3 * Copyright (C) 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
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 * [p.603 of _The Lord of the Rings_, IV/I: "The Taming of Sméagol"]
18 * 'Well do I understand your speech,' he answered in the same language;
19 * 'yet few strangers do so. Why then do you not speak in the Common Tongue,
20 * as is the custom in the West, if you wish to be answered?'
21 * --Gandalf, addressing Théoden's door wardens
23 * [p.508 of _The Lord of the Rings_, III/vi: "The King of the Golden Hall"]
25 * ...the travellers perceived that the floor was paved with stones of many
26 * hues; branching runes and strange devices intertwined beneath their feet.
28 * [p.512 of _The Lord of the Rings_, III/vi: "The King of the Golden Hall"]
32 #define PERL_IN_UTF8_C
36 /* Separate prototypes needed because in ASCII systems these are
37 * usually macros but they still are compiled as code, too. */
38 PERL_CALLCONV UV Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags);
39 PERL_CALLCONV U8* Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv);
42 static const char unees[] =
43 "Malformed UTF-8 character (unexpected end of string)";
46 =head1 Unicode Support
48 This file contains various utility functions for manipulating UTF8-encoded
49 strings. For the uninitiated, this is a method of representing arbitrary
50 Unicode characters as a variable number of bytes, in such a way that
51 characters in the ASCII range are unmodified, and a zero byte never appears
52 within non-zero characters.
58 =for apidoc is_ascii_string
60 Returns true if the first C<len> bytes of the given string are the same whether
61 or not the string is encoded in UTF-8 (or UTF-EBCDIC on EBCDIC machines). That
62 is, if they are invariant. On ASCII-ish machines, only ASCII characters
63 fit this definition, hence the function's name.
65 If C<len> is 0, it will be calculated using C<strlen(s)>.
67 See also is_utf8_string(), is_utf8_string_loclen(), and is_utf8_string_loc().
73 Perl_is_ascii_string(const U8 *s, STRLEN len)
75 const U8* const send = s + (len ? len : strlen((const char *)s));
78 PERL_ARGS_ASSERT_IS_ASCII_STRING;
80 for (; x < send; ++x) {
81 if (!UTF8_IS_INVARIANT(*x))
89 =for apidoc uvuni_to_utf8_flags
91 Adds the UTF-8 representation of the Unicode codepoint C<uv> to the end
92 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
93 bytes available. The return value is the pointer to the byte after the
94 end of the new character. In other words,
96 d = uvuni_to_utf8_flags(d, uv, flags);
100 d = uvuni_to_utf8(d, uv);
102 (which is equivalent to)
104 d = uvuni_to_utf8_flags(d, uv, 0);
106 is the recommended Unicode-aware way of saying
114 Perl_uvuni_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
116 PERL_ARGS_ASSERT_UVUNI_TO_UTF8_FLAGS;
118 if (ckWARN(WARN_UTF8)) {
119 if (UNICODE_IS_SURROGATE(uv) &&
120 !(flags & UNICODE_ALLOW_SURROGATE))
121 Perl_warner(aTHX_ packWARN(WARN_UTF8), "UTF-16 surrogate 0x%04"UVxf, uv);
123 ((uv >= 0xFDD0 && uv <= 0xFDEF &&
124 !(flags & UNICODE_ALLOW_FDD0))
126 ((uv & 0xFFFE) == 0xFFFE && /* Either FFFE or FFFF. */
127 !(flags & UNICODE_ALLOW_FFFF))) &&
128 /* UNICODE_ALLOW_SUPER includes
129 * FFFEs and FFFFs beyond 0x10FFFF. */
130 ((uv <= PERL_UNICODE_MAX) ||
131 !(flags & UNICODE_ALLOW_SUPER))
133 Perl_warner(aTHX_ packWARN(WARN_UTF8),
134 "Unicode non-character 0x%04"UVxf" is illegal for interchange", uv);
136 if (UNI_IS_INVARIANT(uv)) {
137 *d++ = (U8)UTF_TO_NATIVE(uv);
142 STRLEN len = UNISKIP(uv);
145 *p-- = (U8)UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
146 uv >>= UTF_ACCUMULATION_SHIFT;
148 *p = (U8)UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
151 #else /* Non loop style */
153 *d++ = (U8)(( uv >> 6) | 0xc0);
154 *d++ = (U8)(( uv & 0x3f) | 0x80);
158 *d++ = (U8)(( uv >> 12) | 0xe0);
159 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
160 *d++ = (U8)(( uv & 0x3f) | 0x80);
164 *d++ = (U8)(( uv >> 18) | 0xf0);
165 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
166 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
167 *d++ = (U8)(( uv & 0x3f) | 0x80);
170 if (uv < 0x4000000) {
171 *d++ = (U8)(( uv >> 24) | 0xf8);
172 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
173 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
174 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
175 *d++ = (U8)(( uv & 0x3f) | 0x80);
178 if (uv < 0x80000000) {
179 *d++ = (U8)(( uv >> 30) | 0xfc);
180 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
181 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
182 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
183 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
184 *d++ = (U8)(( uv & 0x3f) | 0x80);
188 if (uv < UTF8_QUAD_MAX)
191 *d++ = 0xfe; /* Can't match U+FEFF! */
192 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
193 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
194 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
195 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
196 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
197 *d++ = (U8)(( uv & 0x3f) | 0x80);
202 *d++ = 0xff; /* Can't match U+FFFE! */
203 *d++ = 0x80; /* 6 Reserved bits */
204 *d++ = (U8)(((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
205 *d++ = (U8)(((uv >> 54) & 0x3f) | 0x80);
206 *d++ = (U8)(((uv >> 48) & 0x3f) | 0x80);
207 *d++ = (U8)(((uv >> 42) & 0x3f) | 0x80);
208 *d++ = (U8)(((uv >> 36) & 0x3f) | 0x80);
209 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
210 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
211 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
212 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
213 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
214 *d++ = (U8)(( uv & 0x3f) | 0x80);
218 #endif /* Loop style */
223 Tests if some arbitrary number of bytes begins in a valid UTF-8
224 character. Note that an INVARIANT (i.e. ASCII) character is a valid
225 UTF-8 character. The actual number of bytes in the UTF-8 character
226 will be returned if it is valid, otherwise 0.
228 This is the "slow" version as opposed to the "fast" version which is
229 the "unrolled" IS_UTF8_CHAR(). E.g. for t/uni/class.t the speed
230 difference is a factor of 2 to 3. For lengths (UTF8SKIP(s)) of four
231 or less you should use the IS_UTF8_CHAR(), for lengths of five or more
232 you should use the _slow(). In practice this means that the _slow()
233 will be used very rarely, since the maximum Unicode code point (as of
234 Unicode 4.1) is U+10FFFF, which encodes in UTF-8 to four bytes. Only
235 the "Perl extended UTF-8" (the infamous 'v-strings') will encode into
240 S_is_utf8_char_slow(const U8 *s, const STRLEN len)
246 PERL_ARGS_ASSERT_IS_UTF8_CHAR_SLOW;
248 if (UTF8_IS_INVARIANT(u))
251 if (!UTF8_IS_START(u))
254 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
260 u = NATIVE_TO_UTF(u);
262 u &= UTF_START_MASK(len);
266 if (!UTF8_IS_CONTINUATION(*s))
268 uv = UTF8_ACCUMULATE(uv, *s);
275 if ((STRLEN)UNISKIP(uv) < len)
282 =for apidoc is_utf8_char
284 Tests if some arbitrary number of bytes begins in a valid UTF-8
285 character. Note that an INVARIANT (i.e. ASCII on non-EBCDIC machines)
286 character is a valid UTF-8 character. The actual number of bytes in the UTF-8
287 character will be returned if it is valid, otherwise 0.
291 Perl_is_utf8_char(const U8 *s)
293 const STRLEN len = UTF8SKIP(s);
295 PERL_ARGS_ASSERT_IS_UTF8_CHAR;
297 if (IS_UTF8_CHAR_FAST(len))
298 return IS_UTF8_CHAR(s, len) ? len : 0;
299 #endif /* #ifdef IS_UTF8_CHAR */
300 return is_utf8_char_slow(s, len);
305 =for apidoc is_utf8_string
307 Returns true if first C<len> bytes of the given string form a valid
308 UTF-8 string, false otherwise. If C<len> is 0, it will be calculated
309 using C<strlen(s)>. Note that 'a valid UTF-8 string' does not mean 'a
310 string that contains code points above 0x7F encoded in UTF-8' because a
311 valid ASCII string is a valid UTF-8 string.
313 See also is_ascii_string(), is_utf8_string_loclen(), and is_utf8_string_loc().
319 Perl_is_utf8_string(const U8 *s, STRLEN len)
321 const U8* const send = s + (len ? len : strlen((const char *)s));
324 PERL_ARGS_ASSERT_IS_UTF8_STRING;
328 /* Inline the easy bits of is_utf8_char() here for speed... */
329 if (UTF8_IS_INVARIANT(*x))
331 else if (!UTF8_IS_START(*x))
334 /* ... and call is_utf8_char() only if really needed. */
337 if (IS_UTF8_CHAR_FAST(c)) {
338 if (!IS_UTF8_CHAR(x, c))
342 c = is_utf8_char_slow(x, c);
345 #endif /* #ifdef IS_UTF8_CHAR */
360 Implemented as a macro in utf8.h
362 =for apidoc is_utf8_string_loc
364 Like is_utf8_string() but stores the location of the failure (in the
365 case of "utf8ness failure") or the location s+len (in the case of
366 "utf8ness success") in the C<ep>.
368 See also is_utf8_string_loclen() and is_utf8_string().
370 =for apidoc is_utf8_string_loclen
372 Like is_utf8_string() but stores the location of the failure (in the
373 case of "utf8ness failure") or the location s+len (in the case of
374 "utf8ness success") in the C<ep>, and the number of UTF-8
375 encoded characters in the C<el>.
377 See also is_utf8_string_loc() and is_utf8_string().
383 Perl_is_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
385 const U8* const send = s + (len ? len : strlen((const char *)s));
390 PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN;
393 /* Inline the easy bits of is_utf8_char() here for speed... */
394 if (UTF8_IS_INVARIANT(*x))
396 else if (!UTF8_IS_START(*x))
399 /* ... and call is_utf8_char() only if really needed. */
402 if (IS_UTF8_CHAR_FAST(c)) {
403 if (!IS_UTF8_CHAR(x, c))
406 c = is_utf8_char_slow(x, c);
409 #endif /* #ifdef IS_UTF8_CHAR */
428 =for apidoc utf8n_to_uvuni
430 Bottom level UTF-8 decode routine.
431 Returns the Unicode code point value of the first character in the string C<s>
432 which is assumed to be in UTF-8 encoding and no longer than C<curlen>;
433 C<retlen> will be set to the length, in bytes, of that character.
435 If C<s> does not point to a well-formed UTF-8 character, the behaviour
436 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
437 it is assumed that the caller will raise a warning, and this function
438 will silently just set C<retlen> to C<-1> and return zero. If the
439 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
440 malformations will be given, C<retlen> will be set to the expected
441 length of the UTF-8 character in bytes, and zero will be returned.
443 The C<flags> can also contain various flags to allow deviations from
444 the strict UTF-8 encoding (see F<utf8.h>).
446 Most code should use utf8_to_uvchr() rather than call this directly.
452 Perl_utf8n_to_uvuni(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
455 const U8 * const s0 = s;
458 const bool dowarn = ckWARN_d(WARN_UTF8);
459 const UV startbyte = *s;
460 STRLEN expectlen = 0;
464 PERL_ARGS_ASSERT_UTF8N_TO_UVUNI;
466 /* This list is a superset of the UTF8_ALLOW_XXX. BUT it isn't, eg SUPER missing XXX */
468 #define UTF8_WARN_EMPTY 1
469 #define UTF8_WARN_CONTINUATION 2
470 #define UTF8_WARN_NON_CONTINUATION 3
471 #define UTF8_WARN_FE_FF 4
472 #define UTF8_WARN_SHORT 5
473 #define UTF8_WARN_OVERFLOW 6
474 #define UTF8_WARN_SURROGATE 7
475 #define UTF8_WARN_LONG 8
476 #define UTF8_WARN_FFFF 9 /* Also FFFE. */
479 !(flags & UTF8_ALLOW_EMPTY)) {
480 warning = UTF8_WARN_EMPTY;
484 if (UTF8_IS_INVARIANT(uv)) {
487 return (UV) (NATIVE_TO_UTF(*s));
490 if (UTF8_IS_CONTINUATION(uv) &&
491 !(flags & UTF8_ALLOW_CONTINUATION)) {
492 warning = UTF8_WARN_CONTINUATION;
496 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
497 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
498 warning = UTF8_WARN_NON_CONTINUATION;
503 uv = NATIVE_TO_UTF(uv);
505 if ((uv == 0xfe || uv == 0xff) &&
506 !(flags & UTF8_ALLOW_FE_FF)) {
507 warning = UTF8_WARN_FE_FF;
512 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
513 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
514 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
515 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
517 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
518 else { len = 7; uv &= 0x01; }
520 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
521 else if (!(uv & 0x01)) { len = 7; uv = 0; }
522 else { len = 13; uv = 0; } /* whoa! */
530 if ((curlen < expectlen) &&
531 !(flags & UTF8_ALLOW_SHORT)) {
532 warning = UTF8_WARN_SHORT;
541 if (!UTF8_IS_CONTINUATION(*s) &&
542 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
544 warning = UTF8_WARN_NON_CONTINUATION;
548 uv = UTF8_ACCUMULATE(uv, *s);
550 /* These cannot be allowed. */
552 if (expectlen != 13 && !(flags & UTF8_ALLOW_LONG)) {
553 warning = UTF8_WARN_LONG;
557 else { /* uv < ouv */
558 /* This cannot be allowed. */
559 warning = UTF8_WARN_OVERFLOW;
567 if (UNICODE_IS_SURROGATE(uv) &&
568 !(flags & UTF8_ALLOW_SURROGATE)) {
569 warning = UTF8_WARN_SURROGATE;
571 } else if ((expectlen > (STRLEN)UNISKIP(uv)) &&
572 !(flags & UTF8_ALLOW_LONG)) {
573 warning = UTF8_WARN_LONG;
575 } else if (UNICODE_IS_ILLEGAL(uv) &&
576 !(flags & UTF8_ALLOW_FFFF)) {
577 warning = UTF8_WARN_FFFF;
585 if (flags & UTF8_CHECK_ONLY) {
587 *retlen = ((STRLEN) -1);
592 if (warning == UTF8_WARN_FFFF) {
593 sv = newSVpvs_flags("Unicode non-character ", SVs_TEMP);
594 Perl_sv_catpvf(aTHX_ sv, "0x%04"UVxf" is illegal for interchange", uv);
597 sv = newSVpvs_flags("Malformed UTF-8 character ", SVs_TEMP);
600 case 0: /* Intentionally empty. */ break;
601 case UTF8_WARN_EMPTY:
602 sv_catpvs(sv, "(empty string)");
604 case UTF8_WARN_CONTINUATION:
605 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
607 case UTF8_WARN_NON_CONTINUATION:
609 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
610 (UV)s[1], startbyte);
612 const int len = (int)(s-s0);
613 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
614 (UV)s[1], len, len > 1 ? "s" : "", startbyte, (int)expectlen);
618 case UTF8_WARN_FE_FF:
619 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
621 case UTF8_WARN_SHORT:
622 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
623 (int)curlen, curlen == 1 ? "" : "s", (int)expectlen, startbyte);
624 expectlen = curlen; /* distance for caller to skip */
626 case UTF8_WARN_OVERFLOW:
627 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
630 case UTF8_WARN_SURROGATE:
631 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
634 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
635 (int)expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
638 sv_catpvs(sv, "(unknown reason)");
644 const char * const s = SvPVX_const(sv);
647 Perl_warner(aTHX_ packWARN(WARN_UTF8),
648 "%s in %s", s, OP_DESC(PL_op));
650 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
655 *retlen = expectlen ? expectlen : len;
661 =for apidoc utf8_to_uvchr
663 Returns the native character value of the first character in the string C<s>
664 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
665 length, in bytes, of that character.
667 If C<s> does not point to a well-formed UTF-8 character, zero is
668 returned and retlen is set, if possible, to -1.
674 Perl_utf8_to_uvchr(pTHX_ const U8 *s, STRLEN *retlen)
676 PERL_ARGS_ASSERT_UTF8_TO_UVCHR;
678 return utf8n_to_uvchr(s, UTF8_MAXBYTES, retlen,
679 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
683 =for apidoc utf8_to_uvuni
685 Returns the Unicode code point of the first character in the string C<s>
686 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
687 length, in bytes, of that character.
689 This function should only be used when the returned UV is considered
690 an index into the Unicode semantic tables (e.g. swashes).
692 If C<s> does not point to a well-formed UTF-8 character, zero is
693 returned and retlen is set, if possible, to -1.
699 Perl_utf8_to_uvuni(pTHX_ const U8 *s, STRLEN *retlen)
701 PERL_ARGS_ASSERT_UTF8_TO_UVUNI;
703 /* Call the low level routine asking for checks */
704 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXBYTES, retlen,
705 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
709 =for apidoc utf8_length
711 Return the length of the UTF-8 char encoded string C<s> in characters.
712 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
713 up past C<e>, croaks.
719 Perl_utf8_length(pTHX_ const U8 *s, const U8 *e)
724 PERL_ARGS_ASSERT_UTF8_LENGTH;
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. */
731 goto warn_and_return;
733 if (!UTF8_IS_INVARIANT(*s))
744 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
745 "%s in %s", unees, OP_DESC(PL_op));
747 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8), unees);
754 =for apidoc utf8_distance
756 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
759 WARNING: use only if you *know* that the pointers point inside the
766 Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
768 PERL_ARGS_ASSERT_UTF8_DISTANCE;
770 return (a < b) ? -1 * (IV) utf8_length(a, b) : (IV) utf8_length(b, a);
776 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
779 WARNING: do not use the following unless you *know* C<off> is within
780 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
781 on the first byte of character or just after the last byte of a character.
787 Perl_utf8_hop(pTHX_ const U8 *s, I32 off)
789 PERL_ARGS_ASSERT_UTF8_HOP;
792 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
793 * the bitops (especially ~) can create illegal UTF-8.
794 * In other words: in Perl UTF-8 is not just for Unicode. */
803 while (UTF8_IS_CONTINUATION(*s))
811 =for apidoc bytes_cmp_utf8
813 Compares the sequence of characters (stored as octets) in b, blen with the
814 sequence of characters (stored as UTF-8) in u, ulen. Returns 0 if they are
815 equal, -1 or -2 if the first string is less than the second string, +1 or +2
816 if the first string is greater than the second string.
818 -1 or +1 is returned if the shorter string was identical to the start of the
819 longer string. -2 or +2 is returned if the was a difference between characters
826 Perl_bytes_cmp_utf8(pTHX_ const U8 *b, STRLEN blen, const U8 *u, STRLEN ulen)
828 const U8 *const bend = b + blen;
829 const U8 *const uend = u + ulen;
831 PERL_ARGS_ASSERT_BYTES_CMP_UTF8;
835 while (b < bend && u < uend) {
837 if (!UTF8_IS_INVARIANT(c)) {
838 if (UTF8_IS_DOWNGRADEABLE_START(c)) {
841 if (UTF8_IS_CONTINUATION(c1)) {
842 c = UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(c, c1));
844 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
845 "Malformed UTF-8 character "
846 "(unexpected non-continuation byte 0x%02x"
847 ", immediately after start byte 0x%02x)"
848 /* Dear diag.t, it's in the pod. */
851 PL_op ? OP_DESC(PL_op) : "");
856 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
857 "%s in %s", unees, OP_DESC(PL_op));
859 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8), unees);
860 return -2; /* Really want to return undef :-) */
867 return *b < c ? -2 : +2;
872 if (b == bend && u == uend)
875 return b < bend ? +1 : -1;
879 =for apidoc utf8_to_bytes
881 Converts a string C<s> of length C<len> from UTF-8 into native byte encoding.
882 Unlike C<bytes_to_utf8>, this over-writes the original string, and
883 updates len to contain the new length.
884 Returns zero on failure, setting C<len> to -1.
886 If you need a copy of the string, see C<bytes_from_utf8>.
892 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
895 U8 * const send = s + *len;
898 PERL_ARGS_ASSERT_UTF8_TO_BYTES;
900 /* ensure valid UTF-8 and chars < 256 before updating string */
904 if (!UTF8_IS_INVARIANT(c) &&
905 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
906 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
907 *len = ((STRLEN) -1);
915 *d++ = (U8)utf8_to_uvchr(s, &ulen);
924 =for apidoc bytes_from_utf8
926 Converts a string C<s> of length C<len> from UTF-8 into native byte encoding.
927 Unlike C<utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
928 the newly-created string, and updates C<len> to contain the new
929 length. Returns the original string if no conversion occurs, C<len>
930 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
931 0 if C<s> is converted or consisted entirely of characters that are invariant
932 in utf8 (i.e., US-ASCII on non-EBCDIC machines).
938 Perl_bytes_from_utf8(pTHX_ const U8 *s, STRLEN *len, bool *is_utf8)
945 PERL_ARGS_ASSERT_BYTES_FROM_UTF8;
951 /* ensure valid UTF-8 and chars < 256 before converting string */
952 for (send = s + *len; s < send;) {
954 if (!UTF8_IS_INVARIANT(c)) {
955 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
956 (c = *s++) && UTF8_IS_CONTINUATION(c))
965 Newx(d, (*len) - count + 1, U8);
966 s = start; start = d;
969 if (!UTF8_IS_INVARIANT(c)) {
970 /* Then it is two-byte encoded */
971 c = UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(c, *s++));
981 =for apidoc bytes_to_utf8
983 Converts a string C<s> of length C<len> from the native encoding into UTF-8.
984 Returns a pointer to the newly-created string, and sets C<len> to
985 reflect the new length.
987 A NUL character will be written after the end of the string.
989 If you want to convert to UTF-8 from encodings other than
990 the native (Latin1 or EBCDIC),
991 see sv_recode_to_utf8().
997 Perl_bytes_to_utf8(pTHX_ const U8 *s, STRLEN *len)
999 const U8 * const send = s + (*len);
1003 PERL_ARGS_ASSERT_BYTES_TO_UTF8;
1004 PERL_UNUSED_CONTEXT;
1006 Newx(d, (*len) * 2 + 1, U8);
1010 const UV uv = NATIVE_TO_ASCII(*s++);
1011 if (UNI_IS_INVARIANT(uv))
1012 *d++ = (U8)UTF_TO_NATIVE(uv);
1014 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
1015 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
1024 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
1026 * Destination must be pre-extended to 3/2 source. Do not use in-place.
1027 * We optimize for native, for obvious reasons. */
1030 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
1035 PERL_ARGS_ASSERT_UTF16_TO_UTF8;
1038 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %"UVuf, (UV)bytelen);
1043 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
1047 *d++ = UNI_TO_NATIVE(uv);
1054 *d++ = (U8)(( uv >> 6) | 0xc0);
1055 *d++ = (U8)(( uv & 0x3f) | 0x80);
1058 if (uv >= 0xd800 && uv <= 0xdbff) { /* surrogates */
1060 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
1062 UV low = (p[0] << 8) + p[1];
1064 if (low < 0xdc00 || low > 0xdfff)
1065 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
1066 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
1068 } else if (uv >= 0xdc00 && uv <= 0xdfff) {
1069 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
1072 *d++ = (U8)(( uv >> 12) | 0xe0);
1073 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
1074 *d++ = (U8)(( uv & 0x3f) | 0x80);
1078 *d++ = (U8)(( uv >> 18) | 0xf0);
1079 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
1080 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
1081 *d++ = (U8)(( uv & 0x3f) | 0x80);
1085 *newlen = d - dstart;
1089 /* Note: this one is slightly destructive of the source. */
1092 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
1095 U8* const send = s + bytelen;
1097 PERL_ARGS_ASSERT_UTF16_TO_UTF8_REVERSED;
1100 Perl_croak(aTHX_ "panic: utf16_to_utf8_reversed: odd bytelen %"UVuf,
1104 const U8 tmp = s[0];
1109 return utf16_to_utf8(p, d, bytelen, newlen);
1112 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
1115 Perl_is_uni_alnum(pTHX_ UV c)
1117 U8 tmpbuf[UTF8_MAXBYTES+1];
1118 uvchr_to_utf8(tmpbuf, c);
1119 return is_utf8_alnum(tmpbuf);
1123 Perl_is_uni_idfirst(pTHX_ UV c)
1125 U8 tmpbuf[UTF8_MAXBYTES+1];
1126 uvchr_to_utf8(tmpbuf, c);
1127 return is_utf8_idfirst(tmpbuf);
1131 Perl_is_uni_alpha(pTHX_ UV c)
1133 U8 tmpbuf[UTF8_MAXBYTES+1];
1134 uvchr_to_utf8(tmpbuf, c);
1135 return is_utf8_alpha(tmpbuf);
1139 Perl_is_uni_ascii(pTHX_ UV c)
1141 U8 tmpbuf[UTF8_MAXBYTES+1];
1142 uvchr_to_utf8(tmpbuf, c);
1143 return is_utf8_ascii(tmpbuf);
1147 Perl_is_uni_space(pTHX_ UV c)
1149 U8 tmpbuf[UTF8_MAXBYTES+1];
1150 uvchr_to_utf8(tmpbuf, c);
1151 return is_utf8_space(tmpbuf);
1155 Perl_is_uni_digit(pTHX_ UV c)
1157 U8 tmpbuf[UTF8_MAXBYTES+1];
1158 uvchr_to_utf8(tmpbuf, c);
1159 return is_utf8_digit(tmpbuf);
1163 Perl_is_uni_upper(pTHX_ UV c)
1165 U8 tmpbuf[UTF8_MAXBYTES+1];
1166 uvchr_to_utf8(tmpbuf, c);
1167 return is_utf8_upper(tmpbuf);
1171 Perl_is_uni_lower(pTHX_ UV c)
1173 U8 tmpbuf[UTF8_MAXBYTES+1];
1174 uvchr_to_utf8(tmpbuf, c);
1175 return is_utf8_lower(tmpbuf);
1179 Perl_is_uni_cntrl(pTHX_ UV c)
1181 U8 tmpbuf[UTF8_MAXBYTES+1];
1182 uvchr_to_utf8(tmpbuf, c);
1183 return is_utf8_cntrl(tmpbuf);
1187 Perl_is_uni_graph(pTHX_ UV c)
1189 U8 tmpbuf[UTF8_MAXBYTES+1];
1190 uvchr_to_utf8(tmpbuf, c);
1191 return is_utf8_graph(tmpbuf);
1195 Perl_is_uni_print(pTHX_ UV c)
1197 U8 tmpbuf[UTF8_MAXBYTES+1];
1198 uvchr_to_utf8(tmpbuf, c);
1199 return is_utf8_print(tmpbuf);
1203 Perl_is_uni_punct(pTHX_ UV c)
1205 U8 tmpbuf[UTF8_MAXBYTES+1];
1206 uvchr_to_utf8(tmpbuf, c);
1207 return is_utf8_punct(tmpbuf);
1211 Perl_is_uni_xdigit(pTHX_ UV c)
1213 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1214 uvchr_to_utf8(tmpbuf, c);
1215 return is_utf8_xdigit(tmpbuf);
1219 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
1221 PERL_ARGS_ASSERT_TO_UNI_UPPER;
1223 uvchr_to_utf8(p, c);
1224 return to_utf8_upper(p, p, lenp);
1228 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
1230 PERL_ARGS_ASSERT_TO_UNI_TITLE;
1232 uvchr_to_utf8(p, c);
1233 return to_utf8_title(p, p, lenp);
1237 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
1239 PERL_ARGS_ASSERT_TO_UNI_LOWER;
1241 uvchr_to_utf8(p, c);
1242 return to_utf8_lower(p, p, lenp);
1246 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1248 PERL_ARGS_ASSERT_TO_UNI_FOLD;
1250 uvchr_to_utf8(p, c);
1251 return to_utf8_fold(p, p, lenp);
1254 /* for now these all assume no locale info available for Unicode > 255 */
1257 Perl_is_uni_alnum_lc(pTHX_ UV c)
1259 return is_uni_alnum(c); /* XXX no locale support yet */
1263 Perl_is_uni_idfirst_lc(pTHX_ UV c)
1265 return is_uni_idfirst(c); /* XXX no locale support yet */
1269 Perl_is_uni_alpha_lc(pTHX_ UV c)
1271 return is_uni_alpha(c); /* XXX no locale support yet */
1275 Perl_is_uni_ascii_lc(pTHX_ UV c)
1277 return is_uni_ascii(c); /* XXX no locale support yet */
1281 Perl_is_uni_space_lc(pTHX_ UV c)
1283 return is_uni_space(c); /* XXX no locale support yet */
1287 Perl_is_uni_digit_lc(pTHX_ UV c)
1289 return is_uni_digit(c); /* XXX no locale support yet */
1293 Perl_is_uni_upper_lc(pTHX_ UV c)
1295 return is_uni_upper(c); /* XXX no locale support yet */
1299 Perl_is_uni_lower_lc(pTHX_ UV c)
1301 return is_uni_lower(c); /* XXX no locale support yet */
1305 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1307 return is_uni_cntrl(c); /* XXX no locale support yet */
1311 Perl_is_uni_graph_lc(pTHX_ UV c)
1313 return is_uni_graph(c); /* XXX no locale support yet */
1317 Perl_is_uni_print_lc(pTHX_ UV c)
1319 return is_uni_print(c); /* XXX no locale support yet */
1323 Perl_is_uni_punct_lc(pTHX_ UV c)
1325 return is_uni_punct(c); /* XXX no locale support yet */
1329 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1331 return is_uni_xdigit(c); /* XXX no locale support yet */
1335 Perl_to_uni_upper_lc(pTHX_ U32 c)
1337 /* XXX returns only the first character -- do not use XXX */
1338 /* XXX no locale support yet */
1340 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1341 return (U32)to_uni_upper(c, tmpbuf, &len);
1345 Perl_to_uni_title_lc(pTHX_ U32 c)
1347 /* XXX returns only the first character XXX -- do not use XXX */
1348 /* XXX no locale support yet */
1350 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1351 return (U32)to_uni_title(c, tmpbuf, &len);
1355 Perl_to_uni_lower_lc(pTHX_ U32 c)
1357 /* XXX returns only the first character -- do not use XXX */
1358 /* XXX no locale support yet */
1360 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1361 return (U32)to_uni_lower(c, tmpbuf, &len);
1365 S_is_utf8_common(pTHX_ const U8 *const p, SV **swash,
1366 const char *const swashname)
1370 PERL_ARGS_ASSERT_IS_UTF8_COMMON;
1372 if (!is_utf8_char(p))
1375 *swash = swash_init("utf8", swashname, &PL_sv_undef, 1, 0);
1376 return swash_fetch(*swash, p, TRUE) != 0;
1380 Perl_is_utf8_alnum(pTHX_ const U8 *p)
1384 PERL_ARGS_ASSERT_IS_UTF8_ALNUM;
1386 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1387 * descendant of isalnum(3), in other words, it doesn't
1388 * contain the '_'. --jhi */
1389 return is_utf8_common(p, &PL_utf8_alnum, "IsWord");
1393 Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
1397 PERL_ARGS_ASSERT_IS_UTF8_IDFIRST;
1401 /* is_utf8_idstart would be more logical. */
1402 return is_utf8_common(p, &PL_utf8_idstart, "IdStart");
1406 Perl_is_utf8_idcont(pTHX_ const U8 *p)
1410 PERL_ARGS_ASSERT_IS_UTF8_IDCONT;
1414 return is_utf8_common(p, &PL_utf8_idcont, "IdContinue");
1418 Perl_is_utf8_alpha(pTHX_ const U8 *p)
1422 PERL_ARGS_ASSERT_IS_UTF8_ALPHA;
1424 return is_utf8_common(p, &PL_utf8_alpha, "IsAlpha");
1428 Perl_is_utf8_ascii(pTHX_ const U8 *p)
1432 PERL_ARGS_ASSERT_IS_UTF8_ASCII;
1434 return is_utf8_common(p, &PL_utf8_ascii, "IsAscii");
1438 Perl_is_utf8_space(pTHX_ const U8 *p)
1442 PERL_ARGS_ASSERT_IS_UTF8_SPACE;
1444 return is_utf8_common(p, &PL_utf8_space, "IsSpacePerl");
1448 Perl_is_utf8_perl_space(pTHX_ const U8 *p)
1452 PERL_ARGS_ASSERT_IS_UTF8_PERL_SPACE;
1454 return is_utf8_common(p, &PL_utf8_perl_space, "IsPerlSpace");
1458 Perl_is_utf8_perl_word(pTHX_ const U8 *p)
1462 PERL_ARGS_ASSERT_IS_UTF8_PERL_WORD;
1464 return is_utf8_common(p, &PL_utf8_perl_word, "IsPerlWord");
1468 Perl_is_utf8_digit(pTHX_ const U8 *p)
1472 PERL_ARGS_ASSERT_IS_UTF8_DIGIT;
1474 return is_utf8_common(p, &PL_utf8_digit, "IsDigit");
1478 Perl_is_utf8_posix_digit(pTHX_ const U8 *p)
1482 PERL_ARGS_ASSERT_IS_UTF8_POSIX_DIGIT;
1484 return is_utf8_common(p, &PL_utf8_posix_digit, "IsPosixDigit");
1488 Perl_is_utf8_upper(pTHX_ const U8 *p)
1492 PERL_ARGS_ASSERT_IS_UTF8_UPPER;
1494 return is_utf8_common(p, &PL_utf8_upper, "IsUppercase");
1498 Perl_is_utf8_lower(pTHX_ const U8 *p)
1502 PERL_ARGS_ASSERT_IS_UTF8_LOWER;
1504 return is_utf8_common(p, &PL_utf8_lower, "IsLowercase");
1508 Perl_is_utf8_cntrl(pTHX_ const U8 *p)
1512 PERL_ARGS_ASSERT_IS_UTF8_CNTRL;
1514 return is_utf8_common(p, &PL_utf8_cntrl, "IsCntrl");
1518 Perl_is_utf8_graph(pTHX_ const U8 *p)
1522 PERL_ARGS_ASSERT_IS_UTF8_GRAPH;
1524 return is_utf8_common(p, &PL_utf8_graph, "IsGraph");
1528 Perl_is_utf8_print(pTHX_ const U8 *p)
1532 PERL_ARGS_ASSERT_IS_UTF8_PRINT;
1534 return is_utf8_common(p, &PL_utf8_print, "IsPrint");
1538 Perl_is_utf8_punct(pTHX_ const U8 *p)
1542 PERL_ARGS_ASSERT_IS_UTF8_PUNCT;
1544 return is_utf8_common(p, &PL_utf8_punct, "IsPunct");
1548 Perl_is_utf8_xdigit(pTHX_ const U8 *p)
1552 PERL_ARGS_ASSERT_IS_UTF8_XDIGIT;
1554 return is_utf8_common(p, &PL_utf8_xdigit, "IsXDigit");
1558 Perl_is_utf8_mark(pTHX_ const U8 *p)
1562 PERL_ARGS_ASSERT_IS_UTF8_MARK;
1564 return is_utf8_common(p, &PL_utf8_mark, "IsM");
1568 Perl_is_utf8_X_begin(pTHX_ const U8 *p)
1572 PERL_ARGS_ASSERT_IS_UTF8_X_BEGIN;
1574 return is_utf8_common(p, &PL_utf8_X_begin, "_X_Begin");
1578 Perl_is_utf8_X_extend(pTHX_ const U8 *p)
1582 PERL_ARGS_ASSERT_IS_UTF8_X_EXTEND;
1584 return is_utf8_common(p, &PL_utf8_X_extend, "_X_Extend");
1588 Perl_is_utf8_X_prepend(pTHX_ const U8 *p)
1592 PERL_ARGS_ASSERT_IS_UTF8_X_PREPEND;
1594 return is_utf8_common(p, &PL_utf8_X_prepend, "GCB=Prepend");
1598 Perl_is_utf8_X_non_hangul(pTHX_ const U8 *p)
1602 PERL_ARGS_ASSERT_IS_UTF8_X_NON_HANGUL;
1604 return is_utf8_common(p, &PL_utf8_X_non_hangul, "HST=Not_Applicable");
1608 Perl_is_utf8_X_L(pTHX_ const U8 *p)
1612 PERL_ARGS_ASSERT_IS_UTF8_X_L;
1614 return is_utf8_common(p, &PL_utf8_X_L, "GCB=L");
1618 Perl_is_utf8_X_LV(pTHX_ const U8 *p)
1622 PERL_ARGS_ASSERT_IS_UTF8_X_LV;
1624 return is_utf8_common(p, &PL_utf8_X_LV, "GCB=LV");
1628 Perl_is_utf8_X_LVT(pTHX_ const U8 *p)
1632 PERL_ARGS_ASSERT_IS_UTF8_X_LVT;
1634 return is_utf8_common(p, &PL_utf8_X_LVT, "GCB=LVT");
1638 Perl_is_utf8_X_T(pTHX_ const U8 *p)
1642 PERL_ARGS_ASSERT_IS_UTF8_X_T;
1644 return is_utf8_common(p, &PL_utf8_X_T, "GCB=T");
1648 Perl_is_utf8_X_V(pTHX_ const U8 *p)
1652 PERL_ARGS_ASSERT_IS_UTF8_X_V;
1654 return is_utf8_common(p, &PL_utf8_X_V, "GCB=V");
1658 Perl_is_utf8_X_LV_LVT_V(pTHX_ const U8 *p)
1662 PERL_ARGS_ASSERT_IS_UTF8_X_LV_LVT_V;
1664 return is_utf8_common(p, &PL_utf8_X_LV_LVT_V, "_X_LV_LVT_V");
1668 =for apidoc to_utf8_case
1670 The "p" contains the pointer to the UTF-8 string encoding
1671 the character that is being converted.
1673 The "ustrp" is a pointer to the character buffer to put the
1674 conversion result to. The "lenp" is a pointer to the length
1677 The "swashp" is a pointer to the swash to use.
1679 Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
1680 and loaded by SWASHNEW, using lib/utf8_heavy.pl. The special (usually,
1681 but not always, a multicharacter mapping), is tried first.
1683 The "special" is a string like "utf8::ToSpecLower", which means the
1684 hash %utf8::ToSpecLower. The access to the hash is through
1685 Perl_to_utf8_case().
1687 The "normal" is a string like "ToLower" which means the swash
1693 Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp,
1694 SV **swashp, const char *normal, const char *special)
1697 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1699 const UV uv0 = utf8_to_uvchr(p, NULL);
1700 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1701 * are necessary in EBCDIC, they are redundant no-ops
1702 * in ASCII-ish platforms, and hopefully optimized away. */
1703 const UV uv1 = NATIVE_TO_UNI(uv0);
1705 PERL_ARGS_ASSERT_TO_UTF8_CASE;
1707 uvuni_to_utf8(tmpbuf, uv1);
1709 if (!*swashp) /* load on-demand */
1710 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1711 /* This is the beginnings of a skeleton of code to read the info section
1712 * that is in all the swashes in case we ever want to do that, so one can
1713 * read things whose maps aren't code points, and whose default if missing
1714 * is not to the code point itself. This was just to see if it actually
1715 * worked. Details on what the possibilities are are in perluniprops.pod
1716 HV * const hv = get_hv("utf8::SwashInfo", 0);
1719 svp = hv_fetch(hv, (const char*)normal, strlen(normal), FALSE);
1722 HV * const this_hash = SvRV(*svp);
1723 svp = hv_fetch(this_hash, "type", strlen("type"), FALSE);
1724 s = SvPV_const(*svp, len);
1729 /* It might be "special" (sometimes, but not always,
1730 * a multicharacter mapping) */
1731 HV * const hv = get_hv(special, 0);
1735 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1739 s = SvPV_const(*svp, len);
1741 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
1744 /* If we have EBCDIC we need to remap the characters
1745 * since any characters in the low 256 are Unicode
1746 * code points, not EBCDIC. */
1747 U8 *t = (U8*)s, *tend = t + len, *d;
1754 const UV c = utf8_to_uvchr(t, &tlen);
1756 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1765 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1770 Copy(tmpbuf, ustrp, len, U8);
1772 Copy(s, ustrp, len, U8);
1778 if (!len && *swashp) {
1779 const UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1782 /* It was "normal" (a single character mapping). */
1783 const UV uv3 = UNI_TO_NATIVE(uv2);
1784 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
1788 if (!len) /* Neither: just copy. In other words, there was no mapping
1789 defined, which means that the code point maps to itself */
1790 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1795 return len ? utf8_to_uvchr(ustrp, 0) : 0;
1799 =for apidoc to_utf8_upper
1801 Convert the UTF-8 encoded character at p to its uppercase version and
1802 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1803 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1804 the uppercase version may be longer than the original character.
1806 The first character of the uppercased version is returned
1807 (but note, as explained above, that there may be more.)
1812 Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1816 PERL_ARGS_ASSERT_TO_UTF8_UPPER;
1818 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1819 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1823 =for apidoc to_utf8_title
1825 Convert the UTF-8 encoded character at p to its titlecase version and
1826 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1827 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1828 titlecase version may be longer than the original character.
1830 The first character of the titlecased version is returned
1831 (but note, as explained above, that there may be more.)
1836 Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1840 PERL_ARGS_ASSERT_TO_UTF8_TITLE;
1842 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1843 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1847 =for apidoc to_utf8_lower
1849 Convert the UTF-8 encoded character at p to its lowercase version and
1850 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1851 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1852 lowercase version may be longer than the original character.
1854 The first character of the lowercased version is returned
1855 (but note, as explained above, that there may be more.)
1860 Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1864 PERL_ARGS_ASSERT_TO_UTF8_LOWER;
1866 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1867 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1871 =for apidoc to_utf8_fold
1873 Convert the UTF-8 encoded character at p to its foldcase version and
1874 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1875 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1876 foldcase version may be longer than the original character (up to
1879 The first character of the foldcased version is returned
1880 (but note, as explained above, that there may be more.)
1885 Perl_to_utf8_fold(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1889 PERL_ARGS_ASSERT_TO_UTF8_FOLD;
1891 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1892 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1896 * A "swash" is a swatch hash.
1897 * A "swatch" is a bit vector generated by utf8.c:S_swash_get().
1898 * C<pkg> is a pointer to a package name for SWASHNEW, should be "utf8".
1899 * For other parameters, see utf8::SWASHNEW in lib/utf8_heavy.pl.
1902 Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
1907 const size_t pkg_len = strlen(pkg);
1908 const size_t name_len = strlen(name);
1909 HV * const stash = gv_stashpvn(pkg, pkg_len, 0);
1912 PERL_ARGS_ASSERT_SWASH_INIT;
1914 PUSHSTACKi(PERLSI_MAGIC);
1918 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
1920 errsv_save = newSVsv(ERRSV);
1921 /* It is assumed that callers of this routine are not passing in any
1922 user derived data. */
1923 /* Need to do this after save_re_context() as it will set PL_tainted to
1924 1 while saving $1 etc (see the code after getrx: in Perl_magic_get).
1925 Even line to create errsv_save can turn on PL_tainted. */
1926 SAVEBOOL(PL_tainted);
1928 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
1931 sv_setsv(ERRSV, errsv_save);
1932 SvREFCNT_dec(errsv_save);
1938 mPUSHp(pkg, pkg_len);
1939 mPUSHp(name, name_len);
1944 errsv_save = newSVsv(ERRSV);
1945 if (call_method("SWASHNEW", G_SCALAR))
1946 retval = newSVsv(*PL_stack_sp--);
1948 retval = &PL_sv_undef;
1950 sv_setsv(ERRSV, errsv_save);
1951 SvREFCNT_dec(errsv_save);
1954 if (IN_PERL_COMPILETIME) {
1955 CopHINTS_set(PL_curcop, PL_hints);
1957 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
1959 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
1961 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1967 /* This API is wrong for special case conversions since we may need to
1968 * return several Unicode characters for a single Unicode character
1969 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1970 * the lower-level routine, and it is similarly broken for returning
1971 * multiple values. --jhi */
1972 /* Now SWASHGET is recasted into S_swash_get in this file. */
1975 * Returns the value of property/mapping C<swash> for the first character
1976 * of the string C<ptr>. If C<do_utf8> is true, the string C<ptr> is
1977 * assumed to be in utf8. If C<do_utf8> is false, the string C<ptr> is
1978 * assumed to be in native 8-bit encoding. Caches the swatch in C<swash>.
1981 Perl_swash_fetch(pTHX_ SV *swash, const U8 *ptr, bool do_utf8)
1984 HV *const hv = MUTABLE_HV(SvRV(swash));
1989 const U8 *tmps = NULL;
1993 const UV c = NATIVE_TO_ASCII(*ptr);
1995 PERL_ARGS_ASSERT_SWASH_FETCH;
1997 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1998 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
1999 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
2002 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
2003 * then the "swatch" is a vec() for all the chars which start
2005 * So the key in the hash (klen) is length of encoded char -1
2007 klen = UTF8SKIP(ptr) - 1;
2011 /* If char is invariant then swatch is for all the invariant chars
2012 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
2014 needents = UTF_CONTINUATION_MARK;
2015 off = NATIVE_TO_UTF(ptr[klen]);
2018 /* If char is encoded then swatch is for the prefix */
2019 needents = (1 << UTF_ACCUMULATION_SHIFT);
2020 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
2024 * This single-entry cache saves about 1/3 of the utf8 overhead in test
2025 * suite. (That is, only 7-8% overall over just a hash cache. Still,
2026 * it's nothing to sniff at.) Pity we usually come through at least
2027 * two function calls to get here...
2029 * NB: this code assumes that swatches are never modified, once generated!
2032 if (hv == PL_last_swash_hv &&
2033 klen == PL_last_swash_klen &&
2034 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
2036 tmps = PL_last_swash_tmps;
2037 slen = PL_last_swash_slen;
2040 /* Try our second-level swatch cache, kept in a hash. */
2041 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
2043 /* If not cached, generate it via swash_get */
2044 if (!svp || !SvPOK(*svp)
2045 || !(tmps = (const U8*)SvPV_const(*svp, slen))) {
2046 /* We use utf8n_to_uvuni() as we want an index into
2047 Unicode tables, not a native character number.
2049 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
2051 0 : UTF8_ALLOW_ANY);
2052 swatch = swash_get(swash,
2053 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
2054 (klen) ? (code_point & ~(needents - 1)) : 0,
2057 if (IN_PERL_COMPILETIME)
2058 CopHINTS_set(PL_curcop, PL_hints);
2060 svp = hv_store(hv, (const char *)ptr, klen, swatch, 0);
2062 if (!svp || !(tmps = (U8*)SvPV(*svp, slen))
2063 || (slen << 3) < needents)
2064 Perl_croak(aTHX_ "panic: swash_fetch got improper swatch");
2067 PL_last_swash_hv = hv;
2068 assert(klen <= sizeof(PL_last_swash_key));
2069 PL_last_swash_klen = (U8)klen;
2070 /* FIXME change interpvar.h? */
2071 PL_last_swash_tmps = (U8 *) tmps;
2072 PL_last_swash_slen = slen;
2074 Copy(ptr, PL_last_swash_key, klen, U8);
2077 switch ((int)((slen << 3) / needents)) {
2079 bit = 1 << (off & 7);
2081 return (tmps[off] & bit) != 0;
2086 return (tmps[off] << 8) + tmps[off + 1] ;
2089 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
2091 Perl_croak(aTHX_ "panic: swash_fetch got swatch of unexpected bit width");
2092 NORETURN_FUNCTION_END;
2095 /* Read a single line of the main body of the swash input text. These are of
2098 * where each number is hex. The first two numbers form the minimum and
2099 * maximum of a range, and the third is the value associated with the range.
2100 * Not all swashes should have a third number
2102 * On input: l points to the beginning of the line to be examined; it points
2103 * to somewhere in the string of the whole input text, and is
2104 * terminated by a \n or the null string terminator.
2105 * lend points to the null terminator of that string
2106 * wants_value is non-zero if the swash expects a third number
2107 * typestr is the name of the swash's mapping, like 'ToLower'
2108 * On output: *min, *max, and *val are set to the values read from the line.
2109 * returns a pointer just beyond the line examined. If there was no
2110 * valid min number on the line, returns lend+1
2114 S_swash_scan_list_line(pTHX_ U8* l, U8* const lend, UV* min, UV* max, UV* val,
2115 const bool wants_value, const U8* const typestr)
2117 const int typeto = typestr[0] == 'T' && typestr[1] == 'o';
2118 STRLEN numlen; /* Length of the number */
2119 I32 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
2121 /* nl points to the next \n in the scan */
2122 U8* const nl = (U8*)memchr(l, '\n', lend - l);
2124 /* Get the first number on the line: the range minimum */
2126 *min = grok_hex((char *)l, &numlen, &flags, NULL);
2127 if (numlen) /* If found a hex number, position past it */
2129 else if (nl) { /* Else, go handle next line, if any */
2130 return nl + 1; /* 1 is length of "\n" */
2132 else { /* Else, no next line */
2133 return lend + 1; /* to LIST's end at which \n is not found */
2136 /* The max range value follows, separated by a BLANK */
2139 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
2141 *max = grok_hex((char *)l, &numlen, &flags, NULL);
2144 else /* If no value here, it is a single element range */
2147 /* Non-binary tables have a third entry: what the first element of the
2152 flags = PERL_SCAN_SILENT_ILLDIGIT |
2153 PERL_SCAN_DISALLOW_PREFIX;
2155 *val = grok_hex((char *)l, &numlen, &flags, NULL);
2164 Perl_croak(aTHX_ "%s: illegal mapping '%s'",
2170 *val = 0; /* bits == 1, then any val should be ignored */
2172 else { /* Nothing following range min, should be single element with no
2178 Perl_croak(aTHX_ "%s: illegal mapping '%s'", typestr, l);
2182 *val = 0; /* bits == 1, then val should be ignored */
2185 /* Position to next line if any, or EOF */
2195 * Returns a swatch (a bit vector string) for a code point sequence
2196 * that starts from the value C<start> and comprises the number C<span>.
2197 * A C<swash> must be an object created by SWASHNEW (see lib/utf8_heavy.pl).
2198 * Should be used via swash_fetch, which will cache the swatch in C<swash>.
2201 S_swash_get(pTHX_ SV* swash, UV start, UV span)
2204 U8 *l, *lend, *x, *xend, *s;
2205 STRLEN lcur, xcur, scur;
2206 HV *const hv = MUTABLE_HV(SvRV(swash));
2208 /* The string containing the main body of the table */
2209 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2211 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2212 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2213 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2214 SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);
2215 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
2216 const STRLEN bits = SvUV(*bitssvp);
2217 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2218 const UV none = SvUV(*nonesvp);
2219 const UV end = start + span;
2221 PERL_ARGS_ASSERT_SWASH_GET;
2223 if (bits != 1 && bits != 8 && bits != 16 && bits != 32) {
2224 Perl_croak(aTHX_ "panic: swash_get doesn't expect bits %"UVuf,
2228 /* create and initialize $swatch */
2229 scur = octets ? (span * octets) : (span + 7) / 8;
2230 swatch = newSV(scur);
2232 s = (U8*)SvPVX(swatch);
2233 if (octets && none) {
2234 const U8* const e = s + scur;
2237 *s++ = (U8)(none & 0xff);
2238 else if (bits == 16) {
2239 *s++ = (U8)((none >> 8) & 0xff);
2240 *s++ = (U8)( none & 0xff);
2242 else if (bits == 32) {
2243 *s++ = (U8)((none >> 24) & 0xff);
2244 *s++ = (U8)((none >> 16) & 0xff);
2245 *s++ = (U8)((none >> 8) & 0xff);
2246 *s++ = (U8)( none & 0xff);
2252 (void)memzero((U8*)s, scur + 1);
2254 SvCUR_set(swatch, scur);
2255 s = (U8*)SvPVX(swatch);
2257 /* read $swash->{LIST} */
2258 l = (U8*)SvPV(*listsvp, lcur);
2262 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2263 cBOOL(octets), typestr);
2268 /* If looking for something beyond this range, go try the next one */
2275 if (!none || val < none) {
2280 for (key = min; key <= max; key++) {
2284 /* offset must be non-negative (start <= min <= key < end) */
2285 offset = octets * (key - start);
2287 s[offset] = (U8)(val & 0xff);
2288 else if (bits == 16) {
2289 s[offset ] = (U8)((val >> 8) & 0xff);
2290 s[offset + 1] = (U8)( val & 0xff);
2292 else if (bits == 32) {
2293 s[offset ] = (U8)((val >> 24) & 0xff);
2294 s[offset + 1] = (U8)((val >> 16) & 0xff);
2295 s[offset + 2] = (U8)((val >> 8) & 0xff);
2296 s[offset + 3] = (U8)( val & 0xff);
2299 if (!none || val < none)
2303 else { /* bits == 1, then val should be ignored */
2307 for (key = min; key <= max; key++) {
2308 const STRLEN offset = (STRLEN)(key - start);
2311 s[offset >> 3] |= 1 << (offset & 7);
2317 /* read $swash->{EXTRAS} */
2318 x = (U8*)SvPV(*extssvp, xcur);
2326 SV **otherbitssvp, *other;
2330 const U8 opc = *x++;
2334 nl = (U8*)memchr(x, '\n', xend - x);
2336 if (opc != '-' && opc != '+' && opc != '!' && opc != '&') {
2338 x = nl + 1; /* 1 is length of "\n" */
2342 x = xend; /* to EXTRAS' end at which \n is not found */
2349 namelen = nl - namestr;
2353 namelen = xend - namestr;
2357 othersvp = hv_fetch(hv, (char *)namestr, namelen, FALSE);
2358 otherhv = MUTABLE_HV(SvRV(*othersvp));
2359 otherbitssvp = hv_fetchs(otherhv, "BITS", FALSE);
2360 otherbits = (STRLEN)SvUV(*otherbitssvp);
2361 if (bits < otherbits)
2362 Perl_croak(aTHX_ "panic: swash_get found swatch size mismatch");
2364 /* The "other" swatch must be destroyed after. */
2365 other = swash_get(*othersvp, start, span);
2366 o = (U8*)SvPV(other, olen);
2369 Perl_croak(aTHX_ "panic: swash_get got improper swatch");
2371 s = (U8*)SvPV(swatch, slen);
2372 if (bits == 1 && otherbits == 1) {
2374 Perl_croak(aTHX_ "panic: swash_get found swatch length mismatch");
2398 STRLEN otheroctets = otherbits >> 3;
2400 U8* const send = s + slen;
2405 if (otherbits == 1) {
2406 otherval = (o[offset >> 3] >> (offset & 7)) & 1;
2410 STRLEN vlen = otheroctets;
2418 if (opc == '+' && otherval)
2419 NOOP; /* replace with otherval */
2420 else if (opc == '!' && !otherval)
2422 else if (opc == '-' && otherval)
2424 else if (opc == '&' && !otherval)
2427 s += octets; /* no replacement */
2432 *s++ = (U8)( otherval & 0xff);
2433 else if (bits == 16) {
2434 *s++ = (U8)((otherval >> 8) & 0xff);
2435 *s++ = (U8)( otherval & 0xff);
2437 else if (bits == 32) {
2438 *s++ = (U8)((otherval >> 24) & 0xff);
2439 *s++ = (U8)((otherval >> 16) & 0xff);
2440 *s++ = (U8)((otherval >> 8) & 0xff);
2441 *s++ = (U8)( otherval & 0xff);
2445 sv_free(other); /* through with it! */
2451 Perl__swash_inversion_hash(pTHX_ SV* swash)
2454 /* Subject to change or removal. For use only in one place in regexec.c
2456 * Returns a hash which is the inversion and closure of a swash mapping.
2457 * For example, consider the input lines:
2462 * The returned hash would have two keys, the utf8 for 006B and the utf8 for
2463 * 006C. The value for each key is an array. For 006C, the array would
2464 * have a two elements, the utf8 for itself, and for 004C. For 006B, there
2465 * would be three elements in its array, the utf8 for 006B, 004B and 212A.
2467 * Essentially, for any code point, it gives all the code points that map to
2468 * it, or the list of 'froms' for that point.
2470 * Currently it only looks at the main body of the swash, and ignores any
2471 * additions or deletions from other swashes */
2475 HV *const hv = MUTABLE_HV(SvRV(swash));
2477 /* The string containing the main body of the table */
2478 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2480 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2481 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2482 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2483 /*SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);*/
2484 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
2485 const STRLEN bits = SvUV(*bitssvp);
2486 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2487 const UV none = SvUV(*nonesvp);
2491 PERL_ARGS_ASSERT__SWASH_INVERSION_HASH;
2493 /* Must have at least 8 bits to get the mappings */
2494 if (bits != 8 && bits != 16 && bits != 32) {
2495 Perl_croak(aTHX_ "panic: swash_inversion_hash doesn't expect bits %"UVuf,
2499 /* read $swash->{LIST} */
2500 l = (U8*)SvPV(*listsvp, lcur);
2503 /* Go through each input line */
2507 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2508 cBOOL(octets), typestr);
2513 /* Each element in the range is to be inverted */
2514 for (inverse = min; inverse <= max; inverse++) {
2519 bool found_key = FALSE;
2521 /* The key is the inverse mapping */
2522 char key[UTF8_MAXBYTES+1];
2523 char* key_end = (char *) uvuni_to_utf8((U8*) key, val);
2524 STRLEN key_len = key_end - key;
2526 /* And the value is what the forward mapping is from. */
2527 char utf8_inverse[UTF8_MAXBYTES+1];
2528 char *utf8_inverse_end = (char *) uvuni_to_utf8((U8*) utf8_inverse, inverse);
2530 /* Get the list for the map */
2531 if ((listp = hv_fetch(ret, key, key_len, FALSE))) {
2532 list = (AV*) *listp;
2534 else { /* No entry yet for it: create one */
2536 if (! hv_store(ret, key, key_len, (SV*) list, FALSE)) {
2537 Perl_croak(aTHX_ "panic: hv_store() unexpectedly failed");
2541 for (i = 0; i < av_len(list); i++) {
2542 SV** entryp = av_fetch(list, i, FALSE);
2544 if (entryp == NULL) {
2545 Perl_croak(aTHX_ "panic: av_fetch() unexpectedly failed");
2548 if (SvCUR(entry) != key_len) {
2551 if (memEQ(key, SvPVX(entry), key_len)) {
2557 element = newSVpvn_flags(key, key_len, SVf_UTF8);
2558 av_push(list, element);
2562 /* Simply add the value to the list */
2563 element = newSVpvn_flags(utf8_inverse, utf8_inverse_end - utf8_inverse, SVf_UTF8);
2564 av_push(list, element);
2566 /* swash_get() increments the value of val for each element in the
2567 * range. That makes more compact tables possible. You can
2568 * express the capitalization, for example, of all consecutive
2569 * letters with a single line: 0061\t007A\t0041 This maps 0061 to
2570 * 0041, 0062 to 0042, etc. I (khw) have never understood 'none',
2571 * and it's not documented, and perhaps not even currently used,
2572 * but I copied the semantics from swash_get(), just in case */
2573 if (!none || val < none) {
2583 =for apidoc uvchr_to_utf8
2585 Adds the UTF-8 representation of the Native codepoint C<uv> to the end
2586 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
2587 bytes available. The return value is the pointer to the byte after the
2588 end of the new character. In other words,
2590 d = uvchr_to_utf8(d, uv);
2592 is the recommended wide native character-aware way of saying
2599 /* On ASCII machines this is normally a macro but we want a
2600 real function in case XS code wants it
2603 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
2605 PERL_ARGS_ASSERT_UVCHR_TO_UTF8;
2607 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
2611 Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
2613 PERL_ARGS_ASSERT_UVCHR_TO_UTF8_FLAGS;
2615 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
2619 =for apidoc utf8n_to_uvchr
2622 Returns the native character value of the first character in the string
2624 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
2625 length, in bytes, of that character.
2627 Allows length and flags to be passed to low level routine.
2631 /* On ASCII machines this is normally a macro but we want
2632 a real function in case XS code wants it
2635 Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen,
2638 const UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
2640 PERL_ARGS_ASSERT_UTF8N_TO_UVCHR;
2642 return UNI_TO_NATIVE(uv);
2646 =for apidoc pv_uni_display
2648 Build to the scalar dsv a displayable version of the string spv,
2649 length len, the displayable version being at most pvlim bytes long
2650 (if longer, the rest is truncated and "..." will be appended).
2652 The flags argument can have UNI_DISPLAY_ISPRINT set to display
2653 isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
2654 to display the \\[nrfta\\] as the backslashed versions (like '\n')
2655 (UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
2656 UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
2657 UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
2659 The pointer to the PV of the dsv is returned.
2663 Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
2668 PERL_ARGS_ASSERT_PV_UNI_DISPLAY;
2672 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
2674 /* This serves double duty as a flag and a character to print after
2675 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
2679 if (pvlim && SvCUR(dsv) >= pvlim) {
2683 u = utf8_to_uvchr((U8*)s, 0);
2685 const unsigned char c = (unsigned char)u & 0xFF;
2686 if (flags & UNI_DISPLAY_BACKSLASH) {
2703 const char string = ok;
2704 sv_catpvs(dsv, "\\");
2705 sv_catpvn(dsv, &string, 1);
2708 /* isPRINT() is the locale-blind version. */
2709 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
2710 const char string = c;
2711 sv_catpvn(dsv, &string, 1);
2716 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
2719 sv_catpvs(dsv, "...");
2725 =for apidoc sv_uni_display
2727 Build to the scalar dsv a displayable version of the scalar sv,
2728 the displayable version being at most pvlim bytes long
2729 (if longer, the rest is truncated and "..." will be appended).
2731 The flags argument is as in pv_uni_display().
2733 The pointer to the PV of the dsv is returned.
2738 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
2740 PERL_ARGS_ASSERT_SV_UNI_DISPLAY;
2742 return Perl_pv_uni_display(aTHX_ dsv, (const U8*)SvPVX_const(ssv),
2743 SvCUR(ssv), pvlim, flags);
2747 =for apidoc foldEQ_utf8
2749 Returns true if the leading portions of the strings s1 and s2 (either or both
2750 of which may be in UTF-8) are the same case-insensitively; false otherwise.
2751 How far into the strings to compare is determined by other input parameters.
2753 If u1 is true, the string s1 is assumed to be in UTF-8-encoded Unicode;
2754 otherwise it is assumed to be in native 8-bit encoding. Correspondingly for u2
2757 If the byte length l1 is non-zero, it says how far into s1 to check for fold
2758 equality. In other words, s1+l1 will be used as a goal to reach. The
2759 scan will not be considered to be a match unless the goal is reached, and
2760 scanning won't continue past that goal. Correspondingly for l2 with respect to
2763 If pe1 is non-NULL and the pointer it points to is not NULL, that pointer is
2764 considered an end pointer beyond which scanning of s1 will not continue under
2765 any circumstances. This means that if both l1 and pe1 are specified, and pe1
2766 is less than s1+l1, the match will never be successful because it can never
2767 get as far as its goal (and in fact is asserted against). Correspondingly for
2768 pe2 with respect to s2.
2770 At least one of s1 and s2 must have a goal (at least one of l1 and l2 must be
2771 non-zero), and if both do, both have to be
2772 reached for a successful match. Also, if the fold of a character is multiple
2773 characters, all of them must be matched (see tr21 reference below for
2776 Upon a successful match, if pe1 is non-NULL,
2777 it will be set to point to the beginning of the I<next> character of s1 beyond
2778 what was matched. Correspondingly for pe2 and s2.
2780 For case-insensitiveness, the "casefolding" of Unicode is used
2781 instead of upper/lowercasing both the characters, see
2782 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
2786 Perl_foldEQ_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
2789 register const U8 *p1 = (const U8*)s1; /* Point to current char */
2790 register const U8 *p2 = (const U8*)s2;
2791 register const U8 *g1 = NULL; /* goal for s1 */
2792 register const U8 *g2 = NULL;
2793 register const U8 *e1 = NULL; /* Don't scan s1 past this */
2794 register U8 *f1 = NULL; /* Point to current folded */
2795 register const U8 *e2 = NULL;
2796 register U8 *f2 = NULL;
2797 STRLEN n1 = 0, n2 = 0; /* Number of bytes in current char */
2798 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
2799 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
2800 U8 natbuf[2]; /* Holds native 8-bit char converted to utf8;
2801 these always fit in 2 bytes */
2803 PERL_ARGS_ASSERT_FOLDEQ_UTF8;
2810 g1 = (const U8*)s1 + l1;
2818 g2 = (const U8*)s2 + l2;
2821 /* Must have at least one goal */
2826 /* Will never match if goal is out-of-bounds */
2827 assert(! e1 || e1 >= g1);
2829 /* Here, there isn't an end pointer, or it is beyond the goal. We
2830 * only go as far as the goal */
2834 assert(e1); /* Must have an end for looking at s1 */
2837 /* Same for goal for s2 */
2839 assert(! e2 || e2 >= g2);
2846 /* Look through both strings, a character at a time */
2847 while (p1 < e1 && p2 < e2) {
2849 /* If at the beginning of a new character in s1, get its fold to use
2850 * and the length of the fold */
2853 to_utf8_fold(p1, foldbuf1, &n1);
2855 else { /* Not utf8, convert to it first and then get fold */
2856 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p1)));
2857 to_utf8_fold(natbuf, foldbuf1, &n1);
2862 if (n2 == 0) { /* Same for s2 */
2864 to_utf8_fold(p2, foldbuf2, &n2);
2867 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p2)));
2868 to_utf8_fold(natbuf, foldbuf2, &n2);
2873 /* While there is more to look for in both folds, see if they
2874 * continue to match */
2876 U8 fold_length = UTF8SKIP(f1);
2877 if (fold_length != UTF8SKIP(f2)
2878 || (fold_length == 1 && *f1 != *f2) /* Short circuit memNE
2879 function call for single
2881 || memNE((char*)f1, (char*)f2, fold_length))
2883 return 0; /* mismatch */
2886 /* Here, they matched, advance past them */
2893 /* When reach the end of any fold, advance the input past it */
2895 p1 += u1 ? UTF8SKIP(p1) : 1;
2898 p2 += u2 ? UTF8SKIP(p2) : 1;
2900 } /* End of loop through both strings */
2902 /* A match is defined by each scan that specified an explicit length
2903 * reaching its final goal, and the other not having matched a partial
2904 * character (which can happen when the fold of a character is more than one
2906 if (! ((g1 == 0 || p1 == g1) && (g2 == 0 || p2 == g2)) || n1 || n2) {
2910 /* Successful match. Set output pointers */
2922 * c-indentation-style: bsd
2924 * indent-tabs-mode: t
2927 * ex: set ts=8 sts=4 sw=4 noet: