3 * Copyright (C) 2000, 2001, 2002, 2003, 2004, by Larry Wall and others
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * 'What a fix!' said Sam. 'That's the one place in all the lands we've ever
12 * heard of that we don't want to see any closer; and that's the one place
13 * we're trying to get to! And that's just where we can't get, nohow.'
15 * 'Well do I understand your speech,' he answered in the same language;
16 * 'yet few strangers do so. Why then do you not speak in the Common Tongue,
17 * as is the custom in the West, if you wish to be answered?'
19 * ...the travellers perceived that the floor was paved with stones of many
20 * hues; branching runes and strange devices intertwined beneath their feet.
24 #define PERL_IN_UTF8_C
27 static char unees[] = "Malformed UTF-8 character (unexpected end of string)";
30 =head1 Unicode Support
32 This file contains various utility functions for manipulating UTF8-encoded
33 strings. For the uninitiated, this is a method of representing arbitrary
34 Unicode characters as a variable number of bytes, in such a way that
35 characters in the ASCII range are unmodified, and a zero byte never appears
36 within non-zero characters.
38 =for apidoc A|U8 *|uvuni_to_utf8_flags|U8 *d|UV uv|UV flags
40 Adds the UTF-8 representation of the Unicode codepoint C<uv> to the end
41 of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free
42 bytes available. The return value is the pointer to the byte after the
43 end of the new character. In other words,
45 d = uvuni_to_utf8_flags(d, uv, flags);
49 d = uvuni_to_utf8(d, uv);
51 (which is equivalent to)
53 d = uvuni_to_utf8_flags(d, uv, 0);
55 is the recommended Unicode-aware way of saying
63 Perl_uvuni_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
65 if (ckWARN(WARN_UTF8)) {
66 if (UNICODE_IS_SURROGATE(uv) &&
67 !(flags & UNICODE_ALLOW_SURROGATE))
68 Perl_warner(aTHX_ packWARN(WARN_UTF8), "UTF-16 surrogate 0x%04"UVxf, uv);
70 ((uv >= 0xFDD0 && uv <= 0xFDEF &&
71 !(flags & UNICODE_ALLOW_FDD0))
73 ((uv & 0xFFFE) == 0xFFFE && /* Either FFFE or FFFF. */
74 !(flags & UNICODE_ALLOW_FFFF))) &&
75 /* UNICODE_ALLOW_SUPER includes
76 * FFFEs and FFFFs beyond 0x10FFFF. */
77 ((uv <= PERL_UNICODE_MAX) ||
78 !(flags & UNICODE_ALLOW_SUPER))
80 Perl_warner(aTHX_ packWARN(WARN_UTF8),
81 "Unicode character 0x%04"UVxf" is illegal", uv);
83 if (UNI_IS_INVARIANT(uv)) {
84 *d++ = (U8)UTF_TO_NATIVE(uv);
89 STRLEN len = UNISKIP(uv);
92 *p-- = (U8)UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
93 uv >>= UTF_ACCUMULATION_SHIFT;
95 *p = (U8)UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
98 #else /* Non loop style */
100 *d++ = (U8)(( uv >> 6) | 0xc0);
101 *d++ = (U8)(( uv & 0x3f) | 0x80);
105 *d++ = (U8)(( uv >> 12) | 0xe0);
106 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
107 *d++ = (U8)(( uv & 0x3f) | 0x80);
111 *d++ = (U8)(( uv >> 18) | 0xf0);
112 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
113 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
114 *d++ = (U8)(( uv & 0x3f) | 0x80);
117 if (uv < 0x4000000) {
118 *d++ = (U8)(( uv >> 24) | 0xf8);
119 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
120 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
121 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
122 *d++ = (U8)(( uv & 0x3f) | 0x80);
125 if (uv < 0x80000000) {
126 *d++ = (U8)(( uv >> 30) | 0xfc);
127 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
128 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
129 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
130 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
131 *d++ = (U8)(( uv & 0x3f) | 0x80);
135 if (uv < UTF8_QUAD_MAX)
138 *d++ = 0xfe; /* Can't match U+FEFF! */
139 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
140 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
141 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
142 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
143 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
144 *d++ = (U8)(( uv & 0x3f) | 0x80);
149 *d++ = 0xff; /* Can't match U+FFFE! */
150 *d++ = 0x80; /* 6 Reserved bits */
151 *d++ = (U8)(((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
152 *d++ = (U8)(((uv >> 54) & 0x3f) | 0x80);
153 *d++ = (U8)(((uv >> 48) & 0x3f) | 0x80);
154 *d++ = (U8)(((uv >> 42) & 0x3f) | 0x80);
155 *d++ = (U8)(((uv >> 36) & 0x3f) | 0x80);
156 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
157 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
158 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
159 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
160 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
161 *d++ = (U8)(( uv & 0x3f) | 0x80);
165 #endif /* Loop style */
169 Perl_uvuni_to_utf8(pTHX_ U8 *d, UV uv)
171 return Perl_uvuni_to_utf8_flags(aTHX_ d, uv, 0);
176 =for apidoc A|STRLEN|is_utf8_char|U8 *s
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.
185 Perl_is_utf8_char(pTHX_ U8 *s)
191 if (UTF8_IS_INVARIANT(u))
194 if (!UTF8_IS_START(u))
199 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
204 u &= UTF_START_MASK(len);
208 if (!UTF8_IS_CONTINUATION(*s))
210 uv = UTF8_ACCUMULATE(uv, *s);
217 if ((STRLEN)UNISKIP(uv) < len)
224 =for apidoc A|bool|is_utf8_string|U8 *s|STRLEN len
226 Returns true if first C<len> bytes of the given string form a valid
227 UTF-8 string, false otherwise. Note that 'a valid UTF-8 string' does
228 not mean 'a string that contains code points above 0x7F encoded in UTF-8'
229 because a valid ASCII string is a valid UTF-8 string.
235 Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len)
242 len = strlen((char *)s);
246 /* Inline the easy bits of is_utf8_char() here for speed... */
247 if (UTF8_IS_INVARIANT(*x))
249 else if (!UTF8_IS_START(*x))
252 /* ... and call is_utf8_char() only if really needed. */
266 =for apidoc A|bool|is_utf8_string_loc|U8 *s|STRLEN len|U8 **p
268 Like is_ut8_string but store the location of the failure in
275 Perl_is_utf8_string_loc(pTHX_ U8 *s, STRLEN len, U8 **p)
282 len = strlen((char *)s);
286 /* Inline the easy bits of is_utf8_char() here for speed... */
287 if (UTF8_IS_INVARIANT(*x))
289 else if (!UTF8_IS_START(*x)) {
295 /* ... and call is_utf8_char() only if really needed. */
315 =for apidoc A|UV|utf8n_to_uvuni|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
317 Bottom level UTF-8 decode routine.
318 Returns the unicode code point value of the first character in the string C<s>
319 which is assumed to be in UTF-8 encoding and no longer than C<curlen>;
320 C<retlen> will be set to the length, in bytes, of that character.
322 If C<s> does not point to a well-formed UTF-8 character, the behaviour
323 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
324 it is assumed that the caller will raise a warning, and this function
325 will silently just set C<retlen> to C<-1> and return zero. If the
326 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
327 malformations will be given, C<retlen> will be set to the expected
328 length of the UTF-8 character in bytes, and zero will be returned.
330 The C<flags> can also contain various flags to allow deviations from
331 the strict UTF-8 encoding (see F<utf8.h>).
333 Most code should use utf8_to_uvchr() rather than call this directly.
339 Perl_utf8n_to_uvuni(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
344 bool dowarn = ckWARN_d(WARN_UTF8);
346 STRLEN expectlen = 0;
349 /* This list is a superset of the UTF8_ALLOW_XXX. */
351 #define UTF8_WARN_EMPTY 1
352 #define UTF8_WARN_CONTINUATION 2
353 #define UTF8_WARN_NON_CONTINUATION 3
354 #define UTF8_WARN_FE_FF 4
355 #define UTF8_WARN_SHORT 5
356 #define UTF8_WARN_OVERFLOW 6
357 #define UTF8_WARN_SURROGATE 7
358 #define UTF8_WARN_LONG 8
359 #define UTF8_WARN_FFFF 9 /* Also FFFE. */
362 !(flags & UTF8_ALLOW_EMPTY)) {
363 warning = UTF8_WARN_EMPTY;
367 if (UTF8_IS_INVARIANT(uv)) {
370 return (UV) (NATIVE_TO_UTF(*s));
373 if (UTF8_IS_CONTINUATION(uv) &&
374 !(flags & UTF8_ALLOW_CONTINUATION)) {
375 warning = UTF8_WARN_CONTINUATION;
379 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
380 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
381 warning = UTF8_WARN_NON_CONTINUATION;
386 uv = NATIVE_TO_UTF(uv);
388 if ((uv == 0xfe || uv == 0xff) &&
389 !(flags & UTF8_ALLOW_FE_FF)) {
390 warning = UTF8_WARN_FE_FF;
395 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
396 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
397 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
398 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
400 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
401 else { len = 7; uv &= 0x01; }
403 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
404 else if (!(uv & 0x01)) { len = 7; uv = 0; }
405 else { len = 13; uv = 0; } /* whoa! */
413 if ((curlen < expectlen) &&
414 !(flags & UTF8_ALLOW_SHORT)) {
415 warning = UTF8_WARN_SHORT;
424 if (!UTF8_IS_CONTINUATION(*s) &&
425 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
427 warning = UTF8_WARN_NON_CONTINUATION;
431 uv = UTF8_ACCUMULATE(uv, *s);
433 /* These cannot be allowed. */
435 if (expectlen != 13 && !(flags & UTF8_ALLOW_LONG)) {
436 warning = UTF8_WARN_LONG;
440 else { /* uv < ouv */
441 /* This cannot be allowed. */
442 warning = UTF8_WARN_OVERFLOW;
450 if (UNICODE_IS_SURROGATE(uv) &&
451 !(flags & UTF8_ALLOW_SURROGATE)) {
452 warning = UTF8_WARN_SURROGATE;
454 } else if ((expectlen > (STRLEN)UNISKIP(uv)) &&
455 !(flags & UTF8_ALLOW_LONG)) {
456 warning = UTF8_WARN_LONG;
458 } else if (UNICODE_IS_ILLEGAL(uv) &&
459 !(flags & UTF8_ALLOW_FFFF)) {
460 warning = UTF8_WARN_FFFF;
468 if (flags & UTF8_CHECK_ONLY) {
475 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
478 case 0: /* Intentionally empty. */ break;
479 case UTF8_WARN_EMPTY:
480 Perl_sv_catpv(aTHX_ sv, "(empty string)");
482 case UTF8_WARN_CONTINUATION:
483 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
485 case UTF8_WARN_NON_CONTINUATION:
487 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
488 (UV)s[1], startbyte);
490 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
491 (UV)s[1], s - s0, s - s0 > 1 ? "s" : "", startbyte, expectlen);
494 case UTF8_WARN_FE_FF:
495 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
497 case UTF8_WARN_SHORT:
498 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
499 curlen, curlen == 1 ? "" : "s", expectlen, startbyte);
500 expectlen = curlen; /* distance for caller to skip */
502 case UTF8_WARN_OVERFLOW:
503 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
506 case UTF8_WARN_SURROGATE:
507 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
510 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
511 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
514 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
517 Perl_sv_catpv(aTHX_ sv, "(unknown reason)");
525 Perl_warner(aTHX_ packWARN(WARN_UTF8),
526 "%s in %s", s, OP_DESC(PL_op));
528 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
533 *retlen = expectlen ? expectlen : len;
539 =for apidoc A|UV|utf8_to_uvchr|U8 *s|STRLEN *retlen
541 Returns the native character value of the first character in the string C<s>
542 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
543 length, in bytes, of that character.
545 If C<s> does not point to a well-formed UTF-8 character, zero is
546 returned and retlen is set, if possible, to -1.
552 Perl_utf8_to_uvchr(pTHX_ U8 *s, STRLEN *retlen)
554 return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXLEN, retlen,
555 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
559 =for apidoc A|UV|utf8_to_uvuni|U8 *s|STRLEN *retlen
561 Returns the Unicode code point of the first character in the string C<s>
562 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
563 length, in bytes, of that character.
565 This function should only be used when returned UV is considered
566 an index into the Unicode semantic tables (e.g. swashes).
568 If C<s> does not point to a well-formed UTF-8 character, zero is
569 returned and retlen is set, if possible, to -1.
575 Perl_utf8_to_uvuni(pTHX_ U8 *s, STRLEN *retlen)
577 /* Call the low level routine asking for checks */
578 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXLEN, retlen,
579 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
583 =for apidoc A|STRLEN|utf8_length|U8 *s|U8 *e
585 Return the length of the UTF-8 char encoded string C<s> in characters.
586 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
587 up past C<e>, croaks.
593 Perl_utf8_length(pTHX_ U8 *s, U8 *e)
597 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
598 * the bitops (especially ~) can create illegal UTF-8.
599 * In other words: in Perl UTF-8 is not just for Unicode. */
602 if (ckWARN_d(WARN_UTF8)) {
604 Perl_warner(aTHX_ packWARN(WARN_UTF8),
605 "%s in %s", unees, OP_DESC(PL_op));
607 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
615 if (ckWARN_d(WARN_UTF8)) {
617 Perl_warner(aTHX_ packWARN(WARN_UTF8),
618 unees, OP_DESC(PL_op));
620 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
632 =for apidoc A|IV|utf8_distance|U8 *a|U8 *b
634 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
637 WARNING: use only if you *know* that the pointers point inside the
644 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
648 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
649 * the bitops (especially ~) can create illegal UTF-8.
650 * In other words: in Perl UTF-8 is not just for Unicode. */
657 if (ckWARN_d(WARN_UTF8)) {
659 Perl_warner(aTHX_ packWARN(WARN_UTF8),
660 "%s in %s", unees, OP_DESC(PL_op));
662 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
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);
693 =for apidoc A|U8 *|utf8_hop|U8 *s|I32 off
695 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
698 WARNING: do not use the following unless you *know* C<off> is within
699 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
700 on the first byte of character or just after the last byte of a character.
706 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
708 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
709 * the bitops (especially ~) can create illegal UTF-8.
710 * In other words: in Perl UTF-8 is not just for Unicode. */
719 while (UTF8_IS_CONTINUATION(*s))
727 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
729 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
730 Unlike C<bytes_to_utf8>, this over-writes the original string, and
731 updates len to contain the new length.
732 Returns zero on failure, setting C<len> to -1.
738 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
744 /* ensure valid UTF-8 and chars < 256 before updating string */
745 for (send = s + *len; s < send; ) {
748 if (!UTF8_IS_INVARIANT(c) &&
749 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
750 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
759 *d++ = (U8)utf8_to_uvchr(s, &ulen);
768 =for apidoc A|U8 *|bytes_from_utf8|U8 *s|STRLEN *len|bool *is_utf8
770 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
771 Unlike <utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
772 the newly-created string, and updates C<len> to contain the new
773 length. Returns the original string if no conversion occurs, C<len>
774 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
775 0 if C<s> is converted or contains all 7bit characters.
781 Perl_bytes_from_utf8(pTHX_ U8 *s, STRLEN *len, bool *is_utf8)
791 /* ensure valid UTF-8 and chars < 256 before converting string */
792 for (send = s + *len; s < send;) {
794 if (!UTF8_IS_INVARIANT(c)) {
795 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
796 (c = *s++) && UTF8_IS_CONTINUATION(c))
805 Newz(801, d, (*len) - count + 1, U8);
806 s = start; start = d;
809 if (!UTF8_IS_INVARIANT(c)) {
810 /* Then it is two-byte encoded */
811 c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++);
812 c = ASCII_TO_NATIVE(c);
822 =for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
824 Converts a string C<s> of length C<len> from ASCII into UTF-8 encoding.
825 Returns a pointer to the newly-created string, and sets C<len> to
826 reflect the new length.
828 If you want to convert to UTF-8 from other encodings than ASCII,
829 see sv_recode_to_utf8().
835 Perl_bytes_to_utf8(pTHX_ U8 *s, STRLEN *len)
842 Newz(801, d, (*len) * 2 + 1, U8);
846 UV uv = NATIVE_TO_ASCII(*s++);
847 if (UNI_IS_INVARIANT(uv))
848 *d++ = (U8)UTF_TO_NATIVE(uv);
850 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
851 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
860 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
862 * Destination must be pre-extended to 3/2 source. Do not use in-place.
863 * We optimize for native, for obvious reasons. */
866 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
871 if (bytelen == 1 && p[0] == 0) { /* Be understanding. */
878 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %d", bytelen);
883 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
890 *d++ = (U8)(( uv >> 6) | 0xc0);
891 *d++ = (U8)(( uv & 0x3f) | 0x80);
894 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
895 UV low = (p[0] << 8) + p[1];
897 if (low < 0xdc00 || low >= 0xdfff)
898 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
899 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
902 *d++ = (U8)(( uv >> 12) | 0xe0);
903 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
904 *d++ = (U8)(( uv & 0x3f) | 0x80);
908 *d++ = (U8)(( uv >> 18) | 0xf0);
909 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
910 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
911 *d++ = (U8)(( uv & 0x3f) | 0x80);
915 *newlen = d - dstart;
919 /* Note: this one is slightly destructive of the source. */
922 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
925 U8* send = s + bytelen;
932 return utf16_to_utf8(p, d, bytelen, newlen);
935 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
938 Perl_is_uni_alnum(pTHX_ UV c)
940 U8 tmpbuf[UTF8_MAXLEN+1];
941 uvchr_to_utf8(tmpbuf, c);
942 return is_utf8_alnum(tmpbuf);
946 Perl_is_uni_alnumc(pTHX_ UV c)
948 U8 tmpbuf[UTF8_MAXLEN+1];
949 uvchr_to_utf8(tmpbuf, c);
950 return is_utf8_alnumc(tmpbuf);
954 Perl_is_uni_idfirst(pTHX_ UV c)
956 U8 tmpbuf[UTF8_MAXLEN+1];
957 uvchr_to_utf8(tmpbuf, c);
958 return is_utf8_idfirst(tmpbuf);
962 Perl_is_uni_alpha(pTHX_ UV c)
964 U8 tmpbuf[UTF8_MAXLEN+1];
965 uvchr_to_utf8(tmpbuf, c);
966 return is_utf8_alpha(tmpbuf);
970 Perl_is_uni_ascii(pTHX_ UV c)
972 U8 tmpbuf[UTF8_MAXLEN+1];
973 uvchr_to_utf8(tmpbuf, c);
974 return is_utf8_ascii(tmpbuf);
978 Perl_is_uni_space(pTHX_ UV c)
980 U8 tmpbuf[UTF8_MAXLEN+1];
981 uvchr_to_utf8(tmpbuf, c);
982 return is_utf8_space(tmpbuf);
986 Perl_is_uni_digit(pTHX_ UV c)
988 U8 tmpbuf[UTF8_MAXLEN+1];
989 uvchr_to_utf8(tmpbuf, c);
990 return is_utf8_digit(tmpbuf);
994 Perl_is_uni_upper(pTHX_ UV c)
996 U8 tmpbuf[UTF8_MAXLEN+1];
997 uvchr_to_utf8(tmpbuf, c);
998 return is_utf8_upper(tmpbuf);
1002 Perl_is_uni_lower(pTHX_ UV c)
1004 U8 tmpbuf[UTF8_MAXLEN+1];
1005 uvchr_to_utf8(tmpbuf, c);
1006 return is_utf8_lower(tmpbuf);
1010 Perl_is_uni_cntrl(pTHX_ UV c)
1012 U8 tmpbuf[UTF8_MAXLEN+1];
1013 uvchr_to_utf8(tmpbuf, c);
1014 return is_utf8_cntrl(tmpbuf);
1018 Perl_is_uni_graph(pTHX_ UV c)
1020 U8 tmpbuf[UTF8_MAXLEN+1];
1021 uvchr_to_utf8(tmpbuf, c);
1022 return is_utf8_graph(tmpbuf);
1026 Perl_is_uni_print(pTHX_ UV c)
1028 U8 tmpbuf[UTF8_MAXLEN+1];
1029 uvchr_to_utf8(tmpbuf, c);
1030 return is_utf8_print(tmpbuf);
1034 Perl_is_uni_punct(pTHX_ UV c)
1036 U8 tmpbuf[UTF8_MAXLEN+1];
1037 uvchr_to_utf8(tmpbuf, c);
1038 return is_utf8_punct(tmpbuf);
1042 Perl_is_uni_xdigit(pTHX_ UV c)
1044 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
1045 uvchr_to_utf8(tmpbuf, c);
1046 return is_utf8_xdigit(tmpbuf);
1050 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
1052 uvchr_to_utf8(p, c);
1053 return to_utf8_upper(p, p, lenp);
1057 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
1059 uvchr_to_utf8(p, c);
1060 return to_utf8_title(p, p, lenp);
1064 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
1066 uvchr_to_utf8(p, c);
1067 return to_utf8_lower(p, p, lenp);
1071 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1073 uvchr_to_utf8(p, c);
1074 return to_utf8_fold(p, p, lenp);
1077 /* for now these all assume no locale info available for Unicode > 255 */
1080 Perl_is_uni_alnum_lc(pTHX_ UV c)
1082 return is_uni_alnum(c); /* XXX no locale support yet */
1086 Perl_is_uni_alnumc_lc(pTHX_ UV c)
1088 return is_uni_alnumc(c); /* XXX no locale support yet */
1092 Perl_is_uni_idfirst_lc(pTHX_ UV c)
1094 return is_uni_idfirst(c); /* XXX no locale support yet */
1098 Perl_is_uni_alpha_lc(pTHX_ UV c)
1100 return is_uni_alpha(c); /* XXX no locale support yet */
1104 Perl_is_uni_ascii_lc(pTHX_ UV c)
1106 return is_uni_ascii(c); /* XXX no locale support yet */
1110 Perl_is_uni_space_lc(pTHX_ UV c)
1112 return is_uni_space(c); /* XXX no locale support yet */
1116 Perl_is_uni_digit_lc(pTHX_ UV c)
1118 return is_uni_digit(c); /* XXX no locale support yet */
1122 Perl_is_uni_upper_lc(pTHX_ UV c)
1124 return is_uni_upper(c); /* XXX no locale support yet */
1128 Perl_is_uni_lower_lc(pTHX_ UV c)
1130 return is_uni_lower(c); /* XXX no locale support yet */
1134 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1136 return is_uni_cntrl(c); /* XXX no locale support yet */
1140 Perl_is_uni_graph_lc(pTHX_ UV c)
1142 return is_uni_graph(c); /* XXX no locale support yet */
1146 Perl_is_uni_print_lc(pTHX_ UV c)
1148 return is_uni_print(c); /* XXX no locale support yet */
1152 Perl_is_uni_punct_lc(pTHX_ UV c)
1154 return is_uni_punct(c); /* XXX no locale support yet */
1158 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1160 return is_uni_xdigit(c); /* XXX no locale support yet */
1164 Perl_to_uni_upper_lc(pTHX_ U32 c)
1166 /* XXX returns only the first character -- do not use XXX */
1167 /* XXX no locale support yet */
1169 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
1170 return (U32)to_uni_upper(c, tmpbuf, &len);
1174 Perl_to_uni_title_lc(pTHX_ U32 c)
1176 /* XXX returns only the first character XXX -- do not use XXX */
1177 /* XXX no locale support yet */
1179 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
1180 return (U32)to_uni_title(c, tmpbuf, &len);
1184 Perl_to_uni_lower_lc(pTHX_ U32 c)
1186 /* XXX returns only the first character -- do not use XXX */
1187 /* XXX no locale support yet */
1189 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
1190 return (U32)to_uni_lower(c, tmpbuf, &len);
1194 Perl_is_utf8_alnum(pTHX_ U8 *p)
1196 if (!is_utf8_char(p))
1199 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1200 * descendant of isalnum(3), in other words, it doesn't
1201 * contain the '_'. --jhi */
1202 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1203 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1204 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1205 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1207 PL_utf8_alnum = swash_init("utf8", "",
1208 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1209 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1214 Perl_is_utf8_alnumc(pTHX_ U8 *p)
1216 if (!is_utf8_char(p))
1219 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1220 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1221 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
1222 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1224 PL_utf8_alnum = swash_init("utf8", "",
1225 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1226 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1231 Perl_is_utf8_idfirst(pTHX_ U8 *p) /* The naming is historical. */
1235 if (!is_utf8_char(p))
1237 if (!PL_utf8_idstart) /* is_utf8_idstart would be more logical. */
1238 PL_utf8_idstart = swash_init("utf8", "IdStart", &PL_sv_undef, 0, 0);
1239 return swash_fetch(PL_utf8_idstart, p, TRUE) != 0;
1243 Perl_is_utf8_idcont(pTHX_ U8 *p)
1247 if (!is_utf8_char(p))
1249 if (!PL_utf8_idcont)
1250 PL_utf8_idcont = swash_init("utf8", "IdContinue", &PL_sv_undef, 0, 0);
1251 return swash_fetch(PL_utf8_idcont, p, TRUE) != 0;
1255 Perl_is_utf8_alpha(pTHX_ U8 *p)
1257 if (!is_utf8_char(p))
1260 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1261 return swash_fetch(PL_utf8_alpha, p, TRUE) != 0;
1265 Perl_is_utf8_ascii(pTHX_ U8 *p)
1267 if (!is_utf8_char(p))
1270 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1271 return swash_fetch(PL_utf8_ascii, p, TRUE) != 0;
1275 Perl_is_utf8_space(pTHX_ U8 *p)
1277 if (!is_utf8_char(p))
1280 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1281 return swash_fetch(PL_utf8_space, p, TRUE) != 0;
1285 Perl_is_utf8_digit(pTHX_ U8 *p)
1287 if (!is_utf8_char(p))
1290 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1291 return swash_fetch(PL_utf8_digit, p, TRUE) != 0;
1295 Perl_is_utf8_upper(pTHX_ U8 *p)
1297 if (!is_utf8_char(p))
1300 PL_utf8_upper = swash_init("utf8", "IsUppercase", &PL_sv_undef, 0, 0);
1301 return swash_fetch(PL_utf8_upper, p, TRUE) != 0;
1305 Perl_is_utf8_lower(pTHX_ U8 *p)
1307 if (!is_utf8_char(p))
1310 PL_utf8_lower = swash_init("utf8", "IsLowercase", &PL_sv_undef, 0, 0);
1311 return swash_fetch(PL_utf8_lower, p, TRUE) != 0;
1315 Perl_is_utf8_cntrl(pTHX_ U8 *p)
1317 if (!is_utf8_char(p))
1320 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1321 return swash_fetch(PL_utf8_cntrl, p, TRUE) != 0;
1325 Perl_is_utf8_graph(pTHX_ U8 *p)
1327 if (!is_utf8_char(p))
1330 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1331 return swash_fetch(PL_utf8_graph, p, TRUE) != 0;
1335 Perl_is_utf8_print(pTHX_ U8 *p)
1337 if (!is_utf8_char(p))
1340 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1341 return swash_fetch(PL_utf8_print, p, TRUE) != 0;
1345 Perl_is_utf8_punct(pTHX_ U8 *p)
1347 if (!is_utf8_char(p))
1350 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1351 return swash_fetch(PL_utf8_punct, p, TRUE) != 0;
1355 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1357 if (!is_utf8_char(p))
1359 if (!PL_utf8_xdigit)
1360 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1361 return swash_fetch(PL_utf8_xdigit, p, TRUE) != 0;
1365 Perl_is_utf8_mark(pTHX_ U8 *p)
1367 if (!is_utf8_char(p))
1370 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1371 return swash_fetch(PL_utf8_mark, p, TRUE) != 0;
1375 =for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special
1377 The "p" contains the pointer to the UTF-8 string encoding
1378 the character that is being converted.
1380 The "ustrp" is a pointer to the character buffer to put the
1381 conversion result to. The "lenp" is a pointer to the length
1384 The "swashp" is a pointer to the swash to use.
1386 Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
1387 and loaded by SWASHGET, using lib/utf8_heavy.pl. The special (usually,
1388 but not always, a multicharacter mapping), is tried first.
1390 The "special" is a string like "utf8::ToSpecLower", which means the
1391 hash %utf8::ToSpecLower. The access to the hash is through
1392 Perl_to_utf8_case().
1394 The "normal" is a string like "ToLower" which means the swash
1400 Perl_to_utf8_case(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp, char *normal, char *special)
1403 U8 tmpbuf[UTF8_MAXLEN_FOLD+1];
1406 uv0 = utf8_to_uvchr(p, 0);
1407 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1408 * are necessary in EBCDIC, they are redundant no-ops
1409 * in ASCII-ish platforms, and hopefully optimized away. */
1410 uv1 = NATIVE_TO_UNI(uv0);
1411 uvuni_to_utf8(tmpbuf, uv1);
1413 if (!*swashp) /* load on-demand */
1414 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1416 /* The 0xDF is the only special casing Unicode code point below 0x100. */
1417 if (special && (uv1 == 0xDF || uv1 > 0xFF)) {
1418 /* It might be "special" (sometimes, but not always,
1419 * a multicharacter mapping) */
1423 if ((hv = get_hv(special, FALSE)) &&
1424 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1428 s = SvPV(*svp, len);
1430 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
1433 /* If we have EBCDIC we need to remap the characters
1434 * since any characters in the low 256 are Unicode
1435 * code points, not EBCDIC. */
1436 U8 *t = (U8*)s, *tend = t + len, *d;
1443 UV c = utf8_to_uvchr(t, &tlen);
1445 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1454 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1459 Copy(tmpbuf, ustrp, len, U8);
1461 Copy(s, ustrp, len, U8);
1467 if (!len && *swashp) {
1468 UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1471 /* It was "normal" (a single character mapping). */
1472 UV uv3 = UNI_TO_NATIVE(uv2);
1474 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
1478 if (!len) /* Neither: just copy. */
1479 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1484 return len ? utf8_to_uvchr(ustrp, 0) : 0;
1488 =for apidoc A|UV|to_utf8_upper|U8 *p|U8 *ustrp|STRLEN *lenp
1490 Convert the UTF-8 encoded character at p to its uppercase version and
1491 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1492 that the ustrp needs to be at least UTF8_MAXLEN_UCLC+1 bytes since the
1493 uppercase version may be longer than the original character (up to two
1496 The first character of the uppercased version is returned
1497 (but note, as explained above, that there may be more.)
1502 Perl_to_utf8_upper(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1504 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1505 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1509 =for apidoc A|UV|to_utf8_title|U8 *p|U8 *ustrp|STRLEN *lenp
1511 Convert the UTF-8 encoded character at p to its titlecase version and
1512 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1513 that the ustrp needs to be at least UTF8_MAXLEN_UCLC+1 bytes since the
1514 titlecase version may be longer than the original character (up to two
1517 The first character of the titlecased version is returned
1518 (but note, as explained above, that there may be more.)
1523 Perl_to_utf8_title(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1525 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1526 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1530 =for apidoc A|UV|to_utf8_lower|U8 *p|U8 *ustrp|STRLEN *lenp
1532 Convert the UTF-8 encoded character at p to its lowercase version and
1533 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1534 that the ustrp needs to be at least UTF8_MAXLEN_UCLC+1 bytes since the
1535 lowercase version may be longer than the original character (up to two
1538 The first character of the lowercased version is returned
1539 (but note, as explained above, that there may be more.)
1544 Perl_to_utf8_lower(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1546 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1547 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1551 =for apidoc A|UV|to_utf8_fold|U8 *p|U8 *ustrp|STRLEN *lenp
1553 Convert the UTF-8 encoded character at p to its foldcase version and
1554 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1555 that the ustrp needs to be at least UTF8_MAXLEN_FOLD+1 bytes since the
1556 foldcase version may be longer than the original character (up to
1559 The first character of the foldcased version is returned
1560 (but note, as explained above, that there may be more.)
1565 Perl_to_utf8_fold(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1567 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1568 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1571 /* a "swash" is a swatch hash */
1574 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1577 SV* tokenbufsv = sv_newmortal();
1579 size_t pkg_len = strlen(pkg);
1580 size_t name_len = strlen(name);
1581 HV *stash = gv_stashpvn(pkg, pkg_len, FALSE);
1584 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
1586 errsv_save = newSVsv(ERRSV);
1587 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
1590 sv_setsv(ERRSV, errsv_save);
1591 SvREFCNT_dec(errsv_save);
1595 PUSHSTACKi(PERLSI_MAGIC);
1598 PUSHs(sv_2mortal(newSVpvn(pkg, pkg_len)));
1599 PUSHs(sv_2mortal(newSVpvn(name, name_len)));
1601 PUSHs(sv_2mortal(newSViv(minbits)));
1602 PUSHs(sv_2mortal(newSViv(none)));
1608 if (IN_PERL_COMPILETIME) {
1609 /* XXX ought to be handled by lex_start */
1612 sv_setpv(tokenbufsv, PL_tokenbuf);
1614 errsv_save = newSVsv(ERRSV);
1615 if (call_method("SWASHNEW", G_SCALAR))
1616 retval = newSVsv(*PL_stack_sp--);
1618 retval = &PL_sv_undef;
1620 sv_setsv(ERRSV, errsv_save);
1621 SvREFCNT_dec(errsv_save);
1624 if (IN_PERL_COMPILETIME) {
1626 char* pv = SvPV(tokenbufsv, len);
1628 Copy(pv, PL_tokenbuf, len+1, char);
1629 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1631 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
1633 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
1635 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1641 /* This API is wrong for special case conversions since we may need to
1642 * return several Unicode characters for a single Unicode character
1643 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1644 * the lower-level routine, and it is similarly broken for returning
1645 * multiple values. --jhi */
1647 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr, bool do_utf8)
1649 HV* hv = (HV*)SvRV(sv);
1658 UV c = NATIVE_TO_ASCII(*ptr);
1660 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1661 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
1662 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
1665 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
1666 * then the "swatch" is a vec() for al the chars which start
1668 * So the key in the hash (klen) is length of encoded char -1
1670 klen = UTF8SKIP(ptr) - 1;
1675 /* If char in invariant then swatch is for all the invariant chars
1676 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
1678 needents = UTF_CONTINUATION_MARK;
1679 off = NATIVE_TO_UTF(ptr[klen]);
1683 /* If char is encoded then swatch is for the prefix */
1684 needents = (1 << UTF_ACCUMULATION_SHIFT);
1685 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
1689 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1690 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1691 * it's nothing to sniff at.) Pity we usually come through at least
1692 * two function calls to get here...
1694 * NB: this code assumes that swatches are never modified, once generated!
1697 if (hv == PL_last_swash_hv &&
1698 klen == PL_last_swash_klen &&
1699 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
1701 tmps = PL_last_swash_tmps;
1702 slen = PL_last_swash_slen;
1705 /* Try our second-level swatch cache, kept in a hash. */
1706 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1708 /* If not cached, generate it via utf8::SWASHGET */
1709 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1711 /* We use utf8n_to_uvuni() as we want an index into
1712 Unicode tables, not a native character number.
1714 UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXLEN, 0,
1716 0 : UTF8_ALLOW_ANY);
1721 PUSHSTACKi(PERLSI_MAGIC);
1725 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
1726 PUSHs(sv_2mortal(newSViv((klen) ?
1727 (code_point & ~(needents - 1)) : 0)));
1728 PUSHs(sv_2mortal(newSViv(needents)));
1730 errsv_save = newSVsv(ERRSV);
1731 if (call_method("SWASHGET", G_SCALAR))
1732 retval = newSVsv(*PL_stack_sp--);
1734 retval = &PL_sv_undef;
1736 sv_setsv(ERRSV, errsv_save);
1737 SvREFCNT_dec(errsv_save);
1741 if (IN_PERL_COMPILETIME)
1742 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1744 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1746 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || (slen << 3) < needents)
1747 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1750 PL_last_swash_hv = hv;
1751 PL_last_swash_klen = klen;
1752 PL_last_swash_tmps = tmps;
1753 PL_last_swash_slen = slen;
1755 Copy(ptr, PL_last_swash_key, klen, U8);
1758 switch ((int)((slen << 3) / needents)) {
1760 bit = 1 << (off & 7);
1762 return (tmps[off] & bit) != 0;
1767 return (tmps[off] << 8) + tmps[off + 1] ;
1770 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1772 Perl_croak(aTHX_ "panic: swash_fetch");
1778 =for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv
1780 Adds the UTF-8 representation of the Native codepoint C<uv> to the end
1781 of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free
1782 bytes available. The return value is the pointer to the byte after the
1783 end of the new character. In other words,
1785 d = uvchr_to_utf8(d, uv);
1787 is the recommended wide native character-aware way of saying
1794 /* On ASCII machines this is normally a macro but we want a
1795 real function in case XS code wants it
1797 #undef Perl_uvchr_to_utf8
1799 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
1801 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
1805 Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
1807 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
1811 =for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
1813 Returns the native character value of the first character in the string C<s>
1814 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
1815 length, in bytes, of that character.
1817 Allows length and flags to be passed to low level routine.
1821 /* On ASCII machines this is normally a macro but we want
1822 a real function in case XS code wants it
1824 #undef Perl_utf8n_to_uvchr
1826 Perl_utf8n_to_uvchr(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
1828 UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
1829 return UNI_TO_NATIVE(uv);
1833 =for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags
1835 Build to the scalar dsv a displayable version of the string spv,
1836 length len, the displayable version being at most pvlim bytes long
1837 (if longer, the rest is truncated and "..." will be appended).
1839 The flags argument can have UNI_DISPLAY_ISPRINT set to display
1840 isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
1841 to display the \\[nrfta\\] as the backslashed versions (like '\n')
1842 (UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
1843 UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
1844 UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
1846 The pointer to the PV of the dsv is returned.
1850 Perl_pv_uni_display(pTHX_ SV *dsv, U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
1855 sv_setpvn(dsv, "", 0);
1856 for (s = (char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
1858 /* This serves double duty as a flag and a character to print after
1859 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
1863 if (pvlim && SvCUR(dsv) >= pvlim) {
1867 u = utf8_to_uvchr((U8*)s, 0);
1869 unsigned char c = u & 0xFF;
1870 if (!ok && (flags & UNI_DISPLAY_BACKSLASH)) {
1887 Perl_sv_catpvf(aTHX_ dsv, "\\%c", ok);
1890 /* isPRINT() is the locale-blind version. */
1891 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
1892 Perl_sv_catpvf(aTHX_ dsv, "%c", c);
1897 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
1900 sv_catpvn(dsv, "...", 3);
1906 =for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags
1908 Build to the scalar dsv a displayable version of the scalar sv,
1909 the displayable version being at most pvlim bytes long
1910 (if longer, the rest is truncated and "..." will be appended).
1912 The flags argument is as in pv_uni_display().
1914 The pointer to the PV of the dsv is returned.
1918 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
1920 return Perl_pv_uni_display(aTHX_ dsv, (U8*)SvPVX(ssv), SvCUR(ssv),
1925 =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
1927 Return true if the strings s1 and s2 differ case-insensitively, false
1928 if not (if they are equal case-insensitively). If u1 is true, the
1929 string s1 is assumed to be in UTF-8-encoded Unicode. If u2 is true,
1930 the string s2 is assumed to be in UTF-8-encoded Unicode. If u1 or u2
1931 are false, the respective string is assumed to be in native 8-bit
1934 If the pe1 and pe2 are non-NULL, the scanning pointers will be copied
1935 in there (they will point at the beginning of the I<next> character).
1936 If the pointers behind pe1 or pe2 are non-NULL, they are the end
1937 pointers beyond which scanning will not continue under any
1938 circumstances. If the byte lengths l1 and l2 are non-zero, s1+l1 and
1939 s2+l2 will be used as goal end pointers that will also stop the scan,
1940 and which qualify towards defining a successful match: all the scans
1941 that define an explicit length must reach their goal pointers for
1942 a match to succeed).
1944 For case-insensitiveness, the "casefolding" of Unicode is used
1945 instead of upper/lowercasing both the characters, see
1946 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
1950 Perl_ibcmp_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
1952 register U8 *p1 = (U8*)s1;
1953 register U8 *p2 = (U8*)s2;
1954 register U8 *e1 = 0, *f1 = 0, *q1 = 0;
1955 register U8 *e2 = 0, *f2 = 0, *q2 = 0;
1956 STRLEN n1 = 0, n2 = 0;
1957 U8 foldbuf1[UTF8_MAXLEN_FOLD+1];
1958 U8 foldbuf2[UTF8_MAXLEN_FOLD+1];
1960 STRLEN foldlen1, foldlen2;
1965 if (e1 == 0 || (l1 && l1 < (UV)(e1 - (U8*)s1)))
1969 if (e2 == 0 || (l2 && l2 < (UV)(e2 - (U8*)s2)))
1972 if ((e1 == 0 && f1 == 0) || (e2 == 0 && f2 == 0) || (f1 == 0 && f2 == 0))
1973 return 1; /* mismatch; possible infinite loop or false positive */
1976 natbuf[1] = 0; /* Need to terminate the buffer. */
1978 while ((e1 == 0 || p1 < e1) &&
1979 (f1 == 0 || p1 < f1) &&
1980 (e2 == 0 || p2 < e2) &&
1981 (f2 == 0 || p2 < f2)) {
1984 to_utf8_fold(p1, foldbuf1, &foldlen1);
1987 to_utf8_fold(natbuf, foldbuf1, &foldlen1);
1994 to_utf8_fold(p2, foldbuf2, &foldlen2);
1997 to_utf8_fold(natbuf, foldbuf2, &foldlen2);
2003 if ( UTF8SKIP(q1) != UTF8SKIP(q2) ||
2004 (UTF8SKIP(q1) == 1 && *q1 != *q2) ||
2005 memNE((char*)q1, (char*)q2, UTF8SKIP(q1)) )
2006 return 1; /* mismatch */
2013 p1 += u1 ? UTF8SKIP(p1) : 1;
2015 p2 += u2 ? UTF8SKIP(p2) : 1;
2019 /* A match is defined by all the scans that specified
2020 * an explicit length reaching their final goals. */
2021 match = (f1 == 0 || p1 == f1) && (f2 == 0 || p2 == f2);
2030 return match ? 0 : 1; /* 0 match, 1 mismatch */
2035 * c-indentation-style: bsd
2037 * indent-tabs-mode: t
2040 * vim: shiftwidth=4: