3 * Copyright (C) 2000, 2001, 2002, 2003, 2004, 2005, 2006,
4 * by Larry Wall and others
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 */
172 Tests if some arbitrary number of bytes begins in a valid UTF-8
173 character. Note that an INVARIANT (i.e. ASCII) character is a valid
174 UTF-8 character. The actual number of bytes in the UTF-8 character
175 will be returned if it is valid, otherwise 0.
177 This is the "slow" version as opposed to the "fast" version which is
178 the "unrolled" IS_UTF8_CHAR(). E.g. for t/uni/class.t the speed
179 difference is a factor of 2 to 3. For lengths (UTF8SKIP(s)) of four
180 or less you should use the IS_UTF8_CHAR(), for lengths of five or more
181 you should use the _slow(). In practice this means that the _slow()
182 will be used very rarely, since the maximum Unicode code point (as of
183 Unicode 4.1) is U+10FFFF, which encodes in UTF-8 to four bytes. Only
184 the "Perl extended UTF-8" (the infamous 'v-strings') will encode into
189 S_is_utf8_char_slow(const U8 *s, const STRLEN len)
195 if (UTF8_IS_INVARIANT(u))
198 if (!UTF8_IS_START(u))
201 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
207 u = NATIVE_TO_UTF(u);
209 u &= UTF_START_MASK(len);
213 if (!UTF8_IS_CONTINUATION(*s))
215 uv = UTF8_ACCUMULATE(uv, *s);
222 if ((STRLEN)UNISKIP(uv) < len)
229 =for apidoc A|STRLEN|is_utf8_char|const U8 *s
231 Tests if some arbitrary number of bytes begins in a valid UTF-8
232 character. Note that an INVARIANT (i.e. ASCII) character is a valid
233 UTF-8 character. The actual number of bytes in the UTF-8 character
234 will be returned if it is valid, otherwise 0.
238 Perl_is_utf8_char(pTHX_ const U8 *s)
240 const STRLEN len = UTF8SKIP(s);
243 if (IS_UTF8_CHAR_FAST(len))
244 return IS_UTF8_CHAR(s, len) ? len : 0;
245 #endif /* #ifdef IS_UTF8_CHAR */
246 return is_utf8_char_slow(s, len);
250 =for apidoc A|bool|is_utf8_string|const U8 *s|STRLEN len
252 Returns true if first C<len> bytes of the given string form a valid
253 UTF-8 string, false otherwise. Note that 'a valid UTF-8 string' does
254 not mean 'a string that contains code points above 0x7F encoded in UTF-8'
255 because a valid ASCII string is a valid UTF-8 string.
257 See also is_utf8_string_loclen() and is_utf8_string_loc().
263 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. */
284 if (IS_UTF8_CHAR_FAST(c)) {
285 if (!IS_UTF8_CHAR(x, c))
289 c = is_utf8_char_slow(x, c);
292 #endif /* #ifdef IS_UTF8_CHAR */
307 Implemented as a macro in utf8.h
309 =for apidoc A|bool|is_utf8_string_loc|const U8 *s|STRLEN len|const U8 **ep
311 Like is_utf8_string() but stores the location of the failure (in the
312 case of "utf8ness failure") or the location s+len (in the case of
313 "utf8ness success") in the C<ep>.
315 See also is_utf8_string_loclen() and is_utf8_string().
317 =for apidoc A|bool|is_utf8_string_loclen|const U8 *s|STRLEN len|const U8 **ep|const STRLEN *el
319 Like is_utf8_string() but stores the location of the failure (in the
320 case of "utf8ness failure") or the location s+len (in the case of
321 "utf8ness success") in the C<ep>, and the number of UTF-8
322 encoded characters in the C<el>.
324 See also is_utf8_string_loc() and is_utf8_string().
330 Perl_is_utf8_string_loclen(pTHX_ const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
338 len = strlen((const char *)s);
344 /* Inline the easy bits of is_utf8_char() here for speed... */
345 if (UTF8_IS_INVARIANT(*x))
347 else if (!UTF8_IS_START(*x))
350 /* ... and call is_utf8_char() only if really needed. */
353 if (IS_UTF8_CHAR_FAST(c)) {
354 if (!IS_UTF8_CHAR(x, c))
357 c = is_utf8_char_slow(x, c);
360 #endif /* #ifdef IS_UTF8_CHAR */
380 =for apidoc A|UV|utf8n_to_uvuni|const U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
382 Bottom level UTF-8 decode routine.
383 Returns the unicode code point value of the first character in the string C<s>
384 which is assumed to be in UTF-8 encoding and no longer than C<curlen>;
385 C<retlen> will be set to the length, in bytes, of that character.
387 If C<s> does not point to a well-formed UTF-8 character, the behaviour
388 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
389 it is assumed that the caller will raise a warning, and this function
390 will silently just set C<retlen> to C<-1> and return zero. If the
391 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
392 malformations will be given, C<retlen> will be set to the expected
393 length of the UTF-8 character in bytes, and zero will be returned.
395 The C<flags> can also contain various flags to allow deviations from
396 the strict UTF-8 encoding (see F<utf8.h>).
398 Most code should use utf8_to_uvchr() rather than call this directly.
404 Perl_utf8n_to_uvuni(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
407 const U8 * const s0 = s;
410 const bool dowarn = ckWARN_d(WARN_UTF8);
411 const UV startbyte = *s;
412 STRLEN expectlen = 0;
415 /* This list is a superset of the UTF8_ALLOW_XXX. */
417 #define UTF8_WARN_EMPTY 1
418 #define UTF8_WARN_CONTINUATION 2
419 #define UTF8_WARN_NON_CONTINUATION 3
420 #define UTF8_WARN_FE_FF 4
421 #define UTF8_WARN_SHORT 5
422 #define UTF8_WARN_OVERFLOW 6
423 #define UTF8_WARN_SURROGATE 7
424 #define UTF8_WARN_LONG 8
425 #define UTF8_WARN_FFFF 9 /* Also FFFE. */
428 !(flags & UTF8_ALLOW_EMPTY)) {
429 warning = UTF8_WARN_EMPTY;
433 if (UTF8_IS_INVARIANT(uv)) {
436 return (UV) (NATIVE_TO_UTF(*s));
439 if (UTF8_IS_CONTINUATION(uv) &&
440 !(flags & UTF8_ALLOW_CONTINUATION)) {
441 warning = UTF8_WARN_CONTINUATION;
445 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
446 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
447 warning = UTF8_WARN_NON_CONTINUATION;
452 uv = NATIVE_TO_UTF(uv);
454 if ((uv == 0xfe || uv == 0xff) &&
455 !(flags & UTF8_ALLOW_FE_FF)) {
456 warning = UTF8_WARN_FE_FF;
461 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
462 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
463 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
464 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
466 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
467 else { len = 7; uv &= 0x01; }
469 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
470 else if (!(uv & 0x01)) { len = 7; uv = 0; }
471 else { len = 13; uv = 0; } /* whoa! */
479 if ((curlen < expectlen) &&
480 !(flags & UTF8_ALLOW_SHORT)) {
481 warning = UTF8_WARN_SHORT;
490 if (!UTF8_IS_CONTINUATION(*s) &&
491 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
493 warning = UTF8_WARN_NON_CONTINUATION;
497 uv = UTF8_ACCUMULATE(uv, *s);
499 /* These cannot be allowed. */
501 if (expectlen != 13 && !(flags & UTF8_ALLOW_LONG)) {
502 warning = UTF8_WARN_LONG;
506 else { /* uv < ouv */
507 /* This cannot be allowed. */
508 warning = UTF8_WARN_OVERFLOW;
516 if (UNICODE_IS_SURROGATE(uv) &&
517 !(flags & UTF8_ALLOW_SURROGATE)) {
518 warning = UTF8_WARN_SURROGATE;
520 } else if ((expectlen > (STRLEN)UNISKIP(uv)) &&
521 !(flags & UTF8_ALLOW_LONG)) {
522 warning = UTF8_WARN_LONG;
524 } else if (UNICODE_IS_ILLEGAL(uv) &&
525 !(flags & UTF8_ALLOW_FFFF)) {
526 warning = UTF8_WARN_FFFF;
534 if (flags & UTF8_CHECK_ONLY) {
541 SV* const sv = sv_2mortal(newSVpvs("Malformed UTF-8 character "));
544 case 0: /* Intentionally empty. */ break;
545 case UTF8_WARN_EMPTY:
546 sv_catpvs(sv, "(empty string)");
548 case UTF8_WARN_CONTINUATION:
549 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
551 case UTF8_WARN_NON_CONTINUATION:
553 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
554 (UV)s[1], startbyte);
556 const int len = (int)(s-s0);
557 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
558 (UV)s[1], len, len > 1 ? "s" : "", startbyte, (int)expectlen);
562 case UTF8_WARN_FE_FF:
563 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
565 case UTF8_WARN_SHORT:
566 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
567 (int)curlen, curlen == 1 ? "" : "s", (int)expectlen, startbyte);
568 expectlen = curlen; /* distance for caller to skip */
570 case UTF8_WARN_OVERFLOW:
571 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
574 case UTF8_WARN_SURROGATE:
575 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
578 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
579 (int)expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
582 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
585 sv_catpvs(sv, "(unknown reason)");
590 const char * const s = SvPVX_const(sv);
593 Perl_warner(aTHX_ packWARN(WARN_UTF8),
594 "%s in %s", s, OP_DESC(PL_op));
596 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
601 *retlen = expectlen ? expectlen : len;
607 =for apidoc A|UV|utf8_to_uvchr|const U8 *s|STRLEN *retlen
609 Returns the native character value of the first character in the string C<s>
610 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
611 length, in bytes, of that character.
613 If C<s> does not point to a well-formed UTF-8 character, zero is
614 returned and retlen is set, if possible, to -1.
620 Perl_utf8_to_uvchr(pTHX_ const U8 *s, STRLEN *retlen)
622 return utf8n_to_uvchr(s, UTF8_MAXBYTES, retlen,
623 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
627 =for apidoc A|UV|utf8_to_uvuni|const U8 *s|STRLEN *retlen
629 Returns the Unicode code point of the first character in the string C<s>
630 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
631 length, in bytes, of that character.
633 This function should only be used when returned UV is considered
634 an index into the Unicode semantic tables (e.g. swashes).
636 If C<s> does not point to a well-formed UTF-8 character, zero is
637 returned and retlen is set, if possible, to -1.
643 Perl_utf8_to_uvuni(pTHX_ const U8 *s, STRLEN *retlen)
645 /* Call the low level routine asking for checks */
646 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXBYTES, retlen,
647 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
651 =for apidoc A|STRLEN|utf8_length|const U8 *s|const U8 *e
653 Return the length of the UTF-8 char encoded string C<s> in characters.
654 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
655 up past C<e>, croaks.
661 Perl_utf8_length(pTHX_ const U8 *s, const U8 *e)
666 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
667 * the bitops (especially ~) can create illegal UTF-8.
668 * In other words: in Perl UTF-8 is not just for Unicode. */
671 goto warn_and_return;
673 const U8 t = UTF8SKIP(s);
676 if (ckWARN_d(WARN_UTF8)) {
678 Perl_warner(aTHX_ packWARN(WARN_UTF8),
679 "%s in %s", unees, OP_DESC(PL_op));
681 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
693 =for apidoc A|IV|utf8_distance|const U8 *a|const U8 *b
695 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
698 WARNING: use only if you *know* that the pointers point inside the
705 Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
707 return (a < b) ? -1 * (IV) utf8_length(a, b) : (IV) utf8_length(b, a);
711 =for apidoc A|U8 *|utf8_hop|U8 *s|I32 off
713 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
716 WARNING: do not use the following unless you *know* C<off> is within
717 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
718 on the first byte of character or just after the last byte of a character.
724 Perl_utf8_hop(pTHX_ const U8 *s, I32 off)
727 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
728 * the bitops (especially ~) can create illegal UTF-8.
729 * In other words: in Perl UTF-8 is not just for Unicode. */
738 while (UTF8_IS_CONTINUATION(*s))
746 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
748 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
749 Unlike C<bytes_to_utf8>, this over-writes the original string, and
750 updates len to contain the new length.
751 Returns zero on failure, setting C<len> to -1.
753 If you need a copy of the string, see C<bytes_from_utf8>.
759 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
762 U8 * const send = s + *len;
765 /* ensure valid UTF-8 and chars < 256 before updating string */
769 if (!UTF8_IS_INVARIANT(c) &&
770 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
771 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
780 *d++ = (U8)utf8_to_uvchr(s, &ulen);
789 =for apidoc A|U8 *|bytes_from_utf8|const U8 *s|STRLEN *len|bool *is_utf8
791 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
792 Unlike C<utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
793 the newly-created string, and updates C<len> to contain the new
794 length. Returns the original string if no conversion occurs, C<len>
795 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
796 0 if C<s> is converted or contains all 7bit characters.
802 Perl_bytes_from_utf8(pTHX_ const U8 *s, STRLEN *len, bool *is_utf8)
813 /* ensure valid UTF-8 and chars < 256 before converting string */
814 for (send = s + *len; s < send;) {
816 if (!UTF8_IS_INVARIANT(c)) {
817 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
818 (c = *s++) && UTF8_IS_CONTINUATION(c))
827 Newx(d, (*len) - count + 1, U8);
828 s = start; start = d;
831 if (!UTF8_IS_INVARIANT(c)) {
832 /* Then it is two-byte encoded */
833 c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++);
834 c = ASCII_TO_NATIVE(c);
844 =for apidoc A|U8 *|bytes_to_utf8|const U8 *s|STRLEN *len
846 Converts a string C<s> of length C<len> from ASCII into UTF-8 encoding.
847 Returns a pointer to the newly-created string, and sets C<len> to
848 reflect the new length.
850 If you want to convert to UTF-8 from other encodings than ASCII,
851 see sv_recode_to_utf8().
857 Perl_bytes_to_utf8(pTHX_ const U8 *s, STRLEN *len)
859 const U8 * const send = s + (*len);
864 Newx(d, (*len) * 2 + 1, U8);
868 const UV uv = NATIVE_TO_ASCII(*s++);
869 if (UNI_IS_INVARIANT(uv))
870 *d++ = (U8)UTF_TO_NATIVE(uv);
872 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
873 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
882 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
884 * Destination must be pre-extended to 3/2 source. Do not use in-place.
885 * We optimize for native, for obvious reasons. */
888 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
893 if (bytelen == 1 && p[0] == 0) { /* Be understanding. */
900 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %"UVf, (UV)bytelen);
905 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
912 *d++ = (U8)(( uv >> 6) | 0xc0);
913 *d++ = (U8)(( uv & 0x3f) | 0x80);
916 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
917 UV low = (p[0] << 8) + p[1];
919 if (low < 0xdc00 || low >= 0xdfff)
920 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
921 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
924 *d++ = (U8)(( uv >> 12) | 0xe0);
925 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
926 *d++ = (U8)(( uv & 0x3f) | 0x80);
930 *d++ = (U8)(( uv >> 18) | 0xf0);
931 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
932 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
933 *d++ = (U8)(( uv & 0x3f) | 0x80);
937 *newlen = d - dstart;
941 /* Note: this one is slightly destructive of the source. */
944 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
947 U8* const send = s + bytelen;
954 return utf16_to_utf8(p, d, bytelen, newlen);
957 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
960 Perl_is_uni_alnum(pTHX_ UV c)
962 U8 tmpbuf[UTF8_MAXBYTES+1];
963 uvchr_to_utf8(tmpbuf, c);
964 return is_utf8_alnum(tmpbuf);
968 Perl_is_uni_alnumc(pTHX_ UV c)
970 U8 tmpbuf[UTF8_MAXBYTES+1];
971 uvchr_to_utf8(tmpbuf, c);
972 return is_utf8_alnumc(tmpbuf);
976 Perl_is_uni_idfirst(pTHX_ UV c)
978 U8 tmpbuf[UTF8_MAXBYTES+1];
979 uvchr_to_utf8(tmpbuf, c);
980 return is_utf8_idfirst(tmpbuf);
984 Perl_is_uni_alpha(pTHX_ UV c)
986 U8 tmpbuf[UTF8_MAXBYTES+1];
987 uvchr_to_utf8(tmpbuf, c);
988 return is_utf8_alpha(tmpbuf);
992 Perl_is_uni_ascii(pTHX_ UV c)
994 U8 tmpbuf[UTF8_MAXBYTES+1];
995 uvchr_to_utf8(tmpbuf, c);
996 return is_utf8_ascii(tmpbuf);
1000 Perl_is_uni_space(pTHX_ UV c)
1002 U8 tmpbuf[UTF8_MAXBYTES+1];
1003 uvchr_to_utf8(tmpbuf, c);
1004 return is_utf8_space(tmpbuf);
1008 Perl_is_uni_digit(pTHX_ UV c)
1010 U8 tmpbuf[UTF8_MAXBYTES+1];
1011 uvchr_to_utf8(tmpbuf, c);
1012 return is_utf8_digit(tmpbuf);
1016 Perl_is_uni_upper(pTHX_ UV c)
1018 U8 tmpbuf[UTF8_MAXBYTES+1];
1019 uvchr_to_utf8(tmpbuf, c);
1020 return is_utf8_upper(tmpbuf);
1024 Perl_is_uni_lower(pTHX_ UV c)
1026 U8 tmpbuf[UTF8_MAXBYTES+1];
1027 uvchr_to_utf8(tmpbuf, c);
1028 return is_utf8_lower(tmpbuf);
1032 Perl_is_uni_cntrl(pTHX_ UV c)
1034 U8 tmpbuf[UTF8_MAXBYTES+1];
1035 uvchr_to_utf8(tmpbuf, c);
1036 return is_utf8_cntrl(tmpbuf);
1040 Perl_is_uni_graph(pTHX_ UV c)
1042 U8 tmpbuf[UTF8_MAXBYTES+1];
1043 uvchr_to_utf8(tmpbuf, c);
1044 return is_utf8_graph(tmpbuf);
1048 Perl_is_uni_print(pTHX_ UV c)
1050 U8 tmpbuf[UTF8_MAXBYTES+1];
1051 uvchr_to_utf8(tmpbuf, c);
1052 return is_utf8_print(tmpbuf);
1056 Perl_is_uni_punct(pTHX_ UV c)
1058 U8 tmpbuf[UTF8_MAXBYTES+1];
1059 uvchr_to_utf8(tmpbuf, c);
1060 return is_utf8_punct(tmpbuf);
1064 Perl_is_uni_xdigit(pTHX_ UV c)
1066 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1067 uvchr_to_utf8(tmpbuf, c);
1068 return is_utf8_xdigit(tmpbuf);
1072 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
1074 uvchr_to_utf8(p, c);
1075 return to_utf8_upper(p, p, lenp);
1079 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
1081 uvchr_to_utf8(p, c);
1082 return to_utf8_title(p, p, lenp);
1086 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
1088 uvchr_to_utf8(p, c);
1089 return to_utf8_lower(p, p, lenp);
1093 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1095 uvchr_to_utf8(p, c);
1096 return to_utf8_fold(p, p, lenp);
1099 /* for now these all assume no locale info available for Unicode > 255 */
1102 Perl_is_uni_alnum_lc(pTHX_ UV c)
1104 return is_uni_alnum(c); /* XXX no locale support yet */
1108 Perl_is_uni_alnumc_lc(pTHX_ UV c)
1110 return is_uni_alnumc(c); /* XXX no locale support yet */
1114 Perl_is_uni_idfirst_lc(pTHX_ UV c)
1116 return is_uni_idfirst(c); /* XXX no locale support yet */
1120 Perl_is_uni_alpha_lc(pTHX_ UV c)
1122 return is_uni_alpha(c); /* XXX no locale support yet */
1126 Perl_is_uni_ascii_lc(pTHX_ UV c)
1128 return is_uni_ascii(c); /* XXX no locale support yet */
1132 Perl_is_uni_space_lc(pTHX_ UV c)
1134 return is_uni_space(c); /* XXX no locale support yet */
1138 Perl_is_uni_digit_lc(pTHX_ UV c)
1140 return is_uni_digit(c); /* XXX no locale support yet */
1144 Perl_is_uni_upper_lc(pTHX_ UV c)
1146 return is_uni_upper(c); /* XXX no locale support yet */
1150 Perl_is_uni_lower_lc(pTHX_ UV c)
1152 return is_uni_lower(c); /* XXX no locale support yet */
1156 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1158 return is_uni_cntrl(c); /* XXX no locale support yet */
1162 Perl_is_uni_graph_lc(pTHX_ UV c)
1164 return is_uni_graph(c); /* XXX no locale support yet */
1168 Perl_is_uni_print_lc(pTHX_ UV c)
1170 return is_uni_print(c); /* XXX no locale support yet */
1174 Perl_is_uni_punct_lc(pTHX_ UV c)
1176 return is_uni_punct(c); /* XXX no locale support yet */
1180 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1182 return is_uni_xdigit(c); /* XXX no locale support yet */
1186 Perl_to_uni_upper_lc(pTHX_ U32 c)
1188 /* XXX returns only the first character -- do not use XXX */
1189 /* XXX no locale support yet */
1191 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1192 return (U32)to_uni_upper(c, tmpbuf, &len);
1196 Perl_to_uni_title_lc(pTHX_ U32 c)
1198 /* XXX returns only the first character XXX -- do not use XXX */
1199 /* XXX no locale support yet */
1201 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1202 return (U32)to_uni_title(c, tmpbuf, &len);
1206 Perl_to_uni_lower_lc(pTHX_ U32 c)
1208 /* XXX returns only the first character -- do not use XXX */
1209 /* XXX no locale support yet */
1211 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1212 return (U32)to_uni_lower(c, tmpbuf, &len);
1216 S_is_utf8_common(pTHX_ const U8 *const p, SV **swash,
1217 const char *const swashname)
1220 if (!is_utf8_char(p))
1223 *swash = swash_init("utf8", swashname, &PL_sv_undef, 1, 0);
1224 return swash_fetch(*swash, p, TRUE) != 0;
1228 Perl_is_utf8_alnum(pTHX_ const U8 *p)
1231 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1232 * descendant of isalnum(3), in other words, it doesn't
1233 * contain the '_'. --jhi */
1234 return is_utf8_common(p, &PL_utf8_alnum, "IsWord");
1238 Perl_is_utf8_alnumc(pTHX_ const U8 *p)
1241 return is_utf8_common(p, &PL_utf8_alnumc, "IsAlnumC");
1245 Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
1250 /* is_utf8_idstart would be more logical. */
1251 return is_utf8_common(p, &PL_utf8_idstart, "IdStart");
1255 Perl_is_utf8_idcont(pTHX_ const U8 *p)
1260 return is_utf8_common(p, &PL_utf8_idcont, "IdContinue");
1264 Perl_is_utf8_alpha(pTHX_ const U8 *p)
1267 return is_utf8_common(p, &PL_utf8_alpha, "IsAlpha");
1271 Perl_is_utf8_ascii(pTHX_ const U8 *p)
1274 return is_utf8_common(p, &PL_utf8_ascii, "IsAscii");
1278 Perl_is_utf8_space(pTHX_ const U8 *p)
1281 return is_utf8_common(p, &PL_utf8_space, "IsSpacePerl");
1285 Perl_is_utf8_digit(pTHX_ const U8 *p)
1288 return is_utf8_common(p, &PL_utf8_digit, "IsDigit");
1292 Perl_is_utf8_upper(pTHX_ const U8 *p)
1295 return is_utf8_common(p, &PL_utf8_upper, "IsUppercase");
1299 Perl_is_utf8_lower(pTHX_ const U8 *p)
1302 return is_utf8_common(p, &PL_utf8_lower, "IsLowercase");
1306 Perl_is_utf8_cntrl(pTHX_ const U8 *p)
1309 return is_utf8_common(p, &PL_utf8_cntrl, "IsCntrl");
1313 Perl_is_utf8_graph(pTHX_ const U8 *p)
1316 return is_utf8_common(p, &PL_utf8_graph, "IsGraph");
1320 Perl_is_utf8_print(pTHX_ const U8 *p)
1323 return is_utf8_common(p, &PL_utf8_print, "IsPrint");
1327 Perl_is_utf8_punct(pTHX_ const U8 *p)
1330 return is_utf8_common(p, &PL_utf8_punct, "IsPunct");
1334 Perl_is_utf8_xdigit(pTHX_ const U8 *p)
1337 return is_utf8_common(p, &PL_utf8_xdigit, "Isxdigit");
1341 Perl_is_utf8_mark(pTHX_ const U8 *p)
1344 return is_utf8_common(p, &PL_utf8_mark, "IsM");
1348 =for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special
1350 The "p" contains the pointer to the UTF-8 string encoding
1351 the character that is being converted.
1353 The "ustrp" is a pointer to the character buffer to put the
1354 conversion result to. The "lenp" is a pointer to the length
1357 The "swashp" is a pointer to the swash to use.
1359 Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
1360 and loaded by SWASHNEW, using lib/utf8_heavy.pl. The special (usually,
1361 but not always, a multicharacter mapping), is tried first.
1363 The "special" is a string like "utf8::ToSpecLower", which means the
1364 hash %utf8::ToSpecLower. The access to the hash is through
1365 Perl_to_utf8_case().
1367 The "normal" is a string like "ToLower" which means the swash
1373 Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp,
1374 SV **swashp, const char *normal, const char *special)
1377 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1380 const UV uv0 = utf8_to_uvchr(p, NULL);
1381 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1382 * are necessary in EBCDIC, they are redundant no-ops
1383 * in ASCII-ish platforms, and hopefully optimized away. */
1384 const UV uv1 = NATIVE_TO_UNI(uv0);
1385 uvuni_to_utf8(tmpbuf, uv1);
1387 if (!*swashp) /* load on-demand */
1388 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1390 /* The 0xDF is the only special casing Unicode code point below 0x100. */
1391 if (special && (uv1 == 0xDF || uv1 > 0xFF)) {
1392 /* It might be "special" (sometimes, but not always,
1393 * a multicharacter mapping) */
1397 if ((hv = get_hv(special, FALSE)) &&
1398 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1402 s = SvPV_const(*svp, len);
1404 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
1407 /* If we have EBCDIC we need to remap the characters
1408 * since any characters in the low 256 are Unicode
1409 * code points, not EBCDIC. */
1410 U8 *t = (U8*)s, *tend = t + len, *d;
1417 const UV c = utf8_to_uvchr(t, &tlen);
1419 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1428 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1433 Copy(tmpbuf, ustrp, len, U8);
1435 Copy(s, ustrp, len, U8);
1441 if (!len && *swashp) {
1442 const UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1445 /* It was "normal" (a single character mapping). */
1446 const UV uv3 = UNI_TO_NATIVE(uv2);
1447 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
1451 if (!len) /* Neither: just copy. */
1452 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1457 return len ? utf8_to_uvchr(ustrp, 0) : 0;
1461 =for apidoc A|UV|to_utf8_upper|const U8 *p|U8 *ustrp|STRLEN *lenp
1463 Convert the UTF-8 encoded character at p to its uppercase version and
1464 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1465 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1466 the uppercase version may be longer than the original character.
1468 The first character of the uppercased version is returned
1469 (but note, as explained above, that there may be more.)
1474 Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1477 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1478 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1482 =for apidoc A|UV|to_utf8_title|const U8 *p|U8 *ustrp|STRLEN *lenp
1484 Convert the UTF-8 encoded character at p to its titlecase version and
1485 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1486 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1487 titlecase version may be longer than the original character.
1489 The first character of the titlecased version is returned
1490 (but note, as explained above, that there may be more.)
1495 Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1498 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1499 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1503 =for apidoc A|UV|to_utf8_lower|const U8 *p|U8 *ustrp|STRLEN *lenp
1505 Convert the UTF-8 encoded character at p to its lowercase version and
1506 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1507 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1508 lowercase version may be longer than the original character.
1510 The first character of the lowercased version is returned
1511 (but note, as explained above, that there may be more.)
1516 Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1519 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1520 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1524 =for apidoc A|UV|to_utf8_fold|const U8 *p|U8 *ustrp|STRLEN *lenp
1526 Convert the UTF-8 encoded character at p to its foldcase version and
1527 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1528 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1529 foldcase version may be longer than the original character (up to
1532 The first character of the foldcased version is returned
1533 (but note, as explained above, that there may be more.)
1538 Perl_to_utf8_fold(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1541 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1542 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1546 * A "swash" is a swatch hash.
1547 * A "swatch" is a bit vector generated by utf8.c:S_swash_get().
1548 * C<pkg> is a pointer to a package name for SWASHNEW, should be "utf8".
1549 * For other parameters, see utf8::SWASHNEW in lib/utf8_heavy.pl.
1552 Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
1556 SV* const tokenbufsv = sv_newmortal();
1558 const size_t pkg_len = strlen(pkg);
1559 const size_t name_len = strlen(name);
1560 HV * const stash = gv_stashpvn(pkg, pkg_len, FALSE);
1563 PUSHSTACKi(PERLSI_MAGIC);
1568 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
1570 errsv_save = newSVsv(ERRSV);
1571 /* It is assumed that callers of this routine are not passing in any
1572 user derived data. */
1573 /* Need to do this after save_re_context() as it will set PL_tainted to
1574 1 while saving $1 etc (see the code after getrx: in Perl_magic_get).
1575 Even line to create errsv_save can turn on PL_tainted. */
1576 SAVEBOOL(PL_tainted);
1578 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
1581 sv_setsv(ERRSV, errsv_save);
1582 SvREFCNT_dec(errsv_save);
1588 PUSHs(sv_2mortal(newSVpvn(pkg, pkg_len)));
1589 PUSHs(sv_2mortal(newSVpvn(name, name_len)));
1591 PUSHs(sv_2mortal(newSViv(minbits)));
1592 PUSHs(sv_2mortal(newSViv(none)));
1594 if (IN_PERL_COMPILETIME) {
1595 /* XXX ought to be handled by lex_start */
1598 sv_setpv(tokenbufsv, PL_tokenbuf);
1600 errsv_save = newSVsv(ERRSV);
1601 if (call_method("SWASHNEW", G_SCALAR))
1602 retval = newSVsv(*PL_stack_sp--);
1604 retval = &PL_sv_undef;
1606 sv_setsv(ERRSV, errsv_save);
1607 SvREFCNT_dec(errsv_save);
1610 if (IN_PERL_COMPILETIME) {
1612 const char* const pv = SvPV_const(tokenbufsv, len);
1614 Copy(pv, PL_tokenbuf, len+1, char);
1615 CopHINTS_set(PL_curcop, PL_hints);
1617 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
1619 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
1621 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1627 /* This API is wrong for special case conversions since we may need to
1628 * return several Unicode characters for a single Unicode character
1629 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1630 * the lower-level routine, and it is similarly broken for returning
1631 * multiple values. --jhi */
1632 /* Now SWASHGET is recasted into S_swash_get in this file. */
1635 * Returns the value of property/mapping C<swash> for the first character
1636 * of the string C<ptr>. If C<do_utf8> is true, the string C<ptr> is
1637 * assumed to be in utf8. If C<do_utf8> is false, the string C<ptr> is
1638 * assumed to be in native 8-bit encoding. Caches the swatch in C<swash>.
1641 Perl_swash_fetch(pTHX_ SV *swash, const U8 *ptr, bool do_utf8)
1644 HV* const hv = (HV*)SvRV(swash);
1649 const U8 *tmps = NULL;
1653 UV c = NATIVE_TO_ASCII(*ptr);
1655 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1656 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
1657 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
1660 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
1661 * then the "swatch" is a vec() for al the chars which start
1663 * So the key in the hash (klen) is length of encoded char -1
1665 klen = UTF8SKIP(ptr) - 1;
1669 /* If char in invariant then swatch is for all the invariant chars
1670 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
1672 needents = UTF_CONTINUATION_MARK;
1673 off = NATIVE_TO_UTF(ptr[klen]);
1676 /* If char is encoded then swatch is for the prefix */
1677 needents = (1 << UTF_ACCUMULATION_SHIFT);
1678 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
1682 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1683 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1684 * it's nothing to sniff at.) Pity we usually come through at least
1685 * two function calls to get here...
1687 * NB: this code assumes that swatches are never modified, once generated!
1690 if (hv == PL_last_swash_hv &&
1691 klen == PL_last_swash_klen &&
1692 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
1694 tmps = PL_last_swash_tmps;
1695 slen = PL_last_swash_slen;
1698 /* Try our second-level swatch cache, kept in a hash. */
1699 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
1701 /* If not cached, generate it via swash_get */
1702 if (!svp || !SvPOK(*svp)
1703 || !(tmps = (const U8*)SvPV_const(*svp, slen))) {
1704 /* We use utf8n_to_uvuni() as we want an index into
1705 Unicode tables, not a native character number.
1707 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
1709 0 : UTF8_ALLOW_ANY);
1710 swatch = swash_get(swash,
1711 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
1712 (klen) ? (code_point & ~(needents - 1)) : 0,
1715 if (IN_PERL_COMPILETIME)
1716 CopHINTS_set(PL_curcop, PL_hints);
1718 svp = hv_store(hv, (const char *)ptr, klen, swatch, 0);
1720 if (!svp || !(tmps = (U8*)SvPV(*svp, slen))
1721 || (slen << 3) < needents)
1722 Perl_croak(aTHX_ "panic: swash_fetch got improper swatch");
1725 PL_last_swash_hv = hv;
1726 PL_last_swash_klen = klen;
1727 /* FIXME change interpvar.h? */
1728 PL_last_swash_tmps = (U8 *) tmps;
1729 PL_last_swash_slen = slen;
1731 Copy(ptr, PL_last_swash_key, klen, U8);
1734 switch ((int)((slen << 3) / needents)) {
1736 bit = 1 << (off & 7);
1738 return (tmps[off] & bit) != 0;
1743 return (tmps[off] << 8) + tmps[off + 1] ;
1746 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1748 Perl_croak(aTHX_ "panic: swash_fetch got swatch of unexpected bit width");
1752 * Returns a swatch (a bit vector string) for a code point sequence
1753 * that starts from the value C<start> and comprises the number C<span>.
1754 * A C<swash> must be an object created by SWASHNEW (see lib/utf8_heavy.pl).
1755 * Should be used via swash_fetch, which will cache the swatch in C<swash>.
1758 S_swash_get(pTHX_ SV* swash, UV start, UV span)
1761 U8 *l, *lend, *x, *xend, *s;
1762 STRLEN lcur, xcur, scur;
1764 HV* const hv = (HV*)SvRV(swash);
1765 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
1766 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
1767 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
1768 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
1769 SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);
1770 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
1771 const int typeto = typestr[0] == 'T' && typestr[1] == 'o';
1772 const STRLEN bits = SvUV(*bitssvp);
1773 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
1774 const UV none = SvUV(*nonesvp);
1775 const UV end = start + span;
1777 if (bits != 1 && bits != 8 && bits != 16 && bits != 32) {
1778 Perl_croak(aTHX_ "panic: swash_get doesn't expect bits %"UVuf,
1782 /* create and initialize $swatch */
1783 swatch = newSVpvs("");
1784 scur = octets ? (span * octets) : (span + 7) / 8;
1785 SvGROW(swatch, scur + 1);
1786 s = (U8*)SvPVX(swatch);
1787 if (octets && none) {
1788 const U8* const e = s + scur;
1791 *s++ = (U8)(none & 0xff);
1792 else if (bits == 16) {
1793 *s++ = (U8)((none >> 8) & 0xff);
1794 *s++ = (U8)( none & 0xff);
1796 else if (bits == 32) {
1797 *s++ = (U8)((none >> 24) & 0xff);
1798 *s++ = (U8)((none >> 16) & 0xff);
1799 *s++ = (U8)((none >> 8) & 0xff);
1800 *s++ = (U8)( none & 0xff);
1806 (void)memzero((U8*)s, scur + 1);
1808 SvCUR_set(swatch, scur);
1809 s = (U8*)SvPVX(swatch);
1811 /* read $swash->{LIST} */
1812 l = (U8*)SvPV(*listsvp, lcur);
1815 UV min, max, val, key;
1817 I32 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
1819 U8* const nl = (U8*)memchr(l, '\n', lend - l);
1822 min = grok_hex((char *)l, &numlen, &flags, NULL);
1826 l = nl + 1; /* 1 is length of "\n" */
1830 l = lend; /* to LIST's end at which \n is not found */
1836 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
1838 max = grok_hex((char *)l, &numlen, &flags, NULL);
1847 flags = PERL_SCAN_SILENT_ILLDIGIT |
1848 PERL_SCAN_DISALLOW_PREFIX;
1850 val = grok_hex((char *)l, &numlen, &flags, NULL);
1859 Perl_croak(aTHX_ "%s: illegal mapping '%s'",
1865 val = 0; /* bits == 1, then val should be ignored */
1872 Perl_croak(aTHX_ "%s: illegal mapping '%s'", typestr, l);
1876 val = 0; /* bits == 1, then val should be ignored */
1889 if (!none || val < none) {
1894 for (key = min; key <= max; key++) {
1898 /* offset must be non-negative (start <= min <= key < end) */
1899 offset = octets * (key - start);
1901 s[offset] = (U8)(val & 0xff);
1902 else if (bits == 16) {
1903 s[offset ] = (U8)((val >> 8) & 0xff);
1904 s[offset + 1] = (U8)( val & 0xff);
1906 else if (bits == 32) {
1907 s[offset ] = (U8)((val >> 24) & 0xff);
1908 s[offset + 1] = (U8)((val >> 16) & 0xff);
1909 s[offset + 2] = (U8)((val >> 8) & 0xff);
1910 s[offset + 3] = (U8)( val & 0xff);
1913 if (!none || val < none)
1917 else { /* bits == 1, then val should be ignored */
1920 for (key = min; key <= max; key++) {
1921 const STRLEN offset = (STRLEN)(key - start);
1924 s[offset >> 3] |= 1 << (offset & 7);
1930 /* read $swash->{EXTRAS} */
1931 x = (U8*)SvPV(*extssvp, xcur);
1939 SV **otherbitssvp, *other;
1947 nl = (U8*)memchr(x, '\n', xend - x);
1949 if (opc != '-' && opc != '+' && opc != '!' && opc != '&') {
1951 x = nl + 1; /* 1 is length of "\n" */
1955 x = xend; /* to EXTRAS' end at which \n is not found */
1962 namelen = nl - namestr;
1966 namelen = xend - namestr;
1970 othersvp = hv_fetch(hv, (char *)namestr, namelen, FALSE);
1971 otherhv = (HV*)SvRV(*othersvp);
1972 otherbitssvp = hv_fetchs(otherhv, "BITS", FALSE);
1973 otherbits = (STRLEN)SvUV(*otherbitssvp);
1974 if (bits < otherbits)
1975 Perl_croak(aTHX_ "panic: swash_get found swatch size mismatch");
1977 /* The "other" swatch must be destroyed after. */
1978 other = swash_get(*othersvp, start, span);
1979 o = (U8*)SvPV(other, olen);
1982 Perl_croak(aTHX_ "panic: swash_get got improper swatch");
1984 s = (U8*)SvPV(swatch, slen);
1985 if (bits == 1 && otherbits == 1) {
1987 Perl_croak(aTHX_ "panic: swash_get found swatch length mismatch");
2011 STRLEN otheroctets = otherbits >> 3;
2013 U8* send = s + slen;
2018 if (otherbits == 1) {
2019 otherval = (o[offset >> 3] >> (offset & 7)) & 1;
2023 STRLEN vlen = otheroctets;
2031 if (opc == '+' && otherval)
2032 NOOP; /* replace with otherval */
2033 else if (opc == '!' && !otherval)
2035 else if (opc == '-' && otherval)
2037 else if (opc == '&' && !otherval)
2040 s += octets; /* no replacement */
2045 *s++ = (U8)( otherval & 0xff);
2046 else if (bits == 16) {
2047 *s++ = (U8)((otherval >> 8) & 0xff);
2048 *s++ = (U8)( otherval & 0xff);
2050 else if (bits == 32) {
2051 *s++ = (U8)((otherval >> 24) & 0xff);
2052 *s++ = (U8)((otherval >> 16) & 0xff);
2053 *s++ = (U8)((otherval >> 8) & 0xff);
2054 *s++ = (U8)( otherval & 0xff);
2058 sv_free(other); /* through with it! */
2064 =for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv
2066 Adds the UTF-8 representation of the Native codepoint C<uv> to the end
2067 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
2068 bytes available. The return value is the pointer to the byte after the
2069 end of the new character. In other words,
2071 d = uvchr_to_utf8(d, uv);
2073 is the recommended wide native character-aware way of saying
2080 /* On ASCII machines this is normally a macro but we want a
2081 real function in case XS code wants it
2084 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
2086 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
2090 Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
2092 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
2096 =for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32
2099 Returns the native character value of the first character in the string
2101 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
2102 length, in bytes, of that character.
2104 Allows length and flags to be passed to low level routine.
2108 /* On ASCII machines this is normally a macro but we want
2109 a real function in case XS code wants it
2112 Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen,
2115 const UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
2116 return UNI_TO_NATIVE(uv);
2120 =for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags
2122 Build to the scalar dsv a displayable version of the string spv,
2123 length len, the displayable version being at most pvlim bytes long
2124 (if longer, the rest is truncated and "..." will be appended).
2126 The flags argument can have UNI_DISPLAY_ISPRINT set to display
2127 isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
2128 to display the \\[nrfta\\] as the backslashed versions (like '\n')
2129 (UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
2130 UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
2131 UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
2133 The pointer to the PV of the dsv is returned.
2137 Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
2142 sv_setpvn(dsv, "", 0);
2143 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
2145 /* This serves double duty as a flag and a character to print after
2146 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
2150 if (pvlim && SvCUR(dsv) >= pvlim) {
2154 u = utf8_to_uvchr((U8*)s, 0);
2156 const unsigned char c = (unsigned char)u & 0xFF;
2157 if (flags & UNI_DISPLAY_BACKSLASH) {
2174 Perl_sv_catpvf(aTHX_ dsv, "\\%c", ok);
2177 /* isPRINT() is the locale-blind version. */
2178 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
2179 Perl_sv_catpvf(aTHX_ dsv, "%c", c);
2184 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
2187 sv_catpvs(dsv, "...");
2193 =for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags
2195 Build to the scalar dsv a displayable version of the scalar sv,
2196 the displayable version being at most pvlim bytes long
2197 (if longer, the rest is truncated and "..." will be appended).
2199 The flags argument is as in pv_uni_display().
2201 The pointer to the PV of the dsv is returned.
2206 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
2208 return Perl_pv_uni_display(aTHX_ dsv, (const U8*)SvPVX_const(ssv),
2209 SvCUR(ssv), pvlim, flags);
2213 =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
2215 Return true if the strings s1 and s2 differ case-insensitively, false
2216 if not (if they are equal case-insensitively). If u1 is true, the
2217 string s1 is assumed to be in UTF-8-encoded Unicode. If u2 is true,
2218 the string s2 is assumed to be in UTF-8-encoded Unicode. If u1 or u2
2219 are false, the respective string is assumed to be in native 8-bit
2222 If the pe1 and pe2 are non-NULL, the scanning pointers will be copied
2223 in there (they will point at the beginning of the I<next> character).
2224 If the pointers behind pe1 or pe2 are non-NULL, they are the end
2225 pointers beyond which scanning will not continue under any
2226 circumstances. If the byte lengths l1 and l2 are non-zero, s1+l1 and
2227 s2+l2 will be used as goal end pointers that will also stop the scan,
2228 and which qualify towards defining a successful match: all the scans
2229 that define an explicit length must reach their goal pointers for
2230 a match to succeed).
2232 For case-insensitiveness, the "casefolding" of Unicode is used
2233 instead of upper/lowercasing both the characters, see
2234 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
2238 Perl_ibcmp_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
2241 register const U8 *p1 = (const U8*)s1;
2242 register const U8 *p2 = (const U8*)s2;
2243 register const U8 *f1 = NULL;
2244 register const U8 *f2 = NULL;
2245 register U8 *e1 = NULL;
2246 register U8 *q1 = NULL;
2247 register U8 *e2 = NULL;
2248 register U8 *q2 = NULL;
2249 STRLEN n1 = 0, n2 = 0;
2250 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
2251 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
2253 STRLEN foldlen1, foldlen2;
2258 if (e1 == 0 || (l1 && l1 < (UV)(e1 - (const U8*)s1)))
2259 f1 = (const U8*)s1 + l1;
2262 if (e2 == 0 || (l2 && l2 < (UV)(e2 - (const U8*)s2)))
2263 f2 = (const U8*)s2 + l2;
2265 if ((e1 == 0 && f1 == 0) || (e2 == 0 && f2 == 0) || (f1 == 0 && f2 == 0))
2266 return 1; /* mismatch; possible infinite loop or false positive */
2269 natbuf[1] = 0; /* Need to terminate the buffer. */
2271 while ((e1 == 0 || p1 < e1) &&
2272 (f1 == 0 || p1 < f1) &&
2273 (e2 == 0 || p2 < e2) &&
2274 (f2 == 0 || p2 < f2)) {
2277 to_utf8_fold(p1, foldbuf1, &foldlen1);
2279 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p1)));
2280 to_utf8_fold(natbuf, foldbuf1, &foldlen1);
2287 to_utf8_fold(p2, foldbuf2, &foldlen2);
2289 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p2)));
2290 to_utf8_fold(natbuf, foldbuf2, &foldlen2);
2296 if ( UTF8SKIP(q1) != UTF8SKIP(q2) ||
2297 (UTF8SKIP(q1) == 1 && *q1 != *q2) ||
2298 memNE((char*)q1, (char*)q2, UTF8SKIP(q1)) )
2299 return 1; /* mismatch */
2306 p1 += u1 ? UTF8SKIP(p1) : 1;
2308 p2 += u2 ? UTF8SKIP(p2) : 1;
2312 /* A match is defined by all the scans that specified
2313 * an explicit length reaching their final goals. */
2314 match = (f1 == 0 || p1 == f1) && (f2 == 0 || p2 == f2);
2323 return match ? 0 : 1; /* 0 match, 1 mismatch */
2328 * c-indentation-style: bsd
2330 * indent-tabs-mode: t
2333 * ex: set ts=8 sts=4 sw=4 noet: