3 * Copyright (C) 2000, 2001, 2002, 2003, 2004, 2005 by Larry Wall and
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * 'What a fix!' said Sam. 'That's the one place in all the lands we've ever
13 * heard of that we don't want to see any closer; and that's the one place
14 * we're trying to get to! And that's just where we can't get, nohow.'
16 * 'Well do I understand your speech,' he answered in the same language;
17 * 'yet few strangers do so. Why then do you not speak in the Common Tongue,
18 * as is the custom in the West, if you wish to be answered?'
20 * ...the travellers perceived that the floor was paved with stones of many
21 * hues; branching runes and strange devices intertwined beneath their feet.
25 #define PERL_IN_UTF8_C
28 static const char unees[] =
29 "Malformed UTF-8 character (unexpected end of string)";
32 =head1 Unicode Support
34 This file contains various utility functions for manipulating UTF8-encoded
35 strings. For the uninitiated, this is a method of representing arbitrary
36 Unicode characters as a variable number of bytes, in such a way that
37 characters in the ASCII range are unmodified, and a zero byte never appears
38 within non-zero characters.
40 =for apidoc A|U8 *|uvuni_to_utf8_flags|U8 *d|UV uv|UV flags
42 Adds the UTF-8 representation of the Unicode codepoint C<uv> to the end
43 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
44 bytes available. The return value is the pointer to the byte after the
45 end of the new character. In other words,
47 d = uvuni_to_utf8_flags(d, uv, flags);
51 d = uvuni_to_utf8(d, uv);
53 (which is equivalent to)
55 d = uvuni_to_utf8_flags(d, uv, 0);
57 is the recommended Unicode-aware way of saying
65 Perl_uvuni_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
67 if (ckWARN(WARN_UTF8)) {
68 if (UNICODE_IS_SURROGATE(uv) &&
69 !(flags & UNICODE_ALLOW_SURROGATE))
70 Perl_warner(aTHX_ packWARN(WARN_UTF8), "UTF-16 surrogate 0x%04"UVxf, uv);
72 ((uv >= 0xFDD0 && uv <= 0xFDEF &&
73 !(flags & UNICODE_ALLOW_FDD0))
75 ((uv & 0xFFFE) == 0xFFFE && /* Either FFFE or FFFF. */
76 !(flags & UNICODE_ALLOW_FFFF))) &&
77 /* UNICODE_ALLOW_SUPER includes
78 * FFFEs and FFFFs beyond 0x10FFFF. */
79 ((uv <= PERL_UNICODE_MAX) ||
80 !(flags & UNICODE_ALLOW_SUPER))
82 Perl_warner(aTHX_ packWARN(WARN_UTF8),
83 "Unicode character 0x%04"UVxf" is illegal", uv);
85 if (UNI_IS_INVARIANT(uv)) {
86 *d++ = (U8)UTF_TO_NATIVE(uv);
91 STRLEN len = UNISKIP(uv);
94 *p-- = (U8)UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
95 uv >>= UTF_ACCUMULATION_SHIFT;
97 *p = (U8)UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
100 #else /* Non loop style */
102 *d++ = (U8)(( uv >> 6) | 0xc0);
103 *d++ = (U8)(( uv & 0x3f) | 0x80);
107 *d++ = (U8)(( uv >> 12) | 0xe0);
108 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
109 *d++ = (U8)(( uv & 0x3f) | 0x80);
113 *d++ = (U8)(( uv >> 18) | 0xf0);
114 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
115 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
116 *d++ = (U8)(( uv & 0x3f) | 0x80);
119 if (uv < 0x4000000) {
120 *d++ = (U8)(( uv >> 24) | 0xf8);
121 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
122 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
123 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
124 *d++ = (U8)(( uv & 0x3f) | 0x80);
127 if (uv < 0x80000000) {
128 *d++ = (U8)(( uv >> 30) | 0xfc);
129 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
130 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
131 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
132 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
133 *d++ = (U8)(( uv & 0x3f) | 0x80);
137 if (uv < UTF8_QUAD_MAX)
140 *d++ = 0xfe; /* Can't match U+FEFF! */
141 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
142 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
143 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
144 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
145 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
146 *d++ = (U8)(( uv & 0x3f) | 0x80);
151 *d++ = 0xff; /* Can't match U+FFFE! */
152 *d++ = 0x80; /* 6 Reserved bits */
153 *d++ = (U8)(((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
154 *d++ = (U8)(((uv >> 54) & 0x3f) | 0x80);
155 *d++ = (U8)(((uv >> 48) & 0x3f) | 0x80);
156 *d++ = (U8)(((uv >> 42) & 0x3f) | 0x80);
157 *d++ = (U8)(((uv >> 36) & 0x3f) | 0x80);
158 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
159 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
160 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
161 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
162 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
163 *d++ = (U8)(( uv & 0x3f) | 0x80);
167 #endif /* Loop style */
171 Perl_uvuni_to_utf8(pTHX_ U8 *d, UV uv)
173 return Perl_uvuni_to_utf8_flags(aTHX_ d, uv, 0);
178 =for apidoc A|STRLEN|is_utf8_char|const U8 *s
180 Tests if some arbitrary number of bytes begins in a valid UTF-8
181 character. Note that an INVARIANT (i.e. ASCII) character is a valid
182 UTF-8 character. The actual number of bytes in the UTF-8 character
183 will be returned if it is valid, otherwise 0.
187 Perl_is_utf8_char(pTHX_ const U8 *s)
193 if (UTF8_IS_INVARIANT(u))
196 if (!UTF8_IS_START(u))
201 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
206 u &= UTF_START_MASK(len);
210 if (!UTF8_IS_CONTINUATION(*s))
212 uv = UTF8_ACCUMULATE(uv, *s);
219 if ((STRLEN)UNISKIP(uv) < len)
226 =for apidoc A|bool|is_utf8_string|const U8 *s|STRLEN len
228 Returns true if first C<len> bytes of the given string form a valid
229 UTF-8 string, false otherwise. Note that 'a valid UTF-8 string' does
230 not mean 'a string that contains code points above 0x7F encoded in UTF-8'
231 because a valid ASCII string is a valid UTF-8 string.
237 Perl_is_utf8_string(pTHX_ const U8 *s, STRLEN len)
243 len = strlen((const char *)s);
248 /* Inline the easy bits of is_utf8_char() here for speed... */
249 if (UTF8_IS_INVARIANT(*x))
251 else if (!UTF8_IS_START(*x))
254 /* ... and call is_utf8_char() only if really needed. */
268 =for apidoc A|bool|is_utf8_string_loc|const U8 *s|STRLEN len|const U8 **p
270 Like is_ut8_string but store the location of the failure in
277 Perl_is_utf8_string_loc(pTHX_ const U8 *s, STRLEN len, const U8 **p)
284 len = strlen((const char *)s);
288 /* Inline the easy bits of is_utf8_char() here for speed... */
289 if (UTF8_IS_INVARIANT(*x))
291 else if (!UTF8_IS_START(*x)) {
297 /* ... and call is_utf8_char() only if really needed. */
317 =for apidoc A|UV|utf8n_to_uvuni|const U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
319 Bottom level UTF-8 decode routine.
320 Returns the unicode code point value of the first character in the string C<s>
321 which is assumed to be in UTF-8 encoding and no longer than C<curlen>;
322 C<retlen> will be set to the length, in bytes, of that character.
324 If C<s> does not point to a well-formed UTF-8 character, the behaviour
325 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
326 it is assumed that the caller will raise a warning, and this function
327 will silently just set C<retlen> to C<-1> and return zero. If the
328 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
329 malformations will be given, C<retlen> will be set to the expected
330 length of the UTF-8 character in bytes, and zero will be returned.
332 The C<flags> can also contain various flags to allow deviations from
333 the strict UTF-8 encoding (see F<utf8.h>).
335 Most code should use utf8_to_uvchr() rather than call this directly.
341 Perl_utf8n_to_uvuni(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
346 const bool dowarn = ckWARN_d(WARN_UTF8);
347 const UV startbyte = *s;
348 STRLEN expectlen = 0;
351 /* This list is a superset of the UTF8_ALLOW_XXX. */
353 #define UTF8_WARN_EMPTY 1
354 #define UTF8_WARN_CONTINUATION 2
355 #define UTF8_WARN_NON_CONTINUATION 3
356 #define UTF8_WARN_FE_FF 4
357 #define UTF8_WARN_SHORT 5
358 #define UTF8_WARN_OVERFLOW 6
359 #define UTF8_WARN_SURROGATE 7
360 #define UTF8_WARN_LONG 8
361 #define UTF8_WARN_FFFF 9 /* Also FFFE. */
364 !(flags & UTF8_ALLOW_EMPTY)) {
365 warning = UTF8_WARN_EMPTY;
369 if (UTF8_IS_INVARIANT(uv)) {
372 return (UV) (NATIVE_TO_UTF(*s));
375 if (UTF8_IS_CONTINUATION(uv) &&
376 !(flags & UTF8_ALLOW_CONTINUATION)) {
377 warning = UTF8_WARN_CONTINUATION;
381 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
382 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
383 warning = UTF8_WARN_NON_CONTINUATION;
388 uv = NATIVE_TO_UTF(uv);
390 if ((uv == 0xfe || uv == 0xff) &&
391 !(flags & UTF8_ALLOW_FE_FF)) {
392 warning = UTF8_WARN_FE_FF;
397 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
398 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
399 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
400 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
402 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
403 else { len = 7; uv &= 0x01; }
405 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
406 else if (!(uv & 0x01)) { len = 7; uv = 0; }
407 else { len = 13; uv = 0; } /* whoa! */
415 if ((curlen < expectlen) &&
416 !(flags & UTF8_ALLOW_SHORT)) {
417 warning = UTF8_WARN_SHORT;
426 if (!UTF8_IS_CONTINUATION(*s) &&
427 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
429 warning = UTF8_WARN_NON_CONTINUATION;
433 uv = UTF8_ACCUMULATE(uv, *s);
435 /* These cannot be allowed. */
437 if (expectlen != 13 && !(flags & UTF8_ALLOW_LONG)) {
438 warning = UTF8_WARN_LONG;
442 else { /* uv < ouv */
443 /* This cannot be allowed. */
444 warning = UTF8_WARN_OVERFLOW;
452 if (UNICODE_IS_SURROGATE(uv) &&
453 !(flags & UTF8_ALLOW_SURROGATE)) {
454 warning = UTF8_WARN_SURROGATE;
456 } else if ((expectlen > (STRLEN)UNISKIP(uv)) &&
457 !(flags & UTF8_ALLOW_LONG)) {
458 warning = UTF8_WARN_LONG;
460 } else if (UNICODE_IS_ILLEGAL(uv) &&
461 !(flags & UTF8_ALLOW_FFFF)) {
462 warning = UTF8_WARN_FFFF;
470 if (flags & UTF8_CHECK_ONLY) {
477 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
480 case 0: /* Intentionally empty. */ break;
481 case UTF8_WARN_EMPTY:
482 Perl_sv_catpv(aTHX_ sv, "(empty string)");
484 case UTF8_WARN_CONTINUATION:
485 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
487 case UTF8_WARN_NON_CONTINUATION:
489 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
490 (UV)s[1], startbyte);
492 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
493 (UV)s[1], s - s0, s - s0 > 1 ? "s" : "", startbyte, expectlen);
496 case UTF8_WARN_FE_FF:
497 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
499 case UTF8_WARN_SHORT:
500 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
501 curlen, curlen == 1 ? "" : "s", expectlen, startbyte);
502 expectlen = curlen; /* distance for caller to skip */
504 case UTF8_WARN_OVERFLOW:
505 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
508 case UTF8_WARN_SURROGATE:
509 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
512 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
513 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
516 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
519 Perl_sv_catpv(aTHX_ sv, "(unknown reason)");
527 Perl_warner(aTHX_ packWARN(WARN_UTF8),
528 "%s in %s", s, OP_DESC(PL_op));
530 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
535 *retlen = expectlen ? expectlen : len;
541 =for apidoc A|UV|utf8_to_uvchr|const U8 *s|STRLEN *retlen
543 Returns the native character value of the first character in the string C<s>
544 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
545 length, in bytes, of that character.
547 If C<s> does not point to a well-formed UTF-8 character, zero is
548 returned and retlen is set, if possible, to -1.
554 Perl_utf8_to_uvchr(pTHX_ const U8 *s, STRLEN *retlen)
556 return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXBYTES, retlen,
557 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
561 =for apidoc A|UV|utf8_to_uvuni|const U8 *s|STRLEN *retlen
563 Returns the Unicode code point of the first character in the string C<s>
564 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
565 length, in bytes, of that character.
567 This function should only be used when returned UV is considered
568 an index into the Unicode semantic tables (e.g. swashes).
570 If C<s> does not point to a well-formed UTF-8 character, zero is
571 returned and retlen is set, if possible, to -1.
577 Perl_utf8_to_uvuni(pTHX_ const U8 *s, STRLEN *retlen)
579 /* Call the low level routine asking for checks */
580 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXBYTES, retlen,
581 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
585 =for apidoc A|STRLEN|utf8_length|const U8 *s|const U8 *e
587 Return the length of the UTF-8 char encoded string C<s> in characters.
588 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
589 up past C<e>, croaks.
595 Perl_utf8_length(pTHX_ const U8 *s, const U8 *e)
599 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
600 * the bitops (especially ~) can create illegal UTF-8.
601 * In other words: in Perl UTF-8 is not just for Unicode. */
604 goto warn_and_return;
606 const U8 t = UTF8SKIP(s);
609 if (ckWARN_d(WARN_UTF8)) {
611 Perl_warner(aTHX_ packWARN(WARN_UTF8),
612 "%s in %s", unees, OP_DESC(PL_op));
614 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
626 =for apidoc A|IV|utf8_distance|const U8 *a|const U8 *b
628 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
631 WARNING: use only if you *know* that the pointers point inside the
638 Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
642 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
643 * the bitops (especially ~) can create illegal UTF-8.
644 * In other words: in Perl UTF-8 is not just for Unicode. */
648 const U8 c = UTF8SKIP(a);
650 goto warn_and_return;
657 const U8 c = UTF8SKIP(b);
661 if (ckWARN_d(WARN_UTF8)) {
663 Perl_warner(aTHX_ packWARN(WARN_UTF8),
664 "%s in %s", unees, OP_DESC(PL_op));
666 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
679 =for apidoc A|U8 *|utf8_hop|U8 *s|I32 off
681 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
684 WARNING: do not use the following unless you *know* C<off> is within
685 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
686 on the first byte of character or just after the last byte of a character.
692 Perl_utf8_hop(pTHX_ const U8 *s, I32 off)
694 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
695 * the bitops (especially ~) can create illegal UTF-8.
696 * In other words: in Perl UTF-8 is not just for Unicode. */
705 while (UTF8_IS_CONTINUATION(*s))
713 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
715 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
716 Unlike C<bytes_to_utf8>, this over-writes the original string, and
717 updates len to contain the new length.
718 Returns zero on failure, setting C<len> to -1.
724 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
730 /* ensure valid UTF-8 and chars < 256 before updating string */
731 for (send = s + *len; s < send; ) {
734 if (!UTF8_IS_INVARIANT(c) &&
735 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
736 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
745 *d++ = (U8)utf8_to_uvchr(s, &ulen);
754 =for apidoc A|U8 *|bytes_from_utf8|const U8 *s|STRLEN *len|bool *is_utf8
756 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
757 Unlike C<utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
758 the newly-created string, and updates C<len> to contain the new
759 length. Returns the original string if no conversion occurs, C<len>
760 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
761 0 if C<s> is converted or contains all 7bit characters.
767 Perl_bytes_from_utf8(pTHX_ const U8 *s, STRLEN *len, bool *is_utf8)
777 /* ensure valid UTF-8 and chars < 256 before converting string */
778 for (send = s + *len; s < send;) {
780 if (!UTF8_IS_INVARIANT(c)) {
781 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
782 (c = *s++) && UTF8_IS_CONTINUATION(c))
791 Newz(801, d, (*len) - count + 1, U8);
792 s = start; start = d;
795 if (!UTF8_IS_INVARIANT(c)) {
796 /* Then it is two-byte encoded */
797 c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++);
798 c = ASCII_TO_NATIVE(c);
808 =for apidoc A|U8 *|bytes_to_utf8|const U8 *s|STRLEN *len
810 Converts a string C<s> of length C<len> from ASCII into UTF-8 encoding.
811 Returns a pointer to the newly-created string, and sets C<len> to
812 reflect the new length.
814 If you want to convert to UTF-8 from other encodings than ASCII,
815 see sv_recode_to_utf8().
821 Perl_bytes_to_utf8(pTHX_ const U8 *s, STRLEN *len)
823 const U8 * const send = s + (*len);
827 Newz(801, d, (*len) * 2 + 1, U8);
831 const UV uv = NATIVE_TO_ASCII(*s++);
832 if (UNI_IS_INVARIANT(uv))
833 *d++ = (U8)UTF_TO_NATIVE(uv);
835 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
836 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
845 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
847 * Destination must be pre-extended to 3/2 source. Do not use in-place.
848 * We optimize for native, for obvious reasons. */
851 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
856 if (bytelen == 1 && p[0] == 0) { /* Be understanding. */
863 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %"UVf, (UV)bytelen);
868 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
875 *d++ = (U8)(( uv >> 6) | 0xc0);
876 *d++ = (U8)(( uv & 0x3f) | 0x80);
879 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
880 UV low = (p[0] << 8) + p[1];
882 if (low < 0xdc00 || low >= 0xdfff)
883 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
884 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
887 *d++ = (U8)(( uv >> 12) | 0xe0);
888 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
889 *d++ = (U8)(( uv & 0x3f) | 0x80);
893 *d++ = (U8)(( uv >> 18) | 0xf0);
894 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
895 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
896 *d++ = (U8)(( uv & 0x3f) | 0x80);
900 *newlen = d - dstart;
904 /* Note: this one is slightly destructive of the source. */
907 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
910 U8* send = s + bytelen;
917 return utf16_to_utf8(p, d, bytelen, newlen);
920 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
923 Perl_is_uni_alnum(pTHX_ UV c)
925 U8 tmpbuf[UTF8_MAXBYTES+1];
926 uvchr_to_utf8(tmpbuf, c);
927 return is_utf8_alnum(tmpbuf);
931 Perl_is_uni_alnumc(pTHX_ UV c)
933 U8 tmpbuf[UTF8_MAXBYTES+1];
934 uvchr_to_utf8(tmpbuf, c);
935 return is_utf8_alnumc(tmpbuf);
939 Perl_is_uni_idfirst(pTHX_ UV c)
941 U8 tmpbuf[UTF8_MAXBYTES+1];
942 uvchr_to_utf8(tmpbuf, c);
943 return is_utf8_idfirst(tmpbuf);
947 Perl_is_uni_alpha(pTHX_ UV c)
949 U8 tmpbuf[UTF8_MAXBYTES+1];
950 uvchr_to_utf8(tmpbuf, c);
951 return is_utf8_alpha(tmpbuf);
955 Perl_is_uni_ascii(pTHX_ UV c)
957 U8 tmpbuf[UTF8_MAXBYTES+1];
958 uvchr_to_utf8(tmpbuf, c);
959 return is_utf8_ascii(tmpbuf);
963 Perl_is_uni_space(pTHX_ UV c)
965 U8 tmpbuf[UTF8_MAXBYTES+1];
966 uvchr_to_utf8(tmpbuf, c);
967 return is_utf8_space(tmpbuf);
971 Perl_is_uni_digit(pTHX_ UV c)
973 U8 tmpbuf[UTF8_MAXBYTES+1];
974 uvchr_to_utf8(tmpbuf, c);
975 return is_utf8_digit(tmpbuf);
979 Perl_is_uni_upper(pTHX_ UV c)
981 U8 tmpbuf[UTF8_MAXBYTES+1];
982 uvchr_to_utf8(tmpbuf, c);
983 return is_utf8_upper(tmpbuf);
987 Perl_is_uni_lower(pTHX_ UV c)
989 U8 tmpbuf[UTF8_MAXBYTES+1];
990 uvchr_to_utf8(tmpbuf, c);
991 return is_utf8_lower(tmpbuf);
995 Perl_is_uni_cntrl(pTHX_ UV c)
997 U8 tmpbuf[UTF8_MAXBYTES+1];
998 uvchr_to_utf8(tmpbuf, c);
999 return is_utf8_cntrl(tmpbuf);
1003 Perl_is_uni_graph(pTHX_ UV c)
1005 U8 tmpbuf[UTF8_MAXBYTES+1];
1006 uvchr_to_utf8(tmpbuf, c);
1007 return is_utf8_graph(tmpbuf);
1011 Perl_is_uni_print(pTHX_ UV c)
1013 U8 tmpbuf[UTF8_MAXBYTES+1];
1014 uvchr_to_utf8(tmpbuf, c);
1015 return is_utf8_print(tmpbuf);
1019 Perl_is_uni_punct(pTHX_ UV c)
1021 U8 tmpbuf[UTF8_MAXBYTES+1];
1022 uvchr_to_utf8(tmpbuf, c);
1023 return is_utf8_punct(tmpbuf);
1027 Perl_is_uni_xdigit(pTHX_ UV c)
1029 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1030 uvchr_to_utf8(tmpbuf, c);
1031 return is_utf8_xdigit(tmpbuf);
1035 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
1037 uvchr_to_utf8(p, c);
1038 return to_utf8_upper(p, p, lenp);
1042 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
1044 uvchr_to_utf8(p, c);
1045 return to_utf8_title(p, p, lenp);
1049 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
1051 uvchr_to_utf8(p, c);
1052 return to_utf8_lower(p, p, lenp);
1056 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1058 uvchr_to_utf8(p, c);
1059 return to_utf8_fold(p, p, lenp);
1062 /* for now these all assume no locale info available for Unicode > 255 */
1065 Perl_is_uni_alnum_lc(pTHX_ UV c)
1067 return is_uni_alnum(c); /* XXX no locale support yet */
1071 Perl_is_uni_alnumc_lc(pTHX_ UV c)
1073 return is_uni_alnumc(c); /* XXX no locale support yet */
1077 Perl_is_uni_idfirst_lc(pTHX_ UV c)
1079 return is_uni_idfirst(c); /* XXX no locale support yet */
1083 Perl_is_uni_alpha_lc(pTHX_ UV c)
1085 return is_uni_alpha(c); /* XXX no locale support yet */
1089 Perl_is_uni_ascii_lc(pTHX_ UV c)
1091 return is_uni_ascii(c); /* XXX no locale support yet */
1095 Perl_is_uni_space_lc(pTHX_ UV c)
1097 return is_uni_space(c); /* XXX no locale support yet */
1101 Perl_is_uni_digit_lc(pTHX_ UV c)
1103 return is_uni_digit(c); /* XXX no locale support yet */
1107 Perl_is_uni_upper_lc(pTHX_ UV c)
1109 return is_uni_upper(c); /* XXX no locale support yet */
1113 Perl_is_uni_lower_lc(pTHX_ UV c)
1115 return is_uni_lower(c); /* XXX no locale support yet */
1119 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1121 return is_uni_cntrl(c); /* XXX no locale support yet */
1125 Perl_is_uni_graph_lc(pTHX_ UV c)
1127 return is_uni_graph(c); /* XXX no locale support yet */
1131 Perl_is_uni_print_lc(pTHX_ UV c)
1133 return is_uni_print(c); /* XXX no locale support yet */
1137 Perl_is_uni_punct_lc(pTHX_ UV c)
1139 return is_uni_punct(c); /* XXX no locale support yet */
1143 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1145 return is_uni_xdigit(c); /* XXX no locale support yet */
1149 Perl_to_uni_upper_lc(pTHX_ U32 c)
1151 /* XXX returns only the first character -- do not use XXX */
1152 /* XXX no locale support yet */
1154 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1155 return (U32)to_uni_upper(c, tmpbuf, &len);
1159 Perl_to_uni_title_lc(pTHX_ U32 c)
1161 /* XXX returns only the first character XXX -- do not use XXX */
1162 /* XXX no locale support yet */
1164 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1165 return (U32)to_uni_title(c, tmpbuf, &len);
1169 Perl_to_uni_lower_lc(pTHX_ U32 c)
1171 /* XXX returns only the first character -- do not use XXX */
1172 /* XXX no locale support yet */
1174 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1175 return (U32)to_uni_lower(c, tmpbuf, &len);
1179 Perl_is_utf8_alnum(pTHX_ const U8 *p)
1181 if (!is_utf8_char(p))
1184 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1185 * descendant of isalnum(3), in other words, it doesn't
1186 * contain the '_'. --jhi */
1187 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1188 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1189 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1190 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1192 PL_utf8_alnum = swash_init("utf8", "",
1193 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1194 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1199 Perl_is_utf8_alnumc(pTHX_ const U8 *p)
1201 if (!is_utf8_char(p))
1204 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1205 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1206 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
1207 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1209 PL_utf8_alnum = swash_init("utf8", "",
1210 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1211 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1216 Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
1220 if (!is_utf8_char(p))
1222 if (!PL_utf8_idstart) /* is_utf8_idstart would be more logical. */
1223 PL_utf8_idstart = swash_init("utf8", "IdStart", &PL_sv_undef, 0, 0);
1224 return swash_fetch(PL_utf8_idstart, p, TRUE) != 0;
1228 Perl_is_utf8_idcont(pTHX_ const U8 *p)
1232 if (!is_utf8_char(p))
1234 if (!PL_utf8_idcont)
1235 PL_utf8_idcont = swash_init("utf8", "IdContinue", &PL_sv_undef, 0, 0);
1236 return swash_fetch(PL_utf8_idcont, p, TRUE) != 0;
1240 Perl_is_utf8_alpha(pTHX_ const U8 *p)
1242 if (!is_utf8_char(p))
1245 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1246 return swash_fetch(PL_utf8_alpha, p, TRUE) != 0;
1250 Perl_is_utf8_ascii(pTHX_ const U8 *p)
1252 if (!is_utf8_char(p))
1255 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1256 return swash_fetch(PL_utf8_ascii, p, TRUE) != 0;
1260 Perl_is_utf8_space(pTHX_ const U8 *p)
1262 if (!is_utf8_char(p))
1265 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1266 return swash_fetch(PL_utf8_space, p, TRUE) != 0;
1270 Perl_is_utf8_digit(pTHX_ const U8 *p)
1272 if (!is_utf8_char(p))
1275 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1276 return swash_fetch(PL_utf8_digit, p, TRUE) != 0;
1280 Perl_is_utf8_upper(pTHX_ const U8 *p)
1282 if (!is_utf8_char(p))
1285 PL_utf8_upper = swash_init("utf8", "IsUppercase", &PL_sv_undef, 0, 0);
1286 return swash_fetch(PL_utf8_upper, p, TRUE) != 0;
1290 Perl_is_utf8_lower(pTHX_ const U8 *p)
1292 if (!is_utf8_char(p))
1295 PL_utf8_lower = swash_init("utf8", "IsLowercase", &PL_sv_undef, 0, 0);
1296 return swash_fetch(PL_utf8_lower, p, TRUE) != 0;
1300 Perl_is_utf8_cntrl(pTHX_ const U8 *p)
1302 if (!is_utf8_char(p))
1305 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1306 return swash_fetch(PL_utf8_cntrl, p, TRUE) != 0;
1310 Perl_is_utf8_graph(pTHX_ const U8 *p)
1312 if (!is_utf8_char(p))
1315 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1316 return swash_fetch(PL_utf8_graph, p, TRUE) != 0;
1320 Perl_is_utf8_print(pTHX_ const U8 *p)
1322 if (!is_utf8_char(p))
1325 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1326 return swash_fetch(PL_utf8_print, p, TRUE) != 0;
1330 Perl_is_utf8_punct(pTHX_ const U8 *p)
1332 if (!is_utf8_char(p))
1335 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1336 return swash_fetch(PL_utf8_punct, p, TRUE) != 0;
1340 Perl_is_utf8_xdigit(pTHX_ const U8 *p)
1342 if (!is_utf8_char(p))
1344 if (!PL_utf8_xdigit)
1345 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1346 return swash_fetch(PL_utf8_xdigit, p, TRUE) != 0;
1350 Perl_is_utf8_mark(pTHX_ const U8 *p)
1352 if (!is_utf8_char(p))
1355 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1356 return swash_fetch(PL_utf8_mark, p, TRUE) != 0;
1360 =for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special
1362 The "p" contains the pointer to the UTF-8 string encoding
1363 the character that is being converted.
1365 The "ustrp" is a pointer to the character buffer to put the
1366 conversion result to. The "lenp" is a pointer to the length
1369 The "swashp" is a pointer to the swash to use.
1371 Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
1372 and loaded by SWASHGET, using lib/utf8_heavy.pl. The special (usually,
1373 but not always, a multicharacter mapping), is tried first.
1375 The "special" is a string like "utf8::ToSpecLower", which means the
1376 hash %utf8::ToSpecLower. The access to the hash is through
1377 Perl_to_utf8_case().
1379 The "normal" is a string like "ToLower" which means the swash
1385 Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp, const char *normal, const char *special)
1387 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1390 const UV uv0 = utf8_to_uvchr(p, 0);
1391 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1392 * are necessary in EBCDIC, they are redundant no-ops
1393 * in ASCII-ish platforms, and hopefully optimized away. */
1394 const UV uv1 = NATIVE_TO_UNI(uv0);
1395 uvuni_to_utf8(tmpbuf, uv1);
1397 if (!*swashp) /* load on-demand */
1398 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1400 /* The 0xDF is the only special casing Unicode code point below 0x100. */
1401 if (special && (uv1 == 0xDF || uv1 > 0xFF)) {
1402 /* It might be "special" (sometimes, but not always,
1403 * a multicharacter mapping) */
1407 if ((hv = get_hv(special, FALSE)) &&
1408 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1412 s = SvPV(*svp, len);
1414 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
1417 /* If we have EBCDIC we need to remap the characters
1418 * since any characters in the low 256 are Unicode
1419 * code points, not EBCDIC. */
1420 U8 *t = (U8*)s, *tend = t + len, *d;
1427 UV c = utf8_to_uvchr(t, &tlen);
1429 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1438 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1443 Copy(tmpbuf, ustrp, len, U8);
1445 Copy(s, ustrp, len, U8);
1451 if (!len && *swashp) {
1452 UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1455 /* It was "normal" (a single character mapping). */
1456 UV uv3 = UNI_TO_NATIVE(uv2);
1458 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
1462 if (!len) /* Neither: just copy. */
1463 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1468 return len ? utf8_to_uvchr(ustrp, 0) : 0;
1472 =for apidoc A|UV|to_utf8_upper|const U8 *p|U8 *ustrp|STRLEN *lenp
1474 Convert the UTF-8 encoded character at p to its uppercase version and
1475 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1476 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1477 the uppercase version may be longer than the original character.
1479 The first character of the uppercased version is returned
1480 (but note, as explained above, that there may be more.)
1485 Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1487 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1488 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1492 =for apidoc A|UV|to_utf8_title|const U8 *p|U8 *ustrp|STRLEN *lenp
1494 Convert the UTF-8 encoded character at p to its titlecase version and
1495 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1496 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1497 titlecase version may be longer than the original character.
1499 The first character of the titlecased version is returned
1500 (but note, as explained above, that there may be more.)
1505 Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1507 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1508 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1512 =for apidoc A|UV|to_utf8_lower|const U8 *p|U8 *ustrp|STRLEN *lenp
1514 Convert the UTF-8 encoded character at p to its lowercase version and
1515 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1516 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1517 lowercase version may be longer than the original character.
1519 The first character of the lowercased version is returned
1520 (but note, as explained above, that there may be more.)
1525 Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1527 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1528 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1532 =for apidoc A|UV|to_utf8_fold|const U8 *p|U8 *ustrp|STRLEN *lenp
1534 Convert the UTF-8 encoded character at p to its foldcase version and
1535 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1536 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1537 foldcase version may be longer than the original character (up to
1540 The first character of the foldcased version is returned
1541 (but note, as explained above, that there may be more.)
1546 Perl_to_utf8_fold(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1548 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1549 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1552 /* a "swash" is a swatch hash */
1555 Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
1559 SV* tokenbufsv = sv_newmortal();
1561 const size_t pkg_len = strlen(pkg);
1562 const size_t name_len = strlen(name);
1563 HV *stash = gv_stashpvn(pkg, pkg_len, FALSE);
1566 PUSHSTACKi(PERLSI_MAGIC);
1571 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
1573 errsv_save = newSVsv(ERRSV);
1574 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
1577 sv_setsv(ERRSV, errsv_save);
1578 SvREFCNT_dec(errsv_save);
1584 PUSHs(sv_2mortal(newSVpvn(pkg, pkg_len)));
1585 PUSHs(sv_2mortal(newSVpvn(name, name_len)));
1587 PUSHs(sv_2mortal(newSViv(minbits)));
1588 PUSHs(sv_2mortal(newSViv(none)));
1590 if (IN_PERL_COMPILETIME) {
1591 /* XXX ought to be handled by lex_start */
1594 sv_setpv(tokenbufsv, PL_tokenbuf);
1596 errsv_save = newSVsv(ERRSV);
1597 if (call_method("SWASHNEW", G_SCALAR))
1598 retval = newSVsv(*PL_stack_sp--);
1600 retval = &PL_sv_undef;
1602 sv_setsv(ERRSV, errsv_save);
1603 SvREFCNT_dec(errsv_save);
1606 if (IN_PERL_COMPILETIME) {
1608 const char* pv = SvPV(tokenbufsv, len);
1610 Copy(pv, PL_tokenbuf, len+1, char);
1611 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1613 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
1615 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
1617 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1623 /* This API is wrong for special case conversions since we may need to
1624 * return several Unicode characters for a single Unicode character
1625 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1626 * the lower-level routine, and it is similarly broken for returning
1627 * multiple values. --jhi */
1629 Perl_swash_fetch(pTHX_ SV *sv, const U8 *ptr, bool do_utf8)
1632 HV* hv = (HV*)SvRV(sv);
1641 UV c = NATIVE_TO_ASCII(*ptr);
1643 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1644 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
1645 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
1648 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
1649 * then the "swatch" is a vec() for al the chars which start
1651 * So the key in the hash (klen) is length of encoded char -1
1653 klen = UTF8SKIP(ptr) - 1;
1658 /* If char in invariant then swatch is for all the invariant chars
1659 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
1661 needents = UTF_CONTINUATION_MARK;
1662 off = NATIVE_TO_UTF(ptr[klen]);
1666 /* If char is encoded then swatch is for the prefix */
1667 needents = (1 << UTF_ACCUMULATION_SHIFT);
1668 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
1672 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1673 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1674 * it's nothing to sniff at.) Pity we usually come through at least
1675 * two function calls to get here...
1677 * NB: this code assumes that swatches are never modified, once generated!
1680 if (hv == PL_last_swash_hv &&
1681 klen == PL_last_swash_klen &&
1682 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
1684 tmps = PL_last_swash_tmps;
1685 slen = PL_last_swash_slen;
1688 /* Try our second-level swatch cache, kept in a hash. */
1689 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
1691 /* If not cached, generate it via utf8::SWASHGET */
1692 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1694 /* We use utf8n_to_uvuni() as we want an index into
1695 Unicode tables, not a native character number.
1697 UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
1699 0 : UTF8_ALLOW_ANY);
1704 PUSHSTACKi(PERLSI_MAGIC);
1708 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
1709 PUSHs(sv_2mortal(newSViv((klen) ?
1710 (code_point & ~(needents - 1)) : 0)));
1711 PUSHs(sv_2mortal(newSViv(needents)));
1713 errsv_save = newSVsv(ERRSV);
1714 if (call_method("SWASHGET", G_SCALAR))
1715 retval = newSVsv(*PL_stack_sp--);
1717 retval = &PL_sv_undef;
1719 sv_setsv(ERRSV, errsv_save);
1720 SvREFCNT_dec(errsv_save);
1724 if (IN_PERL_COMPILETIME)
1725 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1727 svp = hv_store(hv, (const char *)ptr, klen, retval, 0);
1729 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || (slen << 3) < needents)
1730 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1733 PL_last_swash_hv = hv;
1734 PL_last_swash_klen = klen;
1735 PL_last_swash_tmps = tmps;
1736 PL_last_swash_slen = slen;
1738 Copy(ptr, PL_last_swash_key, klen, U8);
1741 switch ((int)((slen << 3) / needents)) {
1743 bit = 1 << (off & 7);
1745 return (tmps[off] & bit) != 0;
1750 return (tmps[off] << 8) + tmps[off + 1] ;
1753 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1755 Perl_croak(aTHX_ "panic: swash_fetch");
1761 =for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv
1763 Adds the UTF-8 representation of the Native codepoint C<uv> to the end
1764 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
1765 bytes available. The return value is the pointer to the byte after the
1766 end of the new character. In other words,
1768 d = uvchr_to_utf8(d, uv);
1770 is the recommended wide native character-aware way of saying
1777 /* On ASCII machines this is normally a macro but we want a
1778 real function in case XS code wants it
1780 #undef Perl_uvchr_to_utf8
1782 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
1784 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
1788 Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
1790 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
1794 =for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
1796 Returns the native character value of the first character in the string C<s>
1797 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
1798 length, in bytes, of that character.
1800 Allows length and flags to be passed to low level routine.
1804 /* On ASCII machines this is normally a macro but we want
1805 a real function in case XS code wants it
1807 #undef Perl_utf8n_to_uvchr
1809 Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
1811 UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
1812 return UNI_TO_NATIVE(uv);
1816 =for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags
1818 Build to the scalar dsv a displayable version of the string spv,
1819 length len, the displayable version being at most pvlim bytes long
1820 (if longer, the rest is truncated and "..." will be appended).
1822 The flags argument can have UNI_DISPLAY_ISPRINT set to display
1823 isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
1824 to display the \\[nrfta\\] as the backslashed versions (like '\n')
1825 (UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
1826 UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
1827 UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
1829 The pointer to the PV of the dsv is returned.
1833 Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
1838 sv_setpvn(dsv, "", 0);
1839 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
1841 /* This serves double duty as a flag and a character to print after
1842 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
1846 if (pvlim && SvCUR(dsv) >= pvlim) {
1850 u = utf8_to_uvchr((U8*)s, 0);
1852 const unsigned char c = (unsigned char)u & 0xFF;
1853 if (!ok && (flags & UNI_DISPLAY_BACKSLASH)) {
1870 Perl_sv_catpvf(aTHX_ dsv, "\\%c", ok);
1873 /* isPRINT() is the locale-blind version. */
1874 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
1875 Perl_sv_catpvf(aTHX_ dsv, "%c", c);
1880 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
1883 sv_catpvn(dsv, "...", 3);
1889 =for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags
1891 Build to the scalar dsv a displayable version of the scalar sv,
1892 the displayable version being at most pvlim bytes long
1893 (if longer, the rest is truncated and "..." will be appended).
1895 The flags argument is as in pv_uni_display().
1897 The pointer to the PV of the dsv is returned.
1901 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
1903 return Perl_pv_uni_display(aTHX_ dsv, (U8*)SvPVX(ssv), SvCUR(ssv),
1908 =for apidoc A|I32|ibcmp_utf8|const char *s1|char **pe1|register UV l1|bool u1|const char *s2|char **pe2|register UV l2|bool u2
1910 Return true if the strings s1 and s2 differ case-insensitively, false
1911 if not (if they are equal case-insensitively). If u1 is true, the
1912 string s1 is assumed to be in UTF-8-encoded Unicode. If u2 is true,
1913 the string s2 is assumed to be in UTF-8-encoded Unicode. If u1 or u2
1914 are false, the respective string is assumed to be in native 8-bit
1917 If the pe1 and pe2 are non-NULL, the scanning pointers will be copied
1918 in there (they will point at the beginning of the I<next> character).
1919 If the pointers behind pe1 or pe2 are non-NULL, they are the end
1920 pointers beyond which scanning will not continue under any
1921 circumstances. If the byte lengths l1 and l2 are non-zero, s1+l1 and
1922 s2+l2 will be used as goal end pointers that will also stop the scan,
1923 and which qualify towards defining a successful match: all the scans
1924 that define an explicit length must reach their goal pointers for
1925 a match to succeed).
1927 For case-insensitiveness, the "casefolding" of Unicode is used
1928 instead of upper/lowercasing both the characters, see
1929 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
1933 Perl_ibcmp_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
1935 register const U8 *p1 = (const U8*)s1;
1936 register const U8 *p2 = (const U8*)s2;
1937 register const U8 *f1 = 0, *f2 = 0;
1938 register U8 *e1 = 0, *q1 = 0;
1939 register U8 *e2 = 0, *q2 = 0;
1940 STRLEN n1 = 0, n2 = 0;
1941 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
1942 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
1944 STRLEN foldlen1, foldlen2;
1949 if (e1 == 0 || (l1 && l1 < (UV)(e1 - (const U8*)s1)))
1950 f1 = (const U8*)s1 + l1;
1953 if (e2 == 0 || (l2 && l2 < (UV)(e2 - (const U8*)s2)))
1954 f2 = (const U8*)s2 + l2;
1956 if ((e1 == 0 && f1 == 0) || (e2 == 0 && f2 == 0) || (f1 == 0 && f2 == 0))
1957 return 1; /* mismatch; possible infinite loop or false positive */
1960 natbuf[1] = 0; /* Need to terminate the buffer. */
1962 while ((e1 == 0 || p1 < e1) &&
1963 (f1 == 0 || p1 < f1) &&
1964 (e2 == 0 || p2 < e2) &&
1965 (f2 == 0 || p2 < f2)) {
1968 to_utf8_fold(p1, foldbuf1, &foldlen1);
1971 to_utf8_fold(natbuf, foldbuf1, &foldlen1);
1978 to_utf8_fold(p2, foldbuf2, &foldlen2);
1981 to_utf8_fold(natbuf, foldbuf2, &foldlen2);
1987 if ( UTF8SKIP(q1) != UTF8SKIP(q2) ||
1988 (UTF8SKIP(q1) == 1 && *q1 != *q2) ||
1989 memNE((char*)q1, (char*)q2, UTF8SKIP(q1)) )
1990 return 1; /* mismatch */
1997 p1 += u1 ? UTF8SKIP(p1) : 1;
1999 p2 += u2 ? UTF8SKIP(p2) : 1;
2003 /* A match is defined by all the scans that specified
2004 * an explicit length reaching their final goals. */
2005 match = (f1 == 0 || p1 == f1) && (f2 == 0 || p2 == f2);
2014 return match ? 0 : 1; /* 0 match, 1 mismatch */
2019 * c-indentation-style: bsd
2021 * indent-tabs-mode: t
2024 * ex: set ts=8 sts=4 sw=4 noet: