3 * Copyright (C) 2000, 2001, 2002, 2003, 2004, 2005 by Larry Wall and
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * 'What a fix!' said Sam. 'That's the one place in all the lands we've ever
13 * heard of that we don't want to see any closer; and that's the one place
14 * we're trying to get to! And that's just where we can't get, nohow.'
16 * 'Well do I understand your speech,' he answered in the same language;
17 * 'yet few strangers do so. Why then do you not speak in the Common Tongue,
18 * as is the custom in the West, if you wish to be answered?'
20 * ...the travellers perceived that the floor was paved with stones of many
21 * hues; branching runes and strange devices intertwined beneath their feet.
25 #define PERL_IN_UTF8_C
28 static const char unees[] =
29 "Malformed UTF-8 character (unexpected end of string)";
32 =head1 Unicode Support
34 This file contains various utility functions for manipulating UTF8-encoded
35 strings. For the uninitiated, this is a method of representing arbitrary
36 Unicode characters as a variable number of bytes, in such a way that
37 characters in the ASCII range are unmodified, and a zero byte never appears
38 within non-zero characters.
40 =for apidoc A|U8 *|uvuni_to_utf8_flags|U8 *d|UV uv|UV flags
42 Adds the UTF-8 representation of the Unicode codepoint C<uv> to the end
43 of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
44 bytes available. The return value is the pointer to the byte after the
45 end of the new character. In other words,
47 d = uvuni_to_utf8_flags(d, uv, flags);
51 d = uvuni_to_utf8(d, uv);
53 (which is equivalent to)
55 d = uvuni_to_utf8_flags(d, uv, 0);
57 is the recommended Unicode-aware way of saying
65 Perl_uvuni_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
67 if (ckWARN(WARN_UTF8)) {
68 if (UNICODE_IS_SURROGATE(uv) &&
69 !(flags & UNICODE_ALLOW_SURROGATE))
70 Perl_warner(aTHX_ packWARN(WARN_UTF8), "UTF-16 surrogate 0x%04"UVxf, uv);
72 ((uv >= 0xFDD0 && uv <= 0xFDEF &&
73 !(flags & UNICODE_ALLOW_FDD0))
75 ((uv & 0xFFFE) == 0xFFFE && /* Either FFFE or FFFF. */
76 !(flags & UNICODE_ALLOW_FFFF))) &&
77 /* UNICODE_ALLOW_SUPER includes
78 * FFFEs and FFFFs beyond 0x10FFFF. */
79 ((uv <= PERL_UNICODE_MAX) ||
80 !(flags & UNICODE_ALLOW_SUPER))
82 Perl_warner(aTHX_ packWARN(WARN_UTF8),
83 "Unicode character 0x%04"UVxf" is illegal", uv);
85 if (UNI_IS_INVARIANT(uv)) {
86 *d++ = (U8)UTF_TO_NATIVE(uv);
91 STRLEN len = UNISKIP(uv);
94 *p-- = (U8)UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
95 uv >>= UTF_ACCUMULATION_SHIFT;
97 *p = (U8)UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
100 #else /* Non loop style */
102 *d++ = (U8)(( uv >> 6) | 0xc0);
103 *d++ = (U8)(( uv & 0x3f) | 0x80);
107 *d++ = (U8)(( uv >> 12) | 0xe0);
108 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
109 *d++ = (U8)(( uv & 0x3f) | 0x80);
113 *d++ = (U8)(( uv >> 18) | 0xf0);
114 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
115 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
116 *d++ = (U8)(( uv & 0x3f) | 0x80);
119 if (uv < 0x4000000) {
120 *d++ = (U8)(( uv >> 24) | 0xf8);
121 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
122 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
123 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
124 *d++ = (U8)(( uv & 0x3f) | 0x80);
127 if (uv < 0x80000000) {
128 *d++ = (U8)(( uv >> 30) | 0xfc);
129 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
130 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
131 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
132 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
133 *d++ = (U8)(( uv & 0x3f) | 0x80);
137 if (uv < UTF8_QUAD_MAX)
140 *d++ = 0xfe; /* Can't match U+FEFF! */
141 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
142 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
143 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
144 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
145 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
146 *d++ = (U8)(( uv & 0x3f) | 0x80);
151 *d++ = 0xff; /* Can't match U+FFFE! */
152 *d++ = 0x80; /* 6 Reserved bits */
153 *d++ = (U8)(((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
154 *d++ = (U8)(((uv >> 54) & 0x3f) | 0x80);
155 *d++ = (U8)(((uv >> 48) & 0x3f) | 0x80);
156 *d++ = (U8)(((uv >> 42) & 0x3f) | 0x80);
157 *d++ = (U8)(((uv >> 36) & 0x3f) | 0x80);
158 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
159 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
160 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
161 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
162 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
163 *d++ = (U8)(( uv & 0x3f) | 0x80);
167 #endif /* Loop style */
171 Perl_uvuni_to_utf8(pTHX_ U8 *d, UV uv)
173 return Perl_uvuni_to_utf8_flags(aTHX_ d, uv, 0);
178 Tests if some arbitrary number of bytes begins in a valid UTF-8
179 character. Note that an INVARIANT (i.e. ASCII) character is a valid
180 UTF-8 character. The actual number of bytes in the UTF-8 character
181 will be returned if it is valid, otherwise 0.
183 This is the "slow" version as opposed to the "fast" version which is
184 the "unrolled" IS_UTF8_CHAR(). E.g. for t/uni/class.t the speed
185 difference is a factor of 2 to 3. For lengths (UTF8SKIP(s)) of four
186 or less you should use the IS_UTF8_CHAR(), for lengths of five or more
187 you should use the _slow(). In practice this means that the _slow()
188 will be used very rarely, since the maximum Unicode code point (as of
189 Unicode 4.1) is U+10FFFF, which encodes in UTF-8 to four bytes. Only
190 the "Perl extended UTF-8" (the infamous 'v-strings') will encode into
195 S_is_utf8_char_slow(pTHX_ const U8 *s, const STRLEN len)
201 if (UTF8_IS_INVARIANT(u))
204 if (!UTF8_IS_START(u))
207 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
213 u = NATIVE_TO_UTF(u);
215 u &= UTF_START_MASK(len);
219 if (!UTF8_IS_CONTINUATION(*s))
221 uv = UTF8_ACCUMULATE(uv, *s);
228 if ((STRLEN)UNISKIP(uv) < len)
235 =for apidoc A|STRLEN|is_utf8_char|const U8 *s
237 Tests if some arbitrary number of bytes begins in a valid UTF-8
238 character. Note that an INVARIANT (i.e. ASCII) character is a valid
239 UTF-8 character. The actual number of bytes in the UTF-8 character
240 will be returned if it is valid, otherwise 0.
244 Perl_is_utf8_char(pTHX_ const U8 *s)
246 const STRLEN len = UTF8SKIP(s);
248 if (IS_UTF8_CHAR_FAST(len))
249 return IS_UTF8_CHAR(s, len) ? len : 0;
250 #endif /* #ifdef IS_UTF8_CHAR */
251 return is_utf8_char_slow(s, len);
255 =for apidoc A|bool|is_utf8_string|const U8 *s|STRLEN len
257 Returns true if first C<len> bytes of the given string form a valid
258 UTF-8 string, false otherwise. Note that 'a valid UTF-8 string' does
259 not mean 'a string that contains code points above 0x7F encoded in UTF-8'
260 because a valid ASCII string is a valid UTF-8 string.
262 See also is_utf8_string_loclen() and is_utf8_string_loc().
268 Perl_is_utf8_string(pTHX_ const U8 *s, STRLEN len)
274 len = strlen((const char *)s);
279 /* Inline the easy bits of is_utf8_char() here for speed... */
280 if (UTF8_IS_INVARIANT(*x))
282 else if (!UTF8_IS_START(*x))
285 /* ... and call is_utf8_char() only if really needed. */
288 if (IS_UTF8_CHAR_FAST(c)) {
289 if (!IS_UTF8_CHAR(x, c))
291 } else if (!is_utf8_char_slow(x, c))
295 #endif /* #ifdef IS_UTF8_CHAR */
310 Implemented as a macro in utf8.h
312 =for apidoc A|bool|is_utf8_string_loc|const U8 *s|STRLEN len|const U8 **ep
314 Like is_utf8_string() but stores the location of the failure (in the
315 case of "utf8ness failure") or the location s+len (in the case of
316 "utf8ness success") in the C<ep>.
318 See also is_utf8_string_loclen() and is_utf8_string().
320 =for apidoc A|bool|is_utf8_string_loclen|const U8 *s|STRLEN len|const U8 **ep|const STRLEN *el
322 Like is_utf8_string() but stores the location of the failure (in the
323 case of "utf8ness failure") or the location s+len (in the case of
324 "utf8ness success") in the C<ep>, and the number of UTF-8
325 encoded characters in the C<el>.
327 See also is_utf8_string_loc() and is_utf8_string().
333 Perl_is_utf8_string_loclen(pTHX_ const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
340 len = strlen((const char *)s);
346 /* Inline the easy bits of is_utf8_char() here for speed... */
347 if (UTF8_IS_INVARIANT(*x))
349 else if (!UTF8_IS_START(*x))
352 /* ... and call is_utf8_char() only if really needed. */
355 if (IS_UTF8_CHAR_FAST(c)) {
356 if (!IS_UTF8_CHAR(x, c))
359 c = is_utf8_char_slow(x, c);
362 #endif /* #ifdef IS_UTF8_CHAR */
382 =for apidoc A|UV|utf8n_to_uvuni|const U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
384 Bottom level UTF-8 decode routine.
385 Returns the unicode code point value of the first character in the string C<s>
386 which is assumed to be in UTF-8 encoding and no longer than C<curlen>;
387 C<retlen> will be set to the length, in bytes, of that character.
389 If C<s> does not point to a well-formed UTF-8 character, the behaviour
390 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
391 it is assumed that the caller will raise a warning, and this function
392 will silently just set C<retlen> to C<-1> and return zero. If the
393 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
394 malformations will be given, C<retlen> will be set to the expected
395 length of the UTF-8 character in bytes, and zero will be returned.
397 The C<flags> can also contain various flags to allow deviations from
398 the strict UTF-8 encoding (see F<utf8.h>).
400 Most code should use utf8_to_uvchr() rather than call this directly.
406 Perl_utf8n_to_uvuni(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
411 const bool dowarn = ckWARN_d(WARN_UTF8);
412 const UV startbyte = *s;
413 STRLEN expectlen = 0;
416 /* This list is a superset of the UTF8_ALLOW_XXX. */
418 #define UTF8_WARN_EMPTY 1
419 #define UTF8_WARN_CONTINUATION 2
420 #define UTF8_WARN_NON_CONTINUATION 3
421 #define UTF8_WARN_FE_FF 4
422 #define UTF8_WARN_SHORT 5
423 #define UTF8_WARN_OVERFLOW 6
424 #define UTF8_WARN_SURROGATE 7
425 #define UTF8_WARN_LONG 8
426 #define UTF8_WARN_FFFF 9 /* Also FFFE. */
429 !(flags & UTF8_ALLOW_EMPTY)) {
430 warning = UTF8_WARN_EMPTY;
434 if (UTF8_IS_INVARIANT(uv)) {
437 return (UV) (NATIVE_TO_UTF(*s));
440 if (UTF8_IS_CONTINUATION(uv) &&
441 !(flags & UTF8_ALLOW_CONTINUATION)) {
442 warning = UTF8_WARN_CONTINUATION;
446 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
447 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
448 warning = UTF8_WARN_NON_CONTINUATION;
453 uv = NATIVE_TO_UTF(uv);
455 if ((uv == 0xfe || uv == 0xff) &&
456 !(flags & UTF8_ALLOW_FE_FF)) {
457 warning = UTF8_WARN_FE_FF;
462 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
463 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
464 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
465 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
467 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
468 else { len = 7; uv &= 0x01; }
470 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
471 else if (!(uv & 0x01)) { len = 7; uv = 0; }
472 else { len = 13; uv = 0; } /* whoa! */
480 if ((curlen < expectlen) &&
481 !(flags & UTF8_ALLOW_SHORT)) {
482 warning = UTF8_WARN_SHORT;
491 if (!UTF8_IS_CONTINUATION(*s) &&
492 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
494 warning = UTF8_WARN_NON_CONTINUATION;
498 uv = UTF8_ACCUMULATE(uv, *s);
500 /* These cannot be allowed. */
502 if (expectlen != 13 && !(flags & UTF8_ALLOW_LONG)) {
503 warning = UTF8_WARN_LONG;
507 else { /* uv < ouv */
508 /* This cannot be allowed. */
509 warning = UTF8_WARN_OVERFLOW;
517 if (UNICODE_IS_SURROGATE(uv) &&
518 !(flags & UTF8_ALLOW_SURROGATE)) {
519 warning = UTF8_WARN_SURROGATE;
521 } else if ((expectlen > (STRLEN)UNISKIP(uv)) &&
522 !(flags & UTF8_ALLOW_LONG)) {
523 warning = UTF8_WARN_LONG;
525 } else if (UNICODE_IS_ILLEGAL(uv) &&
526 !(flags & UTF8_ALLOW_FFFF)) {
527 warning = UTF8_WARN_FFFF;
535 if (flags & UTF8_CHECK_ONLY) {
542 SV* const sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
545 case 0: /* Intentionally empty. */ break;
546 case UTF8_WARN_EMPTY:
547 Perl_sv_catpv(aTHX_ sv, "(empty string)");
549 case UTF8_WARN_CONTINUATION:
550 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
552 case UTF8_WARN_NON_CONTINUATION:
554 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
555 (UV)s[1], startbyte);
557 const int len = (int)(s-s0);
558 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
559 (UV)s[1], len, len > 1 ? "s" : "", startbyte, (int)expectlen);
563 case UTF8_WARN_FE_FF:
564 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
566 case UTF8_WARN_SHORT:
567 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
568 (int)curlen, curlen == 1 ? "" : "s", (int)expectlen, startbyte);
569 expectlen = curlen; /* distance for caller to skip */
571 case UTF8_WARN_OVERFLOW:
572 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
575 case UTF8_WARN_SURROGATE:
576 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
579 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
580 (int)expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
583 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
586 Perl_sv_catpv(aTHX_ sv, "(unknown reason)");
591 const char * const s = SvPVX_const(sv);
594 Perl_warner(aTHX_ packWARN(WARN_UTF8),
595 "%s in %s", s, OP_DESC(PL_op));
597 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
602 *retlen = expectlen ? expectlen : len;
608 =for apidoc A|UV|utf8_to_uvchr|const U8 *s|STRLEN *retlen
610 Returns the native character value of the first character in the string C<s>
611 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
612 length, in bytes, of that character.
614 If C<s> does not point to a well-formed UTF-8 character, zero is
615 returned and retlen is set, if possible, to -1.
621 Perl_utf8_to_uvchr(pTHX_ const U8 *s, STRLEN *retlen)
623 return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXBYTES, retlen,
624 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
628 =for apidoc A|UV|utf8_to_uvuni|const U8 *s|STRLEN *retlen
630 Returns the Unicode code point of the first character in the string C<s>
631 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
632 length, in bytes, of that character.
634 This function should only be used when returned UV is considered
635 an index into the Unicode semantic tables (e.g. swashes).
637 If C<s> does not point to a well-formed UTF-8 character, zero is
638 returned and retlen is set, if possible, to -1.
644 Perl_utf8_to_uvuni(pTHX_ const U8 *s, STRLEN *retlen)
646 /* Call the low level routine asking for checks */
647 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXBYTES, retlen,
648 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
652 =for apidoc A|STRLEN|utf8_length|const U8 *s|const U8 *e
654 Return the length of the UTF-8 char encoded string C<s> in characters.
655 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
656 up past C<e>, croaks.
662 Perl_utf8_length(pTHX_ const U8 *s, const U8 *e)
666 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
667 * the bitops (especially ~) can create illegal UTF-8.
668 * In other words: in Perl UTF-8 is not just for Unicode. */
671 goto warn_and_return;
673 const U8 t = UTF8SKIP(s);
676 if (ckWARN_d(WARN_UTF8)) {
678 Perl_warner(aTHX_ packWARN(WARN_UTF8),
679 "%s in %s", unees, OP_DESC(PL_op));
681 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
693 =for apidoc A|IV|utf8_distance|const U8 *a|const U8 *b
695 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
698 WARNING: use only if you *know* that the pointers point inside the
705 Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
709 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
710 * the bitops (especially ~) can create illegal UTF-8.
711 * In other words: in Perl UTF-8 is not just for Unicode. */
715 const U8 c = UTF8SKIP(a);
717 goto warn_and_return;
724 const U8 c = UTF8SKIP(b);
728 if (ckWARN_d(WARN_UTF8)) {
730 Perl_warner(aTHX_ packWARN(WARN_UTF8),
731 "%s in %s", unees, OP_DESC(PL_op));
733 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
746 =for apidoc A|U8 *|utf8_hop|U8 *s|I32 off
748 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
751 WARNING: do not use the following unless you *know* C<off> is within
752 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
753 on the first byte of character or just after the last byte of a character.
759 Perl_utf8_hop(pTHX_ const U8 *s, I32 off)
761 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
762 * the bitops (especially ~) can create illegal UTF-8.
763 * In other words: in Perl UTF-8 is not just for Unicode. */
772 while (UTF8_IS_CONTINUATION(*s))
780 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
782 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
783 Unlike C<bytes_to_utf8>, this over-writes the original string, and
784 updates len to contain the new length.
785 Returns zero on failure, setting C<len> to -1.
791 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
797 /* ensure valid UTF-8 and chars < 256 before updating string */
798 for (send = s + *len; s < send; ) {
801 if (!UTF8_IS_INVARIANT(c) &&
802 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
803 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
812 *d++ = (U8)utf8_to_uvchr(s, &ulen);
821 =for apidoc A|U8 *|bytes_from_utf8|const U8 *s|STRLEN *len|bool *is_utf8
823 Converts a string C<s> of length C<len> from UTF-8 into byte encoding.
824 Unlike C<utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
825 the newly-created string, and updates C<len> to contain the new
826 length. Returns the original string if no conversion occurs, C<len>
827 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
828 0 if C<s> is converted or contains all 7bit characters.
834 Perl_bytes_from_utf8(pTHX_ const U8 *s, STRLEN *len, bool *is_utf8)
844 /* ensure valid UTF-8 and chars < 256 before converting string */
845 for (send = s + *len; s < send;) {
847 if (!UTF8_IS_INVARIANT(c)) {
848 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
849 (c = *s++) && UTF8_IS_CONTINUATION(c))
858 Newxz(d, (*len) - count + 1, U8);
859 s = start; start = d;
862 if (!UTF8_IS_INVARIANT(c)) {
863 /* Then it is two-byte encoded */
864 c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++);
865 c = ASCII_TO_NATIVE(c);
875 =for apidoc A|U8 *|bytes_to_utf8|const U8 *s|STRLEN *len
877 Converts a string C<s> of length C<len> from ASCII into UTF-8 encoding.
878 Returns a pointer to the newly-created string, and sets C<len> to
879 reflect the new length.
881 If you want to convert to UTF-8 from other encodings than ASCII,
882 see sv_recode_to_utf8().
888 Perl_bytes_to_utf8(pTHX_ const U8 *s, STRLEN *len)
890 const U8 * const send = s + (*len);
894 Newxz(d, (*len) * 2 + 1, U8);
898 const UV uv = NATIVE_TO_ASCII(*s++);
899 if (UNI_IS_INVARIANT(uv))
900 *d++ = (U8)UTF_TO_NATIVE(uv);
902 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
903 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
912 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
914 * Destination must be pre-extended to 3/2 source. Do not use in-place.
915 * We optimize for native, for obvious reasons. */
918 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
923 if (bytelen == 1 && p[0] == 0) { /* Be understanding. */
930 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %"UVf, (UV)bytelen);
935 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
942 *d++ = (U8)(( uv >> 6) | 0xc0);
943 *d++ = (U8)(( uv & 0x3f) | 0x80);
946 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
947 UV low = (p[0] << 8) + p[1];
949 if (low < 0xdc00 || low >= 0xdfff)
950 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
951 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
954 *d++ = (U8)(( uv >> 12) | 0xe0);
955 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
956 *d++ = (U8)(( uv & 0x3f) | 0x80);
960 *d++ = (U8)(( uv >> 18) | 0xf0);
961 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
962 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
963 *d++ = (U8)(( uv & 0x3f) | 0x80);
967 *newlen = d - dstart;
971 /* Note: this one is slightly destructive of the source. */
974 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
977 U8* send = s + bytelen;
984 return utf16_to_utf8(p, d, bytelen, newlen);
987 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
990 Perl_is_uni_alnum(pTHX_ UV c)
992 U8 tmpbuf[UTF8_MAXBYTES+1];
993 uvchr_to_utf8(tmpbuf, c);
994 return is_utf8_alnum(tmpbuf);
998 Perl_is_uni_alnumc(pTHX_ UV c)
1000 U8 tmpbuf[UTF8_MAXBYTES+1];
1001 uvchr_to_utf8(tmpbuf, c);
1002 return is_utf8_alnumc(tmpbuf);
1006 Perl_is_uni_idfirst(pTHX_ UV c)
1008 U8 tmpbuf[UTF8_MAXBYTES+1];
1009 uvchr_to_utf8(tmpbuf, c);
1010 return is_utf8_idfirst(tmpbuf);
1014 Perl_is_uni_alpha(pTHX_ UV c)
1016 U8 tmpbuf[UTF8_MAXBYTES+1];
1017 uvchr_to_utf8(tmpbuf, c);
1018 return is_utf8_alpha(tmpbuf);
1022 Perl_is_uni_ascii(pTHX_ UV c)
1024 U8 tmpbuf[UTF8_MAXBYTES+1];
1025 uvchr_to_utf8(tmpbuf, c);
1026 return is_utf8_ascii(tmpbuf);
1030 Perl_is_uni_space(pTHX_ UV c)
1032 U8 tmpbuf[UTF8_MAXBYTES+1];
1033 uvchr_to_utf8(tmpbuf, c);
1034 return is_utf8_space(tmpbuf);
1038 Perl_is_uni_digit(pTHX_ UV c)
1040 U8 tmpbuf[UTF8_MAXBYTES+1];
1041 uvchr_to_utf8(tmpbuf, c);
1042 return is_utf8_digit(tmpbuf);
1046 Perl_is_uni_upper(pTHX_ UV c)
1048 U8 tmpbuf[UTF8_MAXBYTES+1];
1049 uvchr_to_utf8(tmpbuf, c);
1050 return is_utf8_upper(tmpbuf);
1054 Perl_is_uni_lower(pTHX_ UV c)
1056 U8 tmpbuf[UTF8_MAXBYTES+1];
1057 uvchr_to_utf8(tmpbuf, c);
1058 return is_utf8_lower(tmpbuf);
1062 Perl_is_uni_cntrl(pTHX_ UV c)
1064 U8 tmpbuf[UTF8_MAXBYTES+1];
1065 uvchr_to_utf8(tmpbuf, c);
1066 return is_utf8_cntrl(tmpbuf);
1070 Perl_is_uni_graph(pTHX_ UV c)
1072 U8 tmpbuf[UTF8_MAXBYTES+1];
1073 uvchr_to_utf8(tmpbuf, c);
1074 return is_utf8_graph(tmpbuf);
1078 Perl_is_uni_print(pTHX_ UV c)
1080 U8 tmpbuf[UTF8_MAXBYTES+1];
1081 uvchr_to_utf8(tmpbuf, c);
1082 return is_utf8_print(tmpbuf);
1086 Perl_is_uni_punct(pTHX_ UV c)
1088 U8 tmpbuf[UTF8_MAXBYTES+1];
1089 uvchr_to_utf8(tmpbuf, c);
1090 return is_utf8_punct(tmpbuf);
1094 Perl_is_uni_xdigit(pTHX_ UV c)
1096 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1097 uvchr_to_utf8(tmpbuf, c);
1098 return is_utf8_xdigit(tmpbuf);
1102 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
1104 uvchr_to_utf8(p, c);
1105 return to_utf8_upper(p, p, lenp);
1109 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
1111 uvchr_to_utf8(p, c);
1112 return to_utf8_title(p, p, lenp);
1116 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
1118 uvchr_to_utf8(p, c);
1119 return to_utf8_lower(p, p, lenp);
1123 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1125 uvchr_to_utf8(p, c);
1126 return to_utf8_fold(p, p, lenp);
1129 /* for now these all assume no locale info available for Unicode > 255 */
1132 Perl_is_uni_alnum_lc(pTHX_ UV c)
1134 return is_uni_alnum(c); /* XXX no locale support yet */
1138 Perl_is_uni_alnumc_lc(pTHX_ UV c)
1140 return is_uni_alnumc(c); /* XXX no locale support yet */
1144 Perl_is_uni_idfirst_lc(pTHX_ UV c)
1146 return is_uni_idfirst(c); /* XXX no locale support yet */
1150 Perl_is_uni_alpha_lc(pTHX_ UV c)
1152 return is_uni_alpha(c); /* XXX no locale support yet */
1156 Perl_is_uni_ascii_lc(pTHX_ UV c)
1158 return is_uni_ascii(c); /* XXX no locale support yet */
1162 Perl_is_uni_space_lc(pTHX_ UV c)
1164 return is_uni_space(c); /* XXX no locale support yet */
1168 Perl_is_uni_digit_lc(pTHX_ UV c)
1170 return is_uni_digit(c); /* XXX no locale support yet */
1174 Perl_is_uni_upper_lc(pTHX_ UV c)
1176 return is_uni_upper(c); /* XXX no locale support yet */
1180 Perl_is_uni_lower_lc(pTHX_ UV c)
1182 return is_uni_lower(c); /* XXX no locale support yet */
1186 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1188 return is_uni_cntrl(c); /* XXX no locale support yet */
1192 Perl_is_uni_graph_lc(pTHX_ UV c)
1194 return is_uni_graph(c); /* XXX no locale support yet */
1198 Perl_is_uni_print_lc(pTHX_ UV c)
1200 return is_uni_print(c); /* XXX no locale support yet */
1204 Perl_is_uni_punct_lc(pTHX_ UV c)
1206 return is_uni_punct(c); /* XXX no locale support yet */
1210 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1212 return is_uni_xdigit(c); /* XXX no locale support yet */
1216 Perl_to_uni_upper_lc(pTHX_ U32 c)
1218 /* XXX returns only the first character -- do not use XXX */
1219 /* XXX no locale support yet */
1221 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1222 return (U32)to_uni_upper(c, tmpbuf, &len);
1226 Perl_to_uni_title_lc(pTHX_ U32 c)
1228 /* XXX returns only the first character XXX -- do not use XXX */
1229 /* XXX no locale support yet */
1231 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1232 return (U32)to_uni_title(c, tmpbuf, &len);
1236 Perl_to_uni_lower_lc(pTHX_ U32 c)
1238 /* XXX returns only the first character -- do not use XXX */
1239 /* XXX no locale support yet */
1241 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1242 return (U32)to_uni_lower(c, tmpbuf, &len);
1246 Perl_is_utf8_alnum(pTHX_ const U8 *p)
1248 if (!is_utf8_char(p))
1251 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1252 * descendant of isalnum(3), in other words, it doesn't
1253 * contain the '_'. --jhi */
1254 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1255 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1256 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1257 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1259 PL_utf8_alnum = swash_init("utf8", "",
1260 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1261 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1266 Perl_is_utf8_alnumc(pTHX_ const U8 *p)
1268 if (!is_utf8_char(p))
1271 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1272 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1273 /* return is_utf8_alpha(p) || is_utf8_digit(p); */
1274 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1276 PL_utf8_alnum = swash_init("utf8", "",
1277 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1278 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1283 Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
1287 if (!is_utf8_char(p))
1289 if (!PL_utf8_idstart) /* is_utf8_idstart would be more logical. */
1290 PL_utf8_idstart = swash_init("utf8", "IdStart", &PL_sv_undef, 0, 0);
1291 return swash_fetch(PL_utf8_idstart, p, TRUE) != 0;
1295 Perl_is_utf8_idcont(pTHX_ const U8 *p)
1299 if (!is_utf8_char(p))
1301 if (!PL_utf8_idcont)
1302 PL_utf8_idcont = swash_init("utf8", "IdContinue", &PL_sv_undef, 0, 0);
1303 return swash_fetch(PL_utf8_idcont, p, TRUE) != 0;
1307 Perl_is_utf8_alpha(pTHX_ const U8 *p)
1309 if (!is_utf8_char(p))
1312 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1313 return swash_fetch(PL_utf8_alpha, p, TRUE) != 0;
1317 Perl_is_utf8_ascii(pTHX_ const U8 *p)
1319 if (!is_utf8_char(p))
1322 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1323 return swash_fetch(PL_utf8_ascii, p, TRUE) != 0;
1327 Perl_is_utf8_space(pTHX_ const U8 *p)
1329 if (!is_utf8_char(p))
1332 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1333 return swash_fetch(PL_utf8_space, p, TRUE) != 0;
1337 Perl_is_utf8_digit(pTHX_ const U8 *p)
1339 if (!is_utf8_char(p))
1342 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1343 return swash_fetch(PL_utf8_digit, p, TRUE) != 0;
1347 Perl_is_utf8_upper(pTHX_ const U8 *p)
1349 if (!is_utf8_char(p))
1352 PL_utf8_upper = swash_init("utf8", "IsUppercase", &PL_sv_undef, 0, 0);
1353 return swash_fetch(PL_utf8_upper, p, TRUE) != 0;
1357 Perl_is_utf8_lower(pTHX_ const U8 *p)
1359 if (!is_utf8_char(p))
1362 PL_utf8_lower = swash_init("utf8", "IsLowercase", &PL_sv_undef, 0, 0);
1363 return swash_fetch(PL_utf8_lower, p, TRUE) != 0;
1367 Perl_is_utf8_cntrl(pTHX_ const U8 *p)
1369 if (!is_utf8_char(p))
1372 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1373 return swash_fetch(PL_utf8_cntrl, p, TRUE) != 0;
1377 Perl_is_utf8_graph(pTHX_ const U8 *p)
1379 if (!is_utf8_char(p))
1382 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1383 return swash_fetch(PL_utf8_graph, p, TRUE) != 0;
1387 Perl_is_utf8_print(pTHX_ const U8 *p)
1389 if (!is_utf8_char(p))
1392 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1393 return swash_fetch(PL_utf8_print, p, TRUE) != 0;
1397 Perl_is_utf8_punct(pTHX_ const U8 *p)
1399 if (!is_utf8_char(p))
1402 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1403 return swash_fetch(PL_utf8_punct, p, TRUE) != 0;
1407 Perl_is_utf8_xdigit(pTHX_ const U8 *p)
1409 if (!is_utf8_char(p))
1411 if (!PL_utf8_xdigit)
1412 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1413 return swash_fetch(PL_utf8_xdigit, p, TRUE) != 0;
1417 Perl_is_utf8_mark(pTHX_ const U8 *p)
1419 if (!is_utf8_char(p))
1422 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1423 return swash_fetch(PL_utf8_mark, p, TRUE) != 0;
1427 =for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special
1429 The "p" contains the pointer to the UTF-8 string encoding
1430 the character that is being converted.
1432 The "ustrp" is a pointer to the character buffer to put the
1433 conversion result to. The "lenp" is a pointer to the length
1436 The "swashp" is a pointer to the swash to use.
1438 Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
1439 and loaded by SWASHGET, using lib/utf8_heavy.pl. The special (usually,
1440 but not always, a multicharacter mapping), is tried first.
1442 The "special" is a string like "utf8::ToSpecLower", which means the
1443 hash %utf8::ToSpecLower. The access to the hash is through
1444 Perl_to_utf8_case().
1446 The "normal" is a string like "ToLower" which means the swash
1452 Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp,
1453 SV **swashp, const char *normal, const char *special)
1455 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
1458 const UV uv0 = utf8_to_uvchr(p, NULL);
1459 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1460 * are necessary in EBCDIC, they are redundant no-ops
1461 * in ASCII-ish platforms, and hopefully optimized away. */
1462 const UV uv1 = NATIVE_TO_UNI(uv0);
1463 uvuni_to_utf8(tmpbuf, uv1);
1465 if (!*swashp) /* load on-demand */
1466 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1468 /* The 0xDF is the only special casing Unicode code point below 0x100. */
1469 if (special && (uv1 == 0xDF || uv1 > 0xFF)) {
1470 /* It might be "special" (sometimes, but not always,
1471 * a multicharacter mapping) */
1475 if ((hv = get_hv(special, FALSE)) &&
1476 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1480 s = SvPV_const(*svp, len);
1482 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
1485 /* If we have EBCDIC we need to remap the characters
1486 * since any characters in the low 256 are Unicode
1487 * code points, not EBCDIC. */
1488 U8 *t = (U8*)s, *tend = t + len, *d;
1495 UV c = utf8_to_uvchr(t, &tlen);
1497 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1506 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1511 Copy(tmpbuf, ustrp, len, U8);
1513 Copy(s, ustrp, len, U8);
1519 if (!len && *swashp) {
1520 UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1523 /* It was "normal" (a single character mapping). */
1524 UV uv3 = UNI_TO_NATIVE(uv2);
1526 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
1530 if (!len) /* Neither: just copy. */
1531 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1536 return len ? utf8_to_uvchr(ustrp, 0) : 0;
1540 =for apidoc A|UV|to_utf8_upper|const U8 *p|U8 *ustrp|STRLEN *lenp
1542 Convert the UTF-8 encoded character at p to its uppercase version and
1543 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1544 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1545 the uppercase version may be longer than the original character.
1547 The first character of the uppercased version is returned
1548 (but note, as explained above, that there may be more.)
1553 Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1555 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1556 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1560 =for apidoc A|UV|to_utf8_title|const U8 *p|U8 *ustrp|STRLEN *lenp
1562 Convert the UTF-8 encoded character at p to its titlecase version and
1563 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1564 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1565 titlecase version may be longer than the original character.
1567 The first character of the titlecased version is returned
1568 (but note, as explained above, that there may be more.)
1573 Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1575 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1576 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1580 =for apidoc A|UV|to_utf8_lower|const U8 *p|U8 *ustrp|STRLEN *lenp
1582 Convert the UTF-8 encoded character at p to its lowercase version and
1583 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1584 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1585 lowercase version may be longer than the original character.
1587 The first character of the lowercased version is returned
1588 (but note, as explained above, that there may be more.)
1593 Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1595 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1596 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1600 =for apidoc A|UV|to_utf8_fold|const U8 *p|U8 *ustrp|STRLEN *lenp
1602 Convert the UTF-8 encoded character at p to its foldcase version and
1603 store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1604 that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1605 foldcase version may be longer than the original character (up to
1608 The first character of the foldcased version is returned
1609 (but note, as explained above, that there may be more.)
1614 Perl_to_utf8_fold(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
1616 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1617 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1620 /* a "swash" is a swatch hash */
1623 Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
1627 SV* const tokenbufsv = sv_newmortal();
1629 const size_t pkg_len = strlen(pkg);
1630 const size_t name_len = strlen(name);
1631 HV * const stash = gv_stashpvn(pkg, pkg_len, FALSE);
1634 PUSHSTACKi(PERLSI_MAGIC);
1639 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
1641 errsv_save = newSVsv(ERRSV);
1642 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
1645 sv_setsv(ERRSV, errsv_save);
1646 SvREFCNT_dec(errsv_save);
1652 PUSHs(sv_2mortal(newSVpvn(pkg, pkg_len)));
1653 PUSHs(sv_2mortal(newSVpvn(name, name_len)));
1655 PUSHs(sv_2mortal(newSViv(minbits)));
1656 PUSHs(sv_2mortal(newSViv(none)));
1658 if (IN_PERL_COMPILETIME) {
1659 /* XXX ought to be handled by lex_start */
1662 sv_setpv(tokenbufsv, PL_tokenbuf);
1664 errsv_save = newSVsv(ERRSV);
1665 if (call_method("SWASHNEW", G_SCALAR))
1666 retval = newSVsv(*PL_stack_sp--);
1668 retval = &PL_sv_undef;
1670 sv_setsv(ERRSV, errsv_save);
1671 SvREFCNT_dec(errsv_save);
1674 if (IN_PERL_COMPILETIME) {
1676 const char* const pv = SvPV_const(tokenbufsv, len);
1678 Copy(pv, PL_tokenbuf, len+1, char);
1679 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1681 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
1683 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
1685 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1691 /* This API is wrong for special case conversions since we may need to
1692 * return several Unicode characters for a single Unicode character
1693 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1694 * the lower-level routine, and it is similarly broken for returning
1695 * multiple values. --jhi */
1697 Perl_swash_fetch(pTHX_ SV *sv, const U8 *ptr, bool do_utf8)
1700 HV* const hv = (HV*)SvRV(sv);
1705 const U8 *tmps = NULL;
1709 UV c = NATIVE_TO_ASCII(*ptr);
1711 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1712 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
1713 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
1716 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
1717 * then the "swatch" is a vec() for al the chars which start
1719 * So the key in the hash (klen) is length of encoded char -1
1721 klen = UTF8SKIP(ptr) - 1;
1726 /* If char in invariant then swatch is for all the invariant chars
1727 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
1729 needents = UTF_CONTINUATION_MARK;
1730 off = NATIVE_TO_UTF(ptr[klen]);
1734 /* If char is encoded then swatch is for the prefix */
1735 needents = (1 << UTF_ACCUMULATION_SHIFT);
1736 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
1740 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1741 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1742 * it's nothing to sniff at.) Pity we usually come through at least
1743 * two function calls to get here...
1745 * NB: this code assumes that swatches are never modified, once generated!
1748 if (hv == PL_last_swash_hv &&
1749 klen == PL_last_swash_klen &&
1750 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
1752 tmps = PL_last_swash_tmps;
1753 slen = PL_last_swash_slen;
1756 /* Try our second-level swatch cache, kept in a hash. */
1757 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
1759 /* If not cached, generate it via utf8::SWASHGET */
1760 if (!svp || !SvPOK(*svp) || !(tmps = (const U8*)SvPV_const(*svp, slen))) {
1762 /* We use utf8n_to_uvuni() as we want an index into
1763 Unicode tables, not a native character number.
1765 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
1767 0 : UTF8_ALLOW_ANY);
1772 PUSHSTACKi(PERLSI_MAGIC);
1776 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
1777 PUSHs(sv_2mortal(newSViv((klen) ?
1778 (code_point & ~(needents - 1)) : 0)));
1779 PUSHs(sv_2mortal(newSViv(needents)));
1781 errsv_save = newSVsv(ERRSV);
1782 if (call_method("SWASHGET", G_SCALAR))
1783 retval = newSVsv(*PL_stack_sp--);
1785 retval = &PL_sv_undef;
1787 sv_setsv(ERRSV, errsv_save);
1788 SvREFCNT_dec(errsv_save);
1792 if (IN_PERL_COMPILETIME)
1793 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1795 svp = hv_store(hv, (const char *)ptr, klen, retval, 0);
1797 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || (slen << 3) < needents)
1798 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1801 PL_last_swash_hv = hv;
1802 PL_last_swash_klen = klen;
1803 /* FIXME change interpvar.h? */
1804 PL_last_swash_tmps = (U8 *) tmps;
1805 PL_last_swash_slen = slen;
1807 Copy(ptr, PL_last_swash_key, klen, U8);
1810 switch ((int)((slen << 3) / needents)) {
1812 bit = 1 << (off & 7);
1814 return (tmps[off] & bit) != 0;
1819 return (tmps[off] << 8) + tmps[off + 1] ;
1822 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1824 Perl_croak(aTHX_ "panic: swash_fetch");
1829 Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
1831 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
1835 =for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags
1837 Build to the scalar dsv a displayable version of the string spv,
1838 length len, the displayable version being at most pvlim bytes long
1839 (if longer, the rest is truncated and "..." will be appended).
1841 The flags argument can have UNI_DISPLAY_ISPRINT set to display
1842 isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
1843 to display the \\[nrfta\\] as the backslashed versions (like '\n')
1844 (UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
1845 UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
1846 UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
1848 The pointer to the PV of the dsv is returned.
1852 Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
1857 sv_setpvn(dsv, "", 0);
1858 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
1860 /* This serves double duty as a flag and a character to print after
1861 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
1865 if (pvlim && SvCUR(dsv) >= pvlim) {
1869 u = utf8_to_uvchr((U8*)s, 0);
1871 const unsigned char c = (unsigned char)u & 0xFF;
1872 if (!ok && (flags & UNI_DISPLAY_BACKSLASH)) {
1889 Perl_sv_catpvf(aTHX_ dsv, "\\%c", ok);
1892 /* isPRINT() is the locale-blind version. */
1893 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
1894 Perl_sv_catpvf(aTHX_ dsv, "%c", c);
1899 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
1902 sv_catpvn(dsv, "...", 3);
1908 =for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags
1910 Build to the scalar dsv a displayable version of the scalar sv,
1911 the displayable version being at most pvlim bytes long
1912 (if longer, the rest is truncated and "..." will be appended).
1914 The flags argument is as in pv_uni_display().
1916 The pointer to the PV of the dsv is returned.
1920 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
1922 return Perl_pv_uni_display(aTHX_ dsv, (const U8*)SvPVX_const(ssv),
1923 SvCUR(ssv), pvlim, flags);
1927 =for apidoc A|I32|ibcmp_utf8|const char *s1|char **pe1|register UV l1|bool u1|const char *s2|char **pe2|register UV l2|bool u2
1929 Return true if the strings s1 and s2 differ case-insensitively, false
1930 if not (if they are equal case-insensitively). If u1 is true, the
1931 string s1 is assumed to be in UTF-8-encoded Unicode. If u2 is true,
1932 the string s2 is assumed to be in UTF-8-encoded Unicode. If u1 or u2
1933 are false, the respective string is assumed to be in native 8-bit
1936 If the pe1 and pe2 are non-NULL, the scanning pointers will be copied
1937 in there (they will point at the beginning of the I<next> character).
1938 If the pointers behind pe1 or pe2 are non-NULL, they are the end
1939 pointers beyond which scanning will not continue under any
1940 circumstances. If the byte lengths l1 and l2 are non-zero, s1+l1 and
1941 s2+l2 will be used as goal end pointers that will also stop the scan,
1942 and which qualify towards defining a successful match: all the scans
1943 that define an explicit length must reach their goal pointers for
1944 a match to succeed).
1946 For case-insensitiveness, the "casefolding" of Unicode is used
1947 instead of upper/lowercasing both the characters, see
1948 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
1952 Perl_ibcmp_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
1954 register const U8 *p1 = (const U8*)s1;
1955 register const U8 *p2 = (const U8*)s2;
1956 register const U8 *f1 = 0, *f2 = 0;
1957 register U8 *e1 = 0, *q1 = 0;
1958 register U8 *e2 = 0, *q2 = 0;
1959 STRLEN n1 = 0, n2 = 0;
1960 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
1961 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
1963 STRLEN foldlen1, foldlen2;
1968 if (e1 == 0 || (l1 && l1 < (UV)(e1 - (const U8*)s1)))
1969 f1 = (const U8*)s1 + l1;
1972 if (e2 == 0 || (l2 && l2 < (UV)(e2 - (const U8*)s2)))
1973 f2 = (const U8*)s2 + l2;
1975 if ((e1 == 0 && f1 == 0) || (e2 == 0 && f2 == 0) || (f1 == 0 && f2 == 0))
1976 return 1; /* mismatch; possible infinite loop or false positive */
1979 natbuf[1] = 0; /* Need to terminate the buffer. */
1981 while ((e1 == 0 || p1 < e1) &&
1982 (f1 == 0 || p1 < f1) &&
1983 (e2 == 0 || p2 < e2) &&
1984 (f2 == 0 || p2 < f2)) {
1987 to_utf8_fold(p1, foldbuf1, &foldlen1);
1989 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p1)));
1990 to_utf8_fold(natbuf, foldbuf1, &foldlen1);
1997 to_utf8_fold(p2, foldbuf2, &foldlen2);
1999 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p2)));
2000 to_utf8_fold(natbuf, foldbuf2, &foldlen2);
2006 if ( UTF8SKIP(q1) != UTF8SKIP(q2) ||
2007 (UTF8SKIP(q1) == 1 && *q1 != *q2) ||
2008 memNE((char*)q1, (char*)q2, UTF8SKIP(q1)) )
2009 return 1; /* mismatch */
2016 p1 += u1 ? UTF8SKIP(p1) : 1;
2018 p2 += u2 ? UTF8SKIP(p2) : 1;
2022 /* A match is defined by all the scans that specified
2023 * an explicit length reaching their final goals. */
2024 match = (f1 == 0 || p1 == f1) && (f2 == 0 || p2 == f2);
2033 return match ? 0 : 1; /* 0 match, 1 mismatch */
2038 * c-indentation-style: bsd
2040 * indent-tabs-mode: t
2043 * ex: set ts=8 sts=4 sw=4 noet: