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 Tests if some arbitrary number of bytes begins in a valid UTF-8
179 character. Note that an INVARIANT (i.e. ASCII) character is a valid
180 UTF-8 character. The actual number of bytes in the UTF-8 character
181 will be returned if it is valid, otherwise 0.
183 This is the "slow" version as opposed to the "fast" version which is
184 the "unrolled" IS_UTF8_CHAR(). E.g. for t/uni/class.t the speed
185 difference is a factor of 2 to 3. For lengths (UTF8SKIP(s)) of four
186 or less you should use the IS_UTF8_CHAR(), for lengths of five or more
187 you should use the _slow(). In practice this means that the _slow()
188 will be used very rarely, since the maximum Unicode code point (as of
189 Unicode 4.1) is U+10FFFF, which encodes in UTF-8 to four bytes. Only
190 the "Perl extended UTF-8" (the infamous 'v-strings') will encode into
195 S_is_utf8_char_slow(pTHX_ const U8 *s, const STRLEN len)
201 if (UTF8_IS_INVARIANT(u))
204 if (!UTF8_IS_START(u))
207 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
212 u &= UTF_START_MASK(len);
216 if (!UTF8_IS_CONTINUATION(*s))
218 uv = UTF8_ACCUMULATE(uv, *s);
225 if ((STRLEN)UNISKIP(uv) < len)
232 =for apidoc A|STRLEN|is_utf8_char|const U8 *s
234 Tests if some arbitrary number of bytes begins in a valid UTF-8
235 character. Note that an INVARIANT (i.e. ASCII) character is a valid
236 UTF-8 character. The actual number of bytes in the UTF-8 character
237 will be returned if it is valid, otherwise 0.
241 Perl_is_utf8_char(pTHX_ const U8 *s)
247 return IS_UTF8_CHAR(s, len) ? len : 0;
248 #endif /* #ifdef IS_UTF8_CHAR */
249 return S_is_utf8_char_slow(s, len);
253 =for apidoc A|bool|is_utf8_string|const U8 *s|STRLEN len
255 Returns true if first C<len> bytes of the given string form a valid
256 UTF-8 string, false otherwise. Note that 'a valid UTF-8 string' does
257 not mean 'a string that contains code points above 0x7F encoded in UTF-8'
258 because a valid ASCII string is a valid UTF-8 string.
264 Perl_is_utf8_string(pTHX_ const U8 *s, STRLEN len)
270 len = strlen((const char *)s);
275 /* Inline the easy bits of is_utf8_char() here for speed... */
276 if (UTF8_IS_INVARIANT(*x))
278 else if (!UTF8_IS_START(*x))
281 /* ... and call is_utf8_char() only if really needed. */
285 if (!IS_UTF8_CHAR(x, c))
288 if (!S_is_utf8_char_slow(x, c))
293 #endif /* #ifdef IS_UTF8_CHAR */
306 =for apidoc A|bool|is_utf8_string_loc|const U8 *s|STRLEN len|const U8 **p
308 Like is_ut8_string but store the location of the failure in
315 Perl_is_utf8_string_loc(pTHX_ const U8 *s, STRLEN len, const U8 **p)
322 len = strlen((const char *)s);
326 /* Inline the easy bits of is_utf8_char() here for speed... */
327 if (UTF8_IS_INVARIANT(*x))
329 else if (!UTF8_IS_START(*x)) {
335 /* ... and call is_utf8_char() only if really needed. */
355 =for apidoc A|UV|utf8n_to_uvuni|const U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
357 Bottom level UTF-8 decode routine.
358 Returns the unicode code point value of the first character in the string C<s>
359 which is assumed to be in UTF-8 encoding and no longer than C<curlen>;
360 C<retlen> will be set to the length, in bytes, of that character.
362 If C<s> does not point to a well-formed UTF-8 character, the behaviour
363 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
364 it is assumed that the caller will raise a warning, and this function
365 will silently just set C<retlen> to C<-1> and return zero. If the
366 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
367 malformations will be given, C<retlen> will be set to the expected
368 length of the UTF-8 character in bytes, and zero will be returned.
370 The C<flags> can also contain various flags to allow deviations from
371 the strict UTF-8 encoding (see F<utf8.h>).
373 Most code should use utf8_to_uvchr() rather than call this directly.
379 Perl_utf8n_to_uvuni(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
384 const bool dowarn = ckWARN_d(WARN_UTF8);
385 const UV startbyte = *s;
386 STRLEN expectlen = 0;
389 /* This list is a superset of the UTF8_ALLOW_XXX. */
391 #define UTF8_WARN_EMPTY 1
392 #define UTF8_WARN_CONTINUATION 2
393 #define UTF8_WARN_NON_CONTINUATION 3
394 #define UTF8_WARN_FE_FF 4
395 #define UTF8_WARN_SHORT 5
396 #define UTF8_WARN_OVERFLOW 6
397 #define UTF8_WARN_SURROGATE 7
398 #define UTF8_WARN_LONG 8
399 #define UTF8_WARN_FFFF 9 /* Also FFFE. */
402 !(flags & UTF8_ALLOW_EMPTY)) {
403 warning = UTF8_WARN_EMPTY;
407 if (UTF8_IS_INVARIANT(uv)) {
410 return (UV) (NATIVE_TO_UTF(*s));
413 if (UTF8_IS_CONTINUATION(uv) &&
414 !(flags & UTF8_ALLOW_CONTINUATION)) {
415 warning = UTF8_WARN_CONTINUATION;
419 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
420 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
421 warning = UTF8_WARN_NON_CONTINUATION;
426 uv = NATIVE_TO_UTF(uv);
428 if ((uv == 0xfe || uv == 0xff) &&
429 !(flags & UTF8_ALLOW_FE_FF)) {
430 warning = UTF8_WARN_FE_FF;
435 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
436 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
437 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
438 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
440 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
441 else { len = 7; uv &= 0x01; }
443 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
444 else if (!(uv & 0x01)) { len = 7; uv = 0; }
445 else { len = 13; uv = 0; } /* whoa! */
453 if ((curlen < expectlen) &&
454 !(flags & UTF8_ALLOW_SHORT)) {
455 warning = UTF8_WARN_SHORT;
464 if (!UTF8_IS_CONTINUATION(*s) &&
465 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
467 warning = UTF8_WARN_NON_CONTINUATION;
471 uv = UTF8_ACCUMULATE(uv, *s);
473 /* These cannot be allowed. */
475 if (expectlen != 13 && !(flags & UTF8_ALLOW_LONG)) {
476 warning = UTF8_WARN_LONG;
480 else { /* uv < ouv */
481 /* This cannot be allowed. */
482 warning = UTF8_WARN_OVERFLOW;
490 if (UNICODE_IS_SURROGATE(uv) &&
491 !(flags & UTF8_ALLOW_SURROGATE)) {
492 warning = UTF8_WARN_SURROGATE;
494 } else if ((expectlen > (STRLEN)UNISKIP(uv)) &&
495 !(flags & UTF8_ALLOW_LONG)) {
496 warning = UTF8_WARN_LONG;
498 } else if (UNICODE_IS_ILLEGAL(uv) &&
499 !(flags & UTF8_ALLOW_FFFF)) {
500 warning = UTF8_WARN_FFFF;
508 if (flags & UTF8_CHECK_ONLY) {
515 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
518 case 0: /* Intentionally empty. */ break;
519 case UTF8_WARN_EMPTY:
520 Perl_sv_catpv(aTHX_ sv, "(empty string)");
522 case UTF8_WARN_CONTINUATION:
523 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
525 case UTF8_WARN_NON_CONTINUATION:
527 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
528 (UV)s[1], startbyte);
530 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
531 (UV)s[1], s - s0, s - s0 > 1 ? "s" : "", startbyte, (int)expectlen);
534 case UTF8_WARN_FE_FF:
535 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
537 case UTF8_WARN_SHORT:
538 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
539 (int)curlen, curlen == 1 ? "" : "s", (int)expectlen, startbyte);
540 expectlen = curlen; /* distance for caller to skip */
542 case UTF8_WARN_OVERFLOW:
543 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
546 case UTF8_WARN_SURROGATE:
547 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
550 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
551 (int)expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
554 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
557 Perl_sv_catpv(aTHX_ sv, "(unknown reason)");
562 const char *s = SvPVX_const(sv);
565 Perl_warner(aTHX_ packWARN(WARN_UTF8),
566 "%s in %s", s, OP_DESC(PL_op));
568 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
573 *retlen = expectlen ? expectlen : len;
579 =for apidoc A|UV|utf8_to_uvchr|const U8 *s|STRLEN *retlen
581 Returns the native character value of the first character in the string C<s>
582 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
583 length, in bytes, of that character.
585 If C<s> does not point to a well-formed UTF-8 character, zero is
586 returned and retlen is set, if possible, to -1.
592 Perl_utf8_to_uvchr(pTHX_ const U8 *s, STRLEN *retlen)
594 return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXBYTES, retlen,
595 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
599 =for apidoc A|UV|utf8_to_uvuni|const U8 *s|STRLEN *retlen
601 Returns the Unicode code point of the first character in the string C<s>
602 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
603 length, in bytes, of that character.
605 This function should only be used when returned UV is considered
606 an index into the Unicode semantic tables (e.g. swashes).
608 If C<s> does not point to a well-formed UTF-8 character, zero is
609 returned and retlen is set, if possible, to -1.
615 Perl_utf8_to_uvuni(pTHX_ const U8 *s, STRLEN *retlen)
617 /* Call the low level routine asking for checks */
618 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXBYTES, retlen,
619 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
623 =for apidoc A|STRLEN|utf8_length|const U8 *s|const U8 *e
625 Return the length of the UTF-8 char encoded string C<s> in characters.
626 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
627 up past C<e>, croaks.
633 Perl_utf8_length(pTHX_ const U8 *s, const U8 *e)
637 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
638 * the bitops (especially ~) can create illegal UTF-8.
639 * In other words: in Perl UTF-8 is not just for Unicode. */
642 goto warn_and_return;
644 const U8 t = UTF8SKIP(s);
647 if (ckWARN_d(WARN_UTF8)) {
649 Perl_warner(aTHX_ packWARN(WARN_UTF8),
650 "%s in %s", unees, OP_DESC(PL_op));
652 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
664 =for apidoc A|IV|utf8_distance|const U8 *a|const U8 *b
666 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
669 WARNING: use only if you *know* that the pointers point inside the
676 Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
680 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
681 * the bitops (especially ~) can create illegal UTF-8.
682 * In other words: in Perl UTF-8 is not just for Unicode. */
686 const U8 c = UTF8SKIP(a);
688 goto warn_and_return;
695 const U8 c = UTF8SKIP(b);
699 if (ckWARN_d(WARN_UTF8)) {
701 Perl_warner(aTHX_ packWARN(WARN_UTF8),
702 "%s in %s", unees, OP_DESC(PL_op));
704 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
717 =for apidoc A|U8 *|utf8_hop|U8 *s|I32 off
719 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
722 WARNING: do not use the following unless you *know* C<off> is within
723 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
724 on the first byte of character or just after the last byte of a character.
730 Perl_utf8_hop(pTHX_ const U8 *s, I32 off)
732 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
733 * the bitops (especially ~) can create illegal UTF-8.
734 * In other words: in Perl UTF-8 is not just for Unicode. */
743 while (UTF8_IS_CONTINUATION(*s))
751 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
753 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
754 Unlike C<bytes_to_utf8>, this over-writes the original string, and
755 updates len to contain the new length.
756 Returns zero on failure, setting C<len> to -1.
762 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
768 /* ensure valid UTF-8 and chars < 256 before updating string */
769 for (send = s + *len; s < send; ) {
772 if (!UTF8_IS_INVARIANT(c) &&
773 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
774 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
783 *d++ = (U8)utf8_to_uvchr(s, &ulen);
792 =for apidoc A|U8 *|bytes_from_utf8|const U8 *s|STRLEN *len|bool *is_utf8
794 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
795 Unlike C<utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
796 the newly-created string, and updates C<len> to contain the new
797 length. Returns the original string if no conversion occurs, C<len>
798 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
799 0 if C<s> is converted or contains all 7bit characters.
805 Perl_bytes_from_utf8(pTHX_ const U8 *s, STRLEN *len, bool *is_utf8)
815 /* ensure valid UTF-8 and chars < 256 before converting string */
816 for (send = s + *len; s < send;) {
818 if (!UTF8_IS_INVARIANT(c)) {
819 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
820 (c = *s++) && UTF8_IS_CONTINUATION(c))
829 Newz(801, d, (*len) - count + 1, U8);
830 s = start; start = d;
833 if (!UTF8_IS_INVARIANT(c)) {
834 /* Then it is two-byte encoded */
835 c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++);
836 c = ASCII_TO_NATIVE(c);
846 =for apidoc A|U8 *|bytes_to_utf8|const U8 *s|STRLEN *len
848 Converts a string C<s> of length C<len> from ASCII into UTF-8 encoding.
849 Returns a pointer to the newly-created string, and sets C<len> to
850 reflect the new length.
852 If you want to convert to UTF-8 from other encodings than ASCII,
853 see sv_recode_to_utf8().
859 Perl_bytes_to_utf8(pTHX_ const U8 *s, STRLEN *len)
861 const U8 * const send = s + (*len);
865 Newz(801, d, (*len) * 2 + 1, U8);
869 const UV uv = NATIVE_TO_ASCII(*s++);
870 if (UNI_IS_INVARIANT(uv))
871 *d++ = (U8)UTF_TO_NATIVE(uv);
873 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
874 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
883 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
885 * Destination must be pre-extended to 3/2 source. Do not use in-place.
886 * We optimize for native, for obvious reasons. */
889 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
894 if (bytelen == 1 && p[0] == 0) { /* Be understanding. */
901 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %"UVf, (UV)bytelen);
906 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
913 *d++ = (U8)(( uv >> 6) | 0xc0);
914 *d++ = (U8)(( uv & 0x3f) | 0x80);
917 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
918 UV low = (p[0] << 8) + p[1];
920 if (low < 0xdc00 || low >= 0xdfff)
921 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
922 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
925 *d++ = (U8)(( uv >> 12) | 0xe0);
926 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
927 *d++ = (U8)(( uv & 0x3f) | 0x80);
931 *d++ = (U8)(( uv >> 18) | 0xf0);
932 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
933 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
934 *d++ = (U8)(( uv & 0x3f) | 0x80);
938 *newlen = d - dstart;
942 /* Note: this one is slightly destructive of the source. */
945 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
948 U8* send = s + bytelen;
955 return utf16_to_utf8(p, d, bytelen, newlen);
958 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
961 Perl_is_uni_alnum(pTHX_ UV c)
963 U8 tmpbuf[UTF8_MAXBYTES+1];
964 uvchr_to_utf8(tmpbuf, c);
965 return is_utf8_alnum(tmpbuf);
969 Perl_is_uni_alnumc(pTHX_ UV c)
971 U8 tmpbuf[UTF8_MAXBYTES+1];
972 uvchr_to_utf8(tmpbuf, c);
973 return is_utf8_alnumc(tmpbuf);
977 Perl_is_uni_idfirst(pTHX_ UV c)
979 U8 tmpbuf[UTF8_MAXBYTES+1];
980 uvchr_to_utf8(tmpbuf, c);
981 return is_utf8_idfirst(tmpbuf);
985 Perl_is_uni_alpha(pTHX_ UV c)
987 U8 tmpbuf[UTF8_MAXBYTES+1];
988 uvchr_to_utf8(tmpbuf, c);
989 return is_utf8_alpha(tmpbuf);
993 Perl_is_uni_ascii(pTHX_ UV c)
995 U8 tmpbuf[UTF8_MAXBYTES+1];
996 uvchr_to_utf8(tmpbuf, c);
997 return is_utf8_ascii(tmpbuf);
1001 Perl_is_uni_space(pTHX_ UV c)
1003 U8 tmpbuf[UTF8_MAXBYTES+1];
1004 uvchr_to_utf8(tmpbuf, c);
1005 return is_utf8_space(tmpbuf);
1009 Perl_is_uni_digit(pTHX_ UV c)
1011 U8 tmpbuf[UTF8_MAXBYTES+1];
1012 uvchr_to_utf8(tmpbuf, c);
1013 return is_utf8_digit(tmpbuf);
1017 Perl_is_uni_upper(pTHX_ UV c)
1019 U8 tmpbuf[UTF8_MAXBYTES+1];
1020 uvchr_to_utf8(tmpbuf, c);
1021 return is_utf8_upper(tmpbuf);
1025 Perl_is_uni_lower(pTHX_ UV c)
1027 U8 tmpbuf[UTF8_MAXBYTES+1];
1028 uvchr_to_utf8(tmpbuf, c);
1029 return is_utf8_lower(tmpbuf);
1033 Perl_is_uni_cntrl(pTHX_ UV c)
1035 U8 tmpbuf[UTF8_MAXBYTES+1];
1036 uvchr_to_utf8(tmpbuf, c);
1037 return is_utf8_cntrl(tmpbuf);
1041 Perl_is_uni_graph(pTHX_ UV c)
1043 U8 tmpbuf[UTF8_MAXBYTES+1];
1044 uvchr_to_utf8(tmpbuf, c);
1045 return is_utf8_graph(tmpbuf);
1049 Perl_is_uni_print(pTHX_ UV c)
1051 U8 tmpbuf[UTF8_MAXBYTES+1];
1052 uvchr_to_utf8(tmpbuf, c);
1053 return is_utf8_print(tmpbuf);
1057 Perl_is_uni_punct(pTHX_ UV c)
1059 U8 tmpbuf[UTF8_MAXBYTES+1];
1060 uvchr_to_utf8(tmpbuf, c);
1061 return is_utf8_punct(tmpbuf);
1065 Perl_is_uni_xdigit(pTHX_ UV c)
1067 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1068 uvchr_to_utf8(tmpbuf, c);
1069 return is_utf8_xdigit(tmpbuf);
1073 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
1075 uvchr_to_utf8(p, c);
1076 return to_utf8_upper(p, p, lenp);
1080 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
1082 uvchr_to_utf8(p, c);
1083 return to_utf8_title(p, p, lenp);
1087 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
1089 uvchr_to_utf8(p, c);
1090 return to_utf8_lower(p, p, lenp);
1094 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1096 uvchr_to_utf8(p, c);
1097 return to_utf8_fold(p, p, lenp);
1100 /* for now these all assume no locale info available for Unicode > 255 */
1103 Perl_is_uni_alnum_lc(pTHX_ UV c)
1105 return is_uni_alnum(c); /* XXX no locale support yet */
1109 Perl_is_uni_alnumc_lc(pTHX_ UV c)
1111 return is_uni_alnumc(c); /* XXX no locale support yet */
1115 Perl_is_uni_idfirst_lc(pTHX_ UV c)
1117 return is_uni_idfirst(c); /* XXX no locale support yet */
1121 Perl_is_uni_alpha_lc(pTHX_ UV c)
1123 return is_uni_alpha(c); /* XXX no locale support yet */
1127 Perl_is_uni_ascii_lc(pTHX_ UV c)
1129 return is_uni_ascii(c); /* XXX no locale support yet */
1133 Perl_is_uni_space_lc(pTHX_ UV c)
1135 return is_uni_space(c); /* XXX no locale support yet */
1139 Perl_is_uni_digit_lc(pTHX_ UV c)
1141 return is_uni_digit(c); /* XXX no locale support yet */
1145 Perl_is_uni_upper_lc(pTHX_ UV c)
1147 return is_uni_upper(c); /* XXX no locale support yet */
1151 Perl_is_uni_lower_lc(pTHX_ UV c)
1153 return is_uni_lower(c); /* XXX no locale support yet */
1157 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1159 return is_uni_cntrl(c); /* XXX no locale support yet */
1163 Perl_is_uni_graph_lc(pTHX_ UV c)
1165 return is_uni_graph(c); /* XXX no locale support yet */
1169 Perl_is_uni_print_lc(pTHX_ UV c)
1171 return is_uni_print(c); /* XXX no locale support yet */
1175 Perl_is_uni_punct_lc(pTHX_ UV c)
1177 return is_uni_punct(c); /* XXX no locale support yet */
1181 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1183 return is_uni_xdigit(c); /* XXX no locale support yet */
1187 Perl_to_uni_upper_lc(pTHX_ U32 c)
1189 /* XXX returns only the first character -- do not use XXX */
1190 /* XXX no locale support yet */
1192 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1193 return (U32)to_uni_upper(c, tmpbuf, &len);
1197 Perl_to_uni_title_lc(pTHX_ U32 c)
1199 /* XXX returns only the first character XXX -- do not use XXX */
1200 /* XXX no locale support yet */
1202 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1203 return (U32)to_uni_title(c, tmpbuf, &len);
1207 Perl_to_uni_lower_lc(pTHX_ U32 c)
1209 /* XXX returns only the first character -- do not use XXX */
1210 /* XXX no locale support yet */
1212 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1213 return (U32)to_uni_lower(c, tmpbuf, &len);
1217 Perl_is_utf8_alnum(pTHX_ const U8 *p)
1219 if (!is_utf8_char(p))
1222 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1223 * descendant of isalnum(3), in other words, it doesn't
1224 * contain the '_'. --jhi */
1225 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1226 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1227 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1228 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1230 PL_utf8_alnum = swash_init("utf8", "",
1231 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1232 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1237 Perl_is_utf8_alnumc(pTHX_ const U8 *p)
1239 if (!is_utf8_char(p))
1242 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1243 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1244 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
1245 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1247 PL_utf8_alnum = swash_init("utf8", "",
1248 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1249 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1254 Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
1258 if (!is_utf8_char(p))
1260 if (!PL_utf8_idstart) /* is_utf8_idstart would be more logical. */
1261 PL_utf8_idstart = swash_init("utf8", "IdStart", &PL_sv_undef, 0, 0);
1262 return swash_fetch(PL_utf8_idstart, p, TRUE) != 0;
1266 Perl_is_utf8_idcont(pTHX_ const U8 *p)
1270 if (!is_utf8_char(p))
1272 if (!PL_utf8_idcont)
1273 PL_utf8_idcont = swash_init("utf8", "IdContinue", &PL_sv_undef, 0, 0);
1274 return swash_fetch(PL_utf8_idcont, p, TRUE) != 0;
1278 Perl_is_utf8_alpha(pTHX_ const U8 *p)
1280 if (!is_utf8_char(p))
1283 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1284 return swash_fetch(PL_utf8_alpha, p, TRUE) != 0;
1288 Perl_is_utf8_ascii(pTHX_ const U8 *p)
1290 if (!is_utf8_char(p))
1293 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1294 return swash_fetch(PL_utf8_ascii, p, TRUE) != 0;
1298 Perl_is_utf8_space(pTHX_ const U8 *p)
1300 if (!is_utf8_char(p))
1303 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1304 return swash_fetch(PL_utf8_space, p, TRUE) != 0;
1308 Perl_is_utf8_digit(pTHX_ const U8 *p)
1310 if (!is_utf8_char(p))
1313 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1314 return swash_fetch(PL_utf8_digit, p, TRUE) != 0;
1318 Perl_is_utf8_upper(pTHX_ const U8 *p)
1320 if (!is_utf8_char(p))
1323 PL_utf8_upper = swash_init("utf8", "IsUppercase", &PL_sv_undef, 0, 0);
1324 return swash_fetch(PL_utf8_upper, p, TRUE) != 0;
1328 Perl_is_utf8_lower(pTHX_ const U8 *p)
1330 if (!is_utf8_char(p))
1333 PL_utf8_lower = swash_init("utf8", "IsLowercase", &PL_sv_undef, 0, 0);
1334 return swash_fetch(PL_utf8_lower, p, TRUE) != 0;
1338 Perl_is_utf8_cntrl(pTHX_ const U8 *p)
1340 if (!is_utf8_char(p))
1343 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1344 return swash_fetch(PL_utf8_cntrl, p, TRUE) != 0;
1348 Perl_is_utf8_graph(pTHX_ const U8 *p)
1350 if (!is_utf8_char(p))
1353 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1354 return swash_fetch(PL_utf8_graph, p, TRUE) != 0;
1358 Perl_is_utf8_print(pTHX_ const U8 *p)
1360 if (!is_utf8_char(p))
1363 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1364 return swash_fetch(PL_utf8_print, p, TRUE) != 0;
1368 Perl_is_utf8_punct(pTHX_ const U8 *p)
1370 if (!is_utf8_char(p))
1373 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1374 return swash_fetch(PL_utf8_punct, p, TRUE) != 0;
1378 Perl_is_utf8_xdigit(pTHX_ const U8 *p)
1380 if (!is_utf8_char(p))
1382 if (!PL_utf8_xdigit)
1383 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1384 return swash_fetch(PL_utf8_xdigit, p, TRUE) != 0;
1388 Perl_is_utf8_mark(pTHX_ const U8 *p)
1390 if (!is_utf8_char(p))
1393 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1394 return swash_fetch(PL_utf8_mark, p, TRUE) != 0;
1398 =for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special
1400 The "p" contains the pointer to the UTF-8 string encoding
1401 the character that is being converted.
1403 The "ustrp" is a pointer to the character buffer to put the
1404 conversion result to. The "lenp" is a pointer to the length
1407 The "swashp" is a pointer to the swash to use.
1409 Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
1410 and loaded by SWASHGET, using lib/utf8_heavy.pl. The special (usually,
1411 but not always, a multicharacter mapping), is tried first.
1413 The "special" is a string like "utf8::ToSpecLower", which means the
1414 hash %utf8::ToSpecLower. The access to the hash is through
1415 Perl_to_utf8_case().
1417 The "normal" is a string like "ToLower" which means the swash
1423 Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp, const char *normal, const char *special)
1425 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1428 const UV uv0 = utf8_to_uvchr(p, 0);
1429 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1430 * are necessary in EBCDIC, they are redundant no-ops
1431 * in ASCII-ish platforms, and hopefully optimized away. */
1432 const UV uv1 = NATIVE_TO_UNI(uv0);
1433 uvuni_to_utf8(tmpbuf, uv1);
1435 if (!*swashp) /* load on-demand */
1436 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1438 /* The 0xDF is the only special casing Unicode code point below 0x100. */
1439 if (special && (uv1 == 0xDF || uv1 > 0xFF)) {
1440 /* It might be "special" (sometimes, but not always,
1441 * a multicharacter mapping) */
1445 if ((hv = get_hv(special, FALSE)) &&
1446 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1450 s = SvPV(*svp, len);
1452 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
1455 /* If we have EBCDIC we need to remap the characters
1456 * since any characters in the low 256 are Unicode
1457 * code points, not EBCDIC. */
1458 U8 *t = (U8*)s, *tend = t + len, *d;
1465 UV c = utf8_to_uvchr(t, &tlen);
1467 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1476 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1481 Copy(tmpbuf, ustrp, len, U8);
1483 Copy(s, ustrp, len, U8);
1489 if (!len && *swashp) {
1490 UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1493 /* It was "normal" (a single character mapping). */
1494 UV uv3 = UNI_TO_NATIVE(uv2);
1496 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
1500 if (!len) /* Neither: just copy. */
1501 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1506 return len ? utf8_to_uvchr(ustrp, 0) : 0;
1510 =for apidoc A|UV|to_utf8_upper|const U8 *p|U8 *ustrp|STRLEN *lenp
1512 Convert the UTF-8 encoded character at p to its uppercase version and
1513 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1514 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1515 the uppercase version may be longer than the original character.
1517 The first character of the uppercased version is returned
1518 (but note, as explained above, that there may be more.)
1523 Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1525 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1526 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1530 =for apidoc A|UV|to_utf8_title|const U8 *p|U8 *ustrp|STRLEN *lenp
1532 Convert the UTF-8 encoded character at p to its titlecase version and
1533 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1534 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1535 titlecase version may be longer than the original character.
1537 The first character of the titlecased version is returned
1538 (but note, as explained above, that there may be more.)
1543 Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1545 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1546 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1550 =for apidoc A|UV|to_utf8_lower|const U8 *p|U8 *ustrp|STRLEN *lenp
1552 Convert the UTF-8 encoded character at p to its lowercase version and
1553 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1554 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1555 lowercase version may be longer than the original character.
1557 The first character of the lowercased version is returned
1558 (but note, as explained above, that there may be more.)
1563 Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1565 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1566 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1570 =for apidoc A|UV|to_utf8_fold|const U8 *p|U8 *ustrp|STRLEN *lenp
1572 Convert the UTF-8 encoded character at p to its foldcase version and
1573 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1574 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1575 foldcase version may be longer than the original character (up to
1578 The first character of the foldcased version is returned
1579 (but note, as explained above, that there may be more.)
1584 Perl_to_utf8_fold(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1586 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1587 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1590 /* a "swash" is a swatch hash */
1593 Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
1597 SV* tokenbufsv = sv_newmortal();
1599 const size_t pkg_len = strlen(pkg);
1600 const size_t name_len = strlen(name);
1601 HV *stash = gv_stashpvn(pkg, pkg_len, FALSE);
1604 PUSHSTACKi(PERLSI_MAGIC);
1609 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
1611 errsv_save = newSVsv(ERRSV);
1612 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
1615 sv_setsv(ERRSV, errsv_save);
1616 SvREFCNT_dec(errsv_save);
1622 PUSHs(sv_2mortal(newSVpvn(pkg, pkg_len)));
1623 PUSHs(sv_2mortal(newSVpvn(name, name_len)));
1625 PUSHs(sv_2mortal(newSViv(minbits)));
1626 PUSHs(sv_2mortal(newSViv(none)));
1628 if (IN_PERL_COMPILETIME) {
1629 /* XXX ought to be handled by lex_start */
1632 sv_setpv(tokenbufsv, PL_tokenbuf);
1634 errsv_save = newSVsv(ERRSV);
1635 if (call_method("SWASHNEW", G_SCALAR))
1636 retval = newSVsv(*PL_stack_sp--);
1638 retval = &PL_sv_undef;
1640 sv_setsv(ERRSV, errsv_save);
1641 SvREFCNT_dec(errsv_save);
1644 if (IN_PERL_COMPILETIME) {
1646 const char* pv = SvPV(tokenbufsv, len);
1648 Copy(pv, PL_tokenbuf, len+1, char);
1649 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1651 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
1653 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
1655 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1661 /* This API is wrong for special case conversions since we may need to
1662 * return several Unicode characters for a single Unicode character
1663 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1664 * the lower-level routine, and it is similarly broken for returning
1665 * multiple values. --jhi */
1667 Perl_swash_fetch(pTHX_ SV *sv, const U8 *ptr, bool do_utf8)
1670 HV* hv = (HV*)SvRV(sv);
1679 UV c = NATIVE_TO_ASCII(*ptr);
1681 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1682 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
1683 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
1686 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
1687 * then the "swatch" is a vec() for al the chars which start
1689 * So the key in the hash (klen) is length of encoded char -1
1691 klen = UTF8SKIP(ptr) - 1;
1696 /* If char in invariant then swatch is for all the invariant chars
1697 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
1699 needents = UTF_CONTINUATION_MARK;
1700 off = NATIVE_TO_UTF(ptr[klen]);
1704 /* If char is encoded then swatch is for the prefix */
1705 needents = (1 << UTF_ACCUMULATION_SHIFT);
1706 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
1710 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1711 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1712 * it's nothing to sniff at.) Pity we usually come through at least
1713 * two function calls to get here...
1715 * NB: this code assumes that swatches are never modified, once generated!
1718 if (hv == PL_last_swash_hv &&
1719 klen == PL_last_swash_klen &&
1720 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
1722 tmps = PL_last_swash_tmps;
1723 slen = PL_last_swash_slen;
1726 /* Try our second-level swatch cache, kept in a hash. */
1727 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
1729 /* If not cached, generate it via utf8::SWASHGET */
1730 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1732 /* We use utf8n_to_uvuni() as we want an index into
1733 Unicode tables, not a native character number.
1735 UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
1737 0 : UTF8_ALLOW_ANY);
1742 PUSHSTACKi(PERLSI_MAGIC);
1746 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
1747 PUSHs(sv_2mortal(newSViv((klen) ?
1748 (code_point & ~(needents - 1)) : 0)));
1749 PUSHs(sv_2mortal(newSViv(needents)));
1751 errsv_save = newSVsv(ERRSV);
1752 if (call_method("SWASHGET", G_SCALAR))
1753 retval = newSVsv(*PL_stack_sp--);
1755 retval = &PL_sv_undef;
1757 sv_setsv(ERRSV, errsv_save);
1758 SvREFCNT_dec(errsv_save);
1762 if (IN_PERL_COMPILETIME)
1763 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1765 svp = hv_store(hv, (const char *)ptr, klen, retval, 0);
1767 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || (slen << 3) < needents)
1768 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1771 PL_last_swash_hv = hv;
1772 PL_last_swash_klen = klen;
1773 PL_last_swash_tmps = tmps;
1774 PL_last_swash_slen = slen;
1776 Copy(ptr, PL_last_swash_key, klen, U8);
1779 switch ((int)((slen << 3) / needents)) {
1781 bit = 1 << (off & 7);
1783 return (tmps[off] & bit) != 0;
1788 return (tmps[off] << 8) + tmps[off + 1] ;
1791 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1793 Perl_croak(aTHX_ "panic: swash_fetch");
1799 =for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv
1801 Adds the UTF-8 representation of the Native codepoint C<uv> to the end
1802 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
1803 bytes available. The return value is the pointer to the byte after the
1804 end of the new character. In other words,
1806 d = uvchr_to_utf8(d, uv);
1808 is the recommended wide native character-aware way of saying
1815 /* On ASCII machines this is normally a macro but we want a
1816 real function in case XS code wants it
1818 #undef Perl_uvchr_to_utf8
1820 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
1822 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
1826 Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
1828 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
1832 =for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
1834 Returns the native character value of the first character in the string C<s>
1835 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
1836 length, in bytes, of that character.
1838 Allows length and flags to be passed to low level routine.
1842 /* On ASCII machines this is normally a macro but we want
1843 a real function in case XS code wants it
1845 #undef Perl_utf8n_to_uvchr
1847 Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
1849 UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
1850 return UNI_TO_NATIVE(uv);
1854 =for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags
1856 Build to the scalar dsv a displayable version of the string spv,
1857 length len, the displayable version being at most pvlim bytes long
1858 (if longer, the rest is truncated and "..." will be appended).
1860 The flags argument can have UNI_DISPLAY_ISPRINT set to display
1861 isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
1862 to display the \\[nrfta\\] as the backslashed versions (like '\n')
1863 (UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
1864 UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
1865 UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
1867 The pointer to the PV of the dsv is returned.
1871 Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
1876 sv_setpvn(dsv, "", 0);
1877 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
1879 /* This serves double duty as a flag and a character to print after
1880 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
1884 if (pvlim && SvCUR(dsv) >= pvlim) {
1888 u = utf8_to_uvchr((U8*)s, 0);
1890 const unsigned char c = (unsigned char)u & 0xFF;
1891 if (!ok && (flags & UNI_DISPLAY_BACKSLASH)) {
1908 Perl_sv_catpvf(aTHX_ dsv, "\\%c", ok);
1911 /* isPRINT() is the locale-blind version. */
1912 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
1913 Perl_sv_catpvf(aTHX_ dsv, "%c", c);
1918 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
1921 sv_catpvn(dsv, "...", 3);
1927 =for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags
1929 Build to the scalar dsv a displayable version of the scalar sv,
1930 the displayable version being at most pvlim bytes long
1931 (if longer, the rest is truncated and "..." will be appended).
1933 The flags argument is as in pv_uni_display().
1935 The pointer to the PV of the dsv is returned.
1939 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
1941 return Perl_pv_uni_display(aTHX_ dsv, (U8*)SvPVX(ssv), SvCUR(ssv),
1946 =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
1948 Return true if the strings s1 and s2 differ case-insensitively, false
1949 if not (if they are equal case-insensitively). If u1 is true, the
1950 string s1 is assumed to be in UTF-8-encoded Unicode. If u2 is true,
1951 the string s2 is assumed to be in UTF-8-encoded Unicode. If u1 or u2
1952 are false, the respective string is assumed to be in native 8-bit
1955 If the pe1 and pe2 are non-NULL, the scanning pointers will be copied
1956 in there (they will point at the beginning of the I<next> character).
1957 If the pointers behind pe1 or pe2 are non-NULL, they are the end
1958 pointers beyond which scanning will not continue under any
1959 circumstances. If the byte lengths l1 and l2 are non-zero, s1+l1 and
1960 s2+l2 will be used as goal end pointers that will also stop the scan,
1961 and which qualify towards defining a successful match: all the scans
1962 that define an explicit length must reach their goal pointers for
1963 a match to succeed).
1965 For case-insensitiveness, the "casefolding" of Unicode is used
1966 instead of upper/lowercasing both the characters, see
1967 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
1971 Perl_ibcmp_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
1973 register const U8 *p1 = (const U8*)s1;
1974 register const U8 *p2 = (const U8*)s2;
1975 register const U8 *f1 = 0, *f2 = 0;
1976 register U8 *e1 = 0, *q1 = 0;
1977 register U8 *e2 = 0, *q2 = 0;
1978 STRLEN n1 = 0, n2 = 0;
1979 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
1980 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
1982 STRLEN foldlen1, foldlen2;
1987 if (e1 == 0 || (l1 && l1 < (UV)(e1 - (const U8*)s1)))
1988 f1 = (const U8*)s1 + l1;
1991 if (e2 == 0 || (l2 && l2 < (UV)(e2 - (const U8*)s2)))
1992 f2 = (const U8*)s2 + l2;
1994 if ((e1 == 0 && f1 == 0) || (e2 == 0 && f2 == 0) || (f1 == 0 && f2 == 0))
1995 return 1; /* mismatch; possible infinite loop or false positive */
1998 natbuf[1] = 0; /* Need to terminate the buffer. */
2000 while ((e1 == 0 || p1 < e1) &&
2001 (f1 == 0 || p1 < f1) &&
2002 (e2 == 0 || p2 < e2) &&
2003 (f2 == 0 || p2 < f2)) {
2006 to_utf8_fold(p1, foldbuf1, &foldlen1);
2009 to_utf8_fold(natbuf, foldbuf1, &foldlen1);
2016 to_utf8_fold(p2, foldbuf2, &foldlen2);
2019 to_utf8_fold(natbuf, foldbuf2, &foldlen2);
2025 if ( UTF8SKIP(q1) != UTF8SKIP(q2) ||
2026 (UTF8SKIP(q1) == 1 && *q1 != *q2) ||
2027 memNE((char*)q1, (char*)q2, UTF8SKIP(q1)) )
2028 return 1; /* mismatch */
2035 p1 += u1 ? UTF8SKIP(p1) : 1;
2037 p2 += u2 ? UTF8SKIP(p2) : 1;
2041 /* A match is defined by all the scans that specified
2042 * an explicit length reaching their final goals. */
2043 match = (f1 == 0 || p1 == f1) && (f2 == 0 || p2 == f2);
2052 return match ? 0 : 1; /* 0 match, 1 mismatch */
2057 * c-indentation-style: bsd
2059 * indent-tabs-mode: t
2062 * ex: set ts=8 sts=4 sw=4 noet: