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> bytes from the native encoding into
985 Returns a pointer to the newly-created string, and sets C<len> to
986 reflect the new length in bytes.
988 A NUL character will be written after the end of the string.
990 If you want to convert to UTF-8 from encodings other than
991 the native (Latin1 or EBCDIC),
992 see sv_recode_to_utf8().
998 Perl_bytes_to_utf8(pTHX_ const U8 *s, STRLEN *len)
1000 const U8 * const send = s + (*len);
1004 PERL_ARGS_ASSERT_BYTES_TO_UTF8;
1005 PERL_UNUSED_CONTEXT;
1007 Newx(d, (*len) * 2 + 1, U8);
1011 const UV uv = NATIVE_TO_ASCII(*s++);
1012 if (UNI_IS_INVARIANT(uv))
1013 *d++ = (U8)UTF_TO_NATIVE(uv);
1015 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
1016 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
1025 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
1027 * Destination must be pre-extended to 3/2 source. Do not use in-place.
1028 * We optimize for native, for obvious reasons. */
1031 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
1036 PERL_ARGS_ASSERT_UTF16_TO_UTF8;
1039 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %"UVuf, (UV)bytelen);
1044 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
1048 *d++ = UNI_TO_NATIVE(uv);
1055 *d++ = (U8)(( uv >> 6) | 0xc0);
1056 *d++ = (U8)(( uv & 0x3f) | 0x80);
1059 if (uv >= 0xd800 && uv <= 0xdbff) { /* surrogates */
1061 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
1063 UV low = (p[0] << 8) + p[1];
1065 if (low < 0xdc00 || low > 0xdfff)
1066 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
1067 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
1069 } else if (uv >= 0xdc00 && uv <= 0xdfff) {
1070 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
1073 *d++ = (U8)(( uv >> 12) | 0xe0);
1074 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
1075 *d++ = (U8)(( uv & 0x3f) | 0x80);
1079 *d++ = (U8)(( uv >> 18) | 0xf0);
1080 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
1081 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
1082 *d++ = (U8)(( uv & 0x3f) | 0x80);
1086 *newlen = d - dstart;
1090 /* Note: this one is slightly destructive of the source. */
1093 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
1096 U8* const send = s + bytelen;
1098 PERL_ARGS_ASSERT_UTF16_TO_UTF8_REVERSED;
1101 Perl_croak(aTHX_ "panic: utf16_to_utf8_reversed: odd bytelen %"UVuf,
1105 const U8 tmp = s[0];
1110 return utf16_to_utf8(p, d, bytelen, newlen);
1113 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
1116 Perl_is_uni_alnum(pTHX_ UV c)
1118 U8 tmpbuf[UTF8_MAXBYTES+1];
1119 uvchr_to_utf8(tmpbuf, c);
1120 return is_utf8_alnum(tmpbuf);
1124 Perl_is_uni_idfirst(pTHX_ UV c)
1126 U8 tmpbuf[UTF8_MAXBYTES+1];
1127 uvchr_to_utf8(tmpbuf, c);
1128 return is_utf8_idfirst(tmpbuf);
1132 Perl_is_uni_alpha(pTHX_ UV c)
1134 U8 tmpbuf[UTF8_MAXBYTES+1];
1135 uvchr_to_utf8(tmpbuf, c);
1136 return is_utf8_alpha(tmpbuf);
1140 Perl_is_uni_ascii(pTHX_ UV c)
1142 U8 tmpbuf[UTF8_MAXBYTES+1];
1143 uvchr_to_utf8(tmpbuf, c);
1144 return is_utf8_ascii(tmpbuf);
1148 Perl_is_uni_space(pTHX_ UV c)
1150 U8 tmpbuf[UTF8_MAXBYTES+1];
1151 uvchr_to_utf8(tmpbuf, c);
1152 return is_utf8_space(tmpbuf);
1156 Perl_is_uni_digit(pTHX_ UV c)
1158 U8 tmpbuf[UTF8_MAXBYTES+1];
1159 uvchr_to_utf8(tmpbuf, c);
1160 return is_utf8_digit(tmpbuf);
1164 Perl_is_uni_upper(pTHX_ UV c)
1166 U8 tmpbuf[UTF8_MAXBYTES+1];
1167 uvchr_to_utf8(tmpbuf, c);
1168 return is_utf8_upper(tmpbuf);
1172 Perl_is_uni_lower(pTHX_ UV c)
1174 U8 tmpbuf[UTF8_MAXBYTES+1];
1175 uvchr_to_utf8(tmpbuf, c);
1176 return is_utf8_lower(tmpbuf);
1180 Perl_is_uni_cntrl(pTHX_ UV c)
1182 U8 tmpbuf[UTF8_MAXBYTES+1];
1183 uvchr_to_utf8(tmpbuf, c);
1184 return is_utf8_cntrl(tmpbuf);
1188 Perl_is_uni_graph(pTHX_ UV c)
1190 U8 tmpbuf[UTF8_MAXBYTES+1];
1191 uvchr_to_utf8(tmpbuf, c);
1192 return is_utf8_graph(tmpbuf);
1196 Perl_is_uni_print(pTHX_ UV c)
1198 U8 tmpbuf[UTF8_MAXBYTES+1];
1199 uvchr_to_utf8(tmpbuf, c);
1200 return is_utf8_print(tmpbuf);
1204 Perl_is_uni_punct(pTHX_ UV c)
1206 U8 tmpbuf[UTF8_MAXBYTES+1];
1207 uvchr_to_utf8(tmpbuf, c);
1208 return is_utf8_punct(tmpbuf);
1212 Perl_is_uni_xdigit(pTHX_ UV c)
1214 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1215 uvchr_to_utf8(tmpbuf, c);
1216 return is_utf8_xdigit(tmpbuf);
1220 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
1222 PERL_ARGS_ASSERT_TO_UNI_UPPER;
1224 uvchr_to_utf8(p, c);
1225 return to_utf8_upper(p, p, lenp);
1229 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
1231 PERL_ARGS_ASSERT_TO_UNI_TITLE;
1233 uvchr_to_utf8(p, c);
1234 return to_utf8_title(p, p, lenp);
1238 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
1240 PERL_ARGS_ASSERT_TO_UNI_LOWER;
1242 uvchr_to_utf8(p, c);
1243 return to_utf8_lower(p, p, lenp);
1247 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1249 PERL_ARGS_ASSERT_TO_UNI_FOLD;
1251 uvchr_to_utf8(p, c);
1252 return to_utf8_fold(p, p, lenp);
1255 /* for now these all assume no locale info available for Unicode > 255 */
1258 Perl_is_uni_alnum_lc(pTHX_ UV c)
1260 return is_uni_alnum(c); /* XXX no locale support yet */
1264 Perl_is_uni_idfirst_lc(pTHX_ UV c)
1266 return is_uni_idfirst(c); /* XXX no locale support yet */
1270 Perl_is_uni_alpha_lc(pTHX_ UV c)
1272 return is_uni_alpha(c); /* XXX no locale support yet */
1276 Perl_is_uni_ascii_lc(pTHX_ UV c)
1278 return is_uni_ascii(c); /* XXX no locale support yet */
1282 Perl_is_uni_space_lc(pTHX_ UV c)
1284 return is_uni_space(c); /* XXX no locale support yet */
1288 Perl_is_uni_digit_lc(pTHX_ UV c)
1290 return is_uni_digit(c); /* XXX no locale support yet */
1294 Perl_is_uni_upper_lc(pTHX_ UV c)
1296 return is_uni_upper(c); /* XXX no locale support yet */
1300 Perl_is_uni_lower_lc(pTHX_ UV c)
1302 return is_uni_lower(c); /* XXX no locale support yet */
1306 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1308 return is_uni_cntrl(c); /* XXX no locale support yet */
1312 Perl_is_uni_graph_lc(pTHX_ UV c)
1314 return is_uni_graph(c); /* XXX no locale support yet */
1318 Perl_is_uni_print_lc(pTHX_ UV c)
1320 return is_uni_print(c); /* XXX no locale support yet */
1324 Perl_is_uni_punct_lc(pTHX_ UV c)
1326 return is_uni_punct(c); /* XXX no locale support yet */
1330 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1332 return is_uni_xdigit(c); /* XXX no locale support yet */
1336 Perl_to_uni_upper_lc(pTHX_ U32 c)
1338 /* XXX returns only the first character -- do not use XXX */
1339 /* XXX no locale support yet */
1341 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1342 return (U32)to_uni_upper(c, tmpbuf, &len);
1346 Perl_to_uni_title_lc(pTHX_ U32 c)
1348 /* XXX returns only the first character XXX -- do not use XXX */
1349 /* XXX no locale support yet */
1351 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1352 return (U32)to_uni_title(c, tmpbuf, &len);
1356 Perl_to_uni_lower_lc(pTHX_ U32 c)
1358 /* XXX returns only the first character -- do not use XXX */
1359 /* XXX no locale support yet */
1361 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1362 return (U32)to_uni_lower(c, tmpbuf, &len);
1366 S_is_utf8_common(pTHX_ const U8 *const p, SV **swash,
1367 const char *const swashname)
1371 PERL_ARGS_ASSERT_IS_UTF8_COMMON;
1373 if (!is_utf8_char(p))
1376 *swash = swash_init("utf8", swashname, &PL_sv_undef, 1, 0);
1377 return swash_fetch(*swash, p, TRUE) != 0;
1381 Perl_is_utf8_alnum(pTHX_ const U8 *p)
1385 PERL_ARGS_ASSERT_IS_UTF8_ALNUM;
1387 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1388 * descendant of isalnum(3), in other words, it doesn't
1389 * contain the '_'. --jhi */
1390 return is_utf8_common(p, &PL_utf8_alnum, "IsWord");
1394 Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
1398 PERL_ARGS_ASSERT_IS_UTF8_IDFIRST;
1402 /* is_utf8_idstart would be more logical. */
1403 return is_utf8_common(p, &PL_utf8_idstart, "IdStart");
1407 Perl_is_utf8_idcont(pTHX_ const U8 *p)
1411 PERL_ARGS_ASSERT_IS_UTF8_IDCONT;
1415 return is_utf8_common(p, &PL_utf8_idcont, "IdContinue");
1419 Perl_is_utf8_alpha(pTHX_ const U8 *p)
1423 PERL_ARGS_ASSERT_IS_UTF8_ALPHA;
1425 return is_utf8_common(p, &PL_utf8_alpha, "IsAlpha");
1429 Perl_is_utf8_ascii(pTHX_ const U8 *p)
1433 PERL_ARGS_ASSERT_IS_UTF8_ASCII;
1435 return is_utf8_common(p, &PL_utf8_ascii, "IsAscii");
1439 Perl_is_utf8_space(pTHX_ const U8 *p)
1443 PERL_ARGS_ASSERT_IS_UTF8_SPACE;
1445 return is_utf8_common(p, &PL_utf8_space, "IsSpacePerl");
1449 Perl_is_utf8_perl_space(pTHX_ const U8 *p)
1453 PERL_ARGS_ASSERT_IS_UTF8_PERL_SPACE;
1455 return is_utf8_common(p, &PL_utf8_perl_space, "IsPerlSpace");
1459 Perl_is_utf8_perl_word(pTHX_ const U8 *p)
1463 PERL_ARGS_ASSERT_IS_UTF8_PERL_WORD;
1465 return is_utf8_common(p, &PL_utf8_perl_word, "IsPerlWord");
1469 Perl_is_utf8_digit(pTHX_ const U8 *p)
1473 PERL_ARGS_ASSERT_IS_UTF8_DIGIT;
1475 return is_utf8_common(p, &PL_utf8_digit, "IsDigit");
1479 Perl_is_utf8_posix_digit(pTHX_ const U8 *p)
1483 PERL_ARGS_ASSERT_IS_UTF8_POSIX_DIGIT;
1485 return is_utf8_common(p, &PL_utf8_posix_digit, "IsPosixDigit");
1489 Perl_is_utf8_upper(pTHX_ const U8 *p)
1493 PERL_ARGS_ASSERT_IS_UTF8_UPPER;
1495 return is_utf8_common(p, &PL_utf8_upper, "IsUppercase");
1499 Perl_is_utf8_lower(pTHX_ const U8 *p)
1503 PERL_ARGS_ASSERT_IS_UTF8_LOWER;
1505 return is_utf8_common(p, &PL_utf8_lower, "IsLowercase");
1509 Perl_is_utf8_cntrl(pTHX_ const U8 *p)
1513 PERL_ARGS_ASSERT_IS_UTF8_CNTRL;
1515 return is_utf8_common(p, &PL_utf8_cntrl, "IsCntrl");
1519 Perl_is_utf8_graph(pTHX_ const U8 *p)
1523 PERL_ARGS_ASSERT_IS_UTF8_GRAPH;
1525 return is_utf8_common(p, &PL_utf8_graph, "IsGraph");
1529 Perl_is_utf8_print(pTHX_ const U8 *p)
1533 PERL_ARGS_ASSERT_IS_UTF8_PRINT;
1535 return is_utf8_common(p, &PL_utf8_print, "IsPrint");
1539 Perl_is_utf8_punct(pTHX_ const U8 *p)
1543 PERL_ARGS_ASSERT_IS_UTF8_PUNCT;
1545 return is_utf8_common(p, &PL_utf8_punct, "IsPunct");
1549 Perl_is_utf8_xdigit(pTHX_ const U8 *p)
1553 PERL_ARGS_ASSERT_IS_UTF8_XDIGIT;
1555 return is_utf8_common(p, &PL_utf8_xdigit, "IsXDigit");
1559 Perl_is_utf8_mark(pTHX_ const U8 *p)
1563 PERL_ARGS_ASSERT_IS_UTF8_MARK;
1565 return is_utf8_common(p, &PL_utf8_mark, "IsM");
1569 Perl_is_utf8_X_begin(pTHX_ const U8 *p)
1573 PERL_ARGS_ASSERT_IS_UTF8_X_BEGIN;
1575 return is_utf8_common(p, &PL_utf8_X_begin, "_X_Begin");
1579 Perl_is_utf8_X_extend(pTHX_ const U8 *p)
1583 PERL_ARGS_ASSERT_IS_UTF8_X_EXTEND;
1585 return is_utf8_common(p, &PL_utf8_X_extend, "_X_Extend");
1589 Perl_is_utf8_X_prepend(pTHX_ const U8 *p)
1593 PERL_ARGS_ASSERT_IS_UTF8_X_PREPEND;
1595 return is_utf8_common(p, &PL_utf8_X_prepend, "GCB=Prepend");
1599 Perl_is_utf8_X_non_hangul(pTHX_ const U8 *p)
1603 PERL_ARGS_ASSERT_IS_UTF8_X_NON_HANGUL;
1605 return is_utf8_common(p, &PL_utf8_X_non_hangul, "HST=Not_Applicable");
1609 Perl_is_utf8_X_L(pTHX_ const U8 *p)
1613 PERL_ARGS_ASSERT_IS_UTF8_X_L;
1615 return is_utf8_common(p, &PL_utf8_X_L, "GCB=L");
1619 Perl_is_utf8_X_LV(pTHX_ const U8 *p)
1623 PERL_ARGS_ASSERT_IS_UTF8_X_LV;
1625 return is_utf8_common(p, &PL_utf8_X_LV, "GCB=LV");
1629 Perl_is_utf8_X_LVT(pTHX_ const U8 *p)
1633 PERL_ARGS_ASSERT_IS_UTF8_X_LVT;
1635 return is_utf8_common(p, &PL_utf8_X_LVT, "GCB=LVT");
1639 Perl_is_utf8_X_T(pTHX_ const U8 *p)
1643 PERL_ARGS_ASSERT_IS_UTF8_X_T;
1645 return is_utf8_common(p, &PL_utf8_X_T, "GCB=T");
1649 Perl_is_utf8_X_V(pTHX_ const U8 *p)
1653 PERL_ARGS_ASSERT_IS_UTF8_X_V;
1655 return is_utf8_common(p, &PL_utf8_X_V, "GCB=V");
1659 Perl_is_utf8_X_LV_LVT_V(pTHX_ const U8 *p)
1663 PERL_ARGS_ASSERT_IS_UTF8_X_LV_LVT_V;
1665 return is_utf8_common(p, &PL_utf8_X_LV_LVT_V, "_X_LV_LVT_V");
1669 =for apidoc to_utf8_case
1671 The "p" contains the pointer to the UTF-8 string encoding
1672 the character that is being converted.
1674 The "ustrp" is a pointer to the character buffer to put the
1675 conversion result to. The "lenp" is a pointer to the length
1678 The "swashp" is a pointer to the swash to use.
1680 Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
1681 and loaded by SWASHNEW, using lib/utf8_heavy.pl. The special (usually,
1682 but not always, a multicharacter mapping), is tried first.
1684 The "special" is a string like "utf8::ToSpecLower", which means the
1685 hash %utf8::ToSpecLower. The access to the hash is through
1686 Perl_to_utf8_case().
1688 The "normal" is a string like "ToLower" which means the swash
1694 Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp,
1695 SV **swashp, const char *normal, const char *special)
1698 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1700 const UV uv0 = utf8_to_uvchr(p, NULL);
1701 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1702 * are necessary in EBCDIC, they are redundant no-ops
1703 * in ASCII-ish platforms, and hopefully optimized away. */
1704 const UV uv1 = NATIVE_TO_UNI(uv0);
1706 PERL_ARGS_ASSERT_TO_UTF8_CASE;
1708 uvuni_to_utf8(tmpbuf, uv1);
1710 if (!*swashp) /* load on-demand */
1711 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1712 /* This is the beginnings of a skeleton of code to read the info section
1713 * that is in all the swashes in case we ever want to do that, so one can
1714 * read things whose maps aren't code points, and whose default if missing
1715 * is not to the code point itself. This was just to see if it actually
1716 * worked. Details on what the possibilities are are in perluniprops.pod
1717 HV * const hv = get_hv("utf8::SwashInfo", 0);
1720 svp = hv_fetch(hv, (const char*)normal, strlen(normal), FALSE);
1723 HV * const this_hash = SvRV(*svp);
1724 svp = hv_fetch(this_hash, "type", strlen("type"), FALSE);
1725 s = SvPV_const(*svp, len);
1730 /* It might be "special" (sometimes, but not always,
1731 * a multicharacter mapping) */
1732 HV * const hv = get_hv(special, 0);
1736 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1740 s = SvPV_const(*svp, len);
1742 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
1745 /* If we have EBCDIC we need to remap the characters
1746 * since any characters in the low 256 are Unicode
1747 * code points, not EBCDIC. */
1748 U8 *t = (U8*)s, *tend = t + len, *d;
1755 const UV c = utf8_to_uvchr(t, &tlen);
1757 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1766 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1771 Copy(tmpbuf, ustrp, len, U8);
1773 Copy(s, ustrp, len, U8);
1779 if (!len && *swashp) {
1780 const UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1783 /* It was "normal" (a single character mapping). */
1784 const UV uv3 = UNI_TO_NATIVE(uv2);
1785 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
1789 if (!len) /* Neither: just copy. In other words, there was no mapping
1790 defined, which means that the code point maps to itself */
1791 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1796 return len ? utf8_to_uvchr(ustrp, 0) : 0;
1800 =for apidoc to_utf8_upper
1802 Convert the UTF-8 encoded character at p to its uppercase version and
1803 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1804 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1805 the uppercase version may be longer than the original character.
1807 The first character of the uppercased version is returned
1808 (but note, as explained above, that there may be more.)
1813 Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1817 PERL_ARGS_ASSERT_TO_UTF8_UPPER;
1819 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1820 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1824 =for apidoc to_utf8_title
1826 Convert the UTF-8 encoded character at p to its titlecase version and
1827 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1828 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1829 titlecase version may be longer than the original character.
1831 The first character of the titlecased version is returned
1832 (but note, as explained above, that there may be more.)
1837 Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1841 PERL_ARGS_ASSERT_TO_UTF8_TITLE;
1843 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1844 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1848 =for apidoc to_utf8_lower
1850 Convert the UTF-8 encoded character at p to its lowercase version and
1851 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1852 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1853 lowercase version may be longer than the original character.
1855 The first character of the lowercased version is returned
1856 (but note, as explained above, that there may be more.)
1861 Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1865 PERL_ARGS_ASSERT_TO_UTF8_LOWER;
1867 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1868 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1872 =for apidoc to_utf8_fold
1874 Convert the UTF-8 encoded character at p to its foldcase version and
1875 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1876 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1877 foldcase version may be longer than the original character (up to
1880 The first character of the foldcased version is returned
1881 (but note, as explained above, that there may be more.)
1886 Perl_to_utf8_fold(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1890 PERL_ARGS_ASSERT_TO_UTF8_FOLD;
1892 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1893 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1897 * A "swash" is a swatch hash.
1898 * A "swatch" is a bit vector generated by utf8.c:S_swash_get().
1899 * C<pkg> is a pointer to a package name for SWASHNEW, should be "utf8".
1900 * For other parameters, see utf8::SWASHNEW in lib/utf8_heavy.pl.
1903 Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
1908 const size_t pkg_len = strlen(pkg);
1909 const size_t name_len = strlen(name);
1910 HV * const stash = gv_stashpvn(pkg, pkg_len, 0);
1914 PERL_ARGS_ASSERT_SWASH_INIT;
1916 PUSHSTACKi(PERLSI_MAGIC);
1920 method = gv_fetchmeth(stash, "SWASHNEW", 8, -1);
1921 if (!method) { /* demand load utf8 */
1923 errsv_save = newSVsv(ERRSV);
1924 /* It is assumed that callers of this routine are not passing in any
1925 user derived data. */
1926 /* Need to do this after save_re_context() as it will set PL_tainted to
1927 1 while saving $1 etc (see the code after getrx: in Perl_magic_get).
1928 Even line to create errsv_save can turn on PL_tainted. */
1929 SAVEBOOL(PL_tainted);
1931 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
1934 sv_setsv(ERRSV, errsv_save);
1935 SvREFCNT_dec(errsv_save);
1941 mPUSHp(pkg, pkg_len);
1942 mPUSHp(name, name_len);
1947 errsv_save = newSVsv(ERRSV);
1948 /* If we already have a pointer to the method, no need to use call_method()
1949 to repeat the lookup. */
1950 if (method ? call_sv(MUTABLE_SV(method), G_SCALAR)
1951 : call_sv(newSVpvs_flags("SWASHNEW", SVs_TEMP), G_SCALAR | G_METHOD))
1952 retval = newSVsv(*PL_stack_sp--);
1954 retval = &PL_sv_undef;
1956 sv_setsv(ERRSV, errsv_save);
1957 SvREFCNT_dec(errsv_save);
1960 if (IN_PERL_COMPILETIME) {
1961 CopHINTS_set(PL_curcop, PL_hints);
1963 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
1965 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
1967 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1973 /* This API is wrong for special case conversions since we may need to
1974 * return several Unicode characters for a single Unicode character
1975 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1976 * the lower-level routine, and it is similarly broken for returning
1977 * multiple values. --jhi
1978 * For those, you should use to_utf8_case() instead */
1979 /* Now SWASHGET is recasted into S_swash_get in this file. */
1982 * Returns the value of property/mapping C<swash> for the first character
1983 * of the string C<ptr>. If C<do_utf8> is true, the string C<ptr> is
1984 * assumed to be in utf8. If C<do_utf8> is false, the string C<ptr> is
1985 * assumed to be in native 8-bit encoding. Caches the swatch in C<swash>.
1988 Perl_swash_fetch(pTHX_ SV *swash, const U8 *ptr, bool do_utf8)
1991 HV *const hv = MUTABLE_HV(SvRV(swash));
1996 const U8 *tmps = NULL;
2000 const UV c = NATIVE_TO_ASCII(*ptr);
2002 PERL_ARGS_ASSERT_SWASH_FETCH;
2004 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
2005 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
2006 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
2009 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
2010 * then the "swatch" is a vec() for all the chars which start
2012 * So the key in the hash (klen) is length of encoded char -1
2014 klen = UTF8SKIP(ptr) - 1;
2018 /* If char is invariant then swatch is for all the invariant chars
2019 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
2021 needents = UTF_CONTINUATION_MARK;
2022 off = NATIVE_TO_UTF(ptr[klen]);
2025 /* If char is encoded then swatch is for the prefix */
2026 needents = (1 << UTF_ACCUMULATION_SHIFT);
2027 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
2031 * This single-entry cache saves about 1/3 of the utf8 overhead in test
2032 * suite. (That is, only 7-8% overall over just a hash cache. Still,
2033 * it's nothing to sniff at.) Pity we usually come through at least
2034 * two function calls to get here...
2036 * NB: this code assumes that swatches are never modified, once generated!
2039 if (hv == PL_last_swash_hv &&
2040 klen == PL_last_swash_klen &&
2041 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
2043 tmps = PL_last_swash_tmps;
2044 slen = PL_last_swash_slen;
2047 /* Try our second-level swatch cache, kept in a hash. */
2048 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
2050 /* If not cached, generate it via swash_get */
2051 if (!svp || !SvPOK(*svp)
2052 || !(tmps = (const U8*)SvPV_const(*svp, slen))) {
2053 /* We use utf8n_to_uvuni() as we want an index into
2054 Unicode tables, not a native character number.
2056 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
2058 0 : UTF8_ALLOW_ANY);
2059 swatch = swash_get(swash,
2060 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
2061 (klen) ? (code_point & ~(needents - 1)) : 0,
2064 if (IN_PERL_COMPILETIME)
2065 CopHINTS_set(PL_curcop, PL_hints);
2067 svp = hv_store(hv, (const char *)ptr, klen, swatch, 0);
2069 if (!svp || !(tmps = (U8*)SvPV(*svp, slen))
2070 || (slen << 3) < needents)
2071 Perl_croak(aTHX_ "panic: swash_fetch got improper swatch");
2074 PL_last_swash_hv = hv;
2075 assert(klen <= sizeof(PL_last_swash_key));
2076 PL_last_swash_klen = (U8)klen;
2077 /* FIXME change interpvar.h? */
2078 PL_last_swash_tmps = (U8 *) tmps;
2079 PL_last_swash_slen = slen;
2081 Copy(ptr, PL_last_swash_key, klen, U8);
2084 switch ((int)((slen << 3) / needents)) {
2086 bit = 1 << (off & 7);
2088 return (tmps[off] & bit) != 0;
2093 return (tmps[off] << 8) + tmps[off + 1] ;
2096 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
2098 Perl_croak(aTHX_ "panic: swash_fetch got swatch of unexpected bit width");
2099 NORETURN_FUNCTION_END;
2102 /* Read a single line of the main body of the swash input text. These are of
2105 * where each number is hex. The first two numbers form the minimum and
2106 * maximum of a range, and the third is the value associated with the range.
2107 * Not all swashes should have a third number
2109 * On input: l points to the beginning of the line to be examined; it points
2110 * to somewhere in the string of the whole input text, and is
2111 * terminated by a \n or the null string terminator.
2112 * lend points to the null terminator of that string
2113 * wants_value is non-zero if the swash expects a third number
2114 * typestr is the name of the swash's mapping, like 'ToLower'
2115 * On output: *min, *max, and *val are set to the values read from the line.
2116 * returns a pointer just beyond the line examined. If there was no
2117 * valid min number on the line, returns lend+1
2121 S_swash_scan_list_line(pTHX_ U8* l, U8* const lend, UV* min, UV* max, UV* val,
2122 const bool wants_value, const U8* const typestr)
2124 const int typeto = typestr[0] == 'T' && typestr[1] == 'o';
2125 STRLEN numlen; /* Length of the number */
2126 I32 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
2128 /* nl points to the next \n in the scan */
2129 U8* const nl = (U8*)memchr(l, '\n', lend - l);
2131 /* Get the first number on the line: the range minimum */
2133 *min = grok_hex((char *)l, &numlen, &flags, NULL);
2134 if (numlen) /* If found a hex number, position past it */
2136 else if (nl) { /* Else, go handle next line, if any */
2137 return nl + 1; /* 1 is length of "\n" */
2139 else { /* Else, no next line */
2140 return lend + 1; /* to LIST's end at which \n is not found */
2143 /* The max range value follows, separated by a BLANK */
2146 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
2148 *max = grok_hex((char *)l, &numlen, &flags, NULL);
2151 else /* If no value here, it is a single element range */
2154 /* Non-binary tables have a third entry: what the first element of the
2159 flags = PERL_SCAN_SILENT_ILLDIGIT |
2160 PERL_SCAN_DISALLOW_PREFIX;
2162 *val = grok_hex((char *)l, &numlen, &flags, NULL);
2171 Perl_croak(aTHX_ "%s: illegal mapping '%s'",
2177 *val = 0; /* bits == 1, then any val should be ignored */
2179 else { /* Nothing following range min, should be single element with no
2185 Perl_croak(aTHX_ "%s: illegal mapping '%s'", typestr, l);
2189 *val = 0; /* bits == 1, then val should be ignored */
2192 /* Position to next line if any, or EOF */
2202 * Returns a swatch (a bit vector string) for a code point sequence
2203 * that starts from the value C<start> and comprises the number C<span>.
2204 * A C<swash> must be an object created by SWASHNEW (see lib/utf8_heavy.pl).
2205 * Should be used via swash_fetch, which will cache the swatch in C<swash>.
2208 S_swash_get(pTHX_ SV* swash, UV start, UV span)
2211 U8 *l, *lend, *x, *xend, *s;
2212 STRLEN lcur, xcur, scur;
2213 HV *const hv = MUTABLE_HV(SvRV(swash));
2215 /* The string containing the main body of the table */
2216 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2218 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2219 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2220 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2221 SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);
2222 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
2223 const STRLEN bits = SvUV(*bitssvp);
2224 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2225 const UV none = SvUV(*nonesvp);
2226 const UV end = start + span;
2228 PERL_ARGS_ASSERT_SWASH_GET;
2230 if (bits != 1 && bits != 8 && bits != 16 && bits != 32) {
2231 Perl_croak(aTHX_ "panic: swash_get doesn't expect bits %"UVuf,
2235 /* create and initialize $swatch */
2236 scur = octets ? (span * octets) : (span + 7) / 8;
2237 swatch = newSV(scur);
2239 s = (U8*)SvPVX(swatch);
2240 if (octets && none) {
2241 const U8* const e = s + scur;
2244 *s++ = (U8)(none & 0xff);
2245 else if (bits == 16) {
2246 *s++ = (U8)((none >> 8) & 0xff);
2247 *s++ = (U8)( none & 0xff);
2249 else if (bits == 32) {
2250 *s++ = (U8)((none >> 24) & 0xff);
2251 *s++ = (U8)((none >> 16) & 0xff);
2252 *s++ = (U8)((none >> 8) & 0xff);
2253 *s++ = (U8)( none & 0xff);
2259 (void)memzero((U8*)s, scur + 1);
2261 SvCUR_set(swatch, scur);
2262 s = (U8*)SvPVX(swatch);
2264 /* read $swash->{LIST} */
2265 l = (U8*)SvPV(*listsvp, lcur);
2269 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2270 cBOOL(octets), typestr);
2275 /* If looking for something beyond this range, go try the next one */
2282 if (!none || val < none) {
2287 for (key = min; key <= max; key++) {
2291 /* offset must be non-negative (start <= min <= key < end) */
2292 offset = octets * (key - start);
2294 s[offset] = (U8)(val & 0xff);
2295 else if (bits == 16) {
2296 s[offset ] = (U8)((val >> 8) & 0xff);
2297 s[offset + 1] = (U8)( val & 0xff);
2299 else if (bits == 32) {
2300 s[offset ] = (U8)((val >> 24) & 0xff);
2301 s[offset + 1] = (U8)((val >> 16) & 0xff);
2302 s[offset + 2] = (U8)((val >> 8) & 0xff);
2303 s[offset + 3] = (U8)( val & 0xff);
2306 if (!none || val < none)
2310 else { /* bits == 1, then val should be ignored */
2314 for (key = min; key <= max; key++) {
2315 const STRLEN offset = (STRLEN)(key - start);
2318 s[offset >> 3] |= 1 << (offset & 7);
2324 /* read $swash->{EXTRAS} */
2325 x = (U8*)SvPV(*extssvp, xcur);
2333 SV **otherbitssvp, *other;
2337 const U8 opc = *x++;
2341 nl = (U8*)memchr(x, '\n', xend - x);
2343 if (opc != '-' && opc != '+' && opc != '!' && opc != '&') {
2345 x = nl + 1; /* 1 is length of "\n" */
2349 x = xend; /* to EXTRAS' end at which \n is not found */
2356 namelen = nl - namestr;
2360 namelen = xend - namestr;
2364 othersvp = hv_fetch(hv, (char *)namestr, namelen, FALSE);
2365 otherhv = MUTABLE_HV(SvRV(*othersvp));
2366 otherbitssvp = hv_fetchs(otherhv, "BITS", FALSE);
2367 otherbits = (STRLEN)SvUV(*otherbitssvp);
2368 if (bits < otherbits)
2369 Perl_croak(aTHX_ "panic: swash_get found swatch size mismatch");
2371 /* The "other" swatch must be destroyed after. */
2372 other = swash_get(*othersvp, start, span);
2373 o = (U8*)SvPV(other, olen);
2376 Perl_croak(aTHX_ "panic: swash_get got improper swatch");
2378 s = (U8*)SvPV(swatch, slen);
2379 if (bits == 1 && otherbits == 1) {
2381 Perl_croak(aTHX_ "panic: swash_get found swatch length mismatch");
2405 STRLEN otheroctets = otherbits >> 3;
2407 U8* const send = s + slen;
2412 if (otherbits == 1) {
2413 otherval = (o[offset >> 3] >> (offset & 7)) & 1;
2417 STRLEN vlen = otheroctets;
2425 if (opc == '+' && otherval)
2426 NOOP; /* replace with otherval */
2427 else if (opc == '!' && !otherval)
2429 else if (opc == '-' && otherval)
2431 else if (opc == '&' && !otherval)
2434 s += octets; /* no replacement */
2439 *s++ = (U8)( otherval & 0xff);
2440 else if (bits == 16) {
2441 *s++ = (U8)((otherval >> 8) & 0xff);
2442 *s++ = (U8)( otherval & 0xff);
2444 else if (bits == 32) {
2445 *s++ = (U8)((otherval >> 24) & 0xff);
2446 *s++ = (U8)((otherval >> 16) & 0xff);
2447 *s++ = (U8)((otherval >> 8) & 0xff);
2448 *s++ = (U8)( otherval & 0xff);
2452 sv_free(other); /* through with it! */
2458 Perl__swash_inversion_hash(pTHX_ SV* swash)
2461 /* Subject to change or removal. For use only in one place in regexec.c
2463 * Returns a hash which is the inversion and closure of a swash mapping.
2464 * For example, consider the input lines:
2469 * The returned hash would have two keys, the utf8 for 006B and the utf8 for
2470 * 006C. The value for each key is an array. For 006C, the array would
2471 * have a two elements, the utf8 for itself, and for 004C. For 006B, there
2472 * would be three elements in its array, the utf8 for 006B, 004B and 212A.
2474 * Essentially, for any code point, it gives all the code points that map to
2475 * it, or the list of 'froms' for that point.
2477 * Currently it only looks at the main body of the swash, and ignores any
2478 * additions or deletions from other swashes */
2482 HV *const hv = MUTABLE_HV(SvRV(swash));
2484 /* The string containing the main body of the table */
2485 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2487 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2488 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2489 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2490 /*SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);*/
2491 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
2492 const STRLEN bits = SvUV(*bitssvp);
2493 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2494 const UV none = SvUV(*nonesvp);
2498 PERL_ARGS_ASSERT__SWASH_INVERSION_HASH;
2500 /* Must have at least 8 bits to get the mappings */
2501 if (bits != 8 && bits != 16 && bits != 32) {
2502 Perl_croak(aTHX_ "panic: swash_inversion_hash doesn't expect bits %"UVuf,
2506 /* read $swash->{LIST} */
2507 l = (U8*)SvPV(*listsvp, lcur);
2510 /* Go through each input line */
2514 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2515 cBOOL(octets), typestr);
2520 /* Each element in the range is to be inverted */
2521 for (inverse = min; inverse <= max; inverse++) {
2526 bool found_key = FALSE;
2528 /* The key is the inverse mapping */
2529 char key[UTF8_MAXBYTES+1];
2530 char* key_end = (char *) uvuni_to_utf8((U8*) key, val);
2531 STRLEN key_len = key_end - key;
2533 /* And the value is what the forward mapping is from. */
2534 char utf8_inverse[UTF8_MAXBYTES+1];
2535 char *utf8_inverse_end = (char *) uvuni_to_utf8((U8*) utf8_inverse, inverse);
2537 /* Get the list for the map */
2538 if ((listp = hv_fetch(ret, key, key_len, FALSE))) {
2539 list = (AV*) *listp;
2541 else { /* No entry yet for it: create one */
2543 if (! hv_store(ret, key, key_len, (SV*) list, FALSE)) {
2544 Perl_croak(aTHX_ "panic: hv_store() unexpectedly failed");
2548 for (i = 0; i < av_len(list); i++) {
2549 SV** entryp = av_fetch(list, i, FALSE);
2551 if (entryp == NULL) {
2552 Perl_croak(aTHX_ "panic: av_fetch() unexpectedly failed");
2555 if (SvCUR(entry) != key_len) {
2558 if (memEQ(key, SvPVX(entry), key_len)) {
2564 element = newSVpvn_flags(key, key_len, SVf_UTF8);
2565 av_push(list, element);
2569 /* Simply add the value to the list */
2570 element = newSVpvn_flags(utf8_inverse, utf8_inverse_end - utf8_inverse, SVf_UTF8);
2571 av_push(list, element);
2573 /* swash_get() increments the value of val for each element in the
2574 * range. That makes more compact tables possible. You can
2575 * express the capitalization, for example, of all consecutive
2576 * letters with a single line: 0061\t007A\t0041 This maps 0061 to
2577 * 0041, 0062 to 0042, etc. I (khw) have never understood 'none',
2578 * and it's not documented, and perhaps not even currently used,
2579 * but I copied the semantics from swash_get(), just in case */
2580 if (!none || val < none) {
2590 =for apidoc uvchr_to_utf8
2592 Adds the UTF-8 representation of the Native codepoint C<uv> to the end
2593 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
2594 bytes available. The return value is the pointer to the byte after the
2595 end of the new character. In other words,
2597 d = uvchr_to_utf8(d, uv);
2599 is the recommended wide native character-aware way of saying
2606 /* On ASCII machines this is normally a macro but we want a
2607 real function in case XS code wants it
2610 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
2612 PERL_ARGS_ASSERT_UVCHR_TO_UTF8;
2614 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
2618 Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
2620 PERL_ARGS_ASSERT_UVCHR_TO_UTF8_FLAGS;
2622 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
2626 =for apidoc utf8n_to_uvchr
2629 Returns the native character value of the first character in the string
2631 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
2632 length, in bytes, of that character.
2634 Allows length and flags to be passed to low level routine.
2638 /* On ASCII machines this is normally a macro but we want
2639 a real function in case XS code wants it
2642 Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen,
2645 const UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
2647 PERL_ARGS_ASSERT_UTF8N_TO_UVCHR;
2649 return UNI_TO_NATIVE(uv);
2653 Perl_check_utf8_print(pTHX_ register const U8* s, const STRLEN len)
2655 /* May change: warns if surrogates, non-character code points, or
2656 * non-Unicode code points are in s which has length len. Returns TRUE if
2657 * none found; FALSE otherwise. The only other validity check is to make
2658 * sure that this won't exceed the string's length */
2660 const U8* const e = s + len;
2663 PERL_ARGS_ASSERT_CHECK_UTF8_PRINT;
2666 if (UTF8SKIP(s) > len) {
2667 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
2668 "%s in %s", unees, PL_op ? OP_DESC(PL_op) : "print");
2671 if (*s >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE) {
2673 if (UTF8_IS_SUPER(s)) {
2674 UV uv = utf8_to_uvchr(s, &char_len);
2675 Perl_warner(aTHX_ packWARN(WARN_UTF8),
2676 "Code point 0x%04"UVXf" is not Unicode, may not be portable", uv);
2679 else if (UTF8_IS_SURROGATE(s)) {
2680 UV uv = utf8_to_uvchr(s, &char_len);
2681 Perl_warner(aTHX_ packWARN(WARN_UTF8),
2682 "Unicode surrogate U+%04"UVXf" is illegal in UTF-8", uv);
2686 (UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s))
2688 UV uv = utf8_to_uvchr(s, &char_len);
2689 Perl_warner(aTHX_ packWARN(WARN_UTF8),
2690 "Unicode non-character U+%04"UVXf" is illegal for open interchange", uv);
2701 =for apidoc pv_uni_display
2703 Build to the scalar dsv a displayable version of the string spv,
2704 length len, the displayable version being at most pvlim bytes long
2705 (if longer, the rest is truncated and "..." will be appended).
2707 The flags argument can have UNI_DISPLAY_ISPRINT set to display
2708 isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
2709 to display the \\[nrfta\\] as the backslashed versions (like '\n')
2710 (UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
2711 UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
2712 UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
2714 The pointer to the PV of the dsv is returned.
2718 Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
2723 PERL_ARGS_ASSERT_PV_UNI_DISPLAY;
2727 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
2729 /* This serves double duty as a flag and a character to print after
2730 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
2734 if (pvlim && SvCUR(dsv) >= pvlim) {
2738 u = utf8_to_uvchr((U8*)s, 0);
2740 const unsigned char c = (unsigned char)u & 0xFF;
2741 if (flags & UNI_DISPLAY_BACKSLASH) {
2758 const char string = ok;
2759 sv_catpvs(dsv, "\\");
2760 sv_catpvn(dsv, &string, 1);
2763 /* isPRINT() is the locale-blind version. */
2764 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
2765 const char string = c;
2766 sv_catpvn(dsv, &string, 1);
2771 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
2774 sv_catpvs(dsv, "...");
2780 =for apidoc sv_uni_display
2782 Build to the scalar dsv a displayable version of the scalar sv,
2783 the displayable version being at most pvlim bytes long
2784 (if longer, the rest is truncated and "..." will be appended).
2786 The flags argument is as in pv_uni_display().
2788 The pointer to the PV of the dsv is returned.
2793 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
2795 PERL_ARGS_ASSERT_SV_UNI_DISPLAY;
2797 return Perl_pv_uni_display(aTHX_ dsv, (const U8*)SvPVX_const(ssv),
2798 SvCUR(ssv), pvlim, flags);
2802 =for apidoc foldEQ_utf8
2804 Returns true if the leading portions of the strings s1 and s2 (either or both
2805 of which may be in UTF-8) are the same case-insensitively; false otherwise.
2806 How far into the strings to compare is determined by other input parameters.
2808 If u1 is true, the string s1 is assumed to be in UTF-8-encoded Unicode;
2809 otherwise it is assumed to be in native 8-bit encoding. Correspondingly for u2
2812 If the byte length l1 is non-zero, it says how far into s1 to check for fold
2813 equality. In other words, s1+l1 will be used as a goal to reach. The
2814 scan will not be considered to be a match unless the goal is reached, and
2815 scanning won't continue past that goal. Correspondingly for l2 with respect to
2818 If pe1 is non-NULL and the pointer it points to is not NULL, that pointer is
2819 considered an end pointer beyond which scanning of s1 will not continue under
2820 any circumstances. This means that if both l1 and pe1 are specified, and pe1
2821 is less than s1+l1, the match will never be successful because it can never
2822 get as far as its goal (and in fact is asserted against). Correspondingly for
2823 pe2 with respect to s2.
2825 At least one of s1 and s2 must have a goal (at least one of l1 and l2 must be
2826 non-zero), and if both do, both have to be
2827 reached for a successful match. Also, if the fold of a character is multiple
2828 characters, all of them must be matched (see tr21 reference below for
2831 Upon a successful match, if pe1 is non-NULL,
2832 it will be set to point to the beginning of the I<next> character of s1 beyond
2833 what was matched. Correspondingly for pe2 and s2.
2835 For case-insensitiveness, the "casefolding" of Unicode is used
2836 instead of upper/lowercasing both the characters, see
2837 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
2841 Perl_foldEQ_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
2844 register const U8 *p1 = (const U8*)s1; /* Point to current char */
2845 register const U8 *p2 = (const U8*)s2;
2846 register const U8 *g1 = NULL; /* goal for s1 */
2847 register const U8 *g2 = NULL;
2848 register const U8 *e1 = NULL; /* Don't scan s1 past this */
2849 register U8 *f1 = NULL; /* Point to current folded */
2850 register const U8 *e2 = NULL;
2851 register U8 *f2 = NULL;
2852 STRLEN n1 = 0, n2 = 0; /* Number of bytes in current char */
2853 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
2854 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
2855 U8 natbuf[2]; /* Holds native 8-bit char converted to utf8;
2856 these always fit in 2 bytes */
2858 PERL_ARGS_ASSERT_FOLDEQ_UTF8;
2865 g1 = (const U8*)s1 + l1;
2873 g2 = (const U8*)s2 + l2;
2876 /* Must have at least one goal */
2881 /* Will never match if goal is out-of-bounds */
2882 assert(! e1 || e1 >= g1);
2884 /* Here, there isn't an end pointer, or it is beyond the goal. We
2885 * only go as far as the goal */
2889 assert(e1); /* Must have an end for looking at s1 */
2892 /* Same for goal for s2 */
2894 assert(! e2 || e2 >= g2);
2901 /* Look through both strings, a character at a time */
2902 while (p1 < e1 && p2 < e2) {
2904 /* If at the beginning of a new character in s1, get its fold to use
2905 * and the length of the fold */
2908 to_utf8_fold(p1, foldbuf1, &n1);
2910 else { /* Not utf8, convert to it first and then get fold */
2911 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p1)));
2912 to_utf8_fold(natbuf, foldbuf1, &n1);
2917 if (n2 == 0) { /* Same for s2 */
2919 to_utf8_fold(p2, foldbuf2, &n2);
2922 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p2)));
2923 to_utf8_fold(natbuf, foldbuf2, &n2);
2928 /* While there is more to look for in both folds, see if they
2929 * continue to match */
2931 U8 fold_length = UTF8SKIP(f1);
2932 if (fold_length != UTF8SKIP(f2)
2933 || (fold_length == 1 && *f1 != *f2) /* Short circuit memNE
2934 function call for single
2936 || memNE((char*)f1, (char*)f2, fold_length))
2938 return 0; /* mismatch */
2941 /* Here, they matched, advance past them */
2948 /* When reach the end of any fold, advance the input past it */
2950 p1 += u1 ? UTF8SKIP(p1) : 1;
2953 p2 += u2 ? UTF8SKIP(p2) : 1;
2955 } /* End of loop through both strings */
2957 /* A match is defined by each scan that specified an explicit length
2958 * reaching its final goal, and the other not having matched a partial
2959 * character (which can happen when the fold of a character is more than one
2961 if (! ((g1 == 0 || p1 == g1) && (g2 == 0 || p2 == g2)) || n1 || n2) {
2965 /* Successful match. Set output pointers */
2977 * c-indentation-style: bsd
2979 * indent-tabs-mode: t
2982 * ex: set ts=8 sts=4 sw=4 noet: