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);
53 if (UNICODE_IS_SURROGATE(uv))
54 Perl_croak(aTHX_ "UTF-16 surrogate 0x%04"UVxf, uv);
57 STRLEN len = UNISKIP(uv);
60 *p-- = UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
61 uv >>= UTF_ACCUMULATION_SHIFT;
63 *p = UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
66 #else /* Non loop style */
68 *d++ = (( uv >> 6) | 0xc0);
69 *d++ = (( uv & 0x3f) | 0x80);
73 *d++ = (( uv >> 12) | 0xe0);
74 *d++ = (((uv >> 6) & 0x3f) | 0x80);
75 *d++ = (( uv & 0x3f) | 0x80);
79 *d++ = (( uv >> 18) | 0xf0);
80 *d++ = (((uv >> 12) & 0x3f) | 0x80);
81 *d++ = (((uv >> 6) & 0x3f) | 0x80);
82 *d++ = (( uv & 0x3f) | 0x80);
86 *d++ = (( uv >> 24) | 0xf8);
87 *d++ = (((uv >> 18) & 0x3f) | 0x80);
88 *d++ = (((uv >> 12) & 0x3f) | 0x80);
89 *d++ = (((uv >> 6) & 0x3f) | 0x80);
90 *d++ = (( uv & 0x3f) | 0x80);
93 if (uv < 0x80000000) {
94 *d++ = (( uv >> 30) | 0xfc);
95 *d++ = (((uv >> 24) & 0x3f) | 0x80);
96 *d++ = (((uv >> 18) & 0x3f) | 0x80);
97 *d++ = (((uv >> 12) & 0x3f) | 0x80);
98 *d++ = (((uv >> 6) & 0x3f) | 0x80);
99 *d++ = (( uv & 0x3f) | 0x80);
103 if (uv < UTF8_QUAD_MAX)
106 *d++ = 0xfe; /* Can't match U+FEFF! */
107 *d++ = (((uv >> 30) & 0x3f) | 0x80);
108 *d++ = (((uv >> 24) & 0x3f) | 0x80);
109 *d++ = (((uv >> 18) & 0x3f) | 0x80);
110 *d++ = (((uv >> 12) & 0x3f) | 0x80);
111 *d++ = (((uv >> 6) & 0x3f) | 0x80);
112 *d++ = (( uv & 0x3f) | 0x80);
117 *d++ = 0xff; /* Can't match U+FFFE! */
118 *d++ = 0x80; /* 6 Reserved bits */
119 *d++ = (((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
120 *d++ = (((uv >> 54) & 0x3f) | 0x80);
121 *d++ = (((uv >> 48) & 0x3f) | 0x80);
122 *d++ = (((uv >> 42) & 0x3f) | 0x80);
123 *d++ = (((uv >> 36) & 0x3f) | 0x80);
124 *d++ = (((uv >> 30) & 0x3f) | 0x80);
125 *d++ = (((uv >> 24) & 0x3f) | 0x80);
126 *d++ = (((uv >> 18) & 0x3f) | 0x80);
127 *d++ = (((uv >> 12) & 0x3f) | 0x80);
128 *d++ = (((uv >> 6) & 0x3f) | 0x80);
129 *d++ = (( uv & 0x3f) | 0x80);
133 #endif /* Loop style */
139 =for apidoc A|STRLEN|is_utf8_char|U8 *s
141 Tests if some arbitrary number of bytes begins in a valid UTF-8
142 character. Note that an INVARIANT (i.e. ASCII) character is a valid UTF-8 character.
143 The actual number of bytes in the UTF-8 character will be returned if
144 it is valid, otherwise 0.
149 Perl_is_utf8_char(pTHX_ U8 *s)
155 if (UTF8_IS_INVARIANT(u))
158 if (!UTF8_IS_START(u))
163 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
168 u &= UTF_START_MASK(len);
172 if (!UTF8_IS_CONTINUATION(*s))
174 uv = UTF8_ACCUMULATE(uv, *s);
181 if (UNISKIP(uv) < len)
188 =for apidoc A|bool|is_utf8_string|U8 *s|STRLEN len
190 Returns true if first C<len> bytes of the given string form a valid UTF8
191 string, false otherwise. Note that 'a valid UTF8 string' does not mean
192 'a string that contains UTF8' because a valid ASCII string is a valid
199 Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len)
206 len = strlen((char *)s);
222 =for apidoc A|UV|utf8n_to_uvuni|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
224 Bottom level UTF-8 decode routine.
225 Returns the unicode code point value of the first character in the string C<s>
226 which is assumed to be in UTF8 encoding and no longer than C<curlen>;
227 C<retlen> will be set to the length, in bytes, of that character.
229 If C<s> does not point to a well-formed UTF8 character, the behaviour
230 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
231 it is assumed that the caller will raise a warning, and this function
232 will silently just set C<retlen> to C<-1> and return zero. If the
233 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
234 malformations will be given, C<retlen> will be set to the expected
235 length of the UTF-8 character in bytes, and zero will be returned.
237 The C<flags> can also contain various flags to allow deviations from
238 the strict UTF-8 encoding (see F<utf8.h>).
240 Most code should use utf8_to_uvchr() rather than call this directly.
246 Perl_utf8n_to_uvuni(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
250 bool dowarn = ckWARN_d(WARN_UTF8);
251 STRLEN expectlen = 0;
254 /* This list is a superset of the UTF8_ALLOW_XXX. */
256 #define UTF8_WARN_EMPTY 1
257 #define UTF8_WARN_CONTINUATION 2
258 #define UTF8_WARN_NON_CONTINUATION 3
259 #define UTF8_WARN_FE_FF 4
260 #define UTF8_WARN_SHORT 5
261 #define UTF8_WARN_OVERFLOW 6
262 #define UTF8_WARN_SURROGATE 7
263 #define UTF8_WARN_BOM 8
264 #define UTF8_WARN_LONG 9
265 #define UTF8_WARN_FFFF 10
268 !(flags & UTF8_ALLOW_EMPTY)) {
269 warning = UTF8_WARN_EMPTY;
273 if (UTF8_IS_INVARIANT(uv)) {
276 return (UV) (NATIVE_TO_UTF(*s));
279 if (UTF8_IS_CONTINUATION(uv) &&
280 !(flags & UTF8_ALLOW_CONTINUATION)) {
281 warning = UTF8_WARN_CONTINUATION;
285 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
286 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
287 warning = UTF8_WARN_NON_CONTINUATION;
292 uv = NATIVE_TO_UTF(uv);
294 if ((uv == 0xfe || uv == 0xff) &&
295 !(flags & UTF8_ALLOW_FE_FF)) {
296 warning = UTF8_WARN_FE_FF;
301 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
302 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
303 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
304 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
306 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
307 else { len = 7; uv &= 0x01; }
309 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
310 else if (!(uv & 0x01)) { len = 7; uv = 0; }
311 else { len = 13; uv = 0; } /* whoa! */
319 if ((curlen < expectlen) &&
320 !(flags & UTF8_ALLOW_SHORT)) {
321 warning = UTF8_WARN_SHORT;
330 if (!UTF8_IS_CONTINUATION(*s) &&
331 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
333 warning = UTF8_WARN_NON_CONTINUATION;
337 uv = UTF8_ACCUMULATE(uv, *s);
339 /* These cannot be allowed. */
341 if (!(flags & UTF8_ALLOW_LONG)) {
342 warning = UTF8_WARN_LONG;
346 else { /* uv < ouv */
347 /* This cannot be allowed. */
348 warning = UTF8_WARN_OVERFLOW;
356 if (UNICODE_IS_SURROGATE(uv) &&
357 !(flags & UTF8_ALLOW_SURROGATE)) {
358 warning = UTF8_WARN_SURROGATE;
360 } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) &&
361 !(flags & UTF8_ALLOW_BOM)) {
362 warning = UTF8_WARN_BOM;
364 } else if ((expectlen > UNISKIP(uv)) &&
365 !(flags & UTF8_ALLOW_LONG)) {
366 warning = UTF8_WARN_LONG;
368 } else if (UNICODE_IS_ILLEGAL(uv) &&
369 !(flags & UTF8_ALLOW_FFFF)) {
370 warning = UTF8_WARN_FFFF;
378 if (flags & UTF8_CHECK_ONLY) {
385 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
388 case 0: /* Intentionally empty. */ break;
389 case UTF8_WARN_EMPTY:
390 Perl_sv_catpvf(aTHX_ sv, "(empty string)");
392 case UTF8_WARN_CONTINUATION:
393 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf")", uv);
395 case UTF8_WARN_NON_CONTINUATION:
396 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")",
399 case UTF8_WARN_FE_FF:
400 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
402 case UTF8_WARN_SHORT:
403 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
404 curlen, curlen == 1 ? "" : "s", expectlen);
405 expectlen = curlen; /* distance for caller to skip */
407 case UTF8_WARN_OVERFLOW:
408 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x)",
411 case UTF8_WARN_SURROGATE:
412 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
415 Perl_sv_catpvf(aTHX_ sv, "(byte order mark 0x%04"UVxf")", uv);
418 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
419 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
422 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
425 Perl_sv_catpvf(aTHX_ sv, "(unknown reason)");
433 Perl_warner(aTHX_ WARN_UTF8,
434 "%s in %s", s, OP_DESC(PL_op));
436 Perl_warner(aTHX_ WARN_UTF8, "%s", s);
441 *retlen = expectlen ? expectlen : len;
447 =for apidoc A|UV|utf8_to_uvchr|U8 *s|STRLEN *retlen
449 Returns the native character value of the first character in the string C<s>
450 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
451 length, in bytes, of that character.
453 If C<s> does not point to a well-formed UTF8 character, zero is
454 returned and retlen is set, if possible, to -1.
460 Perl_utf8_to_uvchr(pTHX_ U8 *s, STRLEN *retlen)
462 return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXLEN, retlen, 0);
466 =for apidoc A|UV|utf8_to_uvuni|U8 *s|STRLEN *retlen
468 Returns the Unicode code point of the first character in the string C<s>
469 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
470 length, in bytes, of that character.
472 This function should only be used when returned UV is considered
473 an index into the Unicode semantic tables (e.g. swashes).
475 If C<s> does not point to a well-formed UTF8 character, zero is
476 returned and retlen is set, if possible, to -1.
482 Perl_utf8_to_uvuni(pTHX_ U8 *s, STRLEN *retlen)
484 /* Call the low level routine asking for checks */
485 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXLEN, retlen, 0);
489 =for apidoc A|STRLEN|utf8_length|U8 *s|U8 *e
491 Return the length of the UTF-8 char encoded string C<s> in characters.
492 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
493 up past C<e>, croaks.
499 Perl_utf8_length(pTHX_ U8 *s, U8 *e)
503 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
504 * the bitops (especially ~) can create illegal UTF-8.
505 * In other words: in Perl UTF-8 is not just for Unicode. */
508 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
513 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
522 =for apidoc A|IV|utf8_distance|U8 *a|U8 *b
524 Returns the number of UTF8 characters between the UTF-8 pointers C<a>
527 WARNING: use only if you *know* that the pointers point inside the
534 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
538 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
539 * the bitops (especially ~) can create illegal UTF-8.
540 * In other words: in Perl UTF-8 is not just for Unicode. */
547 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
557 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
567 =for apidoc A|U8 *|utf8_hop|U8 *s|I32 off
569 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
572 WARNING: do not use the following unless you *know* C<off> is within
573 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
574 on the first byte of character or just after the last byte of a character.
580 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
582 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
583 * the bitops (especially ~) can create illegal UTF-8.
584 * In other words: in Perl UTF-8 is not just for Unicode. */
593 while (UTF8_IS_CONTINUATION(*s))
601 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
603 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
604 Unlike C<bytes_to_utf8>, this over-writes the original string, and
605 updates len to contain the new length.
606 Returns zero on failure, setting C<len> to -1.
612 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
618 /* ensure valid UTF8 and chars < 256 before updating string */
619 for (send = s + *len; s < send; ) {
622 if (!UTF8_IS_INVARIANT(c) &&
623 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
624 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
633 *d++ = (U8)utf8_to_uvchr(s, &ulen);
642 =for apidoc A|U8 *|bytes_from_utf8|U8 *s|STRLEN *len|bool *is_utf8
644 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
645 Unlike <utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
646 the newly-created string, and updates C<len> to contain the new
647 length. Returns the original string if no conversion occurs, C<len>
648 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
649 0 if C<s> is converted or contains all 7bit characters.
655 Perl_bytes_from_utf8(pTHX_ U8 *s, STRLEN *len, bool *is_utf8)
665 /* ensure valid UTF8 and chars < 256 before converting string */
666 for (send = s + *len; s < send;) {
668 if (!UTF8_IS_INVARIANT(c)) {
669 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
670 (c = *s++) && UTF8_IS_CONTINUATION(c))
679 Newz(801, d, (*len) - count + 1, U8);
680 s = start; start = d;
683 if (!UTF8_IS_INVARIANT(c)) {
684 /* Then it is two-byte encoded */
685 c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++);
686 c = ASCII_TO_NATIVE(c);
696 =for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
698 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
699 Returns a pointer to the newly-created string, and sets C<len> to
700 reflect the new length.
706 Perl_bytes_to_utf8(pTHX_ U8 *s, STRLEN *len)
713 Newz(801, d, (*len) * 2 + 1, U8);
717 UV uv = NATIVE_TO_ASCII(*s++);
718 if (UNI_IS_INVARIANT(uv))
719 *d++ = UTF_TO_NATIVE(uv);
721 *d++ = UTF8_EIGHT_BIT_HI(uv);
722 *d++ = UTF8_EIGHT_BIT_LO(uv);
731 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
733 * Destination must be pre-extended to 3/2 source. Do not use in-place.
734 * We optimize for native, for obvious reasons. */
737 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
743 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
748 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
755 *d++ = (( uv >> 6) | 0xc0);
756 *d++ = (( uv & 0x3f) | 0x80);
759 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
761 if (low < 0xdc00 || low >= 0xdfff)
762 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
763 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
766 *d++ = (( uv >> 12) | 0xe0);
767 *d++ = (((uv >> 6) & 0x3f) | 0x80);
768 *d++ = (( uv & 0x3f) | 0x80);
772 *d++ = (( uv >> 18) | 0xf0);
773 *d++ = (((uv >> 12) & 0x3f) | 0x80);
774 *d++ = (((uv >> 6) & 0x3f) | 0x80);
775 *d++ = (( uv & 0x3f) | 0x80);
779 *newlen = d - dstart;
783 /* Note: this one is slightly destructive of the source. */
786 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
789 U8* send = s + bytelen;
796 return utf16_to_utf8(p, d, bytelen, newlen);
799 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
802 Perl_is_uni_alnum(pTHX_ UV c)
804 U8 tmpbuf[UTF8_MAXLEN+1];
805 uvchr_to_utf8(tmpbuf, (UV)c);
806 return is_utf8_alnum(tmpbuf);
810 Perl_is_uni_alnumc(pTHX_ UV c)
812 U8 tmpbuf[UTF8_MAXLEN+1];
813 uvchr_to_utf8(tmpbuf, (UV)c);
814 return is_utf8_alnumc(tmpbuf);
818 Perl_is_uni_idfirst(pTHX_ UV c)
820 U8 tmpbuf[UTF8_MAXLEN+1];
821 uvchr_to_utf8(tmpbuf, (UV)c);
822 return is_utf8_idfirst(tmpbuf);
826 Perl_is_uni_alpha(pTHX_ UV c)
828 U8 tmpbuf[UTF8_MAXLEN+1];
829 uvchr_to_utf8(tmpbuf, (UV)c);
830 return is_utf8_alpha(tmpbuf);
834 Perl_is_uni_ascii(pTHX_ UV c)
836 U8 tmpbuf[UTF8_MAXLEN+1];
837 uvchr_to_utf8(tmpbuf, (UV)c);
838 return is_utf8_ascii(tmpbuf);
842 Perl_is_uni_space(pTHX_ UV c)
844 U8 tmpbuf[UTF8_MAXLEN+1];
845 uvchr_to_utf8(tmpbuf, (UV)c);
846 return is_utf8_space(tmpbuf);
850 Perl_is_uni_digit(pTHX_ UV c)
852 U8 tmpbuf[UTF8_MAXLEN+1];
853 uvchr_to_utf8(tmpbuf, (UV)c);
854 return is_utf8_digit(tmpbuf);
858 Perl_is_uni_upper(pTHX_ UV c)
860 U8 tmpbuf[UTF8_MAXLEN+1];
861 uvchr_to_utf8(tmpbuf, (UV)c);
862 return is_utf8_upper(tmpbuf);
866 Perl_is_uni_lower(pTHX_ UV c)
868 U8 tmpbuf[UTF8_MAXLEN+1];
869 uvchr_to_utf8(tmpbuf, (UV)c);
870 return is_utf8_lower(tmpbuf);
874 Perl_is_uni_cntrl(pTHX_ UV c)
876 U8 tmpbuf[UTF8_MAXLEN+1];
877 uvchr_to_utf8(tmpbuf, (UV)c);
878 return is_utf8_cntrl(tmpbuf);
882 Perl_is_uni_graph(pTHX_ UV c)
884 U8 tmpbuf[UTF8_MAXLEN+1];
885 uvchr_to_utf8(tmpbuf, (UV)c);
886 return is_utf8_graph(tmpbuf);
890 Perl_is_uni_print(pTHX_ UV c)
892 U8 tmpbuf[UTF8_MAXLEN+1];
893 uvchr_to_utf8(tmpbuf, (UV)c);
894 return is_utf8_print(tmpbuf);
898 Perl_is_uni_punct(pTHX_ UV c)
900 U8 tmpbuf[UTF8_MAXLEN+1];
901 uvchr_to_utf8(tmpbuf, (UV)c);
902 return is_utf8_punct(tmpbuf);
906 Perl_is_uni_xdigit(pTHX_ UV c)
908 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
909 uvchr_to_utf8(tmpbuf, (UV)c);
910 return is_utf8_xdigit(tmpbuf);
914 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
916 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
917 uvchr_to_utf8(tmpbuf, (UV)c);
918 return to_utf8_upper(tmpbuf, p, lenp);
922 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
924 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
925 uvchr_to_utf8(tmpbuf, (UV)c);
926 return to_utf8_title(tmpbuf, p, lenp);
930 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
932 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
933 uvchr_to_utf8(tmpbuf, (UV)c);
934 return to_utf8_lower(tmpbuf, p, lenp);
938 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
940 U8 tmpbuf[UTF8_MAXLEN_FOLD+1];
941 uvchr_to_utf8(tmpbuf, (UV)c);
942 return to_utf8_fold(tmpbuf, p, lenp);
945 /* for now these all assume no locale info available for Unicode > 255 */
948 Perl_is_uni_alnum_lc(pTHX_ UV c)
950 return is_uni_alnum(c); /* XXX no locale support yet */
954 Perl_is_uni_alnumc_lc(pTHX_ UV c)
956 return is_uni_alnumc(c); /* XXX no locale support yet */
960 Perl_is_uni_idfirst_lc(pTHX_ UV c)
962 return is_uni_idfirst(c); /* XXX no locale support yet */
966 Perl_is_uni_alpha_lc(pTHX_ UV c)
968 return is_uni_alpha(c); /* XXX no locale support yet */
972 Perl_is_uni_ascii_lc(pTHX_ UV c)
974 return is_uni_ascii(c); /* XXX no locale support yet */
978 Perl_is_uni_space_lc(pTHX_ UV c)
980 return is_uni_space(c); /* XXX no locale support yet */
984 Perl_is_uni_digit_lc(pTHX_ UV c)
986 return is_uni_digit(c); /* XXX no locale support yet */
990 Perl_is_uni_upper_lc(pTHX_ UV c)
992 return is_uni_upper(c); /* XXX no locale support yet */
996 Perl_is_uni_lower_lc(pTHX_ UV c)
998 return is_uni_lower(c); /* XXX no locale support yet */
1002 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1004 return is_uni_cntrl(c); /* XXX no locale support yet */
1008 Perl_is_uni_graph_lc(pTHX_ UV c)
1010 return is_uni_graph(c); /* XXX no locale support yet */
1014 Perl_is_uni_print_lc(pTHX_ UV c)
1016 return is_uni_print(c); /* XXX no locale support yet */
1020 Perl_is_uni_punct_lc(pTHX_ UV c)
1022 return is_uni_punct(c); /* XXX no locale support yet */
1026 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1028 return is_uni_xdigit(c); /* XXX no locale support yet */
1032 Perl_is_utf8_alnum(pTHX_ U8 *p)
1034 if (!is_utf8_char(p))
1037 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1038 * descendant of isalnum(3), in other words, it doesn't
1039 * contain the '_'. --jhi */
1040 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1041 return swash_fetch(PL_utf8_alnum, p, TRUE);
1042 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1043 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1045 PL_utf8_alnum = swash_init("utf8", "",
1046 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1047 return swash_fetch(PL_utf8_alnum, p, TRUE);
1052 Perl_is_utf8_alnumc(pTHX_ U8 *p)
1054 if (!is_utf8_char(p))
1057 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1058 return swash_fetch(PL_utf8_alnum, p, TRUE);
1059 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
1060 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1062 PL_utf8_alnum = swash_init("utf8", "",
1063 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1064 return swash_fetch(PL_utf8_alnum, p, TRUE);
1069 Perl_is_utf8_idfirst(pTHX_ U8 *p)
1071 return *p == '_' || is_utf8_alpha(p);
1075 Perl_is_utf8_alpha(pTHX_ U8 *p)
1077 if (!is_utf8_char(p))
1080 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1081 return swash_fetch(PL_utf8_alpha, p, TRUE);
1085 Perl_is_utf8_ascii(pTHX_ U8 *p)
1087 if (!is_utf8_char(p))
1090 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1091 return swash_fetch(PL_utf8_ascii, p, TRUE);
1095 Perl_is_utf8_space(pTHX_ U8 *p)
1097 if (!is_utf8_char(p))
1100 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1101 return swash_fetch(PL_utf8_space, p, TRUE);
1105 Perl_is_utf8_digit(pTHX_ U8 *p)
1107 if (!is_utf8_char(p))
1110 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1111 return swash_fetch(PL_utf8_digit, p, TRUE);
1115 Perl_is_utf8_upper(pTHX_ U8 *p)
1117 if (!is_utf8_char(p))
1120 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
1121 return swash_fetch(PL_utf8_upper, p, TRUE);
1125 Perl_is_utf8_lower(pTHX_ U8 *p)
1127 if (!is_utf8_char(p))
1130 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
1131 return swash_fetch(PL_utf8_lower, p, TRUE);
1135 Perl_is_utf8_cntrl(pTHX_ U8 *p)
1137 if (!is_utf8_char(p))
1140 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1141 return swash_fetch(PL_utf8_cntrl, p, TRUE);
1145 Perl_is_utf8_graph(pTHX_ U8 *p)
1147 if (!is_utf8_char(p))
1150 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1151 return swash_fetch(PL_utf8_graph, p, TRUE);
1155 Perl_is_utf8_print(pTHX_ U8 *p)
1157 if (!is_utf8_char(p))
1160 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1161 return swash_fetch(PL_utf8_print, p, TRUE);
1165 Perl_is_utf8_punct(pTHX_ U8 *p)
1167 if (!is_utf8_char(p))
1170 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1171 return swash_fetch(PL_utf8_punct, p, TRUE);
1175 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1177 if (!is_utf8_char(p))
1179 if (!PL_utf8_xdigit)
1180 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1181 return swash_fetch(PL_utf8_xdigit, p, TRUE);
1185 Perl_is_utf8_mark(pTHX_ U8 *p)
1187 if (!is_utf8_char(p))
1190 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1191 return swash_fetch(PL_utf8_mark, p, TRUE);
1195 =for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special
1197 The "p" contains the pointer to the UTF-8 string encoding
1198 the character that is being converted.
1200 The "ustrp" is a pointer to the character buffer to put the
1201 conversion result to. The "lenp" is a pointer to the length
1204 The "swash" is a pointer to the swash to use.
1206 The "normal" is a string like "ToLower" which means the swash
1207 $utf8::ToLower, which is stored in lib/unicore/To/Lower.pl,
1208 and loaded by SWASHGET, using lib/utf8_heavy.pl.
1210 The "special" is a string like "utf8::ToSpecLower", which means
1211 the hash %utf8::ToSpecLower, which is stored in the same file,
1212 lib/unicore/To/Lower.pl, and also loaded by SWASHGET. The access
1213 to the hash is by Perl_to_utf8_case().
1219 Perl_to_utf8_case(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp,char *normal, char *special)
1224 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1225 uv = swash_fetch(*swashp, p, TRUE);
1227 uv = UNI_TO_NATIVE(uv);
1233 uv = utf8_to_uvchr(p, 0);
1235 if ((hv = get_hv(special, FALSE)) &&
1236 (keysv = sv_2mortal(Perl_newSVpvf(aTHX_ "%04"UVXf, uv))) &&
1237 (he = hv_fetch_ent(hv, keysv, FALSE, 0))) {
1238 SV *val = HeVAL(he);
1239 char *s = SvPV(val, *lenp);
1241 if (*lenp > 1 || UNI_IS_INVARIANT(c))
1242 Copy(s, ustrp, *lenp, U8);
1244 /* something in the 0x80..0xFF range */
1245 ustrp[0] = UTF8_EIGHT_BIT_HI(c);
1246 ustrp[1] = UTF8_EIGHT_BIT_LO(c);
1253 *lenp = UNISKIP(uv);
1254 uvuni_to_utf8(ustrp, uv);
1259 Perl_to_utf8_upper(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1261 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1262 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1266 Perl_to_utf8_title(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1268 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1269 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1273 Perl_to_utf8_lower(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1275 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1276 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1280 Perl_to_utf8_fold(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1282 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1283 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1286 /* a "swash" is a swatch hash */
1289 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1292 SV* tokenbufsv = sv_2mortal(NEWSV(0,0));
1294 HV *stash = gv_stashpvn(pkg, strlen(pkg), FALSE);
1297 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
1299 errsv_save = newSVsv(ERRSV);
1300 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1302 sv_setsv(ERRSV, errsv_save);
1303 SvREFCNT_dec(errsv_save);
1307 PUSHSTACKi(PERLSI_MAGIC);
1310 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1311 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1313 PUSHs(sv_2mortal(newSViv(minbits)));
1314 PUSHs(sv_2mortal(newSViv(none)));
1320 if (PL_curcop == &PL_compiling)
1321 /* XXX ought to be handled by lex_start */
1322 sv_setpv(tokenbufsv, PL_tokenbuf);
1323 errsv_save = newSVsv(ERRSV);
1324 if (call_method("SWASHNEW", G_SCALAR))
1325 retval = newSVsv(*PL_stack_sp--);
1327 retval = &PL_sv_undef;
1329 sv_setsv(ERRSV, errsv_save);
1330 SvREFCNT_dec(errsv_save);
1333 if (PL_curcop == &PL_compiling) {
1335 char* pv = SvPV(tokenbufsv, len);
1337 Copy(pv, PL_tokenbuf, len+1, char);
1338 PL_curcop->op_private = PL_hints;
1340 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
1341 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1346 /* This API is wrong for special case conversions since we may need to
1347 * return several Unicode characters for a single Unicode character
1348 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1349 * the lower-level routine, and it is similarly broken for returning
1350 * multiple values. --jhi */
1352 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr, bool do_utf8)
1354 HV* hv = (HV*)SvRV(sv);
1363 UV c = NATIVE_TO_ASCII(*ptr);
1365 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1366 tmputf8[0] = UTF8_EIGHT_BIT_HI(c);
1367 tmputf8[1] = UTF8_EIGHT_BIT_LO(c);
1370 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
1371 * then the "swatch" is a vec() for al the chars which start
1373 * So the key in the hash (klen) is length of encoded char -1
1375 klen = UTF8SKIP(ptr) - 1;
1380 /* If char in invariant then swatch is for all the invariant chars
1381 * In both UTF-8 and UTF8-MOD that happens to be UTF_CONTINUATION_MARK
1383 needents = UTF_CONTINUATION_MARK;
1384 off = NATIVE_TO_UTF(ptr[klen]);
1388 /* If char is encoded then swatch is for the prefix */
1389 needents = (1 << UTF_ACCUMULATION_SHIFT);
1390 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
1394 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1395 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1396 * it's nothing to sniff at.) Pity we usually come through at least
1397 * two function calls to get here...
1399 * NB: this code assumes that swatches are never modified, once generated!
1402 if (hv == PL_last_swash_hv &&
1403 klen == PL_last_swash_klen &&
1404 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
1406 tmps = PL_last_swash_tmps;
1407 slen = PL_last_swash_slen;
1410 /* Try our second-level swatch cache, kept in a hash. */
1411 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1413 /* If not cached, generate it via utf8::SWASHGET */
1414 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1416 /* We use utf8n_to_uvuni() as we want an index into
1417 Unicode tables, not a native character number.
1419 UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXLEN, NULL, 0);
1424 PUSHSTACKi(PERLSI_MAGIC);
1428 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
1429 PUSHs(sv_2mortal(newSViv((klen) ?
1430 (code_point & ~(needents - 1)) : 0)));
1431 PUSHs(sv_2mortal(newSViv(needents)));
1433 errsv_save = newSVsv(ERRSV);
1434 if (call_method("SWASHGET", G_SCALAR))
1435 retval = newSVsv(*PL_stack_sp--);
1437 retval = &PL_sv_undef;
1439 sv_setsv(ERRSV, errsv_save);
1440 SvREFCNT_dec(errsv_save);
1444 if (PL_curcop == &PL_compiling)
1445 PL_curcop->op_private = PL_hints;
1447 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1449 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || (slen << 3) < needents)
1450 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1453 PL_last_swash_hv = hv;
1454 PL_last_swash_klen = klen;
1455 PL_last_swash_tmps = tmps;
1456 PL_last_swash_slen = slen;
1458 Copy(ptr, PL_last_swash_key, klen, U8);
1461 switch ((int)((slen << 3) / needents)) {
1463 bit = 1 << (off & 7);
1465 return (tmps[off] & bit) != 0;
1470 return (tmps[off] << 8) + tmps[off + 1] ;
1473 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1475 Perl_croak(aTHX_ "panic: swash_fetch");
1481 =for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv
1483 Adds the UTF8 representation of the Native codepoint C<uv> to the end
1484 of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free
1485 bytes available. The return value is the pointer to the byte after the
1486 end of the new character. In other words,
1488 d = uvchr_to_utf8(d, uv);
1490 is the recommended wide native character-aware way of saying
1497 /* On ASCII machines this is normally a macro but we want a
1498 real function in case XS code wants it
1500 #undef Perl_uvchr_to_utf8
1502 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
1504 return Perl_uvuni_to_utf8(aTHX_ d, NATIVE_TO_UNI(uv));
1509 =for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
1511 Returns the native character value of the first character in the string C<s>
1512 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
1513 length, in bytes, of that character.
1515 Allows length and flags to be passed to low level routine.
1519 /* On ASCII machines this is normally a macro but we want a
1520 real function in case XS code wants it
1522 #undef Perl_utf8n_to_uvchr
1524 Perl_utf8n_to_uvchr(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
1526 UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
1527 return UNI_TO_NATIVE(uv);
1531 =for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags
1533 Build to the scalar dsv a displayable version of the string spv,
1534 length len, the displayable version being at most pvlim bytes long
1535 (if longer, the rest is truncated and "..." will be appended).
1536 The flags argument is currently unused but available for future extensions.
1537 The pointer to the PV of the dsv is returned.
1541 Perl_pv_uni_display(pTHX_ SV *dsv, U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
1546 sv_setpvn(dsv, "", 0);
1547 for (s = (char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
1549 if (pvlim && SvCUR(dsv) >= pvlim) {
1553 u = utf8_to_uvchr((U8*)s, 0);
1554 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
1557 sv_catpvn(dsv, "...", 3);
1563 =for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags
1565 Build to the scalar dsv a displayable version of the scalar sv,
1566 he displayable version being at most pvlim bytes long
1567 (if longer, the rest is truncated and "..." will be appended).
1568 The flags argument is currently unused but available for future extensions.
1569 The pointer to the PV of the dsv is returned.
1573 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
1575 return Perl_pv_uni_display(aTHX_ dsv, (U8*)SvPVX(ssv), SvCUR(ssv),
1580 =for apidoc A|I32|ibcmp_utf8|const char *s1|bool u1|const char *s2|bool u2|register I32 len
1582 Return true if the strings s1 and s2 differ case-insensitively, false
1583 if not (if they are equal case-insensitively). If u1 is true, the
1584 string s1 is assumed to be in UTF-8-encoded Unicode. If u2 is true,
1585 the string s2 is assumed to be in UTF-8-encoded Unicode.
1587 For case-insensitiveness, the "casefolding" of Unicode is used
1588 instead of upper/lowercasing both the characters, see
1589 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
1593 Perl_ibcmp_utf8(pTHX_ const char *s1, bool u1, register I32 len1, const char *s2, bool u2, register I32 len2)
1595 register U8 *a = (U8*)s1;
1596 register U8 *b = (U8*)s2;
1597 register U8 *ae = b + len1;
1598 register U8 *be = b + len2;
1601 STRLEN ulen1, ulen2;
1602 U8 tmpbuf1[UTF8_MAXLEN_FOLD+1];
1603 U8 tmpbuf2[UTF8_MAXLEN_FOLD+1];
1605 while (a < ae && b < be) {
1607 if (a + UTF8SKIP(a) > ae)
1609 ca = utf8_to_uvchr((U8*)a, &la);
1615 if (b + UTF8SKIP(b) > be)
1617 cb = utf8_to_uvchr((U8*)b, &lb);
1624 to_uni_fold(NATIVE_TO_UNI(ca), tmpbuf1, &ulen1);
1628 to_uni_fold(NATIVE_TO_UNI(cb), tmpbuf2, &ulen2);
1632 || (ulen1 == 1 && PL_fold[ca] != PL_fold[cb])
1633 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1))
1634 return 1; /* mismatch */
1639 return a == ae && b == be ? 0 : 1; /* 0 match, 1 mismatch */