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 See also is_utf8_string(), is_utf8_string_loclen(), and is_utf8_string_loc().
71 Perl_is_ascii_string(const U8 *s, STRLEN len)
73 const U8* const send = s + (len ? len : strlen((const char *)s));
76 PERL_ARGS_ASSERT_IS_ASCII_STRING;
78 for (; x < send; ++x) {
79 if (!UTF8_IS_INVARIANT(*x))
87 =for apidoc uvuni_to_utf8_flags
89 Adds the UTF-8 representation of the Unicode codepoint C<uv> to the end
90 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
91 bytes available. The return value is the pointer to the byte after the
92 end of the new character. In other words,
94 d = uvuni_to_utf8_flags(d, uv, flags);
98 d = uvuni_to_utf8(d, uv);
100 (which is equivalent to)
102 d = uvuni_to_utf8_flags(d, uv, 0);
104 is the recommended Unicode-aware way of saying
112 Perl_uvuni_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
114 PERL_ARGS_ASSERT_UVUNI_TO_UTF8_FLAGS;
116 if (ckWARN(WARN_UTF8)) {
117 if (UNICODE_IS_SURROGATE(uv) &&
118 !(flags & UNICODE_ALLOW_SURROGATE))
119 Perl_warner(aTHX_ packWARN(WARN_UTF8), "UTF-16 surrogate 0x%04"UVxf, uv);
121 ((uv >= 0xFDD0 && uv <= 0xFDEF &&
122 !(flags & UNICODE_ALLOW_FDD0))
124 ((uv & 0xFFFE) == 0xFFFE && /* Either FFFE or FFFF. */
125 !(flags & UNICODE_ALLOW_FFFF))) &&
126 /* UNICODE_ALLOW_SUPER includes
127 * FFFEs and FFFFs beyond 0x10FFFF. */
128 ((uv <= PERL_UNICODE_MAX) ||
129 !(flags & UNICODE_ALLOW_SUPER))
131 Perl_warner(aTHX_ packWARN(WARN_UTF8),
132 "Unicode non-character 0x%04"UVxf" is illegal for interchange", uv);
134 if (UNI_IS_INVARIANT(uv)) {
135 *d++ = (U8)UTF_TO_NATIVE(uv);
140 STRLEN len = UNISKIP(uv);
143 *p-- = (U8)UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
144 uv >>= UTF_ACCUMULATION_SHIFT;
146 *p = (U8)UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
149 #else /* Non loop style */
151 *d++ = (U8)(( uv >> 6) | 0xc0);
152 *d++ = (U8)(( uv & 0x3f) | 0x80);
156 *d++ = (U8)(( uv >> 12) | 0xe0);
157 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
158 *d++ = (U8)(( uv & 0x3f) | 0x80);
162 *d++ = (U8)(( uv >> 18) | 0xf0);
163 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
164 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
165 *d++ = (U8)(( uv & 0x3f) | 0x80);
168 if (uv < 0x4000000) {
169 *d++ = (U8)(( uv >> 24) | 0xf8);
170 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
171 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
172 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
173 *d++ = (U8)(( uv & 0x3f) | 0x80);
176 if (uv < 0x80000000) {
177 *d++ = (U8)(( uv >> 30) | 0xfc);
178 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
179 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
180 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
181 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
182 *d++ = (U8)(( uv & 0x3f) | 0x80);
186 if (uv < UTF8_QUAD_MAX)
189 *d++ = 0xfe; /* Can't match U+FEFF! */
190 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
191 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
192 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
193 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
194 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
195 *d++ = (U8)(( uv & 0x3f) | 0x80);
200 *d++ = 0xff; /* Can't match U+FFFE! */
201 *d++ = 0x80; /* 6 Reserved bits */
202 *d++ = (U8)(((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
203 *d++ = (U8)(((uv >> 54) & 0x3f) | 0x80);
204 *d++ = (U8)(((uv >> 48) & 0x3f) | 0x80);
205 *d++ = (U8)(((uv >> 42) & 0x3f) | 0x80);
206 *d++ = (U8)(((uv >> 36) & 0x3f) | 0x80);
207 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
208 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
209 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
210 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
211 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
212 *d++ = (U8)(( uv & 0x3f) | 0x80);
216 #endif /* Loop style */
221 Tests if some arbitrary number of bytes begins in a valid UTF-8
222 character. Note that an INVARIANT (i.e. ASCII) character is a valid
223 UTF-8 character. The actual number of bytes in the UTF-8 character
224 will be returned if it is valid, otherwise 0.
226 This is the "slow" version as opposed to the "fast" version which is
227 the "unrolled" IS_UTF8_CHAR(). E.g. for t/uni/class.t the speed
228 difference is a factor of 2 to 3. For lengths (UTF8SKIP(s)) of four
229 or less you should use the IS_UTF8_CHAR(), for lengths of five or more
230 you should use the _slow(). In practice this means that the _slow()
231 will be used very rarely, since the maximum Unicode code point (as of
232 Unicode 4.1) is U+10FFFF, which encodes in UTF-8 to four bytes. Only
233 the "Perl extended UTF-8" (the infamous 'v-strings') will encode into
238 S_is_utf8_char_slow(const U8 *s, const STRLEN len)
244 PERL_ARGS_ASSERT_IS_UTF8_CHAR_SLOW;
246 if (UTF8_IS_INVARIANT(u))
249 if (!UTF8_IS_START(u))
252 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
258 u = NATIVE_TO_UTF(u);
260 u &= UTF_START_MASK(len);
264 if (!UTF8_IS_CONTINUATION(*s))
266 uv = UTF8_ACCUMULATE(uv, *s);
273 if ((STRLEN)UNISKIP(uv) < len)
280 =for apidoc is_utf8_char
282 Tests if some arbitrary number of bytes begins in a valid UTF-8
283 character. Note that an INVARIANT (i.e. ASCII on non-EBCDIC machines)
284 character is a valid UTF-8 character. The actual number of bytes in the UTF-8
285 character will be returned if it is valid, otherwise 0.
289 Perl_is_utf8_char(const U8 *s)
291 const STRLEN len = UTF8SKIP(s);
293 PERL_ARGS_ASSERT_IS_UTF8_CHAR;
295 if (IS_UTF8_CHAR_FAST(len))
296 return IS_UTF8_CHAR(s, len) ? len : 0;
297 #endif /* #ifdef IS_UTF8_CHAR */
298 return is_utf8_char_slow(s, len);
303 =for apidoc is_utf8_string
305 Returns true if first C<len> bytes of the given string form a valid
306 UTF-8 string, false otherwise. Note that 'a valid UTF-8 string' does
307 not mean 'a string that contains code points above 0x7F encoded in UTF-8'
308 because a valid ASCII string is a valid UTF-8 string.
310 See also is_ascii_string(), is_utf8_string_loclen(), and is_utf8_string_loc().
316 Perl_is_utf8_string(const U8 *s, STRLEN len)
318 const U8* const send = s + (len ? len : strlen((const char *)s));
321 PERL_ARGS_ASSERT_IS_UTF8_STRING;
325 /* Inline the easy bits of is_utf8_char() here for speed... */
326 if (UTF8_IS_INVARIANT(*x))
328 else if (!UTF8_IS_START(*x))
331 /* ... and call is_utf8_char() only if really needed. */
334 if (IS_UTF8_CHAR_FAST(c)) {
335 if (!IS_UTF8_CHAR(x, c))
339 c = is_utf8_char_slow(x, c);
342 #endif /* #ifdef IS_UTF8_CHAR */
357 Implemented as a macro in utf8.h
359 =for apidoc is_utf8_string_loc
361 Like is_utf8_string() but stores the location of the failure (in the
362 case of "utf8ness failure") or the location s+len (in the case of
363 "utf8ness success") in the C<ep>.
365 See also is_utf8_string_loclen() and is_utf8_string().
367 =for apidoc is_utf8_string_loclen
369 Like is_utf8_string() but stores the location of the failure (in the
370 case of "utf8ness failure") or the location s+len (in the case of
371 "utf8ness success") in the C<ep>, and the number of UTF-8
372 encoded characters in the C<el>.
374 See also is_utf8_string_loc() and is_utf8_string().
380 Perl_is_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
382 const U8* const send = s + (len ? len : strlen((const char *)s));
387 PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN;
390 /* Inline the easy bits of is_utf8_char() here for speed... */
391 if (UTF8_IS_INVARIANT(*x))
393 else if (!UTF8_IS_START(*x))
396 /* ... and call is_utf8_char() only if really needed. */
399 if (IS_UTF8_CHAR_FAST(c)) {
400 if (!IS_UTF8_CHAR(x, c))
403 c = is_utf8_char_slow(x, c);
406 #endif /* #ifdef IS_UTF8_CHAR */
425 =for apidoc utf8n_to_uvuni
427 Bottom level UTF-8 decode routine.
428 Returns the Unicode code point value of the first character in the string C<s>
429 which is assumed to be in UTF-8 encoding and no longer than C<curlen>;
430 C<retlen> will be set to the length, in bytes, of that character.
432 If C<s> does not point to a well-formed UTF-8 character, the behaviour
433 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
434 it is assumed that the caller will raise a warning, and this function
435 will silently just set C<retlen> to C<-1> and return zero. If the
436 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
437 malformations will be given, C<retlen> will be set to the expected
438 length of the UTF-8 character in bytes, and zero will be returned.
440 The C<flags> can also contain various flags to allow deviations from
441 the strict UTF-8 encoding (see F<utf8.h>).
443 Most code should use utf8_to_uvchr() rather than call this directly.
449 Perl_utf8n_to_uvuni(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
452 const U8 * const s0 = s;
455 const bool dowarn = ckWARN_d(WARN_UTF8);
456 const UV startbyte = *s;
457 STRLEN expectlen = 0;
461 PERL_ARGS_ASSERT_UTF8N_TO_UVUNI;
463 /* This list is a superset of the UTF8_ALLOW_XXX. BUT it isn't, eg SUPER missing XXX */
465 #define UTF8_WARN_EMPTY 1
466 #define UTF8_WARN_CONTINUATION 2
467 #define UTF8_WARN_NON_CONTINUATION 3
468 #define UTF8_WARN_FE_FF 4
469 #define UTF8_WARN_SHORT 5
470 #define UTF8_WARN_OVERFLOW 6
471 #define UTF8_WARN_SURROGATE 7
472 #define UTF8_WARN_LONG 8
473 #define UTF8_WARN_FFFF 9 /* Also FFFE. */
476 !(flags & UTF8_ALLOW_EMPTY)) {
477 warning = UTF8_WARN_EMPTY;
481 if (UTF8_IS_INVARIANT(uv)) {
484 return (UV) (NATIVE_TO_UTF(*s));
487 if (UTF8_IS_CONTINUATION(uv) &&
488 !(flags & UTF8_ALLOW_CONTINUATION)) {
489 warning = UTF8_WARN_CONTINUATION;
493 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
494 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
495 warning = UTF8_WARN_NON_CONTINUATION;
500 uv = NATIVE_TO_UTF(uv);
502 if ((uv == 0xfe || uv == 0xff) &&
503 !(flags & UTF8_ALLOW_FE_FF)) {
504 warning = UTF8_WARN_FE_FF;
509 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
510 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
511 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
512 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
514 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
515 else { len = 7; uv &= 0x01; }
517 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
518 else if (!(uv & 0x01)) { len = 7; uv = 0; }
519 else { len = 13; uv = 0; } /* whoa! */
527 if ((curlen < expectlen) &&
528 !(flags & UTF8_ALLOW_SHORT)) {
529 warning = UTF8_WARN_SHORT;
538 if (!UTF8_IS_CONTINUATION(*s) &&
539 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
541 warning = UTF8_WARN_NON_CONTINUATION;
545 uv = UTF8_ACCUMULATE(uv, *s);
547 /* These cannot be allowed. */
549 if (expectlen != 13 && !(flags & UTF8_ALLOW_LONG)) {
550 warning = UTF8_WARN_LONG;
554 else { /* uv < ouv */
555 /* This cannot be allowed. */
556 warning = UTF8_WARN_OVERFLOW;
564 if (UNICODE_IS_SURROGATE(uv) &&
565 !(flags & UTF8_ALLOW_SURROGATE)) {
566 warning = UTF8_WARN_SURROGATE;
568 } else if ((expectlen > (STRLEN)UNISKIP(uv)) &&
569 !(flags & UTF8_ALLOW_LONG)) {
570 warning = UTF8_WARN_LONG;
572 } else if (UNICODE_IS_ILLEGAL(uv) &&
573 !(flags & UTF8_ALLOW_FFFF)) {
574 warning = UTF8_WARN_FFFF;
582 if (flags & UTF8_CHECK_ONLY) {
584 *retlen = ((STRLEN) -1);
589 if (warning == UTF8_WARN_FFFF) {
590 sv = newSVpvs_flags("Unicode non-character ", SVs_TEMP);
591 Perl_sv_catpvf(aTHX_ sv, "0x%04"UVxf" is illegal for interchange", uv);
594 sv = newSVpvs_flags("Malformed UTF-8 character ", SVs_TEMP);
597 case 0: /* Intentionally empty. */ break;
598 case UTF8_WARN_EMPTY:
599 sv_catpvs(sv, "(empty string)");
601 case UTF8_WARN_CONTINUATION:
602 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
604 case UTF8_WARN_NON_CONTINUATION:
606 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
607 (UV)s[1], startbyte);
609 const int len = (int)(s-s0);
610 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
611 (UV)s[1], len, len > 1 ? "s" : "", startbyte, (int)expectlen);
615 case UTF8_WARN_FE_FF:
616 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
618 case UTF8_WARN_SHORT:
619 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
620 (int)curlen, curlen == 1 ? "" : "s", (int)expectlen, startbyte);
621 expectlen = curlen; /* distance for caller to skip */
623 case UTF8_WARN_OVERFLOW:
624 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
627 case UTF8_WARN_SURROGATE:
628 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
631 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
632 (int)expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
635 sv_catpvs(sv, "(unknown reason)");
641 const char * const s = SvPVX_const(sv);
644 Perl_warner(aTHX_ packWARN(WARN_UTF8),
645 "%s in %s", s, OP_DESC(PL_op));
647 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
652 *retlen = expectlen ? expectlen : len;
658 =for apidoc utf8_to_uvchr
660 Returns the native character value of the first character in the string C<s>
661 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
662 length, in bytes, of that character.
664 If C<s> does not point to a well-formed UTF-8 character, zero is
665 returned and retlen is set, if possible, to -1.
671 Perl_utf8_to_uvchr(pTHX_ const U8 *s, STRLEN *retlen)
673 PERL_ARGS_ASSERT_UTF8_TO_UVCHR;
675 return utf8n_to_uvchr(s, UTF8_MAXBYTES, retlen,
676 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
680 =for apidoc utf8_to_uvuni
682 Returns the Unicode code point of the first character in the string C<s>
683 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
684 length, in bytes, of that character.
686 This function should only be used when the returned UV is considered
687 an index into the Unicode semantic tables (e.g. swashes).
689 If C<s> does not point to a well-formed UTF-8 character, zero is
690 returned and retlen is set, if possible, to -1.
696 Perl_utf8_to_uvuni(pTHX_ const U8 *s, STRLEN *retlen)
698 PERL_ARGS_ASSERT_UTF8_TO_UVUNI;
700 /* Call the low level routine asking for checks */
701 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXBYTES, retlen,
702 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
706 =for apidoc utf8_length
708 Return the length of the UTF-8 char encoded string C<s> in characters.
709 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
710 up past C<e>, croaks.
716 Perl_utf8_length(pTHX_ const U8 *s, const U8 *e)
721 PERL_ARGS_ASSERT_UTF8_LENGTH;
723 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
724 * the bitops (especially ~) can create illegal UTF-8.
725 * In other words: in Perl UTF-8 is not just for Unicode. */
728 goto warn_and_return;
730 if (!UTF8_IS_INVARIANT(*s))
741 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
742 "%s in %s", unees, OP_DESC(PL_op));
744 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8), unees);
751 =for apidoc utf8_distance
753 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
756 WARNING: use only if you *know* that the pointers point inside the
763 Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
765 PERL_ARGS_ASSERT_UTF8_DISTANCE;
767 return (a < b) ? -1 * (IV) utf8_length(a, b) : (IV) utf8_length(b, a);
773 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
776 WARNING: do not use the following unless you *know* C<off> is within
777 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
778 on the first byte of character or just after the last byte of a character.
784 Perl_utf8_hop(pTHX_ const U8 *s, I32 off)
786 PERL_ARGS_ASSERT_UTF8_HOP;
789 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
790 * the bitops (especially ~) can create illegal UTF-8.
791 * In other words: in Perl UTF-8 is not just for Unicode. */
800 while (UTF8_IS_CONTINUATION(*s))
808 =for apidoc bytes_cmp_utf8
810 Compares the sequence of characters (stored as octets) in b, blen with the
811 sequence of characters (stored as UTF-8) in u, ulen. Returns 0 if they are
812 equal, -1 or -2 if the first string is less than the second string, +1 or +2
813 if the first string is greater than the second string.
815 -1 or +1 is returned if the shorter string was identical to the start of the
816 longer string. -2 or +2 is returned if the was a difference between characters
823 Perl_bytes_cmp_utf8(pTHX_ const U8 *b, STRLEN blen, const U8 *u, STRLEN ulen)
825 const U8 *const bend = b + blen;
826 const U8 *const uend = u + ulen;
828 PERL_ARGS_ASSERT_BYTES_CMP_UTF8;
832 while (b < bend && u < uend) {
834 if (!UTF8_IS_INVARIANT(c)) {
835 if (UTF8_IS_DOWNGRADEABLE_START(c)) {
838 if (UTF8_IS_CONTINUATION(c1)) {
839 c = UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(c, c1));
841 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
842 "Malformed UTF-8 character "
843 "(unexpected non-continuation byte 0x%02x"
844 ", immediately after start byte 0x%02x)"
845 /* Dear diag.t, it's in the pod. */
848 PL_op ? OP_DESC(PL_op) : "");
853 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
854 "%s in %s", unees, OP_DESC(PL_op));
856 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8), unees);
857 return -2; /* Really want to return undef :-) */
864 return *b < c ? -2 : +2;
869 if (b == bend && u == uend)
872 return b < bend ? +1 : -1;
876 =for apidoc utf8_to_bytes
878 Converts a string C<s> of length C<len> from UTF-8 into native byte encoding.
879 Unlike C<bytes_to_utf8>, this over-writes the original string, and
880 updates len to contain the new length.
881 Returns zero on failure, setting C<len> to -1.
883 If you need a copy of the string, see C<bytes_from_utf8>.
889 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
892 U8 * const send = s + *len;
895 PERL_ARGS_ASSERT_UTF8_TO_BYTES;
897 /* ensure valid UTF-8 and chars < 256 before updating string */
901 if (!UTF8_IS_INVARIANT(c) &&
902 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
903 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
904 *len = ((STRLEN) -1);
912 *d++ = (U8)utf8_to_uvchr(s, &ulen);
921 =for apidoc bytes_from_utf8
923 Converts a string C<s> of length C<len> from UTF-8 into native byte encoding.
924 Unlike C<utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
925 the newly-created string, and updates C<len> to contain the new
926 length. Returns the original string if no conversion occurs, C<len>
927 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
928 0 if C<s> is converted or consisted entirely of characters that are invariant
929 in utf8 (i.e., US-ASCII on non-EBCDIC machines).
935 Perl_bytes_from_utf8(pTHX_ const U8 *s, STRLEN *len, bool *is_utf8)
942 PERL_ARGS_ASSERT_BYTES_FROM_UTF8;
948 /* ensure valid UTF-8 and chars < 256 before converting string */
949 for (send = s + *len; s < send;) {
951 if (!UTF8_IS_INVARIANT(c)) {
952 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
953 (c = *s++) && UTF8_IS_CONTINUATION(c))
962 Newx(d, (*len) - count + 1, U8);
963 s = start; start = d;
966 if (!UTF8_IS_INVARIANT(c)) {
967 /* Then it is two-byte encoded */
968 c = UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(c, *s++));
978 =for apidoc bytes_to_utf8
980 Converts a string C<s> of length C<len> from the native encoding into UTF-8.
981 Returns a pointer to the newly-created string, and sets C<len> to
982 reflect the new length.
984 A NUL character will be written after the end of the string.
986 If you want to convert to UTF-8 from encodings other than
987 the native (Latin1 or EBCDIC),
988 see sv_recode_to_utf8().
994 Perl_bytes_to_utf8(pTHX_ const U8 *s, STRLEN *len)
996 const U8 * const send = s + (*len);
1000 PERL_ARGS_ASSERT_BYTES_TO_UTF8;
1001 PERL_UNUSED_CONTEXT;
1003 Newx(d, (*len) * 2 + 1, U8);
1007 const UV uv = NATIVE_TO_ASCII(*s++);
1008 if (UNI_IS_INVARIANT(uv))
1009 *d++ = (U8)UTF_TO_NATIVE(uv);
1011 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
1012 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
1021 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
1023 * Destination must be pre-extended to 3/2 source. Do not use in-place.
1024 * We optimize for native, for obvious reasons. */
1027 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
1032 PERL_ARGS_ASSERT_UTF16_TO_UTF8;
1035 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %"UVuf, (UV)bytelen);
1040 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
1044 *d++ = UNI_TO_NATIVE(uv);
1051 *d++ = (U8)(( uv >> 6) | 0xc0);
1052 *d++ = (U8)(( uv & 0x3f) | 0x80);
1055 if (uv >= 0xd800 && uv <= 0xdbff) { /* surrogates */
1057 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
1059 UV low = (p[0] << 8) + p[1];
1061 if (low < 0xdc00 || low > 0xdfff)
1062 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
1063 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
1065 } else if (uv >= 0xdc00 && uv <= 0xdfff) {
1066 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
1069 *d++ = (U8)(( uv >> 12) | 0xe0);
1070 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
1071 *d++ = (U8)(( uv & 0x3f) | 0x80);
1075 *d++ = (U8)(( uv >> 18) | 0xf0);
1076 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
1077 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
1078 *d++ = (U8)(( uv & 0x3f) | 0x80);
1082 *newlen = d - dstart;
1086 /* Note: this one is slightly destructive of the source. */
1089 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
1092 U8* const send = s + bytelen;
1094 PERL_ARGS_ASSERT_UTF16_TO_UTF8_REVERSED;
1097 Perl_croak(aTHX_ "panic: utf16_to_utf8_reversed: odd bytelen %"UVuf,
1101 const U8 tmp = s[0];
1106 return utf16_to_utf8(p, d, bytelen, newlen);
1109 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
1112 Perl_is_uni_alnum(pTHX_ UV c)
1114 U8 tmpbuf[UTF8_MAXBYTES+1];
1115 uvchr_to_utf8(tmpbuf, c);
1116 return is_utf8_alnum(tmpbuf);
1120 Perl_is_uni_idfirst(pTHX_ UV c)
1122 U8 tmpbuf[UTF8_MAXBYTES+1];
1123 uvchr_to_utf8(tmpbuf, c);
1124 return is_utf8_idfirst(tmpbuf);
1128 Perl_is_uni_alpha(pTHX_ UV c)
1130 U8 tmpbuf[UTF8_MAXBYTES+1];
1131 uvchr_to_utf8(tmpbuf, c);
1132 return is_utf8_alpha(tmpbuf);
1136 Perl_is_uni_ascii(pTHX_ UV c)
1138 U8 tmpbuf[UTF8_MAXBYTES+1];
1139 uvchr_to_utf8(tmpbuf, c);
1140 return is_utf8_ascii(tmpbuf);
1144 Perl_is_uni_space(pTHX_ UV c)
1146 U8 tmpbuf[UTF8_MAXBYTES+1];
1147 uvchr_to_utf8(tmpbuf, c);
1148 return is_utf8_space(tmpbuf);
1152 Perl_is_uni_digit(pTHX_ UV c)
1154 U8 tmpbuf[UTF8_MAXBYTES+1];
1155 uvchr_to_utf8(tmpbuf, c);
1156 return is_utf8_digit(tmpbuf);
1160 Perl_is_uni_upper(pTHX_ UV c)
1162 U8 tmpbuf[UTF8_MAXBYTES+1];
1163 uvchr_to_utf8(tmpbuf, c);
1164 return is_utf8_upper(tmpbuf);
1168 Perl_is_uni_lower(pTHX_ UV c)
1170 U8 tmpbuf[UTF8_MAXBYTES+1];
1171 uvchr_to_utf8(tmpbuf, c);
1172 return is_utf8_lower(tmpbuf);
1176 Perl_is_uni_cntrl(pTHX_ UV c)
1178 U8 tmpbuf[UTF8_MAXBYTES+1];
1179 uvchr_to_utf8(tmpbuf, c);
1180 return is_utf8_cntrl(tmpbuf);
1184 Perl_is_uni_graph(pTHX_ UV c)
1186 U8 tmpbuf[UTF8_MAXBYTES+1];
1187 uvchr_to_utf8(tmpbuf, c);
1188 return is_utf8_graph(tmpbuf);
1192 Perl_is_uni_print(pTHX_ UV c)
1194 U8 tmpbuf[UTF8_MAXBYTES+1];
1195 uvchr_to_utf8(tmpbuf, c);
1196 return is_utf8_print(tmpbuf);
1200 Perl_is_uni_punct(pTHX_ UV c)
1202 U8 tmpbuf[UTF8_MAXBYTES+1];
1203 uvchr_to_utf8(tmpbuf, c);
1204 return is_utf8_punct(tmpbuf);
1208 Perl_is_uni_xdigit(pTHX_ UV c)
1210 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1211 uvchr_to_utf8(tmpbuf, c);
1212 return is_utf8_xdigit(tmpbuf);
1216 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
1218 PERL_ARGS_ASSERT_TO_UNI_UPPER;
1220 uvchr_to_utf8(p, c);
1221 return to_utf8_upper(p, p, lenp);
1225 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
1227 PERL_ARGS_ASSERT_TO_UNI_TITLE;
1229 uvchr_to_utf8(p, c);
1230 return to_utf8_title(p, p, lenp);
1234 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
1236 PERL_ARGS_ASSERT_TO_UNI_LOWER;
1238 uvchr_to_utf8(p, c);
1239 return to_utf8_lower(p, p, lenp);
1243 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1245 PERL_ARGS_ASSERT_TO_UNI_FOLD;
1247 uvchr_to_utf8(p, c);
1248 return to_utf8_fold(p, p, lenp);
1251 /* for now these all assume no locale info available for Unicode > 255 */
1254 Perl_is_uni_alnum_lc(pTHX_ UV c)
1256 return is_uni_alnum(c); /* XXX no locale support yet */
1260 Perl_is_uni_idfirst_lc(pTHX_ UV c)
1262 return is_uni_idfirst(c); /* XXX no locale support yet */
1266 Perl_is_uni_alpha_lc(pTHX_ UV c)
1268 return is_uni_alpha(c); /* XXX no locale support yet */
1272 Perl_is_uni_ascii_lc(pTHX_ UV c)
1274 return is_uni_ascii(c); /* XXX no locale support yet */
1278 Perl_is_uni_space_lc(pTHX_ UV c)
1280 return is_uni_space(c); /* XXX no locale support yet */
1284 Perl_is_uni_digit_lc(pTHX_ UV c)
1286 return is_uni_digit(c); /* XXX no locale support yet */
1290 Perl_is_uni_upper_lc(pTHX_ UV c)
1292 return is_uni_upper(c); /* XXX no locale support yet */
1296 Perl_is_uni_lower_lc(pTHX_ UV c)
1298 return is_uni_lower(c); /* XXX no locale support yet */
1302 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1304 return is_uni_cntrl(c); /* XXX no locale support yet */
1308 Perl_is_uni_graph_lc(pTHX_ UV c)
1310 return is_uni_graph(c); /* XXX no locale support yet */
1314 Perl_is_uni_print_lc(pTHX_ UV c)
1316 return is_uni_print(c); /* XXX no locale support yet */
1320 Perl_is_uni_punct_lc(pTHX_ UV c)
1322 return is_uni_punct(c); /* XXX no locale support yet */
1326 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1328 return is_uni_xdigit(c); /* XXX no locale support yet */
1332 Perl_to_uni_upper_lc(pTHX_ U32 c)
1334 /* XXX returns only the first character -- do not use XXX */
1335 /* XXX no locale support yet */
1337 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1338 return (U32)to_uni_upper(c, tmpbuf, &len);
1342 Perl_to_uni_title_lc(pTHX_ U32 c)
1344 /* XXX returns only the first character XXX -- do not use XXX */
1345 /* XXX no locale support yet */
1347 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1348 return (U32)to_uni_title(c, tmpbuf, &len);
1352 Perl_to_uni_lower_lc(pTHX_ U32 c)
1354 /* XXX returns only the first character -- do not use XXX */
1355 /* XXX no locale support yet */
1357 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1358 return (U32)to_uni_lower(c, tmpbuf, &len);
1362 S_is_utf8_common(pTHX_ const U8 *const p, SV **swash,
1363 const char *const swashname)
1367 PERL_ARGS_ASSERT_IS_UTF8_COMMON;
1369 if (!is_utf8_char(p))
1372 *swash = swash_init("utf8", swashname, &PL_sv_undef, 1, 0);
1373 return swash_fetch(*swash, p, TRUE) != 0;
1377 Perl_is_utf8_alnum(pTHX_ const U8 *p)
1381 PERL_ARGS_ASSERT_IS_UTF8_ALNUM;
1383 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1384 * descendant of isalnum(3), in other words, it doesn't
1385 * contain the '_'. --jhi */
1386 return is_utf8_common(p, &PL_utf8_alnum, "IsWord");
1390 Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
1394 PERL_ARGS_ASSERT_IS_UTF8_IDFIRST;
1398 /* is_utf8_idstart would be more logical. */
1399 return is_utf8_common(p, &PL_utf8_idstart, "IdStart");
1403 Perl_is_utf8_idcont(pTHX_ const U8 *p)
1407 PERL_ARGS_ASSERT_IS_UTF8_IDCONT;
1411 return is_utf8_common(p, &PL_utf8_idcont, "IdContinue");
1415 Perl_is_utf8_alpha(pTHX_ const U8 *p)
1419 PERL_ARGS_ASSERT_IS_UTF8_ALPHA;
1421 return is_utf8_common(p, &PL_utf8_alpha, "IsAlpha");
1425 Perl_is_utf8_ascii(pTHX_ const U8 *p)
1429 PERL_ARGS_ASSERT_IS_UTF8_ASCII;
1431 return is_utf8_common(p, &PL_utf8_ascii, "IsAscii");
1435 Perl_is_utf8_space(pTHX_ const U8 *p)
1439 PERL_ARGS_ASSERT_IS_UTF8_SPACE;
1441 return is_utf8_common(p, &PL_utf8_space, "IsSpacePerl");
1445 Perl_is_utf8_perl_space(pTHX_ const U8 *p)
1449 PERL_ARGS_ASSERT_IS_UTF8_PERL_SPACE;
1451 return is_utf8_common(p, &PL_utf8_perl_space, "IsPerlSpace");
1455 Perl_is_utf8_perl_word(pTHX_ const U8 *p)
1459 PERL_ARGS_ASSERT_IS_UTF8_PERL_WORD;
1461 return is_utf8_common(p, &PL_utf8_perl_word, "IsPerlWord");
1465 Perl_is_utf8_digit(pTHX_ const U8 *p)
1469 PERL_ARGS_ASSERT_IS_UTF8_DIGIT;
1471 return is_utf8_common(p, &PL_utf8_digit, "IsDigit");
1475 Perl_is_utf8_posix_digit(pTHX_ const U8 *p)
1479 PERL_ARGS_ASSERT_IS_UTF8_POSIX_DIGIT;
1481 return is_utf8_common(p, &PL_utf8_posix_digit, "IsPosixDigit");
1485 Perl_is_utf8_upper(pTHX_ const U8 *p)
1489 PERL_ARGS_ASSERT_IS_UTF8_UPPER;
1491 return is_utf8_common(p, &PL_utf8_upper, "IsUppercase");
1495 Perl_is_utf8_lower(pTHX_ const U8 *p)
1499 PERL_ARGS_ASSERT_IS_UTF8_LOWER;
1501 return is_utf8_common(p, &PL_utf8_lower, "IsLowercase");
1505 Perl_is_utf8_cntrl(pTHX_ const U8 *p)
1509 PERL_ARGS_ASSERT_IS_UTF8_CNTRL;
1511 return is_utf8_common(p, &PL_utf8_cntrl, "IsCntrl");
1515 Perl_is_utf8_graph(pTHX_ const U8 *p)
1519 PERL_ARGS_ASSERT_IS_UTF8_GRAPH;
1521 return is_utf8_common(p, &PL_utf8_graph, "IsGraph");
1525 Perl_is_utf8_print(pTHX_ const U8 *p)
1529 PERL_ARGS_ASSERT_IS_UTF8_PRINT;
1531 return is_utf8_common(p, &PL_utf8_print, "IsPrint");
1535 Perl_is_utf8_punct(pTHX_ const U8 *p)
1539 PERL_ARGS_ASSERT_IS_UTF8_PUNCT;
1541 return is_utf8_common(p, &PL_utf8_punct, "IsPunct");
1545 Perl_is_utf8_xdigit(pTHX_ const U8 *p)
1549 PERL_ARGS_ASSERT_IS_UTF8_XDIGIT;
1551 return is_utf8_common(p, &PL_utf8_xdigit, "IsXDigit");
1555 Perl_is_utf8_mark(pTHX_ const U8 *p)
1559 PERL_ARGS_ASSERT_IS_UTF8_MARK;
1561 return is_utf8_common(p, &PL_utf8_mark, "IsM");
1565 Perl_is_utf8_X_begin(pTHX_ const U8 *p)
1569 PERL_ARGS_ASSERT_IS_UTF8_X_BEGIN;
1571 return is_utf8_common(p, &PL_utf8_X_begin, "_X_Begin");
1575 Perl_is_utf8_X_extend(pTHX_ const U8 *p)
1579 PERL_ARGS_ASSERT_IS_UTF8_X_EXTEND;
1581 return is_utf8_common(p, &PL_utf8_X_extend, "_X_Extend");
1585 Perl_is_utf8_X_prepend(pTHX_ const U8 *p)
1589 PERL_ARGS_ASSERT_IS_UTF8_X_PREPEND;
1591 return is_utf8_common(p, &PL_utf8_X_prepend, "GCB=Prepend");
1595 Perl_is_utf8_X_non_hangul(pTHX_ const U8 *p)
1599 PERL_ARGS_ASSERT_IS_UTF8_X_NON_HANGUL;
1601 return is_utf8_common(p, &PL_utf8_X_non_hangul, "HST=Not_Applicable");
1605 Perl_is_utf8_X_L(pTHX_ const U8 *p)
1609 PERL_ARGS_ASSERT_IS_UTF8_X_L;
1611 return is_utf8_common(p, &PL_utf8_X_L, "GCB=L");
1615 Perl_is_utf8_X_LV(pTHX_ const U8 *p)
1619 PERL_ARGS_ASSERT_IS_UTF8_X_LV;
1621 return is_utf8_common(p, &PL_utf8_X_LV, "GCB=LV");
1625 Perl_is_utf8_X_LVT(pTHX_ const U8 *p)
1629 PERL_ARGS_ASSERT_IS_UTF8_X_LVT;
1631 return is_utf8_common(p, &PL_utf8_X_LVT, "GCB=LVT");
1635 Perl_is_utf8_X_T(pTHX_ const U8 *p)
1639 PERL_ARGS_ASSERT_IS_UTF8_X_T;
1641 return is_utf8_common(p, &PL_utf8_X_T, "GCB=T");
1645 Perl_is_utf8_X_V(pTHX_ const U8 *p)
1649 PERL_ARGS_ASSERT_IS_UTF8_X_V;
1651 return is_utf8_common(p, &PL_utf8_X_V, "GCB=V");
1655 Perl_is_utf8_X_LV_LVT_V(pTHX_ const U8 *p)
1659 PERL_ARGS_ASSERT_IS_UTF8_X_LV_LVT_V;
1661 return is_utf8_common(p, &PL_utf8_X_LV_LVT_V, "_X_LV_LVT_V");
1665 =for apidoc to_utf8_case
1667 The "p" contains the pointer to the UTF-8 string encoding
1668 the character that is being converted.
1670 The "ustrp" is a pointer to the character buffer to put the
1671 conversion result to. The "lenp" is a pointer to the length
1674 The "swashp" is a pointer to the swash to use.
1676 Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
1677 and loaded by SWASHNEW, using lib/utf8_heavy.pl. The special (usually,
1678 but not always, a multicharacter mapping), is tried first.
1680 The "special" is a string like "utf8::ToSpecLower", which means the
1681 hash %utf8::ToSpecLower. The access to the hash is through
1682 Perl_to_utf8_case().
1684 The "normal" is a string like "ToLower" which means the swash
1690 Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp,
1691 SV **swashp, const char *normal, const char *special)
1694 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1696 const UV uv0 = utf8_to_uvchr(p, NULL);
1697 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1698 * are necessary in EBCDIC, they are redundant no-ops
1699 * in ASCII-ish platforms, and hopefully optimized away. */
1700 const UV uv1 = NATIVE_TO_UNI(uv0);
1702 PERL_ARGS_ASSERT_TO_UTF8_CASE;
1704 uvuni_to_utf8(tmpbuf, uv1);
1706 if (!*swashp) /* load on-demand */
1707 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1708 /* This is the beginnings of a skeleton of code to read the info section
1709 * that is in all the swashes in case we ever want to do that, so one can
1710 * read things whose maps aren't code points, and whose default if missing
1711 * is not to the code point itself. This was just to see if it actually
1712 * worked. Details on what the possibilities are are in perluniprops.pod
1713 HV * const hv = get_hv("utf8::SwashInfo", 0);
1716 svp = hv_fetch(hv, (const char*)normal, strlen(normal), FALSE);
1719 HV * const this_hash = SvRV(*svp);
1720 svp = hv_fetch(this_hash, "type", strlen("type"), FALSE);
1721 s = SvPV_const(*svp, len);
1726 /* It might be "special" (sometimes, but not always,
1727 * a multicharacter mapping) */
1728 HV * const hv = get_hv(special, 0);
1732 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1736 s = SvPV_const(*svp, len);
1738 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
1741 /* If we have EBCDIC we need to remap the characters
1742 * since any characters in the low 256 are Unicode
1743 * code points, not EBCDIC. */
1744 U8 *t = (U8*)s, *tend = t + len, *d;
1751 const UV c = utf8_to_uvchr(t, &tlen);
1753 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1762 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1767 Copy(tmpbuf, ustrp, len, U8);
1769 Copy(s, ustrp, len, U8);
1775 if (!len && *swashp) {
1776 const UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1779 /* It was "normal" (a single character mapping). */
1780 const UV uv3 = UNI_TO_NATIVE(uv2);
1781 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
1785 if (!len) /* Neither: just copy. In other words, there was no mapping
1786 defined, which means that the code point maps to itself */
1787 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1792 return len ? utf8_to_uvchr(ustrp, 0) : 0;
1796 =for apidoc to_utf8_upper
1798 Convert the UTF-8 encoded character at p to its uppercase version and
1799 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1800 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1801 the uppercase version may be longer than the original character.
1803 The first character of the uppercased version is returned
1804 (but note, as explained above, that there may be more.)
1809 Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1813 PERL_ARGS_ASSERT_TO_UTF8_UPPER;
1815 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1816 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1820 =for apidoc to_utf8_title
1822 Convert the UTF-8 encoded character at p to its titlecase version and
1823 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1824 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1825 titlecase version may be longer than the original character.
1827 The first character of the titlecased version is returned
1828 (but note, as explained above, that there may be more.)
1833 Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1837 PERL_ARGS_ASSERT_TO_UTF8_TITLE;
1839 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1840 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1844 =for apidoc to_utf8_lower
1846 Convert the UTF-8 encoded character at p to its lowercase version and
1847 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1848 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1849 lowercase version may be longer than the original character.
1851 The first character of the lowercased version is returned
1852 (but note, as explained above, that there may be more.)
1857 Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1861 PERL_ARGS_ASSERT_TO_UTF8_LOWER;
1863 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1864 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1868 =for apidoc to_utf8_fold
1870 Convert the UTF-8 encoded character at p to its foldcase version and
1871 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1872 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1873 foldcase version may be longer than the original character (up to
1876 The first character of the foldcased version is returned
1877 (but note, as explained above, that there may be more.)
1882 Perl_to_utf8_fold(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1886 PERL_ARGS_ASSERT_TO_UTF8_FOLD;
1888 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1889 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1893 * A "swash" is a swatch hash.
1894 * A "swatch" is a bit vector generated by utf8.c:S_swash_get().
1895 * C<pkg> is a pointer to a package name for SWASHNEW, should be "utf8".
1896 * For other parameters, see utf8::SWASHNEW in lib/utf8_heavy.pl.
1899 Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
1904 const size_t pkg_len = strlen(pkg);
1905 const size_t name_len = strlen(name);
1906 HV * const stash = gv_stashpvn(pkg, pkg_len, 0);
1909 PERL_ARGS_ASSERT_SWASH_INIT;
1911 PUSHSTACKi(PERLSI_MAGIC);
1915 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
1917 errsv_save = newSVsv(ERRSV);
1918 /* It is assumed that callers of this routine are not passing in any
1919 user derived data. */
1920 /* Need to do this after save_re_context() as it will set PL_tainted to
1921 1 while saving $1 etc (see the code after getrx: in Perl_magic_get).
1922 Even line to create errsv_save can turn on PL_tainted. */
1923 SAVEBOOL(PL_tainted);
1925 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
1928 sv_setsv(ERRSV, errsv_save);
1929 SvREFCNT_dec(errsv_save);
1935 mPUSHp(pkg, pkg_len);
1936 mPUSHp(name, name_len);
1941 errsv_save = newSVsv(ERRSV);
1942 if (call_method("SWASHNEW", G_SCALAR))
1943 retval = newSVsv(*PL_stack_sp--);
1945 retval = &PL_sv_undef;
1947 sv_setsv(ERRSV, errsv_save);
1948 SvREFCNT_dec(errsv_save);
1951 if (IN_PERL_COMPILETIME) {
1952 CopHINTS_set(PL_curcop, PL_hints);
1954 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
1956 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
1958 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1964 /* This API is wrong for special case conversions since we may need to
1965 * return several Unicode characters for a single Unicode character
1966 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1967 * the lower-level routine, and it is similarly broken for returning
1968 * multiple values. --jhi */
1969 /* Now SWASHGET is recasted into S_swash_get in this file. */
1972 * Returns the value of property/mapping C<swash> for the first character
1973 * of the string C<ptr>. If C<do_utf8> is true, the string C<ptr> is
1974 * assumed to be in utf8. If C<do_utf8> is false, the string C<ptr> is
1975 * assumed to be in native 8-bit encoding. Caches the swatch in C<swash>.
1978 Perl_swash_fetch(pTHX_ SV *swash, const U8 *ptr, bool do_utf8)
1981 HV *const hv = MUTABLE_HV(SvRV(swash));
1986 const U8 *tmps = NULL;
1990 const UV c = NATIVE_TO_ASCII(*ptr);
1992 PERL_ARGS_ASSERT_SWASH_FETCH;
1994 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1995 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
1996 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
1999 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
2000 * then the "swatch" is a vec() for all the chars which start
2002 * So the key in the hash (klen) is length of encoded char -1
2004 klen = UTF8SKIP(ptr) - 1;
2008 /* If char is invariant then swatch is for all the invariant chars
2009 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
2011 needents = UTF_CONTINUATION_MARK;
2012 off = NATIVE_TO_UTF(ptr[klen]);
2015 /* If char is encoded then swatch is for the prefix */
2016 needents = (1 << UTF_ACCUMULATION_SHIFT);
2017 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
2021 * This single-entry cache saves about 1/3 of the utf8 overhead in test
2022 * suite. (That is, only 7-8% overall over just a hash cache. Still,
2023 * it's nothing to sniff at.) Pity we usually come through at least
2024 * two function calls to get here...
2026 * NB: this code assumes that swatches are never modified, once generated!
2029 if (hv == PL_last_swash_hv &&
2030 klen == PL_last_swash_klen &&
2031 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
2033 tmps = PL_last_swash_tmps;
2034 slen = PL_last_swash_slen;
2037 /* Try our second-level swatch cache, kept in a hash. */
2038 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
2040 /* If not cached, generate it via swash_get */
2041 if (!svp || !SvPOK(*svp)
2042 || !(tmps = (const U8*)SvPV_const(*svp, slen))) {
2043 /* We use utf8n_to_uvuni() as we want an index into
2044 Unicode tables, not a native character number.
2046 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
2048 0 : UTF8_ALLOW_ANY);
2049 swatch = swash_get(swash,
2050 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
2051 (klen) ? (code_point & ~(needents - 1)) : 0,
2054 if (IN_PERL_COMPILETIME)
2055 CopHINTS_set(PL_curcop, PL_hints);
2057 svp = hv_store(hv, (const char *)ptr, klen, swatch, 0);
2059 if (!svp || !(tmps = (U8*)SvPV(*svp, slen))
2060 || (slen << 3) < needents)
2061 Perl_croak(aTHX_ "panic: swash_fetch got improper swatch");
2064 PL_last_swash_hv = hv;
2065 assert(klen <= sizeof(PL_last_swash_key));
2066 PL_last_swash_klen = (U8)klen;
2067 /* FIXME change interpvar.h? */
2068 PL_last_swash_tmps = (U8 *) tmps;
2069 PL_last_swash_slen = slen;
2071 Copy(ptr, PL_last_swash_key, klen, U8);
2074 switch ((int)((slen << 3) / needents)) {
2076 bit = 1 << (off & 7);
2078 return (tmps[off] & bit) != 0;
2083 return (tmps[off] << 8) + tmps[off + 1] ;
2086 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
2088 Perl_croak(aTHX_ "panic: swash_fetch got swatch of unexpected bit width");
2089 NORETURN_FUNCTION_END;
2092 /* Read a single line of the main body of the swash input text. These are of
2095 * where each number is hex. The first two numbers form the minimum and
2096 * maximum of a range, and the third is the value associated with the range.
2097 * Not all swashes should have a third number
2099 * On input: l points to the beginning of the line to be examined; it points
2100 * to somewhere in the string of the whole input text, and is
2101 * terminated by a \n or the null string terminator.
2102 * lend points to the null terminator of that string
2103 * wants_value is non-zero if the swash expects a third number
2104 * typestr is the name of the swash's mapping, like 'ToLower'
2105 * On output: *min, *max, and *val are set to the values read from the line.
2106 * returns a pointer just beyond the line examined. If there was no
2107 * valid min number on the line, returns lend+1
2111 S_swash_scan_list_line(pTHX_ U8* l, U8* const lend, UV* min, UV* max, UV* val,
2112 const bool wants_value, const U8* const typestr)
2114 const int typeto = typestr[0] == 'T' && typestr[1] == 'o';
2115 STRLEN numlen; /* Length of the number */
2116 I32 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
2118 /* nl points to the next \n in the scan */
2119 U8* const nl = (U8*)memchr(l, '\n', lend - l);
2121 /* Get the first number on the line: the range minimum */
2123 *min = grok_hex((char *)l, &numlen, &flags, NULL);
2124 if (numlen) /* If found a hex number, position past it */
2126 else if (nl) { /* Else, go handle next line, if any */
2127 return nl + 1; /* 1 is length of "\n" */
2129 else { /* Else, no next line */
2130 return lend + 1; /* to LIST's end at which \n is not found */
2133 /* The max range value follows, separated by a BLANK */
2136 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
2138 *max = grok_hex((char *)l, &numlen, &flags, NULL);
2141 else /* If no value here, it is a single element range */
2144 /* Non-binary tables have a third entry: what the first element of the
2149 flags = PERL_SCAN_SILENT_ILLDIGIT |
2150 PERL_SCAN_DISALLOW_PREFIX;
2152 *val = grok_hex((char *)l, &numlen, &flags, NULL);
2161 Perl_croak(aTHX_ "%s: illegal mapping '%s'",
2167 *val = 0; /* bits == 1, then any val should be ignored */
2169 else { /* Nothing following range min, should be single element with no
2175 Perl_croak(aTHX_ "%s: illegal mapping '%s'", typestr, l);
2179 *val = 0; /* bits == 1, then val should be ignored */
2182 /* Position to next line if any, or EOF */
2192 * Returns a swatch (a bit vector string) for a code point sequence
2193 * that starts from the value C<start> and comprises the number C<span>.
2194 * A C<swash> must be an object created by SWASHNEW (see lib/utf8_heavy.pl).
2195 * Should be used via swash_fetch, which will cache the swatch in C<swash>.
2198 S_swash_get(pTHX_ SV* swash, UV start, UV span)
2201 U8 *l, *lend, *x, *xend, *s;
2202 STRLEN lcur, xcur, scur;
2203 HV *const hv = MUTABLE_HV(SvRV(swash));
2205 /* The string containing the main body of the table */
2206 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2208 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2209 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2210 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2211 SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);
2212 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
2213 const STRLEN bits = SvUV(*bitssvp);
2214 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2215 const UV none = SvUV(*nonesvp);
2216 const UV end = start + span;
2218 PERL_ARGS_ASSERT_SWASH_GET;
2220 if (bits != 1 && bits != 8 && bits != 16 && bits != 32) {
2221 Perl_croak(aTHX_ "panic: swash_get doesn't expect bits %"UVuf,
2225 /* create and initialize $swatch */
2226 scur = octets ? (span * octets) : (span + 7) / 8;
2227 swatch = newSV(scur);
2229 s = (U8*)SvPVX(swatch);
2230 if (octets && none) {
2231 const U8* const e = s + scur;
2234 *s++ = (U8)(none & 0xff);
2235 else if (bits == 16) {
2236 *s++ = (U8)((none >> 8) & 0xff);
2237 *s++ = (U8)( none & 0xff);
2239 else if (bits == 32) {
2240 *s++ = (U8)((none >> 24) & 0xff);
2241 *s++ = (U8)((none >> 16) & 0xff);
2242 *s++ = (U8)((none >> 8) & 0xff);
2243 *s++ = (U8)( none & 0xff);
2249 (void)memzero((U8*)s, scur + 1);
2251 SvCUR_set(swatch, scur);
2252 s = (U8*)SvPVX(swatch);
2254 /* read $swash->{LIST} */
2255 l = (U8*)SvPV(*listsvp, lcur);
2259 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2260 cBOOL(octets), typestr);
2265 /* If looking for something beyond this range, go try the next one */
2272 if (!none || val < none) {
2277 for (key = min; key <= max; key++) {
2281 /* offset must be non-negative (start <= min <= key < end) */
2282 offset = octets * (key - start);
2284 s[offset] = (U8)(val & 0xff);
2285 else if (bits == 16) {
2286 s[offset ] = (U8)((val >> 8) & 0xff);
2287 s[offset + 1] = (U8)( val & 0xff);
2289 else if (bits == 32) {
2290 s[offset ] = (U8)((val >> 24) & 0xff);
2291 s[offset + 1] = (U8)((val >> 16) & 0xff);
2292 s[offset + 2] = (U8)((val >> 8) & 0xff);
2293 s[offset + 3] = (U8)( val & 0xff);
2296 if (!none || val < none)
2300 else { /* bits == 1, then val should be ignored */
2304 for (key = min; key <= max; key++) {
2305 const STRLEN offset = (STRLEN)(key - start);
2308 s[offset >> 3] |= 1 << (offset & 7);
2314 /* read $swash->{EXTRAS} */
2315 x = (U8*)SvPV(*extssvp, xcur);
2323 SV **otherbitssvp, *other;
2327 const U8 opc = *x++;
2331 nl = (U8*)memchr(x, '\n', xend - x);
2333 if (opc != '-' && opc != '+' && opc != '!' && opc != '&') {
2335 x = nl + 1; /* 1 is length of "\n" */
2339 x = xend; /* to EXTRAS' end at which \n is not found */
2346 namelen = nl - namestr;
2350 namelen = xend - namestr;
2354 othersvp = hv_fetch(hv, (char *)namestr, namelen, FALSE);
2355 otherhv = MUTABLE_HV(SvRV(*othersvp));
2356 otherbitssvp = hv_fetchs(otherhv, "BITS", FALSE);
2357 otherbits = (STRLEN)SvUV(*otherbitssvp);
2358 if (bits < otherbits)
2359 Perl_croak(aTHX_ "panic: swash_get found swatch size mismatch");
2361 /* The "other" swatch must be destroyed after. */
2362 other = swash_get(*othersvp, start, span);
2363 o = (U8*)SvPV(other, olen);
2366 Perl_croak(aTHX_ "panic: swash_get got improper swatch");
2368 s = (U8*)SvPV(swatch, slen);
2369 if (bits == 1 && otherbits == 1) {
2371 Perl_croak(aTHX_ "panic: swash_get found swatch length mismatch");
2395 STRLEN otheroctets = otherbits >> 3;
2397 U8* const send = s + slen;
2402 if (otherbits == 1) {
2403 otherval = (o[offset >> 3] >> (offset & 7)) & 1;
2407 STRLEN vlen = otheroctets;
2415 if (opc == '+' && otherval)
2416 NOOP; /* replace with otherval */
2417 else if (opc == '!' && !otherval)
2419 else if (opc == '-' && otherval)
2421 else if (opc == '&' && !otherval)
2424 s += octets; /* no replacement */
2429 *s++ = (U8)( otherval & 0xff);
2430 else if (bits == 16) {
2431 *s++ = (U8)((otherval >> 8) & 0xff);
2432 *s++ = (U8)( otherval & 0xff);
2434 else if (bits == 32) {
2435 *s++ = (U8)((otherval >> 24) & 0xff);
2436 *s++ = (U8)((otherval >> 16) & 0xff);
2437 *s++ = (U8)((otherval >> 8) & 0xff);
2438 *s++ = (U8)( otherval & 0xff);
2442 sv_free(other); /* through with it! */
2448 Perl__swash_inversion_hash(pTHX_ SV* swash)
2451 /* Subject to change or removal. For use only in one place in regexec.c
2453 * Returns a hash which is the inversion and closure of a swash mapping.
2454 * For example, consider the input lines:
2459 * The returned hash would have two keys, the utf8 for 006B and the utf8 for
2460 * 006C. The value for each key is an array. For 006C, the array would
2461 * have a two elements, the utf8 for itself, and for 004C. For 006B, there
2462 * would be three elements in its array, the utf8 for 006B, 004B and 212A.
2464 * Essentially, for any code point, it gives all the code points that map to
2465 * it, or the list of 'froms' for that point.
2467 * Currently it only looks at the main body of the swash, and ignores any
2468 * additions or deletions from other swashes */
2472 HV *const hv = MUTABLE_HV(SvRV(swash));
2474 /* The string containing the main body of the table */
2475 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2477 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2478 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2479 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2480 /*SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);*/
2481 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
2482 const STRLEN bits = SvUV(*bitssvp);
2483 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2484 const UV none = SvUV(*nonesvp);
2488 PERL_ARGS_ASSERT__SWASH_INVERSION_HASH;
2490 /* Must have at least 8 bits to get the mappings */
2491 if (bits != 8 && bits != 16 && bits != 32) {
2492 Perl_croak(aTHX_ "panic: swash_inversion_hash doesn't expect bits %"UVuf,
2496 /* read $swash->{LIST} */
2497 l = (U8*)SvPV(*listsvp, lcur);
2500 /* Go through each input line */
2504 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2505 cBOOL(octets), typestr);
2510 /* Each element in the range is to be inverted */
2511 for (inverse = min; inverse <= max; inverse++) {
2516 bool found_key = FALSE;
2518 /* The key is the inverse mapping */
2519 char key[UTF8_MAXBYTES+1];
2520 char* key_end = (char *) uvuni_to_utf8((U8*) key, val);
2521 STRLEN key_len = key_end - key;
2523 /* And the value is what the forward mapping is from. */
2524 char utf8_inverse[UTF8_MAXBYTES+1];
2525 char *utf8_inverse_end = (char *) uvuni_to_utf8((U8*) utf8_inverse, inverse);
2527 /* Get the list for the map */
2528 if ((listp = hv_fetch(ret, key, key_len, FALSE))) {
2529 list = (AV*) *listp;
2531 else { /* No entry yet for it: create one */
2533 if (! hv_store(ret, key, key_len, (SV*) list, FALSE)) {
2534 Perl_croak(aTHX_ "panic: hv_store() unexpectedly failed");
2538 for (i = 0; i < av_len(list); i++) {
2539 SV** entryp = av_fetch(list, i, FALSE);
2541 if (entryp == NULL) {
2542 Perl_croak(aTHX_ "panic: av_fetch() unexpectedly failed");
2545 if (SvCUR(entry) != key_len) {
2548 if (memEQ(key, SvPVX(entry), key_len)) {
2554 element = newSVpvn_flags(key, key_len, SVf_UTF8);
2555 av_push(list, element);
2559 /* Simply add the value to the list */
2560 element = newSVpvn_flags(utf8_inverse, utf8_inverse_end - utf8_inverse, SVf_UTF8);
2561 av_push(list, element);
2563 /* swash_get() increments the value of val for each element in the
2564 * range. That makes more compact tables possible. You can
2565 * express the capitalization, for example, of all consecutive
2566 * letters with a single line: 0061\t007A\t0041 This maps 0061 to
2567 * 0041, 0062 to 0042, etc. I (khw) have never understood 'none',
2568 * and it's not documented, and perhaps not even currently used,
2569 * but I copied the semantics from swash_get(), just in case */
2570 if (!none || val < none) {
2580 =for apidoc uvchr_to_utf8
2582 Adds the UTF-8 representation of the Native codepoint C<uv> to the end
2583 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
2584 bytes available. The return value is the pointer to the byte after the
2585 end of the new character. In other words,
2587 d = uvchr_to_utf8(d, uv);
2589 is the recommended wide native character-aware way of saying
2596 /* On ASCII machines this is normally a macro but we want a
2597 real function in case XS code wants it
2600 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
2602 PERL_ARGS_ASSERT_UVCHR_TO_UTF8;
2604 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
2608 Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
2610 PERL_ARGS_ASSERT_UVCHR_TO_UTF8_FLAGS;
2612 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
2616 =for apidoc utf8n_to_uvchr
2619 Returns the native character value of the first character in the string
2621 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
2622 length, in bytes, of that character.
2624 Allows length and flags to be passed to low level routine.
2628 /* On ASCII machines this is normally a macro but we want
2629 a real function in case XS code wants it
2632 Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen,
2635 const UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
2637 PERL_ARGS_ASSERT_UTF8N_TO_UVCHR;
2639 return UNI_TO_NATIVE(uv);
2643 =for apidoc pv_uni_display
2645 Build to the scalar dsv a displayable version of the string spv,
2646 length len, the displayable version being at most pvlim bytes long
2647 (if longer, the rest is truncated and "..." will be appended).
2649 The flags argument can have UNI_DISPLAY_ISPRINT set to display
2650 isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
2651 to display the \\[nrfta\\] as the backslashed versions (like '\n')
2652 (UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
2653 UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
2654 UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
2656 The pointer to the PV of the dsv is returned.
2660 Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
2665 PERL_ARGS_ASSERT_PV_UNI_DISPLAY;
2669 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
2671 /* This serves double duty as a flag and a character to print after
2672 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
2676 if (pvlim && SvCUR(dsv) >= pvlim) {
2680 u = utf8_to_uvchr((U8*)s, 0);
2682 const unsigned char c = (unsigned char)u & 0xFF;
2683 if (flags & UNI_DISPLAY_BACKSLASH) {
2700 const char string = ok;
2701 sv_catpvs(dsv, "\\");
2702 sv_catpvn(dsv, &string, 1);
2705 /* isPRINT() is the locale-blind version. */
2706 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
2707 const char string = c;
2708 sv_catpvn(dsv, &string, 1);
2713 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
2716 sv_catpvs(dsv, "...");
2722 =for apidoc sv_uni_display
2724 Build to the scalar dsv a displayable version of the scalar sv,
2725 the displayable version being at most pvlim bytes long
2726 (if longer, the rest is truncated and "..." will be appended).
2728 The flags argument is as in pv_uni_display().
2730 The pointer to the PV of the dsv is returned.
2735 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
2737 PERL_ARGS_ASSERT_SV_UNI_DISPLAY;
2739 return Perl_pv_uni_display(aTHX_ dsv, (const U8*)SvPVX_const(ssv),
2740 SvCUR(ssv), pvlim, flags);
2744 =for apidoc foldEQ_utf8
2746 Returns true if the leading portions of the strings s1 and s2 (either or both
2747 of which may be in UTF-8) are the same case-insensitively; false otherwise.
2748 How far into the strings to compare is determined by other input parameters.
2750 If u1 is true, the string s1 is assumed to be in UTF-8-encoded Unicode;
2751 otherwise it is assumed to be in native 8-bit encoding. Correspondingly for u2
2754 If the byte length l1 is non-zero, it says how far into s1 to check for fold
2755 equality. In other words, s1+l1 will be used as a goal to reach. The
2756 scan will not be considered to be a match unless the goal is reached, and
2757 scanning won't continue past that goal. Correspondingly for l2 with respect to
2760 If pe1 is non-NULL and the pointer it points to is not NULL, that pointer is
2761 considered an end pointer beyond which scanning of s1 will not continue under
2762 any circumstances. This means that if both l1 and pe1 are specified, and pe1
2763 is less than s1+l1, the match will never be successful because it can never
2764 get as far as its goal (and in fact is asserted against). Correspondingly for
2765 pe2 with respect to s2.
2767 At least one of s1 and s2 must have a goal (at least one of l1 and l2 must be
2768 non-zero), and if both do, both have to be
2769 reached for a successful match. Also, if the fold of a character is multiple
2770 characters, all of them must be matched (see tr21 reference below for
2773 Upon a successful match, if pe1 is non-NULL,
2774 it will be set to point to the beginning of the I<next> character of s1 beyond
2775 what was matched. Correspondingly for pe2 and s2.
2777 For case-insensitiveness, the "casefolding" of Unicode is used
2778 instead of upper/lowercasing both the characters, see
2779 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
2783 Perl_foldEQ_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
2786 register const U8 *p1 = (const U8*)s1; /* Point to current char */
2787 register const U8 *p2 = (const U8*)s2;
2788 register const U8 *g1 = NULL; /* goal for s1 */
2789 register const U8 *g2 = NULL;
2790 register const U8 *e1 = NULL; /* Don't scan s1 past this */
2791 register U8 *f1 = NULL; /* Point to current folded */
2792 register const U8 *e2 = NULL;
2793 register U8 *f2 = NULL;
2794 STRLEN n1 = 0, n2 = 0; /* Number of bytes in current char */
2795 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
2796 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
2797 U8 natbuf[2]; /* Holds native 8-bit char converted to utf8;
2798 these always fit in 2 bytes */
2800 PERL_ARGS_ASSERT_FOLDEQ_UTF8;
2807 g1 = (const U8*)s1 + l1;
2815 g2 = (const U8*)s2 + l2;
2818 /* Must have at least one goal */
2823 /* Will never match if goal is out-of-bounds */
2824 assert(! e1 || e1 >= g1);
2826 /* Here, there isn't an end pointer, or it is beyond the goal. We
2827 * only go as far as the goal */
2831 assert(e1); /* Must have an end for looking at s1 */
2834 /* Same for goal for s2 */
2836 assert(! e2 || e2 >= g2);
2843 /* Look through both strings, a character at a time */
2844 while (p1 < e1 && p2 < e2) {
2846 /* If at the beginning of a new character in s1, get its fold to use
2847 * and the length of the fold */
2850 to_utf8_fold(p1, foldbuf1, &n1);
2852 else { /* Not utf8, convert to it first and then get fold */
2853 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p1)));
2854 to_utf8_fold(natbuf, foldbuf1, &n1);
2859 if (n2 == 0) { /* Same for s2 */
2861 to_utf8_fold(p2, foldbuf2, &n2);
2864 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p2)));
2865 to_utf8_fold(natbuf, foldbuf2, &n2);
2870 /* While there is more to look for in both folds, see if they
2871 * continue to match */
2873 U8 fold_length = UTF8SKIP(f1);
2874 if (fold_length != UTF8SKIP(f2)
2875 || (fold_length == 1 && *f1 != *f2) /* Short circuit memNE
2876 function call for single
2878 || memNE((char*)f1, (char*)f2, fold_length))
2880 return 0; /* mismatch */
2883 /* Here, they matched, advance past them */
2890 /* When reach the end of any fold, advance the input past it */
2892 p1 += u1 ? UTF8SKIP(p1) : 1;
2895 p2 += u2 ? UTF8SKIP(p2) : 1;
2897 } /* End of loop through both strings */
2899 /* A match is defined by each scan that specified an explicit length
2900 * reaching its final goal, and the other not having matched a partial
2901 * character (which can happen when the fold of a character is more than one
2903 if (! ((g1 == 0 || p1 == g1) && (g2 == 0 || p2 == g2)) || n1 || n2) {
2907 /* Successful match. Set output pointers */
2919 * c-indentation-style: bsd
2921 * indent-tabs-mode: t
2924 * ex: set ts=8 sts=4 sw=4 noet: