3 * Copyright (C) 2000, 2001, 2002, 2003, 2004, 2005 by Larry Wall and
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * 'What a fix!' said Sam. 'That's the one place in all the lands we've ever
13 * heard of that we don't want to see any closer; and that's the one place
14 * we're trying to get to! And that's just where we can't get, nohow.'
16 * 'Well do I understand your speech,' he answered in the same language;
17 * 'yet few strangers do so. Why then do you not speak in the Common Tongue,
18 * as is the custom in the West, if you wish to be answered?'
20 * ...the travellers perceived that the floor was paved with stones of many
21 * hues; branching runes and strange devices intertwined beneath their feet.
25 #define PERL_IN_UTF8_C
28 static const char unees[] =
29 "Malformed UTF-8 character (unexpected end of string)";
32 =head1 Unicode Support
34 This file contains various utility functions for manipulating UTF8-encoded
35 strings. For the uninitiated, this is a method of representing arbitrary
36 Unicode characters as a variable number of bytes, in such a way that
37 characters in the ASCII range are unmodified, and a zero byte never appears
38 within non-zero characters.
40 =for apidoc A|U8 *|uvuni_to_utf8_flags|U8 *d|UV uv|UV flags
42 Adds the UTF-8 representation of the Unicode codepoint C<uv> to the end
43 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
44 bytes available. The return value is the pointer to the byte after the
45 end of the new character. In other words,
47 d = uvuni_to_utf8_flags(d, uv, flags);
51 d = uvuni_to_utf8(d, uv);
53 (which is equivalent to)
55 d = uvuni_to_utf8_flags(d, uv, 0);
57 is the recommended Unicode-aware way of saying
65 Perl_uvuni_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
67 if (ckWARN(WARN_UTF8)) {
68 if (UNICODE_IS_SURROGATE(uv) &&
69 !(flags & UNICODE_ALLOW_SURROGATE))
70 Perl_warner(aTHX_ packWARN(WARN_UTF8), "UTF-16 surrogate 0x%04"UVxf, uv);
72 ((uv >= 0xFDD0 && uv <= 0xFDEF &&
73 !(flags & UNICODE_ALLOW_FDD0))
75 ((uv & 0xFFFE) == 0xFFFE && /* Either FFFE or FFFF. */
76 !(flags & UNICODE_ALLOW_FFFF))) &&
77 /* UNICODE_ALLOW_SUPER includes
78 * FFFEs and FFFFs beyond 0x10FFFF. */
79 ((uv <= PERL_UNICODE_MAX) ||
80 !(flags & UNICODE_ALLOW_SUPER))
82 Perl_warner(aTHX_ packWARN(WARN_UTF8),
83 "Unicode character 0x%04"UVxf" is illegal", uv);
85 if (UNI_IS_INVARIANT(uv)) {
86 *d++ = (U8)UTF_TO_NATIVE(uv);
91 STRLEN len = UNISKIP(uv);
94 *p-- = (U8)UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
95 uv >>= UTF_ACCUMULATION_SHIFT;
97 *p = (U8)UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
100 #else /* Non loop style */
102 *d++ = (U8)(( uv >> 6) | 0xc0);
103 *d++ = (U8)(( uv & 0x3f) | 0x80);
107 *d++ = (U8)(( uv >> 12) | 0xe0);
108 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
109 *d++ = (U8)(( uv & 0x3f) | 0x80);
113 *d++ = (U8)(( uv >> 18) | 0xf0);
114 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
115 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
116 *d++ = (U8)(( uv & 0x3f) | 0x80);
119 if (uv < 0x4000000) {
120 *d++ = (U8)(( uv >> 24) | 0xf8);
121 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
122 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
123 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
124 *d++ = (U8)(( uv & 0x3f) | 0x80);
127 if (uv < 0x80000000) {
128 *d++ = (U8)(( uv >> 30) | 0xfc);
129 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
130 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
131 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
132 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
133 *d++ = (U8)(( uv & 0x3f) | 0x80);
137 if (uv < UTF8_QUAD_MAX)
140 *d++ = 0xfe; /* Can't match U+FEFF! */
141 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
142 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
143 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
144 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
145 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
146 *d++ = (U8)(( uv & 0x3f) | 0x80);
151 *d++ = 0xff; /* Can't match U+FFFE! */
152 *d++ = 0x80; /* 6 Reserved bits */
153 *d++ = (U8)(((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
154 *d++ = (U8)(((uv >> 54) & 0x3f) | 0x80);
155 *d++ = (U8)(((uv >> 48) & 0x3f) | 0x80);
156 *d++ = (U8)(((uv >> 42) & 0x3f) | 0x80);
157 *d++ = (U8)(((uv >> 36) & 0x3f) | 0x80);
158 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
159 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
160 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
161 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
162 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
163 *d++ = (U8)(( uv & 0x3f) | 0x80);
167 #endif /* Loop style */
171 Perl_uvuni_to_utf8(pTHX_ U8 *d, UV uv)
173 return Perl_uvuni_to_utf8_flags(aTHX_ d, uv, 0);
178 Tests if some arbitrary number of bytes begins in a valid UTF-8
179 character. Note that an INVARIANT (i.e. ASCII) character is a valid
180 UTF-8 character. The actual number of bytes in the UTF-8 character
181 will be returned if it is valid, otherwise 0.
183 This is the "slow" version as opposed to the "fast" version which is
184 the "unrolled" IS_UTF8_CHAR(). E.g. for t/uni/class.t the speed
185 difference is a factor of 2 to 3. For lengths (UTF8SKIP(s)) of four
186 or less you should use the IS_UTF8_CHAR(), for lengths of five or more
187 you should use the _slow(). In practice this means that the _slow()
188 will be used very rarely, since the maximum Unicode code point (as of
189 Unicode 4.1) is U+10FFFF, which encodes in UTF-8 to four bytes. Only
190 the "Perl extended UTF-8" (the infamous 'v-strings') will encode into
195 S_is_utf8_char_slow(pTHX_ const U8 *s, const STRLEN len)
201 if (UTF8_IS_INVARIANT(u))
204 if (!UTF8_IS_START(u))
207 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
212 u &= UTF_START_MASK(len);
216 if (!UTF8_IS_CONTINUATION(*s))
218 uv = UTF8_ACCUMULATE(uv, *s);
225 if ((STRLEN)UNISKIP(uv) < len)
232 =for apidoc A|STRLEN|is_utf8_char|const U8 *s
234 Tests if some arbitrary number of bytes begins in a valid UTF-8
235 character. Note that an INVARIANT (i.e. ASCII) character is a valid
236 UTF-8 character. The actual number of bytes in the UTF-8 character
237 will be returned if it is valid, otherwise 0.
241 Perl_is_utf8_char(pTHX_ const U8 *s)
243 STRLEN len = UTF8SKIP(s);
245 if (IS_UTF8_CHAR_FAST(len))
246 return IS_UTF8_CHAR(s, len) ? len : 0;
247 #endif /* #ifdef IS_UTF8_CHAR */
248 return is_utf8_char_slow(s, len);
252 =for apidoc A|bool|is_utf8_string|const U8 *s|STRLEN len
254 Returns true if first C<len> bytes of the given string form a valid
255 UTF-8 string, false otherwise. Note that 'a valid UTF-8 string' does
256 not mean 'a string that contains code points above 0x7F encoded in UTF-8'
257 because a valid ASCII string is a valid UTF-8 string.
259 See also is_utf8_string_loclen() and is_utf8_string_loc().
265 Perl_is_utf8_string(pTHX_ const U8 *s, STRLEN len)
271 len = strlen((const char *)s);
276 /* Inline the easy bits of is_utf8_char() here for speed... */
277 if (UTF8_IS_INVARIANT(*x))
279 else if (!UTF8_IS_START(*x))
282 /* ... and call is_utf8_char() only if really needed. */
285 if (IS_UTF8_CHAR_FAST(c)) {
286 if (!IS_UTF8_CHAR(x, c))
288 } else if (!is_utf8_char_slow(x, c))
292 #endif /* #ifdef IS_UTF8_CHAR */
307 =for apidoc A|bool|is_utf8_string_loclen|const U8 *s|STRLEN len|const U8 **ep|const STRLEN *el
309 Like is_ut8_string() but stores the location of the failure (in the
310 case of "utf8ness failure") or the location s+len (in the case of
311 "utf8ness success") in the C<ep>, and the number of UTF-8
312 encoded characters in the C<el>.
314 See also is_utf8_string_loc() and is_utf8_string().
320 Perl_is_utf8_string_loclen(pTHX_ const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
327 len = strlen((const char *)s);
333 /* Inline the easy bits of is_utf8_char() here for speed... */
334 if (UTF8_IS_INVARIANT(*x))
336 else if (!UTF8_IS_START(*x))
339 /* ... and call is_utf8_char() only if really needed. */
342 if (IS_UTF8_CHAR_FAST(c)) {
343 if (!IS_UTF8_CHAR(x, c))
346 c = is_utf8_char_slow(x, c);
349 #endif /* #ifdef IS_UTF8_CHAR */
368 =for apidoc A|bool|is_utf8_string_loc|const U8 *s|STRLEN len|const U8 **ep|const STRLEN *el
370 Like is_ut8_string() but stores the location of the failure (in the
371 case of "utf8ness failure") or the location s+len (in the case of
372 "utf8ness success") in the C<ep>.
374 See also is_utf8_string_loclen() and is_utf8_string().
380 Perl_is_utf8_string_loc(pTHX_ const U8 *s, STRLEN len, const U8 **ep)
382 return is_utf8_string_loclen(s, len, ep, 0);
386 =for apidoc A|UV|utf8n_to_uvuni|const U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
388 Bottom level UTF-8 decode routine.
389 Returns the unicode code point value of the first character in the string C<s>
390 which is assumed to be in UTF-8 encoding and no longer than C<curlen>;
391 C<retlen> will be set to the length, in bytes, of that character.
393 If C<s> does not point to a well-formed UTF-8 character, the behaviour
394 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
395 it is assumed that the caller will raise a warning, and this function
396 will silently just set C<retlen> to C<-1> and return zero. If the
397 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
398 malformations will be given, C<retlen> will be set to the expected
399 length of the UTF-8 character in bytes, and zero will be returned.
401 The C<flags> can also contain various flags to allow deviations from
402 the strict UTF-8 encoding (see F<utf8.h>).
404 Most code should use utf8_to_uvchr() rather than call this directly.
410 Perl_utf8n_to_uvuni(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
415 const bool dowarn = ckWARN_d(WARN_UTF8);
416 const UV startbyte = *s;
417 STRLEN expectlen = 0;
420 /* This list is a superset of the UTF8_ALLOW_XXX. */
422 #define UTF8_WARN_EMPTY 1
423 #define UTF8_WARN_CONTINUATION 2
424 #define UTF8_WARN_NON_CONTINUATION 3
425 #define UTF8_WARN_FE_FF 4
426 #define UTF8_WARN_SHORT 5
427 #define UTF8_WARN_OVERFLOW 6
428 #define UTF8_WARN_SURROGATE 7
429 #define UTF8_WARN_LONG 8
430 #define UTF8_WARN_FFFF 9 /* Also FFFE. */
433 !(flags & UTF8_ALLOW_EMPTY)) {
434 warning = UTF8_WARN_EMPTY;
438 if (UTF8_IS_INVARIANT(uv)) {
441 return (UV) (NATIVE_TO_UTF(*s));
444 if (UTF8_IS_CONTINUATION(uv) &&
445 !(flags & UTF8_ALLOW_CONTINUATION)) {
446 warning = UTF8_WARN_CONTINUATION;
450 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
451 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
452 warning = UTF8_WARN_NON_CONTINUATION;
457 uv = NATIVE_TO_UTF(uv);
459 if ((uv == 0xfe || uv == 0xff) &&
460 !(flags & UTF8_ALLOW_FE_FF)) {
461 warning = UTF8_WARN_FE_FF;
466 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
467 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
468 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
469 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
471 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
472 else { len = 7; uv &= 0x01; }
474 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
475 else if (!(uv & 0x01)) { len = 7; uv = 0; }
476 else { len = 13; uv = 0; } /* whoa! */
484 if ((curlen < expectlen) &&
485 !(flags & UTF8_ALLOW_SHORT)) {
486 warning = UTF8_WARN_SHORT;
495 if (!UTF8_IS_CONTINUATION(*s) &&
496 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
498 warning = UTF8_WARN_NON_CONTINUATION;
502 uv = UTF8_ACCUMULATE(uv, *s);
504 /* These cannot be allowed. */
506 if (expectlen != 13 && !(flags & UTF8_ALLOW_LONG)) {
507 warning = UTF8_WARN_LONG;
511 else { /* uv < ouv */
512 /* This cannot be allowed. */
513 warning = UTF8_WARN_OVERFLOW;
521 if (UNICODE_IS_SURROGATE(uv) &&
522 !(flags & UTF8_ALLOW_SURROGATE)) {
523 warning = UTF8_WARN_SURROGATE;
525 } else if ((expectlen > (STRLEN)UNISKIP(uv)) &&
526 !(flags & UTF8_ALLOW_LONG)) {
527 warning = UTF8_WARN_LONG;
529 } else if (UNICODE_IS_ILLEGAL(uv) &&
530 !(flags & UTF8_ALLOW_FFFF)) {
531 warning = UTF8_WARN_FFFF;
539 if (flags & UTF8_CHECK_ONLY) {
546 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
549 case 0: /* Intentionally empty. */ break;
550 case UTF8_WARN_EMPTY:
551 Perl_sv_catpv(aTHX_ sv, "(empty string)");
553 case UTF8_WARN_CONTINUATION:
554 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
556 case UTF8_WARN_NON_CONTINUATION:
558 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
559 (UV)s[1], startbyte);
561 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
562 (UV)s[1], s - s0, s - s0 > 1 ? "s" : "", startbyte, (int)expectlen);
565 case UTF8_WARN_FE_FF:
566 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
568 case UTF8_WARN_SHORT:
569 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
570 (int)curlen, curlen == 1 ? "" : "s", (int)expectlen, startbyte);
571 expectlen = curlen; /* distance for caller to skip */
573 case UTF8_WARN_OVERFLOW:
574 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
577 case UTF8_WARN_SURROGATE:
578 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
581 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
582 (int)expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
585 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
588 Perl_sv_catpv(aTHX_ sv, "(unknown reason)");
593 const char *s = SvPVX_const(sv);
596 Perl_warner(aTHX_ packWARN(WARN_UTF8),
597 "%s in %s", s, OP_DESC(PL_op));
599 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
604 *retlen = expectlen ? expectlen : len;
610 =for apidoc A|UV|utf8_to_uvchr|const U8 *s|STRLEN *retlen
612 Returns the native character value of the first character in the string C<s>
613 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
614 length, in bytes, of that character.
616 If C<s> does not point to a well-formed UTF-8 character, zero is
617 returned and retlen is set, if possible, to -1.
623 Perl_utf8_to_uvchr(pTHX_ const U8 *s, STRLEN *retlen)
625 return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXBYTES, retlen,
626 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
630 =for apidoc A|UV|utf8_to_uvuni|const U8 *s|STRLEN *retlen
632 Returns the Unicode code point of the first character in the string C<s>
633 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
634 length, in bytes, of that character.
636 This function should only be used when returned UV is considered
637 an index into the Unicode semantic tables (e.g. swashes).
639 If C<s> does not point to a well-formed UTF-8 character, zero is
640 returned and retlen is set, if possible, to -1.
646 Perl_utf8_to_uvuni(pTHX_ const U8 *s, STRLEN *retlen)
648 /* Call the low level routine asking for checks */
649 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXBYTES, retlen,
650 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
654 =for apidoc A|STRLEN|utf8_length|const U8 *s|const U8 *e
656 Return the length of the UTF-8 char encoded string C<s> in characters.
657 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
658 up past C<e>, croaks.
664 Perl_utf8_length(pTHX_ const U8 *s, const U8 *e)
668 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
669 * the bitops (especially ~) can create illegal UTF-8.
670 * In other words: in Perl UTF-8 is not just for Unicode. */
673 goto warn_and_return;
675 const U8 t = UTF8SKIP(s);
678 if (ckWARN_d(WARN_UTF8)) {
680 Perl_warner(aTHX_ packWARN(WARN_UTF8),
681 "%s in %s", unees, OP_DESC(PL_op));
683 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
695 =for apidoc A|IV|utf8_distance|const U8 *a|const U8 *b
697 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
700 WARNING: use only if you *know* that the pointers point inside the
707 Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
711 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
712 * the bitops (especially ~) can create illegal UTF-8.
713 * In other words: in Perl UTF-8 is not just for Unicode. */
717 const U8 c = UTF8SKIP(a);
719 goto warn_and_return;
726 const U8 c = UTF8SKIP(b);
730 if (ckWARN_d(WARN_UTF8)) {
732 Perl_warner(aTHX_ packWARN(WARN_UTF8),
733 "%s in %s", unees, OP_DESC(PL_op));
735 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
748 =for apidoc A|U8 *|utf8_hop|U8 *s|I32 off
750 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
753 WARNING: do not use the following unless you *know* C<off> is within
754 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
755 on the first byte of character or just after the last byte of a character.
761 Perl_utf8_hop(pTHX_ const U8 *s, I32 off)
763 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
764 * the bitops (especially ~) can create illegal UTF-8.
765 * In other words: in Perl UTF-8 is not just for Unicode. */
774 while (UTF8_IS_CONTINUATION(*s))
782 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
784 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
785 Unlike C<bytes_to_utf8>, this over-writes the original string, and
786 updates len to contain the new length.
787 Returns zero on failure, setting C<len> to -1.
793 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
799 /* ensure valid UTF-8 and chars < 256 before updating string */
800 for (send = s + *len; s < send; ) {
803 if (!UTF8_IS_INVARIANT(c) &&
804 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
805 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
814 *d++ = (U8)utf8_to_uvchr(s, &ulen);
823 =for apidoc A|U8 *|bytes_from_utf8|const U8 *s|STRLEN *len|bool *is_utf8
825 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
826 Unlike C<utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
827 the newly-created string, and updates C<len> to contain the new
828 length. Returns the original string if no conversion occurs, C<len>
829 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
830 0 if C<s> is converted or contains all 7bit characters.
836 Perl_bytes_from_utf8(pTHX_ const U8 *s, STRLEN *len, bool *is_utf8)
846 /* ensure valid UTF-8 and chars < 256 before converting string */
847 for (send = s + *len; s < send;) {
849 if (!UTF8_IS_INVARIANT(c)) {
850 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
851 (c = *s++) && UTF8_IS_CONTINUATION(c))
860 Newz(801, d, (*len) - count + 1, U8);
861 s = start; start = d;
864 if (!UTF8_IS_INVARIANT(c)) {
865 /* Then it is two-byte encoded */
866 c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++);
867 c = ASCII_TO_NATIVE(c);
877 =for apidoc A|U8 *|bytes_to_utf8|const U8 *s|STRLEN *len
879 Converts a string C<s> of length C<len> from ASCII into UTF-8 encoding.
880 Returns a pointer to the newly-created string, and sets C<len> to
881 reflect the new length.
883 If you want to convert to UTF-8 from other encodings than ASCII,
884 see sv_recode_to_utf8().
890 Perl_bytes_to_utf8(pTHX_ const U8 *s, STRLEN *len)
892 const U8 * const send = s + (*len);
896 Newz(801, d, (*len) * 2 + 1, U8);
900 const UV uv = NATIVE_TO_ASCII(*s++);
901 if (UNI_IS_INVARIANT(uv))
902 *d++ = (U8)UTF_TO_NATIVE(uv);
904 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
905 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
914 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
916 * Destination must be pre-extended to 3/2 source. Do not use in-place.
917 * We optimize for native, for obvious reasons. */
920 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
925 if (bytelen == 1 && p[0] == 0) { /* Be understanding. */
932 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %"UVf, (UV)bytelen);
937 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
944 *d++ = (U8)(( uv >> 6) | 0xc0);
945 *d++ = (U8)(( uv & 0x3f) | 0x80);
948 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
949 UV low = (p[0] << 8) + p[1];
951 if (low < 0xdc00 || low >= 0xdfff)
952 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
953 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
956 *d++ = (U8)(( uv >> 12) | 0xe0);
957 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
958 *d++ = (U8)(( uv & 0x3f) | 0x80);
962 *d++ = (U8)(( uv >> 18) | 0xf0);
963 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
964 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
965 *d++ = (U8)(( uv & 0x3f) | 0x80);
969 *newlen = d - dstart;
973 /* Note: this one is slightly destructive of the source. */
976 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
979 U8* send = s + bytelen;
986 return utf16_to_utf8(p, d, bytelen, newlen);
989 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
992 Perl_is_uni_alnum(pTHX_ UV c)
994 U8 tmpbuf[UTF8_MAXBYTES+1];
995 uvchr_to_utf8(tmpbuf, c);
996 return is_utf8_alnum(tmpbuf);
1000 Perl_is_uni_alnumc(pTHX_ UV c)
1002 U8 tmpbuf[UTF8_MAXBYTES+1];
1003 uvchr_to_utf8(tmpbuf, c);
1004 return is_utf8_alnumc(tmpbuf);
1008 Perl_is_uni_idfirst(pTHX_ UV c)
1010 U8 tmpbuf[UTF8_MAXBYTES+1];
1011 uvchr_to_utf8(tmpbuf, c);
1012 return is_utf8_idfirst(tmpbuf);
1016 Perl_is_uni_alpha(pTHX_ UV c)
1018 U8 tmpbuf[UTF8_MAXBYTES+1];
1019 uvchr_to_utf8(tmpbuf, c);
1020 return is_utf8_alpha(tmpbuf);
1024 Perl_is_uni_ascii(pTHX_ UV c)
1026 U8 tmpbuf[UTF8_MAXBYTES+1];
1027 uvchr_to_utf8(tmpbuf, c);
1028 return is_utf8_ascii(tmpbuf);
1032 Perl_is_uni_space(pTHX_ UV c)
1034 U8 tmpbuf[UTF8_MAXBYTES+1];
1035 uvchr_to_utf8(tmpbuf, c);
1036 return is_utf8_space(tmpbuf);
1040 Perl_is_uni_digit(pTHX_ UV c)
1042 U8 tmpbuf[UTF8_MAXBYTES+1];
1043 uvchr_to_utf8(tmpbuf, c);
1044 return is_utf8_digit(tmpbuf);
1048 Perl_is_uni_upper(pTHX_ UV c)
1050 U8 tmpbuf[UTF8_MAXBYTES+1];
1051 uvchr_to_utf8(tmpbuf, c);
1052 return is_utf8_upper(tmpbuf);
1056 Perl_is_uni_lower(pTHX_ UV c)
1058 U8 tmpbuf[UTF8_MAXBYTES+1];
1059 uvchr_to_utf8(tmpbuf, c);
1060 return is_utf8_lower(tmpbuf);
1064 Perl_is_uni_cntrl(pTHX_ UV c)
1066 U8 tmpbuf[UTF8_MAXBYTES+1];
1067 uvchr_to_utf8(tmpbuf, c);
1068 return is_utf8_cntrl(tmpbuf);
1072 Perl_is_uni_graph(pTHX_ UV c)
1074 U8 tmpbuf[UTF8_MAXBYTES+1];
1075 uvchr_to_utf8(tmpbuf, c);
1076 return is_utf8_graph(tmpbuf);
1080 Perl_is_uni_print(pTHX_ UV c)
1082 U8 tmpbuf[UTF8_MAXBYTES+1];
1083 uvchr_to_utf8(tmpbuf, c);
1084 return is_utf8_print(tmpbuf);
1088 Perl_is_uni_punct(pTHX_ UV c)
1090 U8 tmpbuf[UTF8_MAXBYTES+1];
1091 uvchr_to_utf8(tmpbuf, c);
1092 return is_utf8_punct(tmpbuf);
1096 Perl_is_uni_xdigit(pTHX_ UV c)
1098 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1099 uvchr_to_utf8(tmpbuf, c);
1100 return is_utf8_xdigit(tmpbuf);
1104 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
1106 uvchr_to_utf8(p, c);
1107 return to_utf8_upper(p, p, lenp);
1111 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
1113 uvchr_to_utf8(p, c);
1114 return to_utf8_title(p, p, lenp);
1118 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
1120 uvchr_to_utf8(p, c);
1121 return to_utf8_lower(p, p, lenp);
1125 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1127 uvchr_to_utf8(p, c);
1128 return to_utf8_fold(p, p, lenp);
1131 /* for now these all assume no locale info available for Unicode > 255 */
1134 Perl_is_uni_alnum_lc(pTHX_ UV c)
1136 return is_uni_alnum(c); /* XXX no locale support yet */
1140 Perl_is_uni_alnumc_lc(pTHX_ UV c)
1142 return is_uni_alnumc(c); /* XXX no locale support yet */
1146 Perl_is_uni_idfirst_lc(pTHX_ UV c)
1148 return is_uni_idfirst(c); /* XXX no locale support yet */
1152 Perl_is_uni_alpha_lc(pTHX_ UV c)
1154 return is_uni_alpha(c); /* XXX no locale support yet */
1158 Perl_is_uni_ascii_lc(pTHX_ UV c)
1160 return is_uni_ascii(c); /* XXX no locale support yet */
1164 Perl_is_uni_space_lc(pTHX_ UV c)
1166 return is_uni_space(c); /* XXX no locale support yet */
1170 Perl_is_uni_digit_lc(pTHX_ UV c)
1172 return is_uni_digit(c); /* XXX no locale support yet */
1176 Perl_is_uni_upper_lc(pTHX_ UV c)
1178 return is_uni_upper(c); /* XXX no locale support yet */
1182 Perl_is_uni_lower_lc(pTHX_ UV c)
1184 return is_uni_lower(c); /* XXX no locale support yet */
1188 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1190 return is_uni_cntrl(c); /* XXX no locale support yet */
1194 Perl_is_uni_graph_lc(pTHX_ UV c)
1196 return is_uni_graph(c); /* XXX no locale support yet */
1200 Perl_is_uni_print_lc(pTHX_ UV c)
1202 return is_uni_print(c); /* XXX no locale support yet */
1206 Perl_is_uni_punct_lc(pTHX_ UV c)
1208 return is_uni_punct(c); /* XXX no locale support yet */
1212 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1214 return is_uni_xdigit(c); /* XXX no locale support yet */
1218 Perl_to_uni_upper_lc(pTHX_ U32 c)
1220 /* XXX returns only the first character -- do not use XXX */
1221 /* XXX no locale support yet */
1223 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1224 return (U32)to_uni_upper(c, tmpbuf, &len);
1228 Perl_to_uni_title_lc(pTHX_ U32 c)
1230 /* XXX returns only the first character XXX -- do not use XXX */
1231 /* XXX no locale support yet */
1233 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1234 return (U32)to_uni_title(c, tmpbuf, &len);
1238 Perl_to_uni_lower_lc(pTHX_ U32 c)
1240 /* XXX returns only the first character -- do not use XXX */
1241 /* XXX no locale support yet */
1243 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1244 return (U32)to_uni_lower(c, tmpbuf, &len);
1248 Perl_is_utf8_alnum(pTHX_ const U8 *p)
1250 if (!is_utf8_char(p))
1253 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1254 * descendant of isalnum(3), in other words, it doesn't
1255 * contain the '_'. --jhi */
1256 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1257 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1258 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1259 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1261 PL_utf8_alnum = swash_init("utf8", "",
1262 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1263 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1268 Perl_is_utf8_alnumc(pTHX_ const U8 *p)
1270 if (!is_utf8_char(p))
1273 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1274 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1275 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
1276 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1278 PL_utf8_alnum = swash_init("utf8", "",
1279 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1280 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1285 Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
1289 if (!is_utf8_char(p))
1291 if (!PL_utf8_idstart) /* is_utf8_idstart would be more logical. */
1292 PL_utf8_idstart = swash_init("utf8", "IdStart", &PL_sv_undef, 0, 0);
1293 return swash_fetch(PL_utf8_idstart, p, TRUE) != 0;
1297 Perl_is_utf8_idcont(pTHX_ const U8 *p)
1301 if (!is_utf8_char(p))
1303 if (!PL_utf8_idcont)
1304 PL_utf8_idcont = swash_init("utf8", "IdContinue", &PL_sv_undef, 0, 0);
1305 return swash_fetch(PL_utf8_idcont, p, TRUE) != 0;
1309 Perl_is_utf8_alpha(pTHX_ const U8 *p)
1311 if (!is_utf8_char(p))
1314 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1315 return swash_fetch(PL_utf8_alpha, p, TRUE) != 0;
1319 Perl_is_utf8_ascii(pTHX_ const U8 *p)
1321 if (!is_utf8_char(p))
1324 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1325 return swash_fetch(PL_utf8_ascii, p, TRUE) != 0;
1329 Perl_is_utf8_space(pTHX_ const U8 *p)
1331 if (!is_utf8_char(p))
1334 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1335 return swash_fetch(PL_utf8_space, p, TRUE) != 0;
1339 Perl_is_utf8_digit(pTHX_ const U8 *p)
1341 if (!is_utf8_char(p))
1344 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1345 return swash_fetch(PL_utf8_digit, p, TRUE) != 0;
1349 Perl_is_utf8_upper(pTHX_ const U8 *p)
1351 if (!is_utf8_char(p))
1354 PL_utf8_upper = swash_init("utf8", "IsUppercase", &PL_sv_undef, 0, 0);
1355 return swash_fetch(PL_utf8_upper, p, TRUE) != 0;
1359 Perl_is_utf8_lower(pTHX_ const U8 *p)
1361 if (!is_utf8_char(p))
1364 PL_utf8_lower = swash_init("utf8", "IsLowercase", &PL_sv_undef, 0, 0);
1365 return swash_fetch(PL_utf8_lower, p, TRUE) != 0;
1369 Perl_is_utf8_cntrl(pTHX_ const U8 *p)
1371 if (!is_utf8_char(p))
1374 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1375 return swash_fetch(PL_utf8_cntrl, p, TRUE) != 0;
1379 Perl_is_utf8_graph(pTHX_ const U8 *p)
1381 if (!is_utf8_char(p))
1384 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1385 return swash_fetch(PL_utf8_graph, p, TRUE) != 0;
1389 Perl_is_utf8_print(pTHX_ const U8 *p)
1391 if (!is_utf8_char(p))
1394 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1395 return swash_fetch(PL_utf8_print, p, TRUE) != 0;
1399 Perl_is_utf8_punct(pTHX_ const U8 *p)
1401 if (!is_utf8_char(p))
1404 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1405 return swash_fetch(PL_utf8_punct, p, TRUE) != 0;
1409 Perl_is_utf8_xdigit(pTHX_ const U8 *p)
1411 if (!is_utf8_char(p))
1413 if (!PL_utf8_xdigit)
1414 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1415 return swash_fetch(PL_utf8_xdigit, p, TRUE) != 0;
1419 Perl_is_utf8_mark(pTHX_ const U8 *p)
1421 if (!is_utf8_char(p))
1424 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1425 return swash_fetch(PL_utf8_mark, p, TRUE) != 0;
1429 =for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special
1431 The "p" contains the pointer to the UTF-8 string encoding
1432 the character that is being converted.
1434 The "ustrp" is a pointer to the character buffer to put the
1435 conversion result to. The "lenp" is a pointer to the length
1438 The "swashp" is a pointer to the swash to use.
1440 Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
1441 and loaded by SWASHGET, using lib/utf8_heavy.pl. The special (usually,
1442 but not always, a multicharacter mapping), is tried first.
1444 The "special" is a string like "utf8::ToSpecLower", which means the
1445 hash %utf8::ToSpecLower. The access to the hash is through
1446 Perl_to_utf8_case().
1448 The "normal" is a string like "ToLower" which means the swash
1454 Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp, const char *normal, const char *special)
1456 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1459 const UV uv0 = utf8_to_uvchr(p, 0);
1460 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1461 * are necessary in EBCDIC, they are redundant no-ops
1462 * in ASCII-ish platforms, and hopefully optimized away. */
1463 const UV uv1 = NATIVE_TO_UNI(uv0);
1464 uvuni_to_utf8(tmpbuf, uv1);
1466 if (!*swashp) /* load on-demand */
1467 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1469 /* The 0xDF is the only special casing Unicode code point below 0x100. */
1470 if (special && (uv1 == 0xDF || uv1 > 0xFF)) {
1471 /* It might be "special" (sometimes, but not always,
1472 * a multicharacter mapping) */
1476 if ((hv = get_hv(special, FALSE)) &&
1477 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1481 s = SvPV_const(*svp, len);
1483 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
1486 /* If we have EBCDIC we need to remap the characters
1487 * since any characters in the low 256 are Unicode
1488 * code points, not EBCDIC. */
1489 U8 *t = (U8*)s, *tend = t + len, *d;
1496 UV c = utf8_to_uvchr(t, &tlen);
1498 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1507 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1512 Copy(tmpbuf, ustrp, len, U8);
1514 Copy(s, ustrp, len, U8);
1520 if (!len && *swashp) {
1521 UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1524 /* It was "normal" (a single character mapping). */
1525 UV uv3 = UNI_TO_NATIVE(uv2);
1527 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
1531 if (!len) /* Neither: just copy. */
1532 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1537 return len ? utf8_to_uvchr(ustrp, 0) : 0;
1541 =for apidoc A|UV|to_utf8_upper|const U8 *p|U8 *ustrp|STRLEN *lenp
1543 Convert the UTF-8 encoded character at p to its uppercase version and
1544 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1545 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1546 the uppercase version may be longer than the original character.
1548 The first character of the uppercased version is returned
1549 (but note, as explained above, that there may be more.)
1554 Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1556 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1557 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1561 =for apidoc A|UV|to_utf8_title|const U8 *p|U8 *ustrp|STRLEN *lenp
1563 Convert the UTF-8 encoded character at p to its titlecase version and
1564 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1565 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1566 titlecase version may be longer than the original character.
1568 The first character of the titlecased version is returned
1569 (but note, as explained above, that there may be more.)
1574 Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1576 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1577 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1581 =for apidoc A|UV|to_utf8_lower|const U8 *p|U8 *ustrp|STRLEN *lenp
1583 Convert the UTF-8 encoded character at p to its lowercase version and
1584 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1585 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1586 lowercase version may be longer than the original character.
1588 The first character of the lowercased version is returned
1589 (but note, as explained above, that there may be more.)
1594 Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1596 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1597 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1601 =for apidoc A|UV|to_utf8_fold|const U8 *p|U8 *ustrp|STRLEN *lenp
1603 Convert the UTF-8 encoded character at p to its foldcase version and
1604 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1605 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1606 foldcase version may be longer than the original character (up to
1609 The first character of the foldcased version is returned
1610 (but note, as explained above, that there may be more.)
1615 Perl_to_utf8_fold(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1617 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1618 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1621 /* a "swash" is a swatch hash */
1624 Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
1628 SV* tokenbufsv = sv_newmortal();
1630 const size_t pkg_len = strlen(pkg);
1631 const size_t name_len = strlen(name);
1632 HV *stash = gv_stashpvn(pkg, pkg_len, FALSE);
1635 PUSHSTACKi(PERLSI_MAGIC);
1640 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
1642 errsv_save = newSVsv(ERRSV);
1643 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
1646 sv_setsv(ERRSV, errsv_save);
1647 SvREFCNT_dec(errsv_save);
1653 PUSHs(sv_2mortal(newSVpvn(pkg, pkg_len)));
1654 PUSHs(sv_2mortal(newSVpvn(name, name_len)));
1656 PUSHs(sv_2mortal(newSViv(minbits)));
1657 PUSHs(sv_2mortal(newSViv(none)));
1659 if (IN_PERL_COMPILETIME) {
1660 /* XXX ought to be handled by lex_start */
1663 sv_setpv(tokenbufsv, PL_tokenbuf);
1665 errsv_save = newSVsv(ERRSV);
1666 if (call_method("SWASHNEW", G_SCALAR))
1667 retval = newSVsv(*PL_stack_sp--);
1669 retval = &PL_sv_undef;
1671 sv_setsv(ERRSV, errsv_save);
1672 SvREFCNT_dec(errsv_save);
1675 if (IN_PERL_COMPILETIME) {
1677 const char* pv = SvPV_const(tokenbufsv, len);
1679 Copy(pv, PL_tokenbuf, len+1, char);
1680 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1682 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
1684 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
1686 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1692 /* This API is wrong for special case conversions since we may need to
1693 * return several Unicode characters for a single Unicode character
1694 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1695 * the lower-level routine, and it is similarly broken for returning
1696 * multiple values. --jhi */
1698 Perl_swash_fetch(pTHX_ SV *sv, const U8 *ptr, bool do_utf8)
1701 HV* hv = (HV*)SvRV(sv);
1706 const U8 *tmps = NULL;
1710 UV c = NATIVE_TO_ASCII(*ptr);
1712 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1713 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
1714 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
1717 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
1718 * then the "swatch" is a vec() for al the chars which start
1720 * So the key in the hash (klen) is length of encoded char -1
1722 klen = UTF8SKIP(ptr) - 1;
1727 /* If char in invariant then swatch is for all the invariant chars
1728 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
1730 needents = UTF_CONTINUATION_MARK;
1731 off = NATIVE_TO_UTF(ptr[klen]);
1735 /* If char is encoded then swatch is for the prefix */
1736 needents = (1 << UTF_ACCUMULATION_SHIFT);
1737 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
1741 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1742 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1743 * it's nothing to sniff at.) Pity we usually come through at least
1744 * two function calls to get here...
1746 * NB: this code assumes that swatches are never modified, once generated!
1749 if (hv == PL_last_swash_hv &&
1750 klen == PL_last_swash_klen &&
1751 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
1753 tmps = PL_last_swash_tmps;
1754 slen = PL_last_swash_slen;
1757 /* Try our second-level swatch cache, kept in a hash. */
1758 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
1760 /* If not cached, generate it via utf8::SWASHGET */
1761 if (!svp || !SvPOK(*svp) || !(tmps = (const U8*)SvPV_const(*svp, slen))) {
1763 /* We use utf8n_to_uvuni() as we want an index into
1764 Unicode tables, not a native character number.
1766 UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
1768 0 : UTF8_ALLOW_ANY);
1773 PUSHSTACKi(PERLSI_MAGIC);
1777 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
1778 PUSHs(sv_2mortal(newSViv((klen) ?
1779 (code_point & ~(needents - 1)) : 0)));
1780 PUSHs(sv_2mortal(newSViv(needents)));
1782 errsv_save = newSVsv(ERRSV);
1783 if (call_method("SWASHGET", G_SCALAR))
1784 retval = newSVsv(*PL_stack_sp--);
1786 retval = &PL_sv_undef;
1788 sv_setsv(ERRSV, errsv_save);
1789 SvREFCNT_dec(errsv_save);
1793 if (IN_PERL_COMPILETIME)
1794 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1796 svp = hv_store(hv, (const char *)ptr, klen, retval, 0);
1798 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || (slen << 3) < needents)
1799 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1802 PL_last_swash_hv = hv;
1803 PL_last_swash_klen = klen;
1804 /* FIXME change interpvar.h? */
1805 PL_last_swash_tmps = (U8 *) tmps;
1806 PL_last_swash_slen = slen;
1808 Copy(ptr, PL_last_swash_key, klen, U8);
1811 switch ((int)((slen << 3) / needents)) {
1813 bit = 1 << (off & 7);
1815 return (tmps[off] & bit) != 0;
1820 return (tmps[off] << 8) + tmps[off + 1] ;
1823 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1825 Perl_croak(aTHX_ "panic: swash_fetch");
1831 =for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv
1833 Adds the UTF-8 representation of the Native codepoint C<uv> to the end
1834 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
1835 bytes available. The return value is the pointer to the byte after the
1836 end of the new character. In other words,
1838 d = uvchr_to_utf8(d, uv);
1840 is the recommended wide native character-aware way of saying
1847 /* On ASCII machines this is normally a macro but we want a
1848 real function in case XS code wants it
1850 #undef Perl_uvchr_to_utf8
1852 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
1854 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
1858 Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
1860 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
1864 =for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
1866 Returns the native character value of the first character in the string C<s>
1867 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
1868 length, in bytes, of that character.
1870 Allows length and flags to be passed to low level routine.
1874 /* On ASCII machines this is normally a macro but we want
1875 a real function in case XS code wants it
1877 #undef Perl_utf8n_to_uvchr
1879 Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
1881 UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
1882 return UNI_TO_NATIVE(uv);
1886 =for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags
1888 Build to the scalar dsv a displayable version of the string spv,
1889 length len, the displayable version being at most pvlim bytes long
1890 (if longer, the rest is truncated and "..." will be appended).
1892 The flags argument can have UNI_DISPLAY_ISPRINT set to display
1893 isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
1894 to display the \\[nrfta\\] as the backslashed versions (like '\n')
1895 (UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
1896 UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
1897 UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
1899 The pointer to the PV of the dsv is returned.
1903 Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
1908 sv_setpvn(dsv, "", 0);
1909 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
1911 /* This serves double duty as a flag and a character to print after
1912 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
1916 if (pvlim && SvCUR(dsv) >= pvlim) {
1920 u = utf8_to_uvchr((U8*)s, 0);
1922 const unsigned char c = (unsigned char)u & 0xFF;
1923 if (!ok && (flags & UNI_DISPLAY_BACKSLASH)) {
1940 Perl_sv_catpvf(aTHX_ dsv, "\\%c", ok);
1943 /* isPRINT() is the locale-blind version. */
1944 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
1945 Perl_sv_catpvf(aTHX_ dsv, "%c", c);
1950 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
1953 sv_catpvn(dsv, "...", 3);
1959 =for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags
1961 Build to the scalar dsv a displayable version of the scalar sv,
1962 the displayable version being at most pvlim bytes long
1963 (if longer, the rest is truncated and "..." will be appended).
1965 The flags argument is as in pv_uni_display().
1967 The pointer to the PV of the dsv is returned.
1971 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
1973 return Perl_pv_uni_display(aTHX_ dsv, (const U8*)SvPVX_const(ssv),
1974 SvCUR(ssv), pvlim, flags);
1978 =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
1980 Return true if the strings s1 and s2 differ case-insensitively, false
1981 if not (if they are equal case-insensitively). If u1 is true, the
1982 string s1 is assumed to be in UTF-8-encoded Unicode. If u2 is true,
1983 the string s2 is assumed to be in UTF-8-encoded Unicode. If u1 or u2
1984 are false, the respective string is assumed to be in native 8-bit
1987 If the pe1 and pe2 are non-NULL, the scanning pointers will be copied
1988 in there (they will point at the beginning of the I<next> character).
1989 If the pointers behind pe1 or pe2 are non-NULL, they are the end
1990 pointers beyond which scanning will not continue under any
1991 circumstances. If the byte lengths l1 and l2 are non-zero, s1+l1 and
1992 s2+l2 will be used as goal end pointers that will also stop the scan,
1993 and which qualify towards defining a successful match: all the scans
1994 that define an explicit length must reach their goal pointers for
1995 a match to succeed).
1997 For case-insensitiveness, the "casefolding" of Unicode is used
1998 instead of upper/lowercasing both the characters, see
1999 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
2003 Perl_ibcmp_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
2005 register const U8 *p1 = (const U8*)s1;
2006 register const U8 *p2 = (const U8*)s2;
2007 register const U8 *f1 = 0, *f2 = 0;
2008 register U8 *e1 = 0, *q1 = 0;
2009 register U8 *e2 = 0, *q2 = 0;
2010 STRLEN n1 = 0, n2 = 0;
2011 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
2012 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
2014 STRLEN foldlen1, foldlen2;
2019 if (e1 == 0 || (l1 && l1 < (UV)(e1 - (const U8*)s1)))
2020 f1 = (const U8*)s1 + l1;
2023 if (e2 == 0 || (l2 && l2 < (UV)(e2 - (const U8*)s2)))
2024 f2 = (const U8*)s2 + l2;
2026 if ((e1 == 0 && f1 == 0) || (e2 == 0 && f2 == 0) || (f1 == 0 && f2 == 0))
2027 return 1; /* mismatch; possible infinite loop or false positive */
2030 natbuf[1] = 0; /* Need to terminate the buffer. */
2032 while ((e1 == 0 || p1 < e1) &&
2033 (f1 == 0 || p1 < f1) &&
2034 (e2 == 0 || p2 < e2) &&
2035 (f2 == 0 || p2 < f2)) {
2038 to_utf8_fold(p1, foldbuf1, &foldlen1);
2041 to_utf8_fold(natbuf, foldbuf1, &foldlen1);
2048 to_utf8_fold(p2, foldbuf2, &foldlen2);
2051 to_utf8_fold(natbuf, foldbuf2, &foldlen2);
2057 if ( UTF8SKIP(q1) != UTF8SKIP(q2) ||
2058 (UTF8SKIP(q1) == 1 && *q1 != *q2) ||
2059 memNE((char*)q1, (char*)q2, UTF8SKIP(q1)) )
2060 return 1; /* mismatch */
2067 p1 += u1 ? UTF8SKIP(p1) : 1;
2069 p2 += u2 ? UTF8SKIP(p2) : 1;
2073 /* A match is defined by all the scans that specified
2074 * an explicit length reaching their final goals. */
2075 match = (f1 == 0 || p1 == f1) && (f2 == 0 || p2 == f2);
2084 return match ? 0 : 1; /* 0 match, 1 mismatch */
2089 * c-indentation-style: bsd
2091 * indent-tabs-mode: t
2094 * ex: set ts=8 sts=4 sw=4 noet: