTINUATION;
283 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
284 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
285 warning = UTF8_WARN_NON_CONTINUATION;
290 uv = NATIVE_TO_UTF(uv);
292 if ((uv == 0xfe || uv == 0xff) &&
293 !(flags & UTF8_ALLOW_FE_FF)) {
294 warning = UTF8_WARN_FE_FF;
299 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
300 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
301 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
302 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
304 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
305 else { len = 7; uv &= 0x01; }
307 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
308 else if (!(uv & 0x01)) { len = 7; uv = 0; }
309 else { len = 13; uv = 0; } /* whoa! */
317 if ((curlen < expectlen) &&
318 !(flags & UTF8_ALLOW_SHORT)) {
319 warning = UTF8_WARN_SHORT;
328 if (!UTF8_IS_CONTINUATION(*s) &&
329 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
331 warning = UTF8_WARN_NON_CONTINUATION;
335 uv = UTF8_ACCUMULATE(uv, *s);
337 /* These cannot be allowed. */
339 if (!(flags & UTF8_ALLOW_LONG)) {
340 warning = UTF8_WARN_LONG;
344 else { /* uv < ouv */
345 /* This cannot be allowed. */
346 warning = UTF8_WARN_OVERFLOW;
354 if (UNICODE_IS_SURROGATE(uv) &&
355 !(flags & UTF8_ALLOW_SURROGATE)) {
356 warning = UTF8_WARN_SURROGATE;
358 } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) &&
359 !(flags & UTF8_ALLOW_BOM)) {
360 warning = UTF8_WARN_BOM;
362 } else if ((expectlen > UNISKIP(uv)) &&
363 !(flags & UTF8_ALLOW_LONG)) {
364 warning = UTF8_WARN_LONG;
366 } else if (UNICODE_IS_ILLEGAL(uv) &&
367 !(flags & UTF8_ALLOW_FFFF)) {
368 warning = UTF8_WARN_FFFF;
376 if (flags & UTF8_CHECK_ONLY) {
383 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
386 case 0: /* Intentionally empty. */ break;
387 case UTF8_WARN_EMPTY:
388 Perl_sv_catpvf(aTHX_ sv, "(empty string)");
390 case UTF8_WARN_CONTINUATION:
391 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf")", uv);
393 case UTF8_WARN_NON_CONTINUATION:
394 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")",
397 case UTF8_WARN_FE_FF:
398 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
400 case UTF8_WARN_SHORT:
401 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
402 curlen, curlen == 1 ? "" : "s", expectlen);
403 expectlen = curlen; /* distance for caller to skip */
405 case UTF8_WARN_OVERFLOW:
406 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x)",
409 case UTF8_WARN_SURROGATE:
410 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
413 Perl_sv_catpvf(aTHX_ sv, "(byte order mark 0x%04"UVxf")", uv);
416 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
417 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
420 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
423 Perl_sv_catpvf(aTHX_ sv, "(unknown reason)");
431 Perl_warner(aTHX_ WARN_UTF8,
432 "%s in %s", s, OP_DESC(PL_op));
434 Perl_warner(aTHX_ WARN_UTF8, "%s", s);
439 *retlen = expectlen ? expectlen : len;
445 =for apidoc A|UV|utf8_to_uvchr|U8 *s|STRLEN *retlen
447 Returns the native character value of the first character in the string C<s>
448 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
449 length, in bytes, of that character.
451 If C<s> does not point to a well-formed UTF8 character, zero is
452 returned and retlen is set, if possible, to -1.
458 Perl_utf8_to_uvchr(pTHX_ U8 *s, STRLEN *retlen)
460 return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXLEN, retlen, 0);
464 =for apidoc A|UV|utf8_to_uvuni|U8 *s|STRLEN *retlen
466 Returns the Unicode code point of the first character in the string C<s>
467 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
468 length, in bytes, of that character.
470 This function should only be used when returned UV is considered
471 an index into the Unicode semantic tables (e.g. swashes).
473 If C<s> does not point to a well-formed UTF8 character, zero is
474 returned and retlen is set, if possible, to -1.
480 Perl_utf8_to_uvuni(pTHX_ U8 *s, STRLEN *retlen)
482 /* Call the low level routine asking for checks */
483 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXLEN, retlen, 0);
487 =for apidoc A|STRLEN|utf8_length|U8 *s|U8 *e
489 Return the length of the UTF-8 char encoded string C<s> in characters.
490 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
491 up past C<e>, croaks.
497 Perl_utf8_length(pTHX_ U8 *s, U8 *e)
501 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
502 * the bitops (especially ~) can create illegal UTF-8.
503 * In other words: in Perl UTF-8 is not just for Unicode. */
506 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
511 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
520 =for apidoc A|IV|utf8_distance|U8 *a|U8 *b
522 Returns the number of UTF8 characters between the UTF-8 pointers C<a>
525 WARNING: use only if you *know* that the pointers point inside the
532 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
536 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
537 * the bitops (especially ~) can create illegal UTF-8.
538 * In other words: in Perl UTF-8 is not just for Unicode. */
545 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
555 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
565 =for apidoc A|U8 *|utf8_hop|U8 *s|I32 off
567 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
570 WARNING: do not use the following unless you *know* C<off> is within
571 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
572 on the first byte of character or just after the last byte of a character.
578 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
580 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
581 * the bitops (especially ~) can create illegal UTF-8.
582 * In other words: in Perl UTF-8 is not just for Unicode. */
591 while (UTF8_IS_CONTINUATION(*s))
599 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
601 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
602 Unlike C<bytes_to_utf8>, this over-writes the original string, and
603 updates len to contain the new length.
604 Returns zero on failure, setting C<len> to -1.
610 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
616 /* ensure valid UTF8 and chars < 256 before updating string */
617 for (send = s + *len; s < send; ) {
620 if (!UTF8_IS_INVARIANT(c) &&
621 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
622 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
631 *d++ = (U8)utf8_to_uvchr(s, &ulen);
640 =for apidoc A|U8 *|bytes_from_utf8|U8 *s|STRLEN *len|bool *is_utf8
642 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
643 Unlike <utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
644 the newly-created string, and updates C<len> to contain the new
645 length. Returns the original string if no conversion occurs, C<len>
646 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
647 0 if C<s> is converted or contains all 7bit characters.
653 Perl_bytes_from_utf8(pTHX_ U8 *s, STRLEN *len, bool *is_utf8)
663 /* ensure valid UTF8 and chars < 256 before converting string */
664 for (send = s + *len; s < send;) {
666 if (!UTF8_IS_INVARIANT(c)) {
667 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
668 (c = *s++) && UTF8_IS_CONTINUATION(c))
677 Newz(801, d, (*len) - count + 1, U8);
678 s = start; start = d;
681 if (!UTF8_IS_INVARIANT(c)) {
682 /* Then it is two-byte encoded */
683 c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++);
684 c = ASCII_TO_NATIVE(c);
694 =for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
696 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
697 Returns a pointer to the newly-created string, and sets C<len> to
698 reflect the new length.
704 Perl_bytes_to_utf8(pTHX_ U8 *s, STRLEN *len)
711 Newz(801, d, (*len) * 2 + 1, U8);
715 UV uv = NATIVE_TO_ASCII(*s++);
716 if (UNI_IS_INVARIANT(uv))
717 *d++ = UTF_TO_NATIVE(uv);
719 *d++ = UTF8_EIGHT_BIT_HI(uv);
720 *d++ = UTF8_EIGHT_BIT_LO(uv);
729 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
731 * Destination must be pre-extended to 3/2 source. Do not use in-place.
732 * We optimize for native, for obvious reasons. */
735 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
741 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
746 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
753 *d++ = (( uv >> 6) | 0xc0);
754 *d++ = (( uv & 0x3f) | 0x80);
757 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
759 if (low < 0xdc00 || low >= 0xdfff)
760 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
761 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
764 *d++ = (( uv >> 12) | 0xe0);
765 *d++ = (((uv >> 6) & 0x3f) | 0x80);
766 *d++ = (( uv & 0x3f) | 0x80);
770 *d++ = (( uv >> 18) | 0xf0);
771 *d++ = (((uv >> 12) & 0x3f) | 0x80);
772 *d++ = (((uv >> 6) & 0x3f) | 0x80);
773 *d++ = (( uv & 0x3f) | 0x80);
777 *newlen = d - dstart;
781 /* Note: this one is slightly destructive of the source. */
784 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
787 U8* send = s + bytelen;
794 return utf16_to_utf8(p, d, bytelen, newlen);
797 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
800 Perl_is_uni_alnum(pTHX_ UV c)
802 U8 tmpbuf[UTF8_MAXLEN+1];
803 uvchr_to_utf8(tmpbuf, (UV)c);
804 return is_utf8_alnum(tmpbuf);
808 Perl_is_uni_alnumc(pTHX_ UV c)
810 U8 tmpbuf[UTF8_MAXLEN+1];
811 uvchr_to_utf8(tmpbuf, (UV)c);
812 return is_utf8_alnumc(tmpbuf);
816 Perl_is_uni_idfirst(pTHX_ UV c)
818 U8 tmpbuf[UTF8_MAXLEN+1];
819 uvchr_to_utf8(tmpbuf, (UV)c);
820 return is_utf8_idfirst(tmpbuf);
824 Perl_is_uni_alpha(pTHX_ UV c)
826 U8 tmpbuf[UTF8_MAXLEN+1];
827 uvchr_to_utf8(tmpbuf, (UV)c);
828 return is_utf8_alpha(tmpbuf);
832 Perl_is_uni_ascii(pTHX_ UV c)
834 U8 tmpbuf[UTF8_MAXLEN+1];
835 uvchr_to_utf8(tmpbuf, (UV)c);
836 return is_utf8_ascii(tmpbuf);
840 Perl_is_uni_space(pTHX_ UV c)
842 U8 tmpbuf[UTF8_MAXLEN+1];
843 uvchr_to_utf8(tmpbuf, (UV)c);
844 return is_utf8_space(tmpbuf);
848 Perl_is_uni_digit(pTHX_ UV c)
850 U8 tmpbuf[UTF8_MAXLEN+1];
851 uvchr_to_utf8(tmpbuf, (UV)c);
852 return is_utf8_digit(tmpbuf);
856 Perl_is_uni_upper(pTHX_ UV c)
858 U8 tmpbuf[UTF8_MAXLEN+1];
859 uvchr_to_utf8(tmpbuf, (UV)c);
860 return is_utf8_upper(tmpbuf);
864 Perl_is_uni_lower(pTHX_ UV c)
866 U8 tmpbuf[UTF8_MAXLEN+1];
867 uvchr_to_utf8(tmpbuf, (UV)c);
868 return is_utf8_lower(tmpbuf);
872 Perl_is_uni_cntrl(pTHX_ UV c)
874 U8 tmpbuf[UTF8_MAXLEN+1];
875 uvchr_to_utf8(tmpbuf, (UV)c);
876 return is_utf8_cntrl(tmpbuf);
880 Perl_is_uni_graph(pTHX_ UV c)
882 U8 tmpbuf[UTF8_MAXLEN+1];
883 uvchr_to_utf8(tmpbuf, (UV)c);
884 return is_utf8_graph(tmpbuf);
888 Perl_is_uni_print(pTHX_ UV c)
890 U8 tmpbuf[UTF8_MAXLEN+1];
891 uvchr_to_utf8(tmpbuf, (UV)c);
892 return is_utf8_print(tmpbuf);
896 Perl_is_uni_punct(pTHX_ UV c)
898 U8 tmpbuf[UTF8_MAXLEN+1];
899 uvchr_to_utf8(tmpbuf, (UV)c);
900 return is_utf8_punct(tmpbuf);
904 Perl_is_uni_xdigit(pTHX_ UV c)
906 U8 tmpbuf[UTF8_MAXLEN*2+1];
907 uvchr_to_utf8(tmpbuf, (UV)c);
908 return is_utf8_xdigit(tmpbuf);
912 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
914 U8 tmpbuf[UTF8_MAXLEN*2+1];
915 uvchr_to_utf8(tmpbuf, (UV)c);
916 return to_utf8_upper(tmpbuf, p, lenp);
920 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
922 U8 tmpbuf[UTF8_MAXLEN*2+1];
923 uvchr_to_utf8(tmpbuf, (UV)c);
924 return to_utf8_title(tmpbuf, p, lenp);
928 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
930 U8 tmpbuf[UTF8_MAXLEN+1];
931 uvchr_to_utf8(tmpbuf, (UV)c);
932 return to_utf8_lower(tmpbuf, p, lenp);
936 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
938 U8 tmpbuf[UTF8_MAXLEN+1];
939 uvchr_to_utf8(tmpbuf, (UV)c);
940 return to_utf8_fold(tmpbuf, p, lenp);
943 /* for now these all assume no locale info available for Unicode > 255 */
946 Perl_is_uni_alnum_lc(pTHX_ UV c)
948 return is_uni_alnum(c); /* XXX no locale support yet */
952 Perl_is_uni_alnumc_lc(pTHX_ UV c)
954 return is_uni_alnumc(c); /* XXX no locale support yet */
958 Perl_is_uni_idfirst_lc(pTHX_ UV c)
960 return is_uni_idfirst(c); /* XXX no locale support yet */
964 Perl_is_uni_alpha_lc(pTHX_ UV c)
966 return is_uni_alpha(c); /* XXX no locale support yet */
970 Perl_is_uni_ascii_lc(pTHX_ UV c)
972 return is_uni_ascii(c); /* XXX no locale support yet */
976 Perl_is_uni_space_lc(pTHX_ UV c)
978 return is_uni_space(c); /* XXX no locale support yet */
982 Perl_is_uni_digit_lc(pTHX_ UV c)
984 return is_uni_digit(c); /* XXX no locale support yet */
988 Perl_is_uni_upper_lc(pTHX_ UV c)
990 return is_uni_upper(c); /* XXX no locale support yet */
994 Perl_is_uni_lower_lc(pTHX_ UV c)
996 return is_uni_lower(c); /* XXX no locale support yet */
1000 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1002 return is_uni_cntrl(c); /* XXX no locale support yet */
1006 Perl_is_uni_graph_lc(pTHX_ UV c)
1008 return is_uni_graph(c); /* XXX no locale support yet */
1012 Perl_is_uni_print_lc(pTHX_ UV c)
1014 return is_uni_print(c); /* XXX no locale support yet */
1018 Perl_is_uni_punct_lc(pTHX_ UV c)
1020 return is_uni_punct(c); /* XXX no locale support yet */
1024 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1026 return is_uni_xdigit(c); /* XXX no locale support yet */
1030 Perl_is_utf8_alnum(pTHX_ U8 *p)
1032 if (!is_utf8_char(p))
1035 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1036 * descendant of isalnum(3), in other words, it doesn't
1037 * contain the '_'. --jhi */
1038 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1039 return swash_fetch(PL_utf8_alnum, p, TRUE);
1040 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1041 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1043 PL_utf8_alnum = swash_init("utf8", "",
1044 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1045 return swash_fetch(PL_utf8_alnum, p, TRUE);
1050 Perl_is_utf8_alnumc(pTHX_ U8 *p)
1052 if (!is_utf8_char(p))
1055 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1056 return swash_fetch(PL_utf8_alnum, p, TRUE);
1057 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
1058 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1060 PL_utf8_alnum = swash_init("utf8", "",
1061 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1062 return swash_fetch(PL_utf8_alnum, p, TRUE);
1067 Perl_is_utf8_idfirst(pTHX_ U8 *p)
1069 return *p == '_' || is_utf8_alpha(p);
1073 Perl_is_utf8_alpha(pTHX_ U8 *p)
1075 if (!is_utf8_char(p))
1078 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1079 return swash_fetch(PL_utf8_alpha, p, TRUE);
1083 Perl_is_utf8_ascii(pTHX_ U8 *p)
1085 if (!is_utf8_char(p))
1088 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1089 return swash_fetch(PL_utf8_ascii, p, TRUE);
1093 Perl_is_utf8_space(pTHX_ U8 *p)
1095 if (!is_utf8_char(p))
1098 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1099 return swash_fetch(PL_utf8_space, p, TRUE);
1103 Perl_is_utf8_digit(pTHX_ U8 *p)
1105 if (!is_utf8_char(p))
1108 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1109 return swash_fetch(PL_utf8_digit, p, TRUE);
1113 Perl_is_utf8_upper(pTHX_ U8 *p)
1115 if (!is_utf8_char(p))
1118 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
1119 return swash_fetch(PL_utf8_upper, p, TRUE);
1123 Perl_is_utf8_lower(pTHX_ U8 *p)
1125 if (!is_utf8_char(p))
1128 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
1129 return swash_fetch(PL_utf8_lower, p, TRUE);
1133 Perl_is_utf8_cntrl(pTHX_ U8 *p)
1135 if (!is_utf8_char(p))
1138 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1139 return swash_fetch(PL_utf8_cntrl, p, TRUE);
1143 Perl_is_utf8_graph(pTHX_ U8 *p)
1145 if (!is_utf8_char(p))
1148 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1149 return swash_fetch(PL_utf8_graph, p, TRUE);
1153 Perl_is_utf8_print(pTHX_ U8 *p)
1155 if (!is_utf8_char(p))
1158 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1159 return swash_fetch(PL_utf8_print, p, TRUE);
1163 Perl_is_utf8_punct(pTHX_ U8 *p)
1165 if (!is_utf8_char(p))
1168 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1169 return swash_fetch(PL_utf8_punct, p, TRUE);
1173 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1175 if (!is_utf8_char(p))
1177 if (!PL_utf8_xdigit)
1178 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1179 return swash_fetch(PL_utf8_xdigit, p, TRUE);
1183 Perl_is_utf8_mark(pTHX_ U8 *p)
1185 if (!is_utf8_char(p))
1188 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1189 return swash_fetch(PL_utf8_mark, p, TRUE);
1193 =for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special
1195 The "p" contains the pointer to the UTF-8 string encoding
1196 the character that is being converted.
1198 The "ustrp" is a pointer to the character buffer to put the
1199 conversion result to. The "lenp" is a pointer to the length
1202 The "swash" is a pointer to the swash to use.
1204 The "normal" is a string like "ToLower" which means the swash
1205 $utf8::ToLower, which is stored in lib/unicore/To/Lower.pl,
1206 and loaded by SWASHGET, using lib/utf8_heavy.pl.
1208 The "special" is a string like "utf8::ToSpecLower", which means
1209 the hash %utf8::ToSpecLower, which is stored in the same file,
1210 lib/unicore/To/Lower.pl, and also loaded by SWASHGET. The access
1211 to the hash is by Perl_to_utf8_case().
1217 Perl_to_utf8_case(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp,char *normal, char *special)
1222 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1223 uv = swash_fetch(*swashp, p, TRUE);
1225 uv = UNI_TO_NATIVE(uv);
1231 uv = utf8_to_uvchr(p, 0);
1233 if ((hv = get_hv(special, FALSE)) &&
1234 (keysv = sv_2mortal(Perl_newSVpvf(aTHX_ "%04"UVXf, uv))) &&
1235 (he = hv_fetch_ent(hv, keysv, FALSE, 0))) {
1236 SV *val = HeVAL(he);
1237 char *s = SvPV(val, *lenp);
1239 if (*lenp > 1 || UNI_IS_INVARIANT(c))
1240 Copy(s, ustrp, *lenp, U8);
1242 /* something in the 0x80..0xFF range */
1243 ustrp[0] = UTF8_EIGHT_BIT_HI(c);
1244 ustrp[1] = UTF8_EIGHT_BIT_LO(c);
1250 *lenp = UNISKIP(uv);
1251 uvuni_to_utf8(ustrp, uv);
1256 Perl_to_utf8_upper(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1258 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1259 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1263 Perl_to_utf8_title(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1265 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1266 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1270 Perl_to_utf8_lower(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1272 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1273 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1277 Perl_to_utf8_fold(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1279 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1280 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1283 /* a "swash" is a swatch hash */
1286 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1289 SV* tokenbufsv = sv_2mortal(NEWSV(0,0));
1291 HV *stash = gv_stashpvn(pkg, strlen(pkg), FALSE);
1294 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
1296 errsv_save = newSVsv(ERRSV);
1297 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1299 sv_setsv(ERRSV, errsv_save);
1300 SvREFCNT_dec(errsv_save);
1304 PUSHSTACKi(PERLSI_MAGIC);
1307 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1308 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1310 PUSHs(sv_2mortal(newSViv(minbits)));
1311 PUSHs(sv_2mortal(newSViv(none)));
1317 if (PL_curcop == &PL_compiling)
1318 /* XXX ought to be handled by lex_start */
1319 sv_setpv(tokenbufsv, PL_tokenbuf);
1320 errsv_save = newSVsv(ERRSV);
1321 if (call_method("SWASHNEW", G_SCALAR))
1322 retval = newSVsv(*PL_stack_sp--);
1324 retval = &PL_sv_undef;
1326 sv_setsv(ERRSV, errsv_save);
1327 SvREFCNT_dec(errsv_save);
1330 if (PL_curcop == &PL_compiling) {
1332 char* pv = SvPV(tokenbufsv, len);
1334 Copy(pv, PL_tokenbuf, len+1, char);
1335 PL_curcop->op_private = PL_hints;
1337 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1338 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1343 /* This API is wrong for special case conversions since we may need to
1344 * return several Unicode characters for a single Unicode character
1345 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1346 * the lower-level routine, and it is similarly broken for returning
1347 * multiple values. --jhi */
1349 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr, bool do_utf8)
1351 HV* hv = (HV*)SvRV(sv);
1360 UV c = NATIVE_TO_ASCII(*ptr);
1362 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1363 tmputf8[0] = UTF8_EIGHT_BIT_HI(c);
1364 tmputf8[1] = UTF8_EIGHT_BIT_LO(c);
1367 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
1368 * then the "swatch" is a vec() for al the chars which start
1370 * So the key in the hash (klen) is length of encoded char -1
1372 klen = UTF8SKIP(ptr) - 1;
1377 /* If char in invariant then swatch is for all the invariant chars
1378 * In both UTF-8 and UTF8-MOD that happens to be UTF_CONTINUATION_MARK
1380 needents = UTF_CONTINUATION_MARK;
1381 off = NATIVE_TO_UTF(ptr[klen]);
1385 /* If char is encoded then swatch is for the prefix */
1386 needents = (1 << UTF_ACCUMULATION_SHIFT);
1387 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
1391 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1392 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1393 * it's nothing to sniff at.) Pity we usually come through at least
1394 * two function calls to get here...
1396 * NB: this code assumes that swatches are never modified, once generated!
1399 if (hv == PL_last_swash_hv &&
1400 klen == PL_last_swash_klen &&
1401 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
1403 tmps = PL_last_swash_tmps;
1404 slen = PL_last_swash_slen;
1407 /* Try our second-level swatch cache, kept in a hash. */
1408 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1410 /* If not cached, generate it via utf8::SWASHGET */
1411 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1413 /* We use utf8n_to_uvuni() as we want an index into
1414 Unicode tables, not a native character number.
1416 UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXLEN, NULL, 0);
1421 PUSHSTACKi(PERLSI_MAGIC);
1425 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
1426 PUSHs(sv_2mortal(newSViv((klen) ?
1427 (code_point & ~(needents - 1)) : 0)));
1428 PUSHs(sv_2mortal(newSViv(needents)));
1430 errsv_save = newSVsv(ERRSV);
1431 if (call_method("SWASHGET", G_SCALAR))
1432 retval = newSVsv(*PL_stack_sp--);
1434 retval = &PL_sv_undef;
1436 sv_setsv(ERRSV, errsv_save);
1437 SvREFCNT_dec(errsv_save);
1441 if (PL_curcop == &PL_compiling)
1442 PL_curcop->op_private = PL_hints;
1444 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1446 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || (slen << 3) < needents)
1447 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1450 PL_last_swash_hv = hv;
1451 PL_last_swash_klen = klen;
1452 PL_last_swash_tmps = tmps;
1453 PL_last_swash_slen = slen;
1455 Copy(ptr, PL_last_swash_key, klen, U8);
1458 switch ((int)((slen << 3) / needents)) {
1460 bit = 1 << (off & 7);
1462 return (tmps[off] & bit) != 0;
1467 return (tmps[off] << 8) + tmps[off + 1] ;
1470 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1472 Perl_croak(aTHX_ "panic: swash_fetch");
1478 =for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv
1480 Adds the UTF8 representation of the Native codepoint C<uv> to the end
1481 of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free
1482 bytes available. The return value is the pointer to the byte after the
1483 end of the new character. In other words,
1485 d = uvchr_to_utf8(d, uv);
1487 is the recommended wide native character-aware way of saying
1494 /* On ASCII machines this is normally a macro but we want a
1495 real function in case XS code wants it
1497 #undef Perl_uvchr_to_utf8
1499 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
1501 return Perl_uvuni_to_utf8(aTHX_ d, NATIVE_TO_UNI(uv));
1506 =for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
1508 Returns the native character value of the first character in the string C<s>
1509 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
1510 length, in bytes, of that character.
1512 Allows length and flags to be passed to low level routine.
1516 /* On ASCII machines this is normally a macro but we want a
1517 real function in case XS code wants it
1519 #undef Perl_utf8n_to_uvchr
1521 Perl_utf8n_to_uvchr(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
1523 UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
1524 return UNI_TO_NATIVE(uv);
1528 =for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags
1530 Build to the scalar dsv a displayable version of the string spv,
1531 length len, the displayable version being at most pvlim bytes long
1532 (if longer, the rest is truncated and "..." will be appended).
1533 The flags argument is currently unused but available for future extensions.
1534 The pointer to the PV of the dsv is returned.
1538 Perl_pv_uni_display(pTHX_ SV *dsv, U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
1543 sv_setpvn(dsv, "", 0);
1544 for (s = (char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
1546 if (pvlim && SvCUR(dsv) >= pvlim) {
1550 u = utf8_to_uvchr((U8*)s, 0);
1551 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
1554 sv_catpvn(dsv, "...", 3);
1560 =for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags
1562 Build to the scalar dsv a displayable version of the scalar sv,
1563 he displayable version being at most pvlim bytes long
1564 (if longer, the rest is truncated and "..." will be appended).
1565 The flags argument is currently unused but available for future extensions.
1566 The pointer to the PV of the dsv is returned.
1570 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
1572 return Perl_pv_uni_display(aTHX_ dsv, (U8*)SvPVX(ssv), SvCUR(ssv),
1577 =for apidoc A|I32|ibcmp_utf8|const char *s1|bool u1|const char *s2|bool u2|register I32 len
1579 Return true if the strings s1 and s2 differ case-insensitively, false
1580 if not (if they are equal case-insensitively). If u1 is true, the
1581 string s1 is assumed to be in UTF-8-encoded Unicode. If u2 is true,
1582 the string s2 is assumed to be in UTF-8-encoded Unicode. (If both u1
1583 and u2 are false, ibcmp() is called.)
1585 For case-insensitiveness, the "casefolding" of Unicode is used
1586 instead of upper/lowercasing both the characters, see
1587 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
1591 Perl_ibcmp_utf8(pTHX_ const char *s1, bool u1, const char *s2, bool u2, register I32 len)
1594 register U8 *a = (U8*)s1;
1595 register U8 *b = (U8*)s2;
1598 STRLEN ulen1, ulen2;
1599 U8 tmpbuf1[UTF8_MAXLEN*3+1];
1600 U8 tmpbuf2[UTF8_MAXLEN*3+1];
1604 ca = utf8_to_uvchr((U8*)a, &la);
1610 cb = utf8_to_uvchr((U8*)b, &lb);
1617 to_uni_fold(NATIVE_TO_UNI(ca), tmpbuf1, &ulen1);
1621 to_uni_fold(NATIVE_TO_UNI(cb), tmpbuf2, &ulen2);
1625 || (ulen1 == 1 && PL_fold[ca] != PL_fold[cb])
1626 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1))
1635 return ibcmp(s1, s2, len);