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))
287 } else if (!is_utf8_char_slow(x, c))
291 #endif /* #ifdef IS_UTF8_CHAR */
306 Implemented as a macro in utf8.h
308 =for apidoc A|bool|is_utf8_string_loc|const U8 *s|STRLEN len|const U8 **ep
310 Like is_utf8_string() but stores the location of the failure (in the
311 case of "utf8ness failure") or the location s+len (in the case of
312 "utf8ness success") in the C<ep>.
314 See also is_utf8_string_loclen() and is_utf8_string().
316 =for apidoc A|bool|is_utf8_string_loclen|const U8 *s|STRLEN len|const U8 **ep|const STRLEN *el
318 Like is_utf8_string() but stores the location of the failure (in the
319 case of "utf8ness failure") or the location s+len (in the case of
320 "utf8ness success") in the C<ep>, and the number of UTF-8
321 encoded characters in the C<el>.
323 See also is_utf8_string_loc() and is_utf8_string().
329 Perl_is_utf8_string_loclen(pTHX_ const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
337 len = strlen((const char *)s);
343 /* Inline the easy bits of is_utf8_char() here for speed... */
344 if (UTF8_IS_INVARIANT(*x))
346 else if (!UTF8_IS_START(*x))
349 /* ... and call is_utf8_char() only if really needed. */
352 if (IS_UTF8_CHAR_FAST(c)) {
353 if (!IS_UTF8_CHAR(x, c))
356 c = is_utf8_char_slow(x, c);
359 #endif /* #ifdef IS_UTF8_CHAR */
379 =for apidoc A|UV|utf8n_to_uvuni|const U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
381 Bottom level UTF-8 decode routine.
382 Returns the unicode code point value of the first character in the string C<s>
383 which is assumed to be in UTF-8 encoding and no longer than C<curlen>;
384 C<retlen> will be set to the length, in bytes, of that character.
386 If C<s> does not point to a well-formed UTF-8 character, the behaviour
387 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
388 it is assumed that the caller will raise a warning, and this function
389 will silently just set C<retlen> to C<-1> and return zero. If the
390 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
391 malformations will be given, C<retlen> will be set to the expected
392 length of the UTF-8 character in bytes, and zero will be returned.
394 The C<flags> can also contain various flags to allow deviations from
395 the strict UTF-8 encoding (see F<utf8.h>).
397 Most code should use utf8_to_uvchr() rather than call this directly.
403 Perl_utf8n_to_uvuni(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
406 const U8 * const s0 = s;
409 const bool dowarn = ckWARN_d(WARN_UTF8);
410 const UV startbyte = *s;
411 STRLEN expectlen = 0;
414 /* This list is a superset of the UTF8_ALLOW_XXX. */
416 #define UTF8_WARN_EMPTY 1
417 #define UTF8_WARN_CONTINUATION 2
418 #define UTF8_WARN_NON_CONTINUATION 3
419 #define UTF8_WARN_FE_FF 4
420 #define UTF8_WARN_SHORT 5
421 #define UTF8_WARN_OVERFLOW 6
422 #define UTF8_WARN_SURROGATE 7
423 #define UTF8_WARN_LONG 8
424 #define UTF8_WARN_FFFF 9 /* Also FFFE. */
427 !(flags & UTF8_ALLOW_EMPTY)) {
428 warning = UTF8_WARN_EMPTY;
432 if (UTF8_IS_INVARIANT(uv)) {
435 return (UV) (NATIVE_TO_UTF(*s));
438 if (UTF8_IS_CONTINUATION(uv) &&
439 !(flags & UTF8_ALLOW_CONTINUATION)) {
440 warning = UTF8_WARN_CONTINUATION;
444 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
445 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
446 warning = UTF8_WARN_NON_CONTINUATION;
451 uv = NATIVE_TO_UTF(uv);
453 if ((uv == 0xfe || uv == 0xff) &&
454 !(flags & UTF8_ALLOW_FE_FF)) {
455 warning = UTF8_WARN_FE_FF;
460 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
461 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
462 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
463 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
465 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
466 else { len = 7; uv &= 0x01; }
468 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
469 else if (!(uv & 0x01)) { len = 7; uv = 0; }
470 else { len = 13; uv = 0; } /* whoa! */
478 if ((curlen < expectlen) &&
479 !(flags & UTF8_ALLOW_SHORT)) {
480 warning = UTF8_WARN_SHORT;
489 if (!UTF8_IS_CONTINUATION(*s) &&
490 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
492 warning = UTF8_WARN_NON_CONTINUATION;
496 uv = UTF8_ACCUMULATE(uv, *s);
498 /* These cannot be allowed. */
500 if (expectlen != 13 && !(flags & UTF8_ALLOW_LONG)) {
501 warning = UTF8_WARN_LONG;
505 else { /* uv < ouv */
506 /* This cannot be allowed. */
507 warning = UTF8_WARN_OVERFLOW;
515 if (UNICODE_IS_SURROGATE(uv) &&
516 !(flags & UTF8_ALLOW_SURROGATE)) {
517 warning = UTF8_WARN_SURROGATE;
519 } else if ((expectlen > (STRLEN)UNISKIP(uv)) &&
520 !(flags & UTF8_ALLOW_LONG)) {
521 warning = UTF8_WARN_LONG;
523 } else if (UNICODE_IS_ILLEGAL(uv) &&
524 !(flags & UTF8_ALLOW_FFFF)) {
525 warning = UTF8_WARN_FFFF;
533 if (flags & UTF8_CHECK_ONLY) {
540 SV* const sv = sv_2mortal(newSVpvs("Malformed UTF-8 character "));
543 case 0: /* Intentionally empty. */ break;
544 case UTF8_WARN_EMPTY:
545 sv_catpvs(sv, "(empty string)");
547 case UTF8_WARN_CONTINUATION:
548 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
550 case UTF8_WARN_NON_CONTINUATION:
552 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
553 (UV)s[1], startbyte);
555 const int len = (int)(s-s0);
556 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
557 (UV)s[1], len, len > 1 ? "s" : "", startbyte, (int)expectlen);
561 case UTF8_WARN_FE_FF:
562 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
564 case UTF8_WARN_SHORT:
565 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
566 (int)curlen, curlen == 1 ? "" : "s", (int)expectlen, startbyte);
567 expectlen = curlen; /* distance for caller to skip */
569 case UTF8_WARN_OVERFLOW:
570 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
573 case UTF8_WARN_SURROGATE:
574 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
577 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
578 (int)expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
581 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
584 sv_catpvs(sv, "(unknown reason)");
589 const char * const s = SvPVX_const(sv);
592 Perl_warner(aTHX_ packWARN(WARN_UTF8),
593 "%s in %s", s, OP_DESC(PL_op));
595 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
600 *retlen = expectlen ? expectlen : len;
606 =for apidoc A|UV|utf8_to_uvchr|const U8 *s|STRLEN *retlen
608 Returns the native character value of the first character in the string C<s>
609 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
610 length, in bytes, of that character.
612 If C<s> does not point to a well-formed UTF-8 character, zero is
613 returned and retlen is set, if possible, to -1.
619 Perl_utf8_to_uvchr(pTHX_ const U8 *s, STRLEN *retlen)
621 return utf8n_to_uvchr(s, UTF8_MAXBYTES, retlen,
622 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
626 =for apidoc A|UV|utf8_to_uvuni|const U8 *s|STRLEN *retlen
628 Returns the Unicode code point of the first character in the string C<s>
629 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
630 length, in bytes, of that character.
632 This function should only be used when returned UV is considered
633 an index into the Unicode semantic tables (e.g. swashes).
635 If C<s> does not point to a well-formed UTF-8 character, zero is
636 returned and retlen is set, if possible, to -1.
642 Perl_utf8_to_uvuni(pTHX_ const U8 *s, STRLEN *retlen)
644 /* Call the low level routine asking for checks */
645 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXBYTES, retlen,
646 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
650 =for apidoc A|STRLEN|utf8_length|const U8 *s|const U8 *e
652 Return the length of the UTF-8 char encoded string C<s> in characters.
653 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
654 up past C<e>, croaks.
660 Perl_utf8_length(pTHX_ const U8 *s, const U8 *e)
665 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
666 * the bitops (especially ~) can create illegal UTF-8.
667 * In other words: in Perl UTF-8 is not just for Unicode. */
670 goto warn_and_return;
672 const U8 t = UTF8SKIP(s);
675 if (ckWARN_d(WARN_UTF8)) {
677 Perl_warner(aTHX_ packWARN(WARN_UTF8),
678 "%s in %s", unees, OP_DESC(PL_op));
680 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
692 =for apidoc A|IV|utf8_distance|const U8 *a|const U8 *b
694 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
697 WARNING: use only if you *know* that the pointers point inside the
704 Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
706 return (a < b) ? -1 * (IV) utf8_length(a, b) : (IV) utf8_length(b, a);
710 =for apidoc A|U8 *|utf8_hop|U8 *s|I32 off
712 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
715 WARNING: do not use the following unless you *know* C<off> is within
716 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
717 on the first byte of character or just after the last byte of a character.
723 Perl_utf8_hop(pTHX_ const U8 *s, I32 off)
726 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
727 * the bitops (especially ~) can create illegal UTF-8.
728 * In other words: in Perl UTF-8 is not just for Unicode. */
737 while (UTF8_IS_CONTINUATION(*s))
745 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
747 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
748 Unlike C<bytes_to_utf8>, this over-writes the original string, and
749 updates len to contain the new length.
750 Returns zero on failure, setting C<len> to -1.
756 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
759 U8 * const send = s + *len;
762 /* ensure valid UTF-8 and chars < 256 before updating string */
766 if (!UTF8_IS_INVARIANT(c) &&
767 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
768 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
777 *d++ = (U8)utf8_to_uvchr(s, &ulen);
786 =for apidoc A|U8 *|bytes_from_utf8|const U8 *s|STRLEN *len|bool *is_utf8
788 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
789 Unlike C<utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
790 the newly-created string, and updates C<len> to contain the new
791 length. Returns the original string if no conversion occurs, C<len>
792 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
793 0 if C<s> is converted or contains all 7bit characters.
799 Perl_bytes_from_utf8(pTHX_ const U8 *s, STRLEN *len, bool *is_utf8)
810 /* ensure valid UTF-8 and chars < 256 before converting string */
811 for (send = s + *len; s < send;) {
813 if (!UTF8_IS_INVARIANT(c)) {
814 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
815 (c = *s++) && UTF8_IS_CONTINUATION(c))
824 Newx(d, (*len) - count + 1, U8);
825 s = start; start = d;
828 if (!UTF8_IS_INVARIANT(c)) {
829 /* Then it is two-byte encoded */
830 c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++);
831 c = ASCII_TO_NATIVE(c);
841 =for apidoc A|U8 *|bytes_to_utf8|const U8 *s|STRLEN *len
843 Converts a string C<s> of length C<len> from ASCII into UTF-8 encoding.
844 Returns a pointer to the newly-created string, and sets C<len> to
845 reflect the new length.
847 If you want to convert to UTF-8 from other encodings than ASCII,
848 see sv_recode_to_utf8().
854 Perl_bytes_to_utf8(pTHX_ const U8 *s, STRLEN *len)
856 const U8 * const send = s + (*len);
861 Newx(d, (*len) * 2 + 1, U8);
865 const UV uv = NATIVE_TO_ASCII(*s++);
866 if (UNI_IS_INVARIANT(uv))
867 *d++ = (U8)UTF_TO_NATIVE(uv);
869 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
870 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
879 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
881 * Destination must be pre-extended to 3/2 source. Do not use in-place.
882 * We optimize for native, for obvious reasons. */
885 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
890 if (bytelen == 1 && p[0] == 0) { /* Be understanding. */
897 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %"UVf, (UV)bytelen);
902 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
909 *d++ = (U8)(( uv >> 6) | 0xc0);
910 *d++ = (U8)(( uv & 0x3f) | 0x80);
913 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
914 UV low = (p[0] << 8) + p[1];
916 if (low < 0xdc00 || low >= 0xdfff)
917 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
918 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
921 *d++ = (U8)(( uv >> 12) | 0xe0);
922 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
923 *d++ = (U8)(( uv & 0x3f) | 0x80);
927 *d++ = (U8)(( uv >> 18) | 0xf0);
928 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
929 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
930 *d++ = (U8)(( uv & 0x3f) | 0x80);
934 *newlen = d - dstart;
938 /* Note: this one is slightly destructive of the source. */
941 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
944 U8* const send = s + bytelen;
951 return utf16_to_utf8(p, d, bytelen, newlen);
954 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
957 Perl_is_uni_alnum(pTHX_ UV c)
959 U8 tmpbuf[UTF8_MAXBYTES+1];
960 uvchr_to_utf8(tmpbuf, c);
961 return is_utf8_alnum(tmpbuf);
965 Perl_is_uni_alnumc(pTHX_ UV c)
967 U8 tmpbuf[UTF8_MAXBYTES+1];
968 uvchr_to_utf8(tmpbuf, c);
969 return is_utf8_alnumc(tmpbuf);
973 Perl_is_uni_idfirst(pTHX_ UV c)
975 U8 tmpbuf[UTF8_MAXBYTES+1];
976 uvchr_to_utf8(tmpbuf, c);
977 return is_utf8_idfirst(tmpbuf);
981 Perl_is_uni_alpha(pTHX_ UV c)
983 U8 tmpbuf[UTF8_MAXBYTES+1];
984 uvchr_to_utf8(tmpbuf, c);
985 return is_utf8_alpha(tmpbuf);
989 Perl_is_uni_ascii(pTHX_ UV c)
991 U8 tmpbuf[UTF8_MAXBYTES+1];
992 uvchr_to_utf8(tmpbuf, c);
993 return is_utf8_ascii(tmpbuf);
997 Perl_is_uni_space(pTHX_ UV c)
999 U8 tmpbuf[UTF8_MAXBYTES+1];
1000 uvchr_to_utf8(tmpbuf, c);
1001 return is_utf8_space(tmpbuf);
1005 Perl_is_uni_digit(pTHX_ UV c)
1007 U8 tmpbuf[UTF8_MAXBYTES+1];
1008 uvchr_to_utf8(tmpbuf, c);
1009 return is_utf8_digit(tmpbuf);
1013 Perl_is_uni_upper(pTHX_ UV c)
1015 U8 tmpbuf[UTF8_MAXBYTES+1];
1016 uvchr_to_utf8(tmpbuf, c);
1017 return is_utf8_upper(tmpbuf);
1021 Perl_is_uni_lower(pTHX_ UV c)
1023 U8 tmpbuf[UTF8_MAXBYTES+1];
1024 uvchr_to_utf8(tmpbuf, c);
1025 return is_utf8_lower(tmpbuf);
1029 Perl_is_uni_cntrl(pTHX_ UV c)
1031 U8 tmpbuf[UTF8_MAXBYTES+1];
1032 uvchr_to_utf8(tmpbuf, c);
1033 return is_utf8_cntrl(tmpbuf);
1037 Perl_is_uni_graph(pTHX_ UV c)
1039 U8 tmpbuf[UTF8_MAXBYTES+1];
1040 uvchr_to_utf8(tmpbuf, c);
1041 return is_utf8_graph(tmpbuf);
1045 Perl_is_uni_print(pTHX_ UV c)
1047 U8 tmpbuf[UTF8_MAXBYTES+1];
1048 uvchr_to_utf8(tmpbuf, c);
1049 return is_utf8_print(tmpbuf);
1053 Perl_is_uni_punct(pTHX_ UV c)
1055 U8 tmpbuf[UTF8_MAXBYTES+1];
1056 uvchr_to_utf8(tmpbuf, c);
1057 return is_utf8_punct(tmpbuf);
1061 Perl_is_uni_xdigit(pTHX_ UV c)
1063 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1064 uvchr_to_utf8(tmpbuf, c);
1065 return is_utf8_xdigit(tmpbuf);
1069 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
1071 uvchr_to_utf8(p, c);
1072 return to_utf8_upper(p, p, lenp);
1076 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
1078 uvchr_to_utf8(p, c);
1079 return to_utf8_title(p, p, lenp);
1083 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
1085 uvchr_to_utf8(p, c);
1086 return to_utf8_lower(p, p, lenp);
1090 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1092 uvchr_to_utf8(p, c);
1093 return to_utf8_fold(p, p, lenp);
1096 /* for now these all assume no locale info available for Unicode > 255 */
1099 Perl_is_uni_alnum_lc(pTHX_ UV c)
1101 return is_uni_alnum(c); /* XXX no locale support yet */
1105 Perl_is_uni_alnumc_lc(pTHX_ UV c)
1107 return is_uni_alnumc(c); /* XXX no locale support yet */
1111 Perl_is_uni_idfirst_lc(pTHX_ UV c)
1113 return is_uni_idfirst(c); /* XXX no locale support yet */
1117 Perl_is_uni_alpha_lc(pTHX_ UV c)
1119 return is_uni_alpha(c); /* XXX no locale support yet */
1123 Perl_is_uni_ascii_lc(pTHX_ UV c)
1125 return is_uni_ascii(c); /* XXX no locale support yet */
1129 Perl_is_uni_space_lc(pTHX_ UV c)
1131 return is_uni_space(c); /* XXX no locale support yet */
1135 Perl_is_uni_digit_lc(pTHX_ UV c)
1137 return is_uni_digit(c); /* XXX no locale support yet */
1141 Perl_is_uni_upper_lc(pTHX_ UV c)
1143 return is_uni_upper(c); /* XXX no locale support yet */
1147 Perl_is_uni_lower_lc(pTHX_ UV c)
1149 return is_uni_lower(c); /* XXX no locale support yet */
1153 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1155 return is_uni_cntrl(c); /* XXX no locale support yet */
1159 Perl_is_uni_graph_lc(pTHX_ UV c)
1161 return is_uni_graph(c); /* XXX no locale support yet */
1165 Perl_is_uni_print_lc(pTHX_ UV c)
1167 return is_uni_print(c); /* XXX no locale support yet */
1171 Perl_is_uni_punct_lc(pTHX_ UV c)
1173 return is_uni_punct(c); /* XXX no locale support yet */
1177 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1179 return is_uni_xdigit(c); /* XXX no locale support yet */
1183 Perl_to_uni_upper_lc(pTHX_ U32 c)
1185 /* XXX returns only the first character -- do not use XXX */
1186 /* XXX no locale support yet */
1188 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1189 return (U32)to_uni_upper(c, tmpbuf, &len);
1193 Perl_to_uni_title_lc(pTHX_ U32 c)
1195 /* XXX returns only the first character XXX -- do not use XXX */
1196 /* XXX no locale support yet */
1198 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1199 return (U32)to_uni_title(c, tmpbuf, &len);
1203 Perl_to_uni_lower_lc(pTHX_ U32 c)
1205 /* XXX returns only the first character -- do not use XXX */
1206 /* XXX no locale support yet */
1208 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1209 return (U32)to_uni_lower(c, tmpbuf, &len);
1213 S_is_utf8_common(pTHX_ const U8 *const p, SV **swash,
1214 const char *const swashname)
1217 if (!is_utf8_char(p))
1220 *swash = swash_init("utf8", swashname, &PL_sv_undef, 1, 0);
1221 return swash_fetch(*swash, p, TRUE) != 0;
1225 Perl_is_utf8_alnum(pTHX_ const U8 *p)
1228 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1229 * descendant of isalnum(3), in other words, it doesn't
1230 * contain the '_'. --jhi */
1231 return is_utf8_common(p, &PL_utf8_alnum, "IsWord");
1235 Perl_is_utf8_alnumc(pTHX_ const U8 *p)
1238 return is_utf8_common(p, &PL_utf8_alnumc, "IsAlnumC");
1242 Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
1247 /* is_utf8_idstart would be more logical. */
1248 return is_utf8_common(p, &PL_utf8_idstart, "IdStart");
1252 Perl_is_utf8_idcont(pTHX_ const U8 *p)
1257 return is_utf8_common(p, &PL_utf8_idcont, "IdContinue");
1261 Perl_is_utf8_alpha(pTHX_ const U8 *p)
1264 return is_utf8_common(p, &PL_utf8_alpha, "IsAlpha");
1268 Perl_is_utf8_ascii(pTHX_ const U8 *p)
1271 return is_utf8_common(p, &PL_utf8_ascii, "IsAscii");
1275 Perl_is_utf8_space(pTHX_ const U8 *p)
1278 return is_utf8_common(p, &PL_utf8_space, "IsSpacePerl");
1282 Perl_is_utf8_digit(pTHX_ const U8 *p)
1285 return is_utf8_common(p, &PL_utf8_digit, "IsDigit");
1289 Perl_is_utf8_upper(pTHX_ const U8 *p)
1292 return is_utf8_common(p, &PL_utf8_upper, "IsUppercase");
1296 Perl_is_utf8_lower(pTHX_ const U8 *p)
1299 return is_utf8_common(p, &PL_utf8_lower, "IsLowercase");
1303 Perl_is_utf8_cntrl(pTHX_ const U8 *p)
1306 return is_utf8_common(p, &PL_utf8_cntrl, "IsCntrl");
1310 Perl_is_utf8_graph(pTHX_ const U8 *p)
1313 return is_utf8_common(p, &PL_utf8_graph, "IsGraph");
1317 Perl_is_utf8_print(pTHX_ const U8 *p)
1320 return is_utf8_common(p, &PL_utf8_print, "IsPrint");
1324 Perl_is_utf8_punct(pTHX_ const U8 *p)
1327 return is_utf8_common(p, &PL_utf8_punct, "IsPunct");
1331 Perl_is_utf8_xdigit(pTHX_ const U8 *p)
1334 return is_utf8_common(p, &PL_utf8_xdigit, "Isxdigit");
1338 Perl_is_utf8_mark(pTHX_ const U8 *p)
1341 return is_utf8_common(p, &PL_utf8_mark, "IsM");
1345 =for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special
1347 The "p" contains the pointer to the UTF-8 string encoding
1348 the character that is being converted.
1350 The "ustrp" is a pointer to the character buffer to put the
1351 conversion result to. The "lenp" is a pointer to the length
1354 The "swashp" is a pointer to the swash to use.
1356 Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
1357 and loaded by SWASHNEW, using lib/utf8_heavy.pl. The special (usually,
1358 but not always, a multicharacter mapping), is tried first.
1360 The "special" is a string like "utf8::ToSpecLower", which means the
1361 hash %utf8::ToSpecLower. The access to the hash is through
1362 Perl_to_utf8_case().
1364 The "normal" is a string like "ToLower" which means the swash
1370 Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp,
1371 SV **swashp, const char *normal, const char *special)
1374 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1377 const UV uv0 = utf8_to_uvchr(p, NULL);
1378 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1379 * are necessary in EBCDIC, they are redundant no-ops
1380 * in ASCII-ish platforms, and hopefully optimized away. */
1381 const UV uv1 = NATIVE_TO_UNI(uv0);
1382 uvuni_to_utf8(tmpbuf, uv1);
1384 if (!*swashp) /* load on-demand */
1385 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1387 /* The 0xDF is the only special casing Unicode code point below 0x100. */
1388 if (special && (uv1 == 0xDF || uv1 > 0xFF)) {
1389 /* It might be "special" (sometimes, but not always,
1390 * a multicharacter mapping) */
1394 if ((hv = get_hv(special, FALSE)) &&
1395 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1399 s = SvPV_const(*svp, len);
1401 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
1404 /* If we have EBCDIC we need to remap the characters
1405 * since any characters in the low 256 are Unicode
1406 * code points, not EBCDIC. */
1407 U8 *t = (U8*)s, *tend = t + len, *d;
1414 const UV c = utf8_to_uvchr(t, &tlen);
1416 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1425 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1430 Copy(tmpbuf, ustrp, len, U8);
1432 Copy(s, ustrp, len, U8);
1438 if (!len && *swashp) {
1439 const UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1442 /* It was "normal" (a single character mapping). */
1443 const UV uv3 = UNI_TO_NATIVE(uv2);
1444 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
1448 if (!len) /* Neither: just copy. */
1449 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1454 return len ? utf8_to_uvchr(ustrp, 0) : 0;
1458 =for apidoc A|UV|to_utf8_upper|const U8 *p|U8 *ustrp|STRLEN *lenp
1460 Convert the UTF-8 encoded character at p to its uppercase version and
1461 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1462 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1463 the uppercase version may be longer than the original character.
1465 The first character of the uppercased version is returned
1466 (but note, as explained above, that there may be more.)
1471 Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1474 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1475 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1479 =for apidoc A|UV|to_utf8_title|const U8 *p|U8 *ustrp|STRLEN *lenp
1481 Convert the UTF-8 encoded character at p to its titlecase version and
1482 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1483 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1484 titlecase version may be longer than the original character.
1486 The first character of the titlecased version is returned
1487 (but note, as explained above, that there may be more.)
1492 Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1495 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1496 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1500 =for apidoc A|UV|to_utf8_lower|const U8 *p|U8 *ustrp|STRLEN *lenp
1502 Convert the UTF-8 encoded character at p to its lowercase version and
1503 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1504 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1505 lowercase version may be longer than the original character.
1507 The first character of the lowercased version is returned
1508 (but note, as explained above, that there may be more.)
1513 Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1516 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1517 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1521 =for apidoc A|UV|to_utf8_fold|const U8 *p|U8 *ustrp|STRLEN *lenp
1523 Convert the UTF-8 encoded character at p to its foldcase version and
1524 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1525 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1526 foldcase version may be longer than the original character (up to
1529 The first character of the foldcased version is returned
1530 (but note, as explained above, that there may be more.)
1535 Perl_to_utf8_fold(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1538 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1539 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1543 * A "swash" is a swatch hash.
1544 * A "swatch" is a bit vector generated by utf8.c:S_swash_get().
1545 * C<pkg> is a pointer to a package name for SWASHNEW, should be "utf8".
1546 * For other parameters, see utf8::SWASHNEW in lib/utf8_heavy.pl.
1549 Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
1553 SV* const tokenbufsv = sv_newmortal();
1555 const size_t pkg_len = strlen(pkg);
1556 const size_t name_len = strlen(name);
1557 HV * const stash = gv_stashpvn(pkg, pkg_len, FALSE);
1560 PUSHSTACKi(PERLSI_MAGIC);
1565 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
1567 errsv_save = newSVsv(ERRSV);
1568 /* It is assumed that callers of this routine are not passing in any
1569 user derived data. */
1570 /* Need to do this after save_re_context() as it will set PL_tainted to
1571 1 while saving $1 etc (see the code after getrx: in Perl_magic_get).
1572 Even line to create errsv_save can turn on PL_tainted. */
1573 SAVEBOOL(PL_tainted);
1575 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
1578 sv_setsv(ERRSV, errsv_save);
1579 SvREFCNT_dec(errsv_save);
1585 PUSHs(sv_2mortal(newSVpvn(pkg, pkg_len)));
1586 PUSHs(sv_2mortal(newSVpvn(name, name_len)));
1588 PUSHs(sv_2mortal(newSViv(minbits)));
1589 PUSHs(sv_2mortal(newSViv(none)));
1591 if (IN_PERL_COMPILETIME) {
1592 /* XXX ought to be handled by lex_start */
1595 sv_setpv(tokenbufsv, PL_tokenbuf);
1597 errsv_save = newSVsv(ERRSV);
1598 if (call_method("SWASHNEW", G_SCALAR))
1599 retval = newSVsv(*PL_stack_sp--);
1601 retval = &PL_sv_undef;
1603 sv_setsv(ERRSV, errsv_save);
1604 SvREFCNT_dec(errsv_save);
1607 if (IN_PERL_COMPILETIME) {
1609 const char* const pv = SvPV_const(tokenbufsv, len);
1611 Copy(pv, PL_tokenbuf, len+1, char);
1612 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1614 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
1616 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
1618 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1624 /* This API is wrong for special case conversions since we may need to
1625 * return several Unicode characters for a single Unicode character
1626 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1627 * the lower-level routine, and it is similarly broken for returning
1628 * multiple values. --jhi */
1629 /* Now SWASHGET is recasted into S_swash_get in this file. */
1632 * Returns the value of property/mapping C<swash> for the first character
1633 * of the string C<ptr>. If C<do_utf8> is true, the string C<ptr> is
1634 * assumed to be in utf8. If C<do_utf8> is false, the string C<ptr> is
1635 * assumed to be in native 8-bit encoding. Caches the swatch in C<swash>.
1638 Perl_swash_fetch(pTHX_ SV *swash, const U8 *ptr, bool do_utf8)
1641 HV* const hv = (HV*)SvRV(swash);
1646 const U8 *tmps = NULL;
1650 UV c = NATIVE_TO_ASCII(*ptr);
1652 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1653 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
1654 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
1657 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
1658 * then the "swatch" is a vec() for al the chars which start
1660 * So the key in the hash (klen) is length of encoded char -1
1662 klen = UTF8SKIP(ptr) - 1;
1666 /* If char in invariant then swatch is for all the invariant chars
1667 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
1669 needents = UTF_CONTINUATION_MARK;
1670 off = NATIVE_TO_UTF(ptr[klen]);
1673 /* If char is encoded then swatch is for the prefix */
1674 needents = (1 << UTF_ACCUMULATION_SHIFT);
1675 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
1679 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1680 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1681 * it's nothing to sniff at.) Pity we usually come through at least
1682 * two function calls to get here...
1684 * NB: this code assumes that swatches are never modified, once generated!
1687 if (hv == PL_last_swash_hv &&
1688 klen == PL_last_swash_klen &&
1689 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
1691 tmps = PL_last_swash_tmps;
1692 slen = PL_last_swash_slen;
1695 /* Try our second-level swatch cache, kept in a hash. */
1696 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
1698 /* If not cached, generate it via swash_get */
1699 if (!svp || !SvPOK(*svp)
1700 || !(tmps = (const U8*)SvPV_const(*svp, slen))) {
1701 /* We use utf8n_to_uvuni() as we want an index into
1702 Unicode tables, not a native character number.
1704 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
1706 0 : UTF8_ALLOW_ANY);
1707 swatch = swash_get(swash,
1708 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
1709 (klen) ? (code_point & ~(needents - 1)) : 0,
1712 if (IN_PERL_COMPILETIME)
1713 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1715 svp = hv_store(hv, (const char *)ptr, klen, swatch, 0);
1717 if (!svp || !(tmps = (U8*)SvPV(*svp, slen))
1718 || (slen << 3) < needents)
1719 Perl_croak(aTHX_ "panic: swash_fetch got improper swatch");
1722 PL_last_swash_hv = hv;
1723 PL_last_swash_klen = klen;
1724 /* FIXME change interpvar.h? */
1725 PL_last_swash_tmps = (U8 *) tmps;
1726 PL_last_swash_slen = slen;
1728 Copy(ptr, PL_last_swash_key, klen, U8);
1731 switch ((int)((slen << 3) / needents)) {
1733 bit = 1 << (off & 7);
1735 return (tmps[off] & bit) != 0;
1740 return (tmps[off] << 8) + tmps[off + 1] ;
1743 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1745 Perl_croak(aTHX_ "panic: swash_fetch got swatch of unexpected bit width");
1749 * Returns a swatch (a bit vector string) for a code point sequence
1750 * that starts from the value C<start> and comprises the number C<span>.
1751 * A C<swash> must be an object created by SWASHNEW (see lib/utf8_heavy.pl).
1752 * Should be used via swash_fetch, which will cache the swatch in C<swash>.
1755 S_swash_get(pTHX_ SV* swash, UV start, UV span)
1758 U8 *l, *lend, *x, *xend, *s;
1759 STRLEN lcur, xcur, scur;
1761 HV* const hv = (HV*)SvRV(swash);
1762 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
1763 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
1764 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
1765 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
1766 SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);
1767 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
1768 const int typeto = typestr[0] == 'T' && typestr[1] == 'o';
1769 const STRLEN bits = SvUV(*bitssvp);
1770 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
1771 const UV none = SvUV(*nonesvp);
1772 const UV end = start + span;
1774 if (bits != 1 && bits != 8 && bits != 16 && bits != 32) {
1775 Perl_croak(aTHX_ "panic: swash_get doesn't expect bits %"UVuf,
1779 /* create and initialize $swatch */
1780 swatch = newSVpvs("");
1781 scur = octets ? (span * octets) : (span + 7) / 8;
1782 SvGROW(swatch, scur + 1);
1783 s = (U8*)SvPVX(swatch);
1784 if (octets && none) {
1785 const U8* const e = s + scur;
1788 *s++ = (U8)(none & 0xff);
1789 else if (bits == 16) {
1790 *s++ = (U8)((none >> 8) & 0xff);
1791 *s++ = (U8)( none & 0xff);
1793 else if (bits == 32) {
1794 *s++ = (U8)((none >> 24) & 0xff);
1795 *s++ = (U8)((none >> 16) & 0xff);
1796 *s++ = (U8)((none >> 8) & 0xff);
1797 *s++ = (U8)( none & 0xff);
1803 (void)memzero((U8*)s, scur + 1);
1805 SvCUR_set(swatch, scur);
1806 s = (U8*)SvPVX(swatch);
1808 /* read $swash->{LIST} */
1809 l = (U8*)SvPV(*listsvp, lcur);
1812 UV min, max, val, key;
1814 I32 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
1816 U8* const nl = (U8*)memchr(l, '\n', lend - l);
1819 min = grok_hex((char *)l, &numlen, &flags, NULL);
1823 l = nl + 1; /* 1 is length of "\n" */
1827 l = lend; /* to LIST's end at which \n is not found */
1833 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
1835 max = grok_hex((char *)l, &numlen, &flags, NULL);
1844 flags = PERL_SCAN_SILENT_ILLDIGIT |
1845 PERL_SCAN_DISALLOW_PREFIX;
1847 val = grok_hex((char *)l, &numlen, &flags, NULL);
1856 Perl_croak(aTHX_ "%s: illegal mapping '%s'",
1862 val = 0; /* bits == 1, then val should be ignored */
1869 Perl_croak(aTHX_ "%s: illegal mapping '%s'", typestr, l);
1873 val = 0; /* bits == 1, then val should be ignored */
1886 if (!none || val < none) {
1891 for (key = min; key <= max; key++) {
1895 /* offset must be non-negative (start <= min <= key < end) */
1896 offset = octets * (key - start);
1898 s[offset] = (U8)(val & 0xff);
1899 else if (bits == 16) {
1900 s[offset ] = (U8)((val >> 8) & 0xff);
1901 s[offset + 1] = (U8)( val & 0xff);
1903 else if (bits == 32) {
1904 s[offset ] = (U8)((val >> 24) & 0xff);
1905 s[offset + 1] = (U8)((val >> 16) & 0xff);
1906 s[offset + 2] = (U8)((val >> 8) & 0xff);
1907 s[offset + 3] = (U8)( val & 0xff);
1910 if (!none || val < none)
1914 else { /* bits == 1, then val should be ignored */
1917 for (key = min; key <= max; key++) {
1918 const STRLEN offset = (STRLEN)(key - start);
1921 s[offset >> 3] |= 1 << (offset & 7);
1927 /* read $swash->{EXTRAS} */
1928 x = (U8*)SvPV(*extssvp, xcur);
1936 SV **otherbitssvp, *other;
1944 nl = (U8*)memchr(x, '\n', xend - x);
1946 if (opc != '-' && opc != '+' && opc != '!' && opc != '&') {
1948 x = nl + 1; /* 1 is length of "\n" */
1952 x = xend; /* to EXTRAS' end at which \n is not found */
1959 namelen = nl - namestr;
1963 namelen = xend - namestr;
1967 othersvp = hv_fetch(hv, (char *)namestr, namelen, FALSE);
1968 otherhv = (HV*)SvRV(*othersvp);
1969 otherbitssvp = hv_fetchs(otherhv, "BITS", FALSE);
1970 otherbits = (STRLEN)SvUV(*otherbitssvp);
1971 if (bits < otherbits)
1972 Perl_croak(aTHX_ "panic: swash_get found swatch size mismatch");
1974 /* The "other" swatch must be destroyed after. */
1975 other = swash_get(*othersvp, start, span);
1976 o = (U8*)SvPV(other, olen);
1979 Perl_croak(aTHX_ "panic: swash_get got improper swatch");
1981 s = (U8*)SvPV(swatch, slen);
1982 if (bits == 1 && otherbits == 1) {
1984 Perl_croak(aTHX_ "panic: swash_get found swatch length mismatch");
2008 STRLEN otheroctets = otherbits >> 3;
2010 U8* send = s + slen;
2015 if (otherbits == 1) {
2016 otherval = (o[offset >> 3] >> (offset & 7)) & 1;
2020 STRLEN vlen = otheroctets;
2028 if (opc == '+' && otherval)
2029 /*EMPTY*/; /* replace with otherval */
2030 else if (opc == '!' && !otherval)
2032 else if (opc == '-' && otherval)
2034 else if (opc == '&' && !otherval)
2037 s += octets; /* no replacement */
2042 *s++ = (U8)( otherval & 0xff);
2043 else if (bits == 16) {
2044 *s++ = (U8)((otherval >> 8) & 0xff);
2045 *s++ = (U8)( otherval & 0xff);
2047 else if (bits == 32) {
2048 *s++ = (U8)((otherval >> 24) & 0xff);
2049 *s++ = (U8)((otherval >> 16) & 0xff);
2050 *s++ = (U8)((otherval >> 8) & 0xff);
2051 *s++ = (U8)( otherval & 0xff);
2055 sv_free(other); /* through with it! */
2061 =for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv
2063 Adds the UTF-8 representation of the Native codepoint C<uv> to the end
2064 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
2065 bytes available. The return value is the pointer to the byte after the
2066 end of the new character. In other words,
2068 d = uvchr_to_utf8(d, uv);
2070 is the recommended wide native character-aware way of saying
2077 /* On ASCII machines this is normally a macro but we want a
2078 real function in case XS code wants it
2081 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
2083 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
2087 Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
2089 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
2093 =for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32
2096 Returns the native character value of the first character in the string
2098 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
2099 length, in bytes, of that character.
2101 Allows length and flags to be passed to low level routine.
2105 /* On ASCII machines this is normally a macro but we want
2106 a real function in case XS code wants it
2109 Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen,
2112 const UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
2113 return UNI_TO_NATIVE(uv);
2117 =for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags
2119 Build to the scalar dsv a displayable version of the string spv,
2120 length len, the displayable version being at most pvlim bytes long
2121 (if longer, the rest is truncated and "..." will be appended).
2123 The flags argument can have UNI_DISPLAY_ISPRINT set to display
2124 isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
2125 to display the \\[nrfta\\] as the backslashed versions (like '\n')
2126 (UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
2127 UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
2128 UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
2130 The pointer to the PV of the dsv is returned.
2134 Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
2139 sv_setpvn(dsv, "", 0);
2140 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
2142 /* This serves double duty as a flag and a character to print after
2143 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
2147 if (pvlim && SvCUR(dsv) >= pvlim) {
2151 u = utf8_to_uvchr((U8*)s, 0);
2153 const unsigned char c = (unsigned char)u & 0xFF;
2154 if (flags & UNI_DISPLAY_BACKSLASH) {
2171 Perl_sv_catpvf(aTHX_ dsv, "\\%c", ok);
2174 /* isPRINT() is the locale-blind version. */
2175 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
2176 Perl_sv_catpvf(aTHX_ dsv, "%c", c);
2181 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
2184 sv_catpvs(dsv, "...");
2190 =for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags
2192 Build to the scalar dsv a displayable version of the scalar sv,
2193 the displayable version being at most pvlim bytes long
2194 (if longer, the rest is truncated and "..." will be appended).
2196 The flags argument is as in pv_uni_display().
2198 The pointer to the PV of the dsv is returned.
2203 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
2205 return Perl_pv_uni_display(aTHX_ dsv, (const U8*)SvPVX_const(ssv),
2206 SvCUR(ssv), pvlim, flags);
2210 =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
2212 Return true if the strings s1 and s2 differ case-insensitively, false
2213 if not (if they are equal case-insensitively). If u1 is true, the
2214 string s1 is assumed to be in UTF-8-encoded Unicode. If u2 is true,
2215 the string s2 is assumed to be in UTF-8-encoded Unicode. If u1 or u2
2216 are false, the respective string is assumed to be in native 8-bit
2219 If the pe1 and pe2 are non-NULL, the scanning pointers will be copied
2220 in there (they will point at the beginning of the I<next> character).
2221 If the pointers behind pe1 or pe2 are non-NULL, they are the end
2222 pointers beyond which scanning will not continue under any
2223 circumstances. If the byte lengths l1 and l2 are non-zero, s1+l1 and
2224 s2+l2 will be used as goal end pointers that will also stop the scan,
2225 and which qualify towards defining a successful match: all the scans
2226 that define an explicit length must reach their goal pointers for
2227 a match to succeed).
2229 For case-insensitiveness, the "casefolding" of Unicode is used
2230 instead of upper/lowercasing both the characters, see
2231 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
2235 Perl_ibcmp_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
2238 register const U8 *p1 = (const U8*)s1;
2239 register const U8 *p2 = (const U8*)s2;
2240 register const U8 *f1 = NULL;
2241 register const U8 *f2 = NULL;
2242 register U8 *e1 = NULL;
2243 register U8 *q1 = NULL;
2244 register U8 *e2 = NULL;
2245 register U8 *q2 = NULL;
2246 STRLEN n1 = 0, n2 = 0;
2247 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
2248 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
2250 STRLEN foldlen1, foldlen2;
2255 if (e1 == 0 || (l1 && l1 < (UV)(e1 - (const U8*)s1)))
2256 f1 = (const U8*)s1 + l1;
2259 if (e2 == 0 || (l2 && l2 < (UV)(e2 - (const U8*)s2)))
2260 f2 = (const U8*)s2 + l2;
2262 if ((e1 == 0 && f1 == 0) || (e2 == 0 && f2 == 0) || (f1 == 0 && f2 == 0))
2263 return 1; /* mismatch; possible infinite loop or false positive */
2266 natbuf[1] = 0; /* Need to terminate the buffer. */
2268 while ((e1 == 0 || p1 < e1) &&
2269 (f1 == 0 || p1 < f1) &&
2270 (e2 == 0 || p2 < e2) &&
2271 (f2 == 0 || p2 < f2)) {
2274 to_utf8_fold(p1, foldbuf1, &foldlen1);
2276 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p1)));
2277 to_utf8_fold(natbuf, foldbuf1, &foldlen1);
2284 to_utf8_fold(p2, foldbuf2, &foldlen2);
2286 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p2)));
2287 to_utf8_fold(natbuf, foldbuf2, &foldlen2);
2293 if ( UTF8SKIP(q1) != UTF8SKIP(q2) ||
2294 (UTF8SKIP(q1) == 1 && *q1 != *q2) ||
2295 memNE((char*)q1, (char*)q2, UTF8SKIP(q1)) )
2296 return 1; /* mismatch */
2303 p1 += u1 ? UTF8SKIP(p1) : 1;
2305 p2 += u2 ? UTF8SKIP(p2) : 1;
2309 /* A match is defined by all the scans that specified
2310 * an explicit length reaching their final goals. */
2311 match = (f1 == 0 || p1 == f1) && (f2 == 0 || p2 == f2);
2320 return match ? 0 : 1; /* 0 match, 1 mismatch */
2325 * c-indentation-style: bsd
2327 * indent-tabs-mode: t
2330 * ex: set ts=8 sts=4 sw=4 noet: