This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
perldiag.pod: Remove more format precisions
[perl5.git] / utf8.c
CommitLineData
a0ed51b3
LW
1/* utf8.c
2 *
1129b882 3 * Copyright (C) 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
b94e2f88 4 * by Larry Wall and others
a0ed51b3
LW
5 *
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.
8 *
9 */
10
11/*
4ac71550
TC
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.'
15 *
16 * [p.603 of _The Lord of the Rings_, IV/I: "The Taming of Sméagol"]
a0ed51b3
LW
17 *
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,
4ac71550
TC
20 * as is the custom in the West, if you wish to be answered?'
21 * --Gandalf, addressing Théoden's door wardens
22 *
23 * [p.508 of _The Lord of the Rings_, III/vi: "The King of the Golden Hall"]
a0ed51b3
LW
24 *
25 * ...the travellers perceived that the floor was paved with stones of many
26 * hues; branching runes and strange devices intertwined beneath their feet.
4ac71550
TC
27 *
28 * [p.512 of _The Lord of the Rings_, III/vi: "The King of the Golden Hall"]
a0ed51b3
LW
29 */
30
31#include "EXTERN.h"
864dbfa3 32#define PERL_IN_UTF8_C
a0ed51b3
LW
33#include "perl.h"
34
a0c21aa1 35#ifndef EBCDIC
970ea3cb 36/* Separate prototypes needed because in ASCII systems these are
a0c21aa1
JH
37 * usually macros but they still are compiled as code, too. */
38PERL_CALLCONV UV Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags);
39PERL_CALLCONV U8* Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv);
40#endif
41
27da23d5
JH
42static const char unees[] =
43 "Malformed UTF-8 character (unexpected end of string)";
901b21bf 44
48ef279e 45/*
ccfc67b7 46=head1 Unicode Support
a0ed51b3 47
166f8a29
DM
48This file contains various utility functions for manipulating UTF8-encoded
49strings. For the uninitiated, this is a method of representing arbitrary
61296642 50Unicode characters as a variable number of bytes, in such a way that
56da48f7
DM
51characters in the ASCII range are unmodified, and a zero byte never appears
52within non-zero characters.
166f8a29 53
eaf7a4d2
CS
54=cut
55*/
56
57/*
58=for apidoc is_ascii_string
59
970ea3cb
KW
60Returns true if the first C<len> bytes of the given string are the same whether
61or not the string is encoded in UTF-8 (or UTF-EBCDIC on EBCDIC machines). That
62is, if they are invariant. On ASCII-ish machines, only ASCII characters
63fit this definition, hence the function's name.
eaf7a4d2 64
9f7e3d64
MH
65If C<len> is 0, it will be calculated using C<strlen(s)>.
66
eaf7a4d2
CS
67See also is_utf8_string(), is_utf8_string_loclen(), and is_utf8_string_loc().
68
69=cut
70*/
71
72bool
668b6d8d 73Perl_is_ascii_string(const U8 *s, STRLEN len)
eaf7a4d2
CS
74{
75 const U8* const send = s + (len ? len : strlen((const char *)s));
76 const U8* x = s;
77
78 PERL_ARGS_ASSERT_IS_ASCII_STRING;
eaf7a4d2
CS
79
80 for (; x < send; ++x) {
81 if (!UTF8_IS_INVARIANT(*x))
82 break;
83 }
84
85 return x == send;
86}
87
88/*
87cea99e 89=for apidoc uvuni_to_utf8_flags
eebe1485 90
1e54db1a 91Adds the UTF-8 representation of the Unicode codepoint C<uv> to the end
89ebb4a3 92of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
eebe1485 93bytes available. The return value is the pointer to the byte after the
9041c2e3 94end of the new character. In other words,
eebe1485 95
b851fbc1
JH
96 d = uvuni_to_utf8_flags(d, uv, flags);
97
98or, in most cases,
99
9041c2e3 100 d = uvuni_to_utf8(d, uv);
eebe1485 101
b851fbc1
JH
102(which is equivalent to)
103
104 d = uvuni_to_utf8_flags(d, uv, 0);
105
eebe1485
SC
106is the recommended Unicode-aware way of saying
107
108 *(d++) = uv;
109
110=cut
111*/
112
dfe13c55 113U8 *
b851fbc1 114Perl_uvuni_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
a0ed51b3 115{
7918f24d
NC
116 PERL_ARGS_ASSERT_UVUNI_TO_UTF8_FLAGS;
117
62961d2e 118 if (ckWARN(WARN_UTF8)) {
b851fbc1
JH
119 if (UNICODE_IS_SURROGATE(uv) &&
120 !(flags & UNICODE_ALLOW_SURROGATE))
9014280d 121 Perl_warner(aTHX_ packWARN(WARN_UTF8), "UTF-16 surrogate 0x%04"UVxf, uv);
b851fbc1
JH
122 else if (
123 ((uv >= 0xFDD0 && uv <= 0xFDEF &&
124 !(flags & UNICODE_ALLOW_FDD0))
125 ||
c867b360 126 ((uv & 0xFFFE) == 0xFFFE && /* Either FFFE or FFFF. */
b851fbc1
JH
127 !(flags & UNICODE_ALLOW_FFFF))) &&
128 /* UNICODE_ALLOW_SUPER includes
2a20b9da 129 * FFFEs and FFFFs beyond 0x10FFFF. */
b851fbc1
JH
130 ((uv <= PERL_UNICODE_MAX) ||
131 !(flags & UNICODE_ALLOW_SUPER))
132 )
9014280d 133 Perl_warner(aTHX_ packWARN(WARN_UTF8),
6f6ac1de 134 "Unicode non-character 0x%04"UVxf" is illegal for interchange", uv);
507b9800 135 }
c4d5f83a 136 if (UNI_IS_INVARIANT(uv)) {
eb160463 137 *d++ = (U8)UTF_TO_NATIVE(uv);
a0ed51b3
LW
138 return d;
139 }
2d331972 140#if defined(EBCDIC)
1d72bdf6
NIS
141 else {
142 STRLEN len = UNISKIP(uv);
143 U8 *p = d+len-1;
144 while (p > d) {
eb160463 145 *p-- = (U8)UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
1d72bdf6
NIS
146 uv >>= UTF_ACCUMULATION_SHIFT;
147 }
eb160463 148 *p = (U8)UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
1d72bdf6
NIS
149 return d+len;
150 }
151#else /* Non loop style */
a0ed51b3 152 if (uv < 0x800) {
eb160463
GS
153 *d++ = (U8)(( uv >> 6) | 0xc0);
154 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
155 return d;
156 }
157 if (uv < 0x10000) {
eb160463
GS
158 *d++ = (U8)(( uv >> 12) | 0xe0);
159 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
160 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
161 return d;
162 }
163 if (uv < 0x200000) {
eb160463
GS
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);
a0ed51b3
LW
168 return d;
169 }
170 if (uv < 0x4000000) {
eb160463
GS
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);
a0ed51b3
LW
176 return d;
177 }
178 if (uv < 0x80000000) {
eb160463
GS
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);
a0ed51b3
LW
185 return d;
186 }
6b8eaf93 187#ifdef HAS_QUAD
d7578b48 188 if (uv < UTF8_QUAD_MAX)
a0ed51b3
LW
189#endif
190 {
eb160463
GS
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);
a0ed51b3
LW
198 return d;
199 }
6b8eaf93 200#ifdef HAS_QUAD
a0ed51b3 201 {
eb160463
GS
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);
a0ed51b3
LW
215 return d;
216 }
217#endif
1d72bdf6 218#endif /* Loop style */
a0ed51b3 219}
9041c2e3 220
646ca15d
JH
221/*
222
223Tests if some arbitrary number of bytes begins in a valid UTF-8
224character. Note that an INVARIANT (i.e. ASCII) character is a valid
225UTF-8 character. The actual number of bytes in the UTF-8 character
226will be returned if it is valid, otherwise 0.
227
228This is the "slow" version as opposed to the "fast" version which is
229the "unrolled" IS_UTF8_CHAR(). E.g. for t/uni/class.t the speed
230difference is a factor of 2 to 3. For lengths (UTF8SKIP(s)) of four
231or less you should use the IS_UTF8_CHAR(), for lengths of five or more
232you should use the _slow(). In practice this means that the _slow()
233will be used very rarely, since the maximum Unicode code point (as of
234Unicode 4.1) is U+10FFFF, which encodes in UTF-8 to four bytes. Only
235the "Perl extended UTF-8" (the infamous 'v-strings') will encode into
236five bytes or more.
237
238=cut */
c053b435 239STATIC STRLEN
5f66b61c 240S_is_utf8_char_slow(const U8 *s, const STRLEN len)
646ca15d
JH
241{
242 U8 u = *s;
243 STRLEN slen;
244 UV uv, ouv;
245
7918f24d
NC
246 PERL_ARGS_ASSERT_IS_UTF8_CHAR_SLOW;
247
646ca15d
JH
248 if (UTF8_IS_INVARIANT(u))
249 return 1;
250
251 if (!UTF8_IS_START(u))
252 return 0;
253
254 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
255 return 0;
256
257 slen = len - 1;
258 s++;
77263263
ST
259#ifdef EBCDIC
260 u = NATIVE_TO_UTF(u);
261#endif
646ca15d
JH
262 u &= UTF_START_MASK(len);
263 uv = u;
264 ouv = uv;
265 while (slen--) {
266 if (!UTF8_IS_CONTINUATION(*s))
267 return 0;
268 uv = UTF8_ACCUMULATE(uv, *s);
48ef279e 269 if (uv < ouv)
646ca15d
JH
270 return 0;
271 ouv = uv;
272 s++;
273 }
274
275 if ((STRLEN)UNISKIP(uv) < len)
276 return 0;
277
278 return len;
279}
9041c2e3
NIS
280
281/*
87cea99e 282=for apidoc is_utf8_char
eebe1485 283
5da9da9e 284Tests if some arbitrary number of bytes begins in a valid UTF-8
2bbc8d55
SP
285character. Note that an INVARIANT (i.e. ASCII on non-EBCDIC machines)
286character is a valid UTF-8 character. The actual number of bytes in the UTF-8
287character will be returned if it is valid, otherwise 0.
9041c2e3 288
82686b01 289=cut */
067a85ef 290STRLEN
668b6d8d 291Perl_is_utf8_char(const U8 *s)
386d01d6 292{
44f8325f 293 const STRLEN len = UTF8SKIP(s);
7918f24d
NC
294
295 PERL_ARGS_ASSERT_IS_UTF8_CHAR;
3b0fc154 296#ifdef IS_UTF8_CHAR
768c67ee 297 if (IS_UTF8_CHAR_FAST(len))
3b0fc154
JH
298 return IS_UTF8_CHAR(s, len) ? len : 0;
299#endif /* #ifdef IS_UTF8_CHAR */
2c0c5f92 300 return is_utf8_char_slow(s, len);
386d01d6
GS
301}
302
eaf7a4d2 303
6662521e 304/*
87cea99e 305=for apidoc is_utf8_string
6662521e 306
c9ada85f 307Returns true if first C<len> bytes of the given string form a valid
9f7e3d64
MH
308UTF-8 string, false otherwise. If C<len> is 0, it will be calculated
309using C<strlen(s)>. Note that 'a valid UTF-8 string' does not mean 'a
310string that contains code points above 0x7F encoded in UTF-8' because a
311valid ASCII string is a valid UTF-8 string.
6662521e 312
eaf7a4d2 313See also is_ascii_string(), is_utf8_string_loclen(), and is_utf8_string_loc().
768c67ee 314
6662521e
GS
315=cut
316*/
317
8e84507e 318bool
668b6d8d 319Perl_is_utf8_string(const U8 *s, STRLEN len)
6662521e 320{
35da51f7 321 const U8* const send = s + (len ? len : strlen((const char *)s));
7fc63493 322 const U8* x = s;
067a85ef 323
7918f24d 324 PERL_ARGS_ASSERT_IS_UTF8_STRING;
1aa99e6b 325
6662521e 326 while (x < send) {
a3b680e6 327 STRLEN c;
1acdb0da
JH
328 /* Inline the easy bits of is_utf8_char() here for speed... */
329 if (UTF8_IS_INVARIANT(*x))
330 c = 1;
331 else if (!UTF8_IS_START(*x))
768c67ee 332 goto out;
1acdb0da
JH
333 else {
334 /* ... and call is_utf8_char() only if really needed. */
646ca15d
JH
335#ifdef IS_UTF8_CHAR
336 c = UTF8SKIP(x);
768c67ee
JH
337 if (IS_UTF8_CHAR_FAST(c)) {
338 if (!IS_UTF8_CHAR(x, c))
3c614e38
AD
339 c = 0;
340 }
341 else
342 c = is_utf8_char_slow(x, c);
646ca15d
JH
343#else
344 c = is_utf8_char(x);
345#endif /* #ifdef IS_UTF8_CHAR */
1acdb0da 346 if (!c)
768c67ee 347 goto out;
1acdb0da 348 }
6662521e 349 x += c;
6662521e 350 }
768c67ee
JH
351
352 out:
60006e79
JH
353 if (x != send)
354 return FALSE;
067a85ef
A
355
356 return TRUE;
6662521e
GS
357}
358
67e989fb 359/*
814fafa7
NC
360Implemented as a macro in utf8.h
361
87cea99e 362=for apidoc is_utf8_string_loc
814fafa7
NC
363
364Like is_utf8_string() but stores the location of the failure (in the
365case of "utf8ness failure") or the location s+len (in the case of
366"utf8ness success") in the C<ep>.
367
368See also is_utf8_string_loclen() and is_utf8_string().
369
87cea99e 370=for apidoc is_utf8_string_loclen
81cd54e3 371
e3e4599f 372Like is_utf8_string() but stores the location of the failure (in the
768c67ee
JH
373case 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
375encoded characters in the C<el>.
376
377See also is_utf8_string_loc() and is_utf8_string().
81cd54e3
JH
378
379=cut
380*/
381
382bool
668b6d8d 383Perl_is_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
81cd54e3 384{
35da51f7 385 const U8* const send = s + (len ? len : strlen((const char *)s));
7fc63493 386 const U8* x = s;
81cd54e3 387 STRLEN c;
3ebfea28 388 STRLEN outlen = 0;
7918f24d
NC
389
390 PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN;
81cd54e3 391
81cd54e3
JH
392 while (x < send) {
393 /* Inline the easy bits of is_utf8_char() here for speed... */
394 if (UTF8_IS_INVARIANT(*x))
768c67ee
JH
395 c = 1;
396 else if (!UTF8_IS_START(*x))
397 goto out;
81cd54e3 398 else {
768c67ee
JH
399 /* ... and call is_utf8_char() only if really needed. */
400#ifdef IS_UTF8_CHAR
401 c = UTF8SKIP(x);
402 if (IS_UTF8_CHAR_FAST(c)) {
403 if (!IS_UTF8_CHAR(x, c))
404 c = 0;
405 } else
406 c = is_utf8_char_slow(x, c);
407#else
408 c = is_utf8_char(x);
409#endif /* #ifdef IS_UTF8_CHAR */
410 if (!c)
411 goto out;
81cd54e3 412 }
768c67ee 413 x += c;
3ebfea28 414 outlen++;
81cd54e3 415 }
768c67ee
JH
416
417 out:
3ebfea28
AL
418 if (el)
419 *el = outlen;
420
768c67ee
JH
421 if (ep)
422 *ep = x;
3ebfea28 423 return (x == send);
81cd54e3
JH
424}
425
426/*
768c67ee 427
87cea99e 428=for apidoc utf8n_to_uvuni
67e989fb 429
9041c2e3 430Bottom level UTF-8 decode routine.
38a44b82 431Returns the Unicode code point value of the first character in the string C<s>
1e54db1a 432which is assumed to be in UTF-8 encoding and no longer than C<curlen>;
7df053ec 433C<retlen> will be set to the length, in bytes, of that character.
67e989fb 434
1e54db1a 435If C<s> does not point to a well-formed UTF-8 character, the behaviour
dcad2880
JH
436is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
437it is assumed that the caller will raise a warning, and this function
28d3d195
JH
438will silently just set C<retlen> to C<-1> and return zero. If the
439C<flags> does not contain UTF8_CHECK_ONLY, warnings about
440malformations will be given, C<retlen> will be set to the expected
441length of the UTF-8 character in bytes, and zero will be returned.
442
443The C<flags> can also contain various flags to allow deviations from
444the strict UTF-8 encoding (see F<utf8.h>).
67e989fb 445
9041c2e3
NIS
446Most code should use utf8_to_uvchr() rather than call this directly.
447
37607a96
PK
448=cut
449*/
67e989fb 450
a0ed51b3 451UV
7fc63493 452Perl_utf8n_to_uvuni(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
a0ed51b3 453{
97aff369 454 dVAR;
d4c19fe8 455 const U8 * const s0 = s;
9c5ffd7c 456 UV uv = *s, ouv = 0;
ba210ebe 457 STRLEN len = 1;
7fc63493
AL
458 const bool dowarn = ckWARN_d(WARN_UTF8);
459 const UV startbyte = *s;
ba210ebe 460 STRLEN expectlen = 0;
a0dbb045 461 U32 warning = 0;
5b311467 462 SV* sv;
a0dbb045 463
7918f24d
NC
464 PERL_ARGS_ASSERT_UTF8N_TO_UVUNI;
465
5b311467 466/* This list is a superset of the UTF8_ALLOW_XXX. BUT it isn't, eg SUPER missing XXX */
a0dbb045
JH
467
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
c867b360
JH
475#define UTF8_WARN_LONG 8
476#define UTF8_WARN_FFFF 9 /* Also FFFE. */
a0dbb045
JH
477
478 if (curlen == 0 &&
479 !(flags & UTF8_ALLOW_EMPTY)) {
480 warning = UTF8_WARN_EMPTY;
0c443dc2
JH
481 goto malformed;
482 }
483
1d72bdf6 484 if (UTF8_IS_INVARIANT(uv)) {
a0ed51b3
LW
485 if (retlen)
486 *retlen = 1;
c4d5f83a 487 return (UV) (NATIVE_TO_UTF(*s));
a0ed51b3 488 }
67e989fb 489
421a8bf2 490 if (UTF8_IS_CONTINUATION(uv) &&
fcc8fcf6 491 !(flags & UTF8_ALLOW_CONTINUATION)) {
a0dbb045 492 warning = UTF8_WARN_CONTINUATION;
ba210ebe
JH
493 goto malformed;
494 }
495
421a8bf2 496 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
fcc8fcf6 497 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
a0dbb045 498 warning = UTF8_WARN_NON_CONTINUATION;
ba210ebe
JH
499 goto malformed;
500 }
9041c2e3 501
1d72bdf6 502#ifdef EBCDIC
75383841 503 uv = NATIVE_TO_UTF(uv);
1d72bdf6 504#else
fcc8fcf6
JH
505 if ((uv == 0xfe || uv == 0xff) &&
506 !(flags & UTF8_ALLOW_FE_FF)) {
a0dbb045 507 warning = UTF8_WARN_FE_FF;
ba210ebe 508 goto malformed;
a0ed51b3 509 }
1d72bdf6
NIS
510#endif
511
ba210ebe
JH
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; }
1d72bdf6
NIS
516#ifdef EBCDIC
517 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
518 else { len = 7; uv &= 0x01; }
519#else
ba210ebe
JH
520 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
521 else if (!(uv & 0x01)) { len = 7; uv = 0; }
1d72bdf6
NIS
522 else { len = 13; uv = 0; } /* whoa! */
523#endif
524
a0ed51b3
LW
525 if (retlen)
526 *retlen = len;
9041c2e3 527
ba210ebe
JH
528 expectlen = len;
529
fcc8fcf6
JH
530 if ((curlen < expectlen) &&
531 !(flags & UTF8_ALLOW_SHORT)) {
a0dbb045 532 warning = UTF8_WARN_SHORT;
ba210ebe
JH
533 goto malformed;
534 }
535
536 len--;
a0ed51b3 537 s++;
ba210ebe
JH
538 ouv = uv;
539
a0ed51b3 540 while (len--) {
421a8bf2
JH
541 if (!UTF8_IS_CONTINUATION(*s) &&
542 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
a0dbb045
JH
543 s--;
544 warning = UTF8_WARN_NON_CONTINUATION;
ba210ebe 545 goto malformed;
a0ed51b3
LW
546 }
547 else
8850bf83 548 uv = UTF8_ACCUMULATE(uv, *s);
a0dbb045
JH
549 if (!(uv > ouv)) {
550 /* These cannot be allowed. */
551 if (uv == ouv) {
75dbc644 552 if (expectlen != 13 && !(flags & UTF8_ALLOW_LONG)) {
a0dbb045
JH
553 warning = UTF8_WARN_LONG;
554 goto malformed;
555 }
556 }
557 else { /* uv < ouv */
558 /* This cannot be allowed. */
559 warning = UTF8_WARN_OVERFLOW;
560 goto malformed;
561 }
ba210ebe
JH
562 }
563 s++;
564 ouv = uv;
565 }
566
421a8bf2 567 if (UNICODE_IS_SURROGATE(uv) &&
fcc8fcf6 568 !(flags & UTF8_ALLOW_SURROGATE)) {
a0dbb045 569 warning = UTF8_WARN_SURROGATE;
ba210ebe 570 goto malformed;
eb160463 571 } else if ((expectlen > (STRLEN)UNISKIP(uv)) &&
fcc8fcf6 572 !(flags & UTF8_ALLOW_LONG)) {
a0dbb045 573 warning = UTF8_WARN_LONG;
ba210ebe 574 goto malformed;
421a8bf2 575 } else if (UNICODE_IS_ILLEGAL(uv) &&
a9917092 576 !(flags & UTF8_ALLOW_FFFF)) {
a0dbb045 577 warning = UTF8_WARN_FFFF;
a9917092 578 goto malformed;
a0ed51b3 579 }
ba210ebe 580
a0ed51b3 581 return uv;
ba210ebe
JH
582
583malformed:
584
fcc8fcf6 585 if (flags & UTF8_CHECK_ONLY) {
ba210ebe 586 if (retlen)
10edeb5d 587 *retlen = ((STRLEN) -1);
ba210ebe
JH
588 return 0;
589 }
590
a0dbb045 591 if (dowarn) {
5b311467
KW
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);
595 }
596 else {
597 sv = newSVpvs_flags("Malformed UTF-8 character ", SVs_TEMP);
598
599 switch (warning) {
600 case 0: /* Intentionally empty. */ break;
601 case UTF8_WARN_EMPTY:
602 sv_catpvs(sv, "(empty string)");
603 break;
604 case UTF8_WARN_CONTINUATION:
605 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
606 break;
607 case UTF8_WARN_NON_CONTINUATION:
608 if (s == s0)
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);
611 else {
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);
615 }
616
617 break;
618 case UTF8_WARN_FE_FF:
619 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
620 break;
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 */
625 break;
626 case UTF8_WARN_OVERFLOW:
627 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
628 ouv, *s, startbyte);
629 break;
630 case UTF8_WARN_SURROGATE:
631 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
632 break;
633 case UTF8_WARN_LONG:
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);
636 break;
637 default:
638 sv_catpvs(sv, "(unknown reason)");
639 break;
551405c4 640 }
a0dbb045
JH
641 }
642
643 if (warning) {
44f8325f 644 const char * const s = SvPVX_const(sv);
a0dbb045
JH
645
646 if (PL_op)
9014280d 647 Perl_warner(aTHX_ packWARN(WARN_UTF8),
53e06cf0 648 "%s in %s", s, OP_DESC(PL_op));
a0dbb045 649 else
9014280d 650 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
a0dbb045
JH
651 }
652 }
653
ba210ebe 654 if (retlen)
28d3d195 655 *retlen = expectlen ? expectlen : len;
ba210ebe 656
28d3d195 657 return 0;
a0ed51b3
LW
658}
659
8e84507e 660/*
87cea99e 661=for apidoc utf8_to_uvchr
9041c2e3
NIS
662
663Returns the native character value of the first character in the string C<s>
1e54db1a 664which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
9041c2e3
NIS
665length, in bytes, of that character.
666
1e54db1a 667If C<s> does not point to a well-formed UTF-8 character, zero is
9041c2e3
NIS
668returned and retlen is set, if possible, to -1.
669
670=cut
671*/
672
673UV
7fc63493 674Perl_utf8_to_uvchr(pTHX_ const U8 *s, STRLEN *retlen)
9041c2e3 675{
7918f24d
NC
676 PERL_ARGS_ASSERT_UTF8_TO_UVCHR;
677
1754c1a1
NC
678 return utf8n_to_uvchr(s, UTF8_MAXBYTES, retlen,
679 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
9041c2e3
NIS
680}
681
682/*
87cea99e 683=for apidoc utf8_to_uvuni
9041c2e3
NIS
684
685Returns the Unicode code point of the first character in the string C<s>
1e54db1a 686which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
9041c2e3
NIS
687length, in bytes, of that character.
688
2bbc8d55 689This function should only be used when the returned UV is considered
9041c2e3
NIS
690an index into the Unicode semantic tables (e.g. swashes).
691
1e54db1a 692If C<s> does not point to a well-formed UTF-8 character, zero is
ba210ebe 693returned and retlen is set, if possible, to -1.
8e84507e
NIS
694
695=cut
696*/
697
698UV
7fc63493 699Perl_utf8_to_uvuni(pTHX_ const U8 *s, STRLEN *retlen)
8e84507e 700{
7918f24d
NC
701 PERL_ARGS_ASSERT_UTF8_TO_UVUNI;
702
9041c2e3 703 /* Call the low level routine asking for checks */
89ebb4a3 704 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXBYTES, retlen,
872c91ae 705 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
8e84507e
NIS
706}
707
b76347f2 708/*
87cea99e 709=for apidoc utf8_length
b76347f2
JH
710
711Return the length of the UTF-8 char encoded string C<s> in characters.
02eb7b47
JH
712Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
713up past C<e>, croaks.
b76347f2
JH
714
715=cut
716*/
717
718STRLEN
35a4481c 719Perl_utf8_length(pTHX_ const U8 *s, const U8 *e)
b76347f2 720{
97aff369 721 dVAR;
b76347f2
JH
722 STRLEN len = 0;
723
7918f24d
NC
724 PERL_ARGS_ASSERT_UTF8_LENGTH;
725
8850bf83
JH
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. */
729
a3b680e6
AL
730 if (e < s)
731 goto warn_and_return;
b76347f2 732 while (s < e) {
8e91ec7f
AV
733 if (!UTF8_IS_INVARIANT(*s))
734 s += UTF8SKIP(s);
735 else
736 s++;
737 len++;
738 }
739
740 if (e != s) {
741 len--;
742 warn_and_return:
9b387841
NC
743 if (PL_op)
744 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
745 "%s in %s", unees, OP_DESC(PL_op));
746 else
747 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8), unees);
b76347f2
JH
748 }
749
750 return len;
751}
752
b06226ff 753/*
87cea99e 754=for apidoc utf8_distance
b06226ff 755
1e54db1a 756Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
b06226ff
JH
757and C<b>.
758
759WARNING: use only if you *know* that the pointers point inside the
760same UTF-8 buffer.
761
37607a96
PK
762=cut
763*/
a0ed51b3 764
02eb7b47 765IV
35a4481c 766Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
a0ed51b3 767{
7918f24d
NC
768 PERL_ARGS_ASSERT_UTF8_DISTANCE;
769
bf1665bc 770 return (a < b) ? -1 * (IV) utf8_length(a, b) : (IV) utf8_length(b, a);
a0ed51b3
LW
771}
772
b06226ff 773/*
87cea99e 774=for apidoc utf8_hop
b06226ff 775
8850bf83
JH
776Return the UTF-8 pointer C<s> displaced by C<off> characters, either
777forward or backward.
b06226ff
JH
778
779WARNING: do not use the following unless you *know* C<off> is within
8850bf83
JH
780the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
781on the first byte of character or just after the last byte of a character.
b06226ff 782
37607a96
PK
783=cut
784*/
a0ed51b3
LW
785
786U8 *
4373e329 787Perl_utf8_hop(pTHX_ const U8 *s, I32 off)
a0ed51b3 788{
7918f24d
NC
789 PERL_ARGS_ASSERT_UTF8_HOP;
790
96a5add6 791 PERL_UNUSED_CONTEXT;
8850bf83
JH
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. */
795
a0ed51b3
LW
796 if (off >= 0) {
797 while (off--)
798 s += UTF8SKIP(s);
799 }
800 else {
801 while (off++) {
802 s--;
8850bf83
JH
803 while (UTF8_IS_CONTINUATION(*s))
804 s--;
a0ed51b3
LW
805 }
806 }
4373e329 807 return (U8 *)s;
a0ed51b3
LW
808}
809
6940069f 810/*
fed3ba5d
NC
811=for apidoc bytes_cmp_utf8
812
813Compares the sequence of characters (stored as octets) in b, blen with the
814sequence of characters (stored as UTF-8) in u, ulen. Returns 0 if they are
815equal, -1 or -2 if the first string is less than the second string, +1 or +2
816if the first string is greater than the second string.
817
818-1 or +1 is returned if the shorter string was identical to the start of the
819longer string. -2 or +2 is returned if the was a difference between characters
820within the strings.
821
822=cut
823*/
824
825int
826Perl_bytes_cmp_utf8(pTHX_ const U8 *b, STRLEN blen, const U8 *u, STRLEN ulen)
827{
828 const U8 *const bend = b + blen;
829 const U8 *const uend = u + ulen;
830
831 PERL_ARGS_ASSERT_BYTES_CMP_UTF8;
832
833 PERL_UNUSED_CONTEXT;
834
835 while (b < bend && u < uend) {
836 U8 c = *u++;
837 if (!UTF8_IS_INVARIANT(c)) {
838 if (UTF8_IS_DOWNGRADEABLE_START(c)) {
839 if (u < uend) {
840 U8 c1 = *u++;
841 if (UTF8_IS_CONTINUATION(c1)) {
356979f4 842 c = UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(c, c1));
fed3ba5d
NC
843 } else {
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. */
849 "%s%s", c1, c,
850 PL_op ? " in " : "",
851 PL_op ? OP_DESC(PL_op) : "");
852 return -2;
853 }
854 } else {
855 if (PL_op)
856 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
857 "%s in %s", unees, OP_DESC(PL_op));
858 else
859 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8), unees);
860 return -2; /* Really want to return undef :-) */
861 }
862 } else {
863 return -2;
864 }
865 }
866 if (*b != c) {
867 return *b < c ? -2 : +2;
868 }
869 ++b;
870 }
871
872 if (b == bend && u == uend)
873 return 0;
874
875 return b < bend ? +1 : -1;
876}
877
878/*
87cea99e 879=for apidoc utf8_to_bytes
6940069f 880
2bbc8d55 881Converts a string C<s> of length C<len> from UTF-8 into native byte encoding.
246fae53
MG
882Unlike C<bytes_to_utf8>, this over-writes the original string, and
883updates len to contain the new length.
67e989fb 884Returns zero on failure, setting C<len> to -1.
6940069f 885
95be277c
NC
886If you need a copy of the string, see C<bytes_from_utf8>.
887
6940069f
GS
888=cut
889*/
890
891U8 *
37607a96 892Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
6940069f 893{
d4c19fe8
AL
894 U8 * const save = s;
895 U8 * const send = s + *len;
6940069f 896 U8 *d;
246fae53 897
7918f24d
NC
898 PERL_ARGS_ASSERT_UTF8_TO_BYTES;
899
1e54db1a 900 /* ensure valid UTF-8 and chars < 256 before updating string */
d4c19fe8 901 while (s < send) {
dcad2880
JH
902 U8 c = *s++;
903
1d72bdf6
NIS
904 if (!UTF8_IS_INVARIANT(c) &&
905 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
906 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
10edeb5d 907 *len = ((STRLEN) -1);
dcad2880
JH
908 return 0;
909 }
246fae53 910 }
dcad2880
JH
911
912 d = s = save;
6940069f 913 while (s < send) {
ed646e6e 914 STRLEN ulen;
9041c2e3 915 *d++ = (U8)utf8_to_uvchr(s, &ulen);
ed646e6e 916 s += ulen;
6940069f
GS
917 }
918 *d = '\0';
246fae53 919 *len = d - save;
6940069f
GS
920 return save;
921}
922
923/*
87cea99e 924=for apidoc bytes_from_utf8
f9a63242 925
2bbc8d55 926Converts a string C<s> of length C<len> from UTF-8 into native byte encoding.
35a4481c 927Unlike C<utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
ef9edfd0
JH
928the newly-created string, and updates C<len> to contain the new
929length. Returns the original string if no conversion occurs, C<len>
930is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
2bbc8d55
SP
9310 if C<s> is converted or consisted entirely of characters that are invariant
932in utf8 (i.e., US-ASCII on non-EBCDIC machines).
f9a63242 933
37607a96
PK
934=cut
935*/
f9a63242
JH
936
937U8 *
e1ec3a88 938Perl_bytes_from_utf8(pTHX_ const U8 *s, STRLEN *len, bool *is_utf8)
f9a63242 939{
f9a63242 940 U8 *d;
e1ec3a88
AL
941 const U8 *start = s;
942 const U8 *send;
f9a63242
JH
943 I32 count = 0;
944
7918f24d
NC
945 PERL_ARGS_ASSERT_BYTES_FROM_UTF8;
946
96a5add6 947 PERL_UNUSED_CONTEXT;
f9a63242 948 if (!*is_utf8)
73d840c0 949 return (U8 *)start;
f9a63242 950
1e54db1a 951 /* ensure valid UTF-8 and chars < 256 before converting string */
f9a63242 952 for (send = s + *len; s < send;) {
e1ec3a88 953 U8 c = *s++;
1d72bdf6 954 if (!UTF8_IS_INVARIANT(c)) {
db42d148
NIS
955 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
956 (c = *s++) && UTF8_IS_CONTINUATION(c))
957 count++;
958 else
73d840c0 959 return (U8 *)start;
db42d148 960 }
f9a63242
JH
961 }
962
35da51f7 963 *is_utf8 = FALSE;
f9a63242 964
212542aa 965 Newx(d, (*len) - count + 1, U8);
ef9edfd0 966 s = start; start = d;
f9a63242
JH
967 while (s < send) {
968 U8 c = *s++;
c4d5f83a
NIS
969 if (!UTF8_IS_INVARIANT(c)) {
970 /* Then it is two-byte encoded */
356979f4 971 c = UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(c, *s++));
c4d5f83a
NIS
972 }
973 *d++ = c;
f9a63242
JH
974 }
975 *d = '\0';
976 *len = d - start;
73d840c0 977 return (U8 *)start;
f9a63242
JH
978}
979
980/*
87cea99e 981=for apidoc bytes_to_utf8
6940069f 982
ff97e5cf
KW
983Converts a string C<s> of length C<len> bytes from the native encoding into
984UTF-8.
6662521e 985Returns a pointer to the newly-created string, and sets C<len> to
ff97e5cf 986reflect the new length in bytes.
6940069f 987
2bbc8d55
SP
988A NUL character will be written after the end of the string.
989
990If you want to convert to UTF-8 from encodings other than
991the native (Latin1 or EBCDIC),
c9ada85f
JH
992see sv_recode_to_utf8().
993
497711e7 994=cut
6940069f
GS
995*/
996
997U8*
35a4481c 998Perl_bytes_to_utf8(pTHX_ const U8 *s, STRLEN *len)
6940069f 999{
35a4481c 1000 const U8 * const send = s + (*len);
6940069f
GS
1001 U8 *d;
1002 U8 *dst;
7918f24d
NC
1003
1004 PERL_ARGS_ASSERT_BYTES_TO_UTF8;
96a5add6 1005 PERL_UNUSED_CONTEXT;
6940069f 1006
212542aa 1007 Newx(d, (*len) * 2 + 1, U8);
6940069f
GS
1008 dst = d;
1009
1010 while (s < send) {
35a4481c 1011 const UV uv = NATIVE_TO_ASCII(*s++);
c4d5f83a 1012 if (UNI_IS_INVARIANT(uv))
eb160463 1013 *d++ = (U8)UTF_TO_NATIVE(uv);
6940069f 1014 else {
eb160463
GS
1015 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
1016 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
6940069f
GS
1017 }
1018 }
1019 *d = '\0';
6662521e 1020 *len = d-dst;
6940069f
GS
1021 return dst;
1022}
1023
a0ed51b3 1024/*
dea0fc0b 1025 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
a0ed51b3
LW
1026 *
1027 * Destination must be pre-extended to 3/2 source. Do not use in-place.
1028 * We optimize for native, for obvious reasons. */
1029
1030U8*
dea0fc0b 1031Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
a0ed51b3 1032{
dea0fc0b
JH
1033 U8* pend;
1034 U8* dstart = d;
1035
7918f24d
NC
1036 PERL_ARGS_ASSERT_UTF16_TO_UTF8;
1037
dea0fc0b 1038 if (bytelen & 1)
f5992bc4 1039 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %"UVuf, (UV)bytelen);
dea0fc0b
JH
1040
1041 pend = p + bytelen;
1042
a0ed51b3 1043 while (p < pend) {
dea0fc0b
JH
1044 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
1045 p += 2;
a0ed51b3 1046 if (uv < 0x80) {
e294cc5d
JH
1047#ifdef EBCDIC
1048 *d++ = UNI_TO_NATIVE(uv);
1049#else
eb160463 1050 *d++ = (U8)uv;
e294cc5d 1051#endif
a0ed51b3
LW
1052 continue;
1053 }
1054 if (uv < 0x800) {
eb160463
GS
1055 *d++ = (U8)(( uv >> 6) | 0xc0);
1056 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
1057 continue;
1058 }
52b9aa85 1059 if (uv >= 0xd800 && uv <= 0xdbff) { /* surrogates */
01ea242b 1060 if (p >= pend) {
dea0fc0b 1061 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
01ea242b
NC
1062 } else {
1063 UV low = (p[0] << 8) + p[1];
1064 p += 2;
52b9aa85 1065 if (low < 0xdc00 || low > 0xdfff)
01ea242b
NC
1066 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
1067 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
1068 }
dbde1951
NC
1069 } else if (uv >= 0xdc00 && uv <= 0xdfff) {
1070 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
a0ed51b3
LW
1071 }
1072 if (uv < 0x10000) {
eb160463
GS
1073 *d++ = (U8)(( uv >> 12) | 0xe0);
1074 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
1075 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
1076 continue;
1077 }
1078 else {
eb160463
GS
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);
a0ed51b3
LW
1083 continue;
1084 }
1085 }
dea0fc0b 1086 *newlen = d - dstart;
a0ed51b3
LW
1087 return d;
1088}
1089
1090/* Note: this one is slightly destructive of the source. */
1091
1092U8*
dea0fc0b 1093Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
a0ed51b3
LW
1094{
1095 U8* s = (U8*)p;
d4c19fe8 1096 U8* const send = s + bytelen;
7918f24d
NC
1097
1098 PERL_ARGS_ASSERT_UTF16_TO_UTF8_REVERSED;
1099
e0ea5e2d
NC
1100 if (bytelen & 1)
1101 Perl_croak(aTHX_ "panic: utf16_to_utf8_reversed: odd bytelen %"UVuf,
1102 (UV)bytelen);
1103
a0ed51b3 1104 while (s < send) {
d4c19fe8 1105 const U8 tmp = s[0];
a0ed51b3
LW
1106 s[0] = s[1];
1107 s[1] = tmp;
1108 s += 2;
1109 }
dea0fc0b 1110 return utf16_to_utf8(p, d, bytelen, newlen);
a0ed51b3
LW
1111}
1112
1113/* for now these are all defined (inefficiently) in terms of the utf8 versions */
1114
1115bool
84afefe6 1116Perl_is_uni_alnum(pTHX_ UV c)
a0ed51b3 1117{
89ebb4a3 1118 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1119 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1120 return is_utf8_alnum(tmpbuf);
1121}
1122
1123bool
84afefe6 1124Perl_is_uni_idfirst(pTHX_ UV c)
a0ed51b3 1125{
89ebb4a3 1126 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1127 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1128 return is_utf8_idfirst(tmpbuf);
1129}
1130
1131bool
84afefe6 1132Perl_is_uni_alpha(pTHX_ UV c)
a0ed51b3 1133{
89ebb4a3 1134 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1135 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1136 return is_utf8_alpha(tmpbuf);
1137}
1138
1139bool
84afefe6 1140Perl_is_uni_ascii(pTHX_ UV c)
4d61ec05 1141{
89ebb4a3 1142 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1143 uvchr_to_utf8(tmpbuf, c);
4d61ec05
GS
1144 return is_utf8_ascii(tmpbuf);
1145}
1146
1147bool
84afefe6 1148Perl_is_uni_space(pTHX_ UV c)
a0ed51b3 1149{
89ebb4a3 1150 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1151 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1152 return is_utf8_space(tmpbuf);
1153}
1154
1155bool
84afefe6 1156Perl_is_uni_digit(pTHX_ UV c)
a0ed51b3 1157{
89ebb4a3 1158 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1159 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1160 return is_utf8_digit(tmpbuf);
1161}
1162
1163bool
84afefe6 1164Perl_is_uni_upper(pTHX_ UV c)
a0ed51b3 1165{
89ebb4a3 1166 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1167 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1168 return is_utf8_upper(tmpbuf);
1169}
1170
1171bool
84afefe6 1172Perl_is_uni_lower(pTHX_ UV c)
a0ed51b3 1173{
89ebb4a3 1174 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1175 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1176 return is_utf8_lower(tmpbuf);
1177}
1178
1179bool
84afefe6 1180Perl_is_uni_cntrl(pTHX_ UV c)
b8c5462f 1181{
89ebb4a3 1182 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1183 uvchr_to_utf8(tmpbuf, c);
b8c5462f
JH
1184 return is_utf8_cntrl(tmpbuf);
1185}
1186
1187bool
84afefe6 1188Perl_is_uni_graph(pTHX_ UV c)
b8c5462f 1189{
89ebb4a3 1190 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1191 uvchr_to_utf8(tmpbuf, c);
b8c5462f
JH
1192 return is_utf8_graph(tmpbuf);
1193}
1194
1195bool
84afefe6 1196Perl_is_uni_print(pTHX_ UV c)
a0ed51b3 1197{
89ebb4a3 1198 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1199 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1200 return is_utf8_print(tmpbuf);
1201}
1202
b8c5462f 1203bool
84afefe6 1204Perl_is_uni_punct(pTHX_ UV c)
b8c5462f 1205{
89ebb4a3 1206 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1207 uvchr_to_utf8(tmpbuf, c);
b8c5462f
JH
1208 return is_utf8_punct(tmpbuf);
1209}
1210
4d61ec05 1211bool
84afefe6 1212Perl_is_uni_xdigit(pTHX_ UV c)
4d61ec05 1213{
89ebb4a3 1214 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
230880c1 1215 uvchr_to_utf8(tmpbuf, c);
4d61ec05
GS
1216 return is_utf8_xdigit(tmpbuf);
1217}
1218
84afefe6
JH
1219UV
1220Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 1221{
7918f24d
NC
1222 PERL_ARGS_ASSERT_TO_UNI_UPPER;
1223
0ebc6274
JH
1224 uvchr_to_utf8(p, c);
1225 return to_utf8_upper(p, p, lenp);
a0ed51b3
LW
1226}
1227
84afefe6
JH
1228UV
1229Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 1230{
7918f24d
NC
1231 PERL_ARGS_ASSERT_TO_UNI_TITLE;
1232
0ebc6274
JH
1233 uvchr_to_utf8(p, c);
1234 return to_utf8_title(p, p, lenp);
a0ed51b3
LW
1235}
1236
84afefe6
JH
1237UV
1238Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 1239{
7918f24d
NC
1240 PERL_ARGS_ASSERT_TO_UNI_LOWER;
1241
0ebc6274
JH
1242 uvchr_to_utf8(p, c);
1243 return to_utf8_lower(p, p, lenp);
a0ed51b3
LW
1244}
1245
84afefe6
JH
1246UV
1247Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1248{
7918f24d
NC
1249 PERL_ARGS_ASSERT_TO_UNI_FOLD;
1250
0ebc6274
JH
1251 uvchr_to_utf8(p, c);
1252 return to_utf8_fold(p, p, lenp);
84afefe6
JH
1253}
1254
a0ed51b3
LW
1255/* for now these all assume no locale info available for Unicode > 255 */
1256
1257bool
84afefe6 1258Perl_is_uni_alnum_lc(pTHX_ UV c)
a0ed51b3
LW
1259{
1260 return is_uni_alnum(c); /* XXX no locale support yet */
1261}
1262
1263bool
84afefe6 1264Perl_is_uni_idfirst_lc(pTHX_ UV c)
a0ed51b3
LW
1265{
1266 return is_uni_idfirst(c); /* XXX no locale support yet */
1267}
1268
1269bool
84afefe6 1270Perl_is_uni_alpha_lc(pTHX_ UV c)
a0ed51b3
LW
1271{
1272 return is_uni_alpha(c); /* XXX no locale support yet */
1273}
1274
1275bool
84afefe6 1276Perl_is_uni_ascii_lc(pTHX_ UV c)
4d61ec05
GS
1277{
1278 return is_uni_ascii(c); /* XXX no locale support yet */
1279}
1280
1281bool
84afefe6 1282Perl_is_uni_space_lc(pTHX_ UV c)
a0ed51b3
LW
1283{
1284 return is_uni_space(c); /* XXX no locale support yet */
1285}
1286
1287bool
84afefe6 1288Perl_is_uni_digit_lc(pTHX_ UV c)
a0ed51b3
LW
1289{
1290 return is_uni_digit(c); /* XXX no locale support yet */
1291}
1292
1293bool
84afefe6 1294Perl_is_uni_upper_lc(pTHX_ UV c)
a0ed51b3
LW
1295{
1296 return is_uni_upper(c); /* XXX no locale support yet */
1297}
1298
1299bool
84afefe6 1300Perl_is_uni_lower_lc(pTHX_ UV c)
a0ed51b3
LW
1301{
1302 return is_uni_lower(c); /* XXX no locale support yet */
1303}
1304
1305bool
84afefe6 1306Perl_is_uni_cntrl_lc(pTHX_ UV c)
b8c5462f
JH
1307{
1308 return is_uni_cntrl(c); /* XXX no locale support yet */
1309}
1310
1311bool
84afefe6 1312Perl_is_uni_graph_lc(pTHX_ UV c)
b8c5462f
JH
1313{
1314 return is_uni_graph(c); /* XXX no locale support yet */
1315}
1316
1317bool
84afefe6 1318Perl_is_uni_print_lc(pTHX_ UV c)
a0ed51b3
LW
1319{
1320 return is_uni_print(c); /* XXX no locale support yet */
1321}
1322
b8c5462f 1323bool
84afefe6 1324Perl_is_uni_punct_lc(pTHX_ UV c)
b8c5462f
JH
1325{
1326 return is_uni_punct(c); /* XXX no locale support yet */
1327}
1328
4d61ec05 1329bool
84afefe6 1330Perl_is_uni_xdigit_lc(pTHX_ UV c)
4d61ec05
GS
1331{
1332 return is_uni_xdigit(c); /* XXX no locale support yet */
1333}
1334
b7ac61fa
JH
1335U32
1336Perl_to_uni_upper_lc(pTHX_ U32 c)
1337{
ee099d14
JH
1338 /* XXX returns only the first character -- do not use XXX */
1339 /* XXX no locale support yet */
1340 STRLEN len;
89ebb4a3 1341 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
ee099d14 1342 return (U32)to_uni_upper(c, tmpbuf, &len);
b7ac61fa
JH
1343}
1344
1345U32
1346Perl_to_uni_title_lc(pTHX_ U32 c)
1347{
ee099d14
JH
1348 /* XXX returns only the first character XXX -- do not use XXX */
1349 /* XXX no locale support yet */
1350 STRLEN len;
89ebb4a3 1351 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
ee099d14 1352 return (U32)to_uni_title(c, tmpbuf, &len);
b7ac61fa
JH
1353}
1354
1355U32
1356Perl_to_uni_lower_lc(pTHX_ U32 c)
1357{
ee099d14
JH
1358 /* XXX returns only the first character -- do not use XXX */
1359 /* XXX no locale support yet */
1360 STRLEN len;
89ebb4a3 1361 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
ee099d14 1362 return (U32)to_uni_lower(c, tmpbuf, &len);
b7ac61fa
JH
1363}
1364
7452cf6a 1365static bool
5141f98e 1366S_is_utf8_common(pTHX_ const U8 *const p, SV **swash,
bde6a22d
NC
1367 const char *const swashname)
1368{
97aff369 1369 dVAR;
7918f24d
NC
1370
1371 PERL_ARGS_ASSERT_IS_UTF8_COMMON;
1372
bde6a22d
NC
1373 if (!is_utf8_char(p))
1374 return FALSE;
1375 if (!*swash)
711a919c 1376 *swash = swash_init("utf8", swashname, &PL_sv_undef, 1, 0);
bde6a22d
NC
1377 return swash_fetch(*swash, p, TRUE) != 0;
1378}
1379
1380bool
7fc63493 1381Perl_is_utf8_alnum(pTHX_ const U8 *p)
a0ed51b3 1382{
97aff369 1383 dVAR;
7918f24d
NC
1384
1385 PERL_ARGS_ASSERT_IS_UTF8_ALNUM;
1386
671c33bf
NC
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 */
d4c19fe8 1390 return is_utf8_common(p, &PL_utf8_alnum, "IsWord");
a0ed51b3
LW
1391}
1392
1393bool
7fc63493 1394Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
a0ed51b3 1395{
97aff369 1396 dVAR;
7918f24d
NC
1397
1398 PERL_ARGS_ASSERT_IS_UTF8_IDFIRST;
1399
82686b01
JH
1400 if (*p == '_')
1401 return TRUE;
bde6a22d 1402 /* is_utf8_idstart would be more logical. */
d4c19fe8 1403 return is_utf8_common(p, &PL_utf8_idstart, "IdStart");
82686b01
JH
1404}
1405
1406bool
7fc63493 1407Perl_is_utf8_idcont(pTHX_ const U8 *p)
82686b01 1408{
97aff369 1409 dVAR;
7918f24d
NC
1410
1411 PERL_ARGS_ASSERT_IS_UTF8_IDCONT;
1412
82686b01
JH
1413 if (*p == '_')
1414 return TRUE;
d4c19fe8 1415 return is_utf8_common(p, &PL_utf8_idcont, "IdContinue");
a0ed51b3
LW
1416}
1417
1418bool
7fc63493 1419Perl_is_utf8_alpha(pTHX_ const U8 *p)
a0ed51b3 1420{
97aff369 1421 dVAR;
7918f24d
NC
1422
1423 PERL_ARGS_ASSERT_IS_UTF8_ALPHA;
1424
d4c19fe8 1425 return is_utf8_common(p, &PL_utf8_alpha, "IsAlpha");
a0ed51b3
LW
1426}
1427
1428bool
7fc63493 1429Perl_is_utf8_ascii(pTHX_ const U8 *p)
b8c5462f 1430{
97aff369 1431 dVAR;
7918f24d
NC
1432
1433 PERL_ARGS_ASSERT_IS_UTF8_ASCII;
1434
d4c19fe8 1435 return is_utf8_common(p, &PL_utf8_ascii, "IsAscii");
b8c5462f
JH
1436}
1437
1438bool
7fc63493 1439Perl_is_utf8_space(pTHX_ const U8 *p)
a0ed51b3 1440{
97aff369 1441 dVAR;
7918f24d
NC
1442
1443 PERL_ARGS_ASSERT_IS_UTF8_SPACE;
1444
d4c19fe8 1445 return is_utf8_common(p, &PL_utf8_space, "IsSpacePerl");
a0ed51b3
LW
1446}
1447
1448bool
d1eb3177
YO
1449Perl_is_utf8_perl_space(pTHX_ const U8 *p)
1450{
1451 dVAR;
1452
1453 PERL_ARGS_ASSERT_IS_UTF8_PERL_SPACE;
1454
1455 return is_utf8_common(p, &PL_utf8_perl_space, "IsPerlSpace");
1456}
1457
1458bool
1459Perl_is_utf8_perl_word(pTHX_ const U8 *p)
1460{
1461 dVAR;
1462
1463 PERL_ARGS_ASSERT_IS_UTF8_PERL_WORD;
1464
1465 return is_utf8_common(p, &PL_utf8_perl_word, "IsPerlWord");
1466}
1467
1468bool
7fc63493 1469Perl_is_utf8_digit(pTHX_ const U8 *p)
a0ed51b3 1470{
97aff369 1471 dVAR;
7918f24d
NC
1472
1473 PERL_ARGS_ASSERT_IS_UTF8_DIGIT;
1474
d4c19fe8 1475 return is_utf8_common(p, &PL_utf8_digit, "IsDigit");
a0ed51b3
LW
1476}
1477
1478bool
d1eb3177
YO
1479Perl_is_utf8_posix_digit(pTHX_ const U8 *p)
1480{
1481 dVAR;
1482
1483 PERL_ARGS_ASSERT_IS_UTF8_POSIX_DIGIT;
1484
1485 return is_utf8_common(p, &PL_utf8_posix_digit, "IsPosixDigit");
1486}
1487
1488bool
7fc63493 1489Perl_is_utf8_upper(pTHX_ const U8 *p)
a0ed51b3 1490{
97aff369 1491 dVAR;
7918f24d
NC
1492
1493 PERL_ARGS_ASSERT_IS_UTF8_UPPER;
1494
d4c19fe8 1495 return is_utf8_common(p, &PL_utf8_upper, "IsUppercase");
a0ed51b3
LW
1496}
1497
1498bool
7fc63493 1499Perl_is_utf8_lower(pTHX_ const U8 *p)
a0ed51b3 1500{
97aff369 1501 dVAR;
7918f24d
NC
1502
1503 PERL_ARGS_ASSERT_IS_UTF8_LOWER;
1504
d4c19fe8 1505 return is_utf8_common(p, &PL_utf8_lower, "IsLowercase");
a0ed51b3
LW
1506}
1507
1508bool
7fc63493 1509Perl_is_utf8_cntrl(pTHX_ const U8 *p)
b8c5462f 1510{
97aff369 1511 dVAR;
7918f24d
NC
1512
1513 PERL_ARGS_ASSERT_IS_UTF8_CNTRL;
1514
d4c19fe8 1515 return is_utf8_common(p, &PL_utf8_cntrl, "IsCntrl");
b8c5462f
JH
1516}
1517
1518bool
7fc63493 1519Perl_is_utf8_graph(pTHX_ const U8 *p)
b8c5462f 1520{
97aff369 1521 dVAR;
7918f24d
NC
1522
1523 PERL_ARGS_ASSERT_IS_UTF8_GRAPH;
1524
d4c19fe8 1525 return is_utf8_common(p, &PL_utf8_graph, "IsGraph");
b8c5462f
JH
1526}
1527
1528bool
7fc63493 1529Perl_is_utf8_print(pTHX_ const U8 *p)
a0ed51b3 1530{
97aff369 1531 dVAR;
7918f24d
NC
1532
1533 PERL_ARGS_ASSERT_IS_UTF8_PRINT;
1534
d4c19fe8 1535 return is_utf8_common(p, &PL_utf8_print, "IsPrint");
a0ed51b3
LW
1536}
1537
1538bool
7fc63493 1539Perl_is_utf8_punct(pTHX_ const U8 *p)
b8c5462f 1540{
97aff369 1541 dVAR;
7918f24d
NC
1542
1543 PERL_ARGS_ASSERT_IS_UTF8_PUNCT;
1544
d4c19fe8 1545 return is_utf8_common(p, &PL_utf8_punct, "IsPunct");
b8c5462f
JH
1546}
1547
1548bool
7fc63493 1549Perl_is_utf8_xdigit(pTHX_ const U8 *p)
b8c5462f 1550{
97aff369 1551 dVAR;
7918f24d
NC
1552
1553 PERL_ARGS_ASSERT_IS_UTF8_XDIGIT;
1554
d1eb3177 1555 return is_utf8_common(p, &PL_utf8_xdigit, "IsXDigit");
b8c5462f
JH
1556}
1557
1558bool
7fc63493 1559Perl_is_utf8_mark(pTHX_ const U8 *p)
a0ed51b3 1560{
97aff369 1561 dVAR;
7918f24d
NC
1562
1563 PERL_ARGS_ASSERT_IS_UTF8_MARK;
1564
d4c19fe8 1565 return is_utf8_common(p, &PL_utf8_mark, "IsM");
a0ed51b3
LW
1566}
1567
37e2e78e
KW
1568bool
1569Perl_is_utf8_X_begin(pTHX_ const U8 *p)
1570{
1571 dVAR;
1572
1573 PERL_ARGS_ASSERT_IS_UTF8_X_BEGIN;
1574
1575 return is_utf8_common(p, &PL_utf8_X_begin, "_X_Begin");
1576}
1577
1578bool
1579Perl_is_utf8_X_extend(pTHX_ const U8 *p)
1580{
1581 dVAR;
1582
1583 PERL_ARGS_ASSERT_IS_UTF8_X_EXTEND;
1584
1585 return is_utf8_common(p, &PL_utf8_X_extend, "_X_Extend");
1586}
1587
1588bool
1589Perl_is_utf8_X_prepend(pTHX_ const U8 *p)
1590{
1591 dVAR;
1592
1593 PERL_ARGS_ASSERT_IS_UTF8_X_PREPEND;
1594
1595 return is_utf8_common(p, &PL_utf8_X_prepend, "GCB=Prepend");
1596}
1597
1598bool
1599Perl_is_utf8_X_non_hangul(pTHX_ const U8 *p)
1600{
1601 dVAR;
1602
1603 PERL_ARGS_ASSERT_IS_UTF8_X_NON_HANGUL;
1604
1605 return is_utf8_common(p, &PL_utf8_X_non_hangul, "HST=Not_Applicable");
1606}
1607
1608bool
1609Perl_is_utf8_X_L(pTHX_ const U8 *p)
1610{
1611 dVAR;
1612
1613 PERL_ARGS_ASSERT_IS_UTF8_X_L;
1614
1615 return is_utf8_common(p, &PL_utf8_X_L, "GCB=L");
1616}
1617
1618bool
1619Perl_is_utf8_X_LV(pTHX_ const U8 *p)
1620{
1621 dVAR;
1622
1623 PERL_ARGS_ASSERT_IS_UTF8_X_LV;
1624
1625 return is_utf8_common(p, &PL_utf8_X_LV, "GCB=LV");
1626}
1627
1628bool
1629Perl_is_utf8_X_LVT(pTHX_ const U8 *p)
1630{
1631 dVAR;
1632
1633 PERL_ARGS_ASSERT_IS_UTF8_X_LVT;
1634
1635 return is_utf8_common(p, &PL_utf8_X_LVT, "GCB=LVT");
1636}
1637
1638bool
1639Perl_is_utf8_X_T(pTHX_ const U8 *p)
1640{
1641 dVAR;
1642
1643 PERL_ARGS_ASSERT_IS_UTF8_X_T;
1644
1645 return is_utf8_common(p, &PL_utf8_X_T, "GCB=T");
1646}
1647
1648bool
1649Perl_is_utf8_X_V(pTHX_ const U8 *p)
1650{
1651 dVAR;
1652
1653 PERL_ARGS_ASSERT_IS_UTF8_X_V;
1654
1655 return is_utf8_common(p, &PL_utf8_X_V, "GCB=V");
1656}
1657
1658bool
1659Perl_is_utf8_X_LV_LVT_V(pTHX_ const U8 *p)
1660{
1661 dVAR;
1662
1663 PERL_ARGS_ASSERT_IS_UTF8_X_LV_LVT_V;
1664
1665 return is_utf8_common(p, &PL_utf8_X_LV_LVT_V, "_X_LV_LVT_V");
1666}
1667
6b5c0936 1668/*
87cea99e 1669=for apidoc to_utf8_case
6b5c0936
JH
1670
1671The "p" contains the pointer to the UTF-8 string encoding
1672the character that is being converted.
1673
1674The "ustrp" is a pointer to the character buffer to put the
1675conversion result to. The "lenp" is a pointer to the length
1676of the result.
1677
0134edef 1678The "swashp" is a pointer to the swash to use.
6b5c0936 1679
0134edef 1680Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
8fe4d5b2 1681and loaded by SWASHNEW, using lib/utf8_heavy.pl. The special (usually,
0134edef 1682but not always, a multicharacter mapping), is tried first.
6b5c0936 1683
0134edef
JH
1684The "special" is a string like "utf8::ToSpecLower", which means the
1685hash %utf8::ToSpecLower. The access to the hash is through
1686Perl_to_utf8_case().
6b5c0936 1687
0134edef
JH
1688The "normal" is a string like "ToLower" which means the swash
1689%utf8::ToLower.
1690
1691=cut */
6b5c0936 1692
2104c8d9 1693UV
9a957fbc
AL
1694Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp,
1695 SV **swashp, const char *normal, const char *special)
a0ed51b3 1696{
97aff369 1697 dVAR;
89ebb4a3 1698 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
0134edef 1699 STRLEN len = 0;
aec46f14 1700 const UV uv0 = utf8_to_uvchr(p, NULL);
1feea2c7
JH
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. */
f54cb97a 1704 const UV uv1 = NATIVE_TO_UNI(uv0);
7918f24d
NC
1705
1706 PERL_ARGS_ASSERT_TO_UTF8_CASE;
1707
1feea2c7 1708 uvuni_to_utf8(tmpbuf, uv1);
0134edef
JH
1709
1710 if (!*swashp) /* load on-demand */
1711 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
37e2e78e
KW
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);
1718 if (hv) {
1719 SV **svp;
1720 svp = hv_fetch(hv, (const char*)normal, strlen(normal), FALSE);
1721 const char *s;
1722
1723 HV * const this_hash = SvRV(*svp);
1724 svp = hv_fetch(this_hash, "type", strlen("type"), FALSE);
1725 s = SvPV_const(*svp, len);
1726 }
1727 }*/
0134edef 1728
a6f87d8c 1729 if (special) {
0134edef 1730 /* It might be "special" (sometimes, but not always,
2a37f04d 1731 * a multicharacter mapping) */
6673a63c 1732 HV * const hv = get_hv(special, 0);
b08cf34e
JH
1733 SV **svp;
1734
35da51f7 1735 if (hv &&
b08cf34e
JH
1736 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1737 (*svp)) {
cfd0369c 1738 const char *s;
47654450 1739
cfd0369c 1740 s = SvPV_const(*svp, len);
47654450
JH
1741 if (len == 1)
1742 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
2a37f04d 1743 else {
2f9475ad
JH
1744#ifdef EBCDIC
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. */
7cda7a3d 1748 U8 *t = (U8*)s, *tend = t + len, *d;
2f9475ad
JH
1749
1750 d = tmpbuf;
b08cf34e 1751 if (SvUTF8(*svp)) {
2f9475ad
JH
1752 STRLEN tlen = 0;
1753
1754 while (t < tend) {
d4c19fe8 1755 const UV c = utf8_to_uvchr(t, &tlen);
2f9475ad
JH
1756 if (tlen > 0) {
1757 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1758 t += tlen;
1759 }
1760 else
1761 break;
1762 }
1763 }
1764 else {
36fec512
JH
1765 while (t < tend) {
1766 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1767 t++;
1768 }
2f9475ad
JH
1769 }
1770 len = d - tmpbuf;
1771 Copy(tmpbuf, ustrp, len, U8);
1772#else
d2dcd0fb 1773 Copy(s, ustrp, len, U8);
2f9475ad 1774#endif
29e98929 1775 }
983ffd37 1776 }
0134edef
JH
1777 }
1778
1779 if (!len && *swashp) {
d4c19fe8
AL
1780 const UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1781
0134edef
JH
1782 if (uv2) {
1783 /* It was "normal" (a single character mapping). */
d4c19fe8 1784 const UV uv3 = UNI_TO_NATIVE(uv2);
e9101d72 1785 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
2a37f04d
JH
1786 }
1787 }
1feea2c7 1788
37e2e78e
KW
1789 if (!len) /* Neither: just copy. In other words, there was no mapping
1790 defined, which means that the code point maps to itself */
0134edef
JH
1791 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1792
2a37f04d
JH
1793 if (lenp)
1794 *lenp = len;
1795
0134edef 1796 return len ? utf8_to_uvchr(ustrp, 0) : 0;
a0ed51b3
LW
1797}
1798
d3e79532 1799/*
87cea99e 1800=for apidoc to_utf8_upper
d3e79532
JH
1801
1802Convert the UTF-8 encoded character at p to its uppercase version and
1803store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
1804that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1805the uppercase version may be longer than the original character.
d3e79532
JH
1806
1807The first character of the uppercased version is returned
1808(but note, as explained above, that there may be more.)
1809
1810=cut */
1811
2104c8d9 1812UV
7fc63493 1813Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
a0ed51b3 1814{
97aff369 1815 dVAR;
7918f24d
NC
1816
1817 PERL_ARGS_ASSERT_TO_UTF8_UPPER;
1818
983ffd37 1819 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
b4e400f9 1820 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
983ffd37 1821}
a0ed51b3 1822
d3e79532 1823/*
87cea99e 1824=for apidoc to_utf8_title
d3e79532
JH
1825
1826Convert the UTF-8 encoded character at p to its titlecase version and
1827store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
1828that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1829titlecase version may be longer than the original character.
d3e79532
JH
1830
1831The first character of the titlecased version is returned
1832(but note, as explained above, that there may be more.)
1833
1834=cut */
1835
983ffd37 1836UV
7fc63493 1837Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
983ffd37 1838{
97aff369 1839 dVAR;
7918f24d
NC
1840
1841 PERL_ARGS_ASSERT_TO_UTF8_TITLE;
1842
983ffd37 1843 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
b4e400f9 1844 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
a0ed51b3
LW
1845}
1846
d3e79532 1847/*
87cea99e 1848=for apidoc to_utf8_lower
d3e79532
JH
1849
1850Convert the UTF-8 encoded character at p to its lowercase version and
1851store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
1852that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1853lowercase version may be longer than the original character.
d3e79532
JH
1854
1855The first character of the lowercased version is returned
1856(but note, as explained above, that there may be more.)
1857
1858=cut */
1859
2104c8d9 1860UV
7fc63493 1861Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
a0ed51b3 1862{
97aff369 1863 dVAR;
7918f24d
NC
1864
1865 PERL_ARGS_ASSERT_TO_UTF8_LOWER;
1866
983ffd37 1867 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
b4e400f9
JH
1868 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1869}
1870
d3e79532 1871/*
87cea99e 1872=for apidoc to_utf8_fold
d3e79532
JH
1873
1874Convert the UTF-8 encoded character at p to its foldcase version and
1875store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3 1876that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
d3e79532
JH
1877foldcase version may be longer than the original character (up to
1878three characters).
1879
1880The first character of the foldcased version is returned
1881(but note, as explained above, that there may be more.)
1882
1883=cut */
1884
b4e400f9 1885UV
7fc63493 1886Perl_to_utf8_fold(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
b4e400f9 1887{
97aff369 1888 dVAR;
7918f24d
NC
1889
1890 PERL_ARGS_ASSERT_TO_UTF8_FOLD;
1891
b4e400f9
JH
1892 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1893 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
a0ed51b3
LW
1894}
1895
711a919c
ST
1896/* Note:
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.
1901 */
a0ed51b3 1902SV*
7fc63493 1903Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
a0ed51b3 1904{
27da23d5 1905 dVAR;
a0ed51b3 1906 SV* retval;
8e84507e 1907 dSP;
7fc63493
AL
1908 const size_t pkg_len = strlen(pkg);
1909 const size_t name_len = strlen(name);
da51bb9b 1910 HV * const stash = gv_stashpvn(pkg, pkg_len, 0);
f8be5cf0 1911 SV* errsv_save;
88c57d91 1912 GV *method;
ce3b816e 1913
7918f24d
NC
1914 PERL_ARGS_ASSERT_SWASH_INIT;
1915
96ca9f55
DM
1916 PUSHSTACKi(PERLSI_MAGIC);
1917 ENTER;
ec34a119 1918 SAVEHINTS();
96ca9f55 1919 save_re_context();
88c57d91
NC
1920 method = gv_fetchmeth(stash, "SWASHNEW", 8, -1);
1921 if (!method) { /* demand load utf8 */
ce3b816e 1922 ENTER;
f8be5cf0 1923 errsv_save = newSVsv(ERRSV);
dc0c6abb
NC
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);
1930 PL_tainted = 0;
71bed85a 1931 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
a0714e2c 1932 NULL);
f8be5cf0
JH
1933 if (!SvTRUE(ERRSV))
1934 sv_setsv(ERRSV, errsv_save);
1935 SvREFCNT_dec(errsv_save);
ce3b816e
GS
1936 LEAVE;
1937 }
1938 SPAGAIN;
a0ed51b3
LW
1939 PUSHMARK(SP);
1940 EXTEND(SP,5);
6e449a3a
MHM
1941 mPUSHp(pkg, pkg_len);
1942 mPUSHp(name, name_len);
a0ed51b3 1943 PUSHs(listsv);
6e449a3a
MHM
1944 mPUSHi(minbits);
1945 mPUSHi(none);
a0ed51b3 1946 PUTBACK;
f8be5cf0 1947 errsv_save = newSVsv(ERRSV);
88c57d91
NC
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)
8b563eef 1951 : call_sv(newSVpvs_flags("SWASHNEW", SVs_TEMP), G_SCALAR | G_METHOD))
8e84507e 1952 retval = newSVsv(*PL_stack_sp--);
a0ed51b3 1953 else
e24b16f9 1954 retval = &PL_sv_undef;
f8be5cf0
JH
1955 if (!SvTRUE(ERRSV))
1956 sv_setsv(ERRSV, errsv_save);
1957 SvREFCNT_dec(errsv_save);
a0ed51b3
LW
1958 LEAVE;
1959 POPSTACK;
923e4eb5 1960 if (IN_PERL_COMPILETIME) {
623e6609 1961 CopHINTS_set(PL_curcop, PL_hints);
a0ed51b3 1962 }
bc45ce41
JH
1963 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
1964 if (SvPOK(retval))
35c1215d 1965 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
be2597df 1966 SVfARG(retval));
cea2e8a9 1967 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
bc45ce41 1968 }
a0ed51b3
LW
1969 return retval;
1970}
1971
035d37be
JH
1972
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
38684baa
KW
1977 * multiple values. --jhi
1978 * For those, you should use to_utf8_case() instead */
979f2922 1979/* Now SWASHGET is recasted into S_swash_get in this file. */
680c470c
ST
1980
1981/* Note:
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>.
1986 */
a0ed51b3 1987UV
680c470c 1988Perl_swash_fetch(pTHX_ SV *swash, const U8 *ptr, bool do_utf8)
a0ed51b3 1989{
27da23d5 1990 dVAR;
ef8f7699 1991 HV *const hv = MUTABLE_HV(SvRV(swash));
3568d838
JH
1992 U32 klen;
1993 U32 off;
a0ed51b3 1994 STRLEN slen;
7d85a32c 1995 STRLEN needents;
cfd0369c 1996 const U8 *tmps = NULL;
a0ed51b3 1997 U32 bit;
979f2922 1998 SV *swatch;
3568d838 1999 U8 tmputf8[2];
35da51f7 2000 const UV c = NATIVE_TO_ASCII(*ptr);
3568d838 2001
7918f24d
NC
2002 PERL_ARGS_ASSERT_SWASH_FETCH;
2003
3568d838 2004 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
979f2922
ST
2005 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
2006 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
2007 ptr = tmputf8;
3568d838
JH
2008 }
2009 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
37e2e78e 2010 * then the "swatch" is a vec() for all the chars which start
3568d838
JH
2011 * with 0xAA..0xYY
2012 * So the key in the hash (klen) is length of encoded char -1
2013 */
2014 klen = UTF8SKIP(ptr) - 1;
2015 off = ptr[klen];
a0ed51b3 2016
979f2922 2017 if (klen == 0) {
37e2e78e 2018 /* If char is invariant then swatch is for all the invariant chars
1e54db1a 2019 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
7d85a32c 2020 */
979f2922
ST
2021 needents = UTF_CONTINUATION_MARK;
2022 off = NATIVE_TO_UTF(ptr[klen]);
2023 }
2024 else {
7d85a32c 2025 /* If char is encoded then swatch is for the prefix */
979f2922
ST
2026 needents = (1 << UTF_ACCUMULATION_SHIFT);
2027 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
2028 }
7d85a32c 2029
a0ed51b3
LW
2030 /*
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...
2035 *
2036 * NB: this code assumes that swatches are never modified, once generated!
2037 */
2038
3568d838 2039 if (hv == PL_last_swash_hv &&
a0ed51b3 2040 klen == PL_last_swash_klen &&
27da23d5 2041 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
a0ed51b3
LW
2042 {
2043 tmps = PL_last_swash_tmps;
2044 slen = PL_last_swash_slen;
2045 }
2046 else {
2047 /* Try our second-level swatch cache, kept in a hash. */
e1ec3a88 2048 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
a0ed51b3 2049
979f2922
ST
2050 /* If not cached, generate it via swash_get */
2051 if (!svp || !SvPOK(*svp)
2052 || !(tmps = (const U8*)SvPV_const(*svp, slen))) {
2b9d42f0
NIS
2053 /* We use utf8n_to_uvuni() as we want an index into
2054 Unicode tables, not a native character number.
2055 */
aec46f14 2056 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
872c91ae
JH
2057 ckWARN(WARN_UTF8) ?
2058 0 : UTF8_ALLOW_ANY);
680c470c 2059 swatch = swash_get(swash,
979f2922
ST
2060 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
2061 (klen) ? (code_point & ~(needents - 1)) : 0,
2062 needents);
2063
923e4eb5 2064 if (IN_PERL_COMPILETIME)
623e6609 2065 CopHINTS_set(PL_curcop, PL_hints);
a0ed51b3 2066
979f2922 2067 svp = hv_store(hv, (const char *)ptr, klen, swatch, 0);
a0ed51b3 2068
979f2922
ST
2069 if (!svp || !(tmps = (U8*)SvPV(*svp, slen))
2070 || (slen << 3) < needents)
660a4616 2071 Perl_croak(aTHX_ "panic: swash_fetch got improper swatch");
a0ed51b3
LW
2072 }
2073
2074 PL_last_swash_hv = hv;
16d8f38a 2075 assert(klen <= sizeof(PL_last_swash_key));
eac04b2e 2076 PL_last_swash_klen = (U8)klen;
cfd0369c
NC
2077 /* FIXME change interpvar.h? */
2078 PL_last_swash_tmps = (U8 *) tmps;
a0ed51b3
LW
2079 PL_last_swash_slen = slen;
2080 if (klen)
2081 Copy(ptr, PL_last_swash_key, klen, U8);
2082 }
2083
9faf8d75 2084 switch ((int)((slen << 3) / needents)) {
a0ed51b3
LW
2085 case 1:
2086 bit = 1 << (off & 7);
2087 off >>= 3;
2088 return (tmps[off] & bit) != 0;
2089 case 8:
2090 return tmps[off];
2091 case 16:
2092 off <<= 1;
2093 return (tmps[off] << 8) + tmps[off + 1] ;
2094 case 32:
2095 off <<= 2;
2096 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
2097 }
660a4616 2098 Perl_croak(aTHX_ "panic: swash_fetch got swatch of unexpected bit width");
670f1322 2099 NORETURN_FUNCTION_END;
a0ed51b3 2100}
2b9d42f0 2101
319009ee
KW
2102/* Read a single line of the main body of the swash input text. These are of
2103 * the form:
2104 * 0053 0056 0073
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
2108 *
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
2118 */
2119
2120STATIC U8*
2121S_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)
2123{
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;
2127
2128 /* nl points to the next \n in the scan */
2129 U8* const nl = (U8*)memchr(l, '\n', lend - l);
2130
2131 /* Get the first number on the line: the range minimum */
2132 numlen = lend - l;
2133 *min = grok_hex((char *)l, &numlen, &flags, NULL);
2134 if (numlen) /* If found a hex number, position past it */
2135 l += numlen;
2136 else if (nl) { /* Else, go handle next line, if any */
2137 return nl + 1; /* 1 is length of "\n" */
2138 }
2139 else { /* Else, no next line */
2140 return lend + 1; /* to LIST's end at which \n is not found */
2141 }
2142
2143 /* The max range value follows, separated by a BLANK */
2144 if (isBLANK(*l)) {
2145 ++l;
2146 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
2147 numlen = lend - l;
2148 *max = grok_hex((char *)l, &numlen, &flags, NULL);
2149 if (numlen)
2150 l += numlen;
2151 else /* If no value here, it is a single element range */
2152 *max = *min;
2153
2154 /* Non-binary tables have a third entry: what the first element of the
2155 * range maps to */
2156 if (wants_value) {
2157 if (isBLANK(*l)) {
2158 ++l;
2159 flags = PERL_SCAN_SILENT_ILLDIGIT |
2160 PERL_SCAN_DISALLOW_PREFIX;
2161 numlen = lend - l;
2162 *val = grok_hex((char *)l, &numlen, &flags, NULL);
2163 if (numlen)
2164 l += numlen;
2165 else
2166 *val = 0;
2167 }
2168 else {
2169 *val = 0;
2170 if (typeto) {
2171 Perl_croak(aTHX_ "%s: illegal mapping '%s'",
2172 typestr, l);
2173 }
2174 }
2175 }
2176 else
2177 *val = 0; /* bits == 1, then any val should be ignored */
2178 }
2179 else { /* Nothing following range min, should be single element with no
2180 mapping expected */
2181 *max = *min;
2182 if (wants_value) {
2183 *val = 0;
2184 if (typeto) {
2185 Perl_croak(aTHX_ "%s: illegal mapping '%s'", typestr, l);
2186 }
2187 }
2188 else
2189 *val = 0; /* bits == 1, then val should be ignored */
2190 }
2191
2192 /* Position to next line if any, or EOF */
2193 if (nl)
2194 l = nl + 1;
2195 else
2196 l = lend;
2197
2198 return l;
2199}
2200
979f2922
ST
2201/* Note:
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>.
2206 */
2207STATIC SV*
2208S_swash_get(pTHX_ SV* swash, UV start, UV span)
2209{
2210 SV *swatch;
711a919c 2211 U8 *l, *lend, *x, *xend, *s;
979f2922 2212 STRLEN lcur, xcur, scur;
ef8f7699 2213 HV *const hv = MUTABLE_HV(SvRV(swash));
972dd592
KW
2214
2215 /* The string containing the main body of the table */
017a3ce5 2216 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
972dd592 2217
017a3ce5
GA
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);
0bd48802 2222 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
0bd48802
AL
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;
979f2922 2227
7918f24d
NC
2228 PERL_ARGS_ASSERT_SWASH_GET;
2229
979f2922 2230 if (bits != 1 && bits != 8 && bits != 16 && bits != 32) {
660a4616
ST
2231 Perl_croak(aTHX_ "panic: swash_get doesn't expect bits %"UVuf,
2232 (UV)bits);
979f2922
ST
2233 }
2234
2235 /* create and initialize $swatch */
979f2922 2236 scur = octets ? (span * octets) : (span + 7) / 8;
e524fe40
NC
2237 swatch = newSV(scur);
2238 SvPOK_on(swatch);
979f2922
ST
2239 s = (U8*)SvPVX(swatch);
2240 if (octets && none) {
0bd48802 2241 const U8* const e = s + scur;
979f2922
ST
2242 while (s < e) {
2243 if (bits == 8)
2244 *s++ = (U8)(none & 0xff);
2245 else if (bits == 16) {
2246 *s++ = (U8)((none >> 8) & 0xff);
2247 *s++ = (U8)( none & 0xff);
2248 }
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);
2254 }
2255 }
2256 *s = '\0';
2257 }
2258 else {
2259 (void)memzero((U8*)s, scur + 1);
2260 }
2261 SvCUR_set(swatch, scur);
2262 s = (U8*)SvPVX(swatch);
2263
2264 /* read $swash->{LIST} */
2265 l = (U8*)SvPV(*listsvp, lcur);
2266 lend = l + lcur;
2267 while (l < lend) {
35da51f7 2268 UV min, max, val;
319009ee
KW
2269 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2270 cBOOL(octets), typestr);
2271 if (l > lend) {
979f2922
ST
2272 break;
2273 }
2274
972dd592 2275 /* If looking for something beyond this range, go try the next one */
979f2922
ST
2276 if (max < start)
2277 continue;
2278
2279 if (octets) {
35da51f7 2280 UV key;
979f2922
ST
2281 if (min < start) {
2282 if (!none || val < none) {
2283 val += start - min;
2284 }
2285 min = start;
2286 }
2287 for (key = min; key <= max; key++) {
2288 STRLEN offset;
2289 if (key >= end)
2290 goto go_out_list;
2291 /* offset must be non-negative (start <= min <= key < end) */
2292 offset = octets * (key - start);
2293 if (bits == 8)
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);
2298 }
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);
2304 }
2305
2306 if (!none || val < none)
2307 ++val;
2308 }
2309 }
711a919c 2310 else { /* bits == 1, then val should be ignored */
35da51f7 2311 UV key;
979f2922
ST
2312 if (min < start)
2313 min = start;
2314 for (key = min; key <= max; key++) {
0bd48802 2315 const STRLEN offset = (STRLEN)(key - start);
979f2922
ST
2316 if (key >= end)
2317 goto go_out_list;
2318 s[offset >> 3] |= 1 << (offset & 7);
2319 }
2320 }
2321 } /* while */
2322 go_out_list:
2323
2324 /* read $swash->{EXTRAS} */
2325 x = (U8*)SvPV(*extssvp, xcur);
2326 xend = x + xcur;
2327 while (x < xend) {
2328 STRLEN namelen;
2329 U8 *namestr;
2330 SV** othersvp;
2331 HV* otherhv;
2332 STRLEN otherbits;
2333 SV **otherbitssvp, *other;
711a919c 2334 U8 *s, *o, *nl;
979f2922
ST
2335 STRLEN slen, olen;
2336
35da51f7 2337 const U8 opc = *x++;
979f2922
ST
2338 if (opc == '\n')
2339 continue;
2340
2341 nl = (U8*)memchr(x, '\n', xend - x);
2342
2343 if (opc != '-' && opc != '+' && opc != '!' && opc != '&') {
2344 if (nl) {
2345 x = nl + 1; /* 1 is length of "\n" */
2346 continue;
2347 }
2348 else {
2349 x = xend; /* to EXTRAS' end at which \n is not found */
2350 break;
2351 }
2352 }
2353
2354 namestr = x;
2355 if (nl) {
2356 namelen = nl - namestr;
2357 x = nl + 1;
2358 }
2359 else {
2360 namelen = xend - namestr;
2361 x = xend;
2362 }
2363
2364 othersvp = hv_fetch(hv, (char *)namestr, namelen, FALSE);
ef8f7699 2365 otherhv = MUTABLE_HV(SvRV(*othersvp));
017a3ce5 2366 otherbitssvp = hv_fetchs(otherhv, "BITS", FALSE);
979f2922
ST
2367 otherbits = (STRLEN)SvUV(*otherbitssvp);
2368 if (bits < otherbits)
660a4616 2369 Perl_croak(aTHX_ "panic: swash_get found swatch size mismatch");
979f2922
ST
2370
2371 /* The "other" swatch must be destroyed after. */
2372 other = swash_get(*othersvp, start, span);
2373 o = (U8*)SvPV(other, olen);
2374
2375 if (!olen)
660a4616 2376 Perl_croak(aTHX_ "panic: swash_get got improper swatch");
979f2922
ST
2377
2378 s = (U8*)SvPV(swatch, slen);
2379 if (bits == 1 && otherbits == 1) {
2380 if (slen != olen)
660a4616 2381 Perl_croak(aTHX_ "panic: swash_get found swatch length mismatch");
979f2922
ST
2382
2383 switch (opc) {
2384 case '+':
2385 while (slen--)
2386 *s++ |= *o++;
2387 break;
2388 case '!':
2389 while (slen--)
2390 *s++ |= ~*o++;
2391 break;
2392 case '-':
2393 while (slen--)
2394 *s++ &= ~*o++;
2395 break;
2396 case '&':
2397 while (slen--)
2398 *s++ &= *o++;
2399 break;
2400 default:
2401 break;
2402 }
2403 }
711a919c 2404 else {
979f2922
ST
2405 STRLEN otheroctets = otherbits >> 3;
2406 STRLEN offset = 0;
35da51f7 2407 U8* const send = s + slen;
979f2922
ST
2408
2409 while (s < send) {
2410 UV otherval = 0;
2411
2412 if (otherbits == 1) {
2413 otherval = (o[offset >> 3] >> (offset & 7)) & 1;
2414 ++offset;
2415 }
2416 else {
2417 STRLEN vlen = otheroctets;
2418 otherval = *o++;
2419 while (--vlen) {
2420 otherval <<= 8;
2421 otherval |= *o++;
2422 }
2423 }
2424
711a919c 2425 if (opc == '+' && otherval)
6f207bd3 2426 NOOP; /* replace with otherval */
979f2922
ST
2427 else if (opc == '!' && !otherval)
2428 otherval = 1;
2429 else if (opc == '-' && otherval)
2430 otherval = 0;
2431 else if (opc == '&' && !otherval)
2432 otherval = 0;
2433 else {
711a919c 2434 s += octets; /* no replacement */
979f2922
ST
2435 continue;
2436 }
2437
2438 if (bits == 8)
2439 *s++ = (U8)( otherval & 0xff);
2440 else if (bits == 16) {
2441 *s++ = (U8)((otherval >> 8) & 0xff);
2442 *s++ = (U8)( otherval & 0xff);
2443 }
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);
2449 }
2450 }
2451 }
2452 sv_free(other); /* through with it! */
2453 } /* while */
2454 return swatch;
2455}
2456
064c021d
KW
2457HV*
2458Perl__swash_inversion_hash(pTHX_ SV* swash)
2459{
2460
2461 /* Subject to change or removal. For use only in one place in regexec.c
2462 *
2463 * Returns a hash which is the inversion and closure of a swash mapping.
2464 * For example, consider the input lines:
2465 * 004B 006B
2466 * 004C 006C
2467 * 212A 006B
2468 *
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.
2473 *
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.
2476 *
2477 * Currently it only looks at the main body of the swash, and ignores any
2478 * additions or deletions from other swashes */
2479
2480 U8 *l, *lend;
2481 STRLEN lcur;
2482 HV *const hv = MUTABLE_HV(SvRV(swash));
2483
2484 /* The string containing the main body of the table */
2485 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2486
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);
2495
2496 HV* ret = newHV();
2497
2498 PERL_ARGS_ASSERT__SWASH_INVERSION_HASH;
2499
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,
2503 (UV)bits);
2504 }
2505
2506 /* read $swash->{LIST} */
2507 l = (U8*)SvPV(*listsvp, lcur);
2508 lend = l + lcur;
2509
2510 /* Go through each input line */
2511 while (l < lend) {
2512 UV min, max, val;
2513 UV inverse;
2514 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2515 cBOOL(octets), typestr);
2516 if (l > lend) {
2517 break;
2518 }
2519
2520 /* Each element in the range is to be inverted */
2521 for (inverse = min; inverse <= max; inverse++) {
2522 AV* list;
2523 SV* element;
2524 SV** listp;
2525 IV i;
2526 bool found_key = FALSE;
2527
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;
2532
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);
2536
2537 /* Get the list for the map */
2538 if ((listp = hv_fetch(ret, key, key_len, FALSE))) {
2539 list = (AV*) *listp;
2540 }
2541 else { /* No entry yet for it: create one */
2542 list = newAV();
2543 if (! hv_store(ret, key, key_len, (SV*) list, FALSE)) {
2544 Perl_croak(aTHX_ "panic: hv_store() unexpectedly failed");
2545 }
2546 }
2547
2548 for (i = 0; i < av_len(list); i++) {
2549 SV** entryp = av_fetch(list, i, FALSE);
2550 SV* entry;
2551 if (entryp == NULL) {
2552 Perl_croak(aTHX_ "panic: av_fetch() unexpectedly failed");
2553 }
2554 entry = *entryp;
2555 if (SvCUR(entry) != key_len) {
2556 continue;
2557 }
2558 if (memEQ(key, SvPVX(entry), key_len)) {
2559 found_key = TRUE;
2560 break;
2561 }
2562 }
2563 if (! found_key) {
2564 element = newSVpvn_flags(key, key_len, SVf_UTF8);
2565 av_push(list, element);
2566 }
2567
2568
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);
2572
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) {
2581 ++val;
2582 }
2583 }
2584 }
2585
2586 return ret;
2587}
2588
0f830e0b 2589/*
87cea99e 2590=for apidoc uvchr_to_utf8
0f830e0b
NC
2591
2592Adds the UTF-8 representation of the Native codepoint C<uv> to the end
2593of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
2594bytes available. The return value is the pointer to the byte after the
2595end of the new character. In other words,
2596
2597 d = uvchr_to_utf8(d, uv);
2598
2599is the recommended wide native character-aware way of saying
2600
2601 *(d++) = uv;
2602
2603=cut
2604*/
2605
2606/* On ASCII machines this is normally a macro but we want a
2607 real function in case XS code wants it
2608*/
0f830e0b
NC
2609U8 *
2610Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
2611{
7918f24d
NC
2612 PERL_ARGS_ASSERT_UVCHR_TO_UTF8;
2613
0f830e0b
NC
2614 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
2615}
2616
b851fbc1
JH
2617U8 *
2618Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
2619{
7918f24d
NC
2620 PERL_ARGS_ASSERT_UVCHR_TO_UTF8_FLAGS;
2621
b851fbc1
JH
2622 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
2623}
2b9d42f0
NIS
2624
2625/*
87cea99e 2626=for apidoc utf8n_to_uvchr
0f830e0b
NC
2627flags
2628
48ef279e 2629Returns the native character value of the first character in the string
0f830e0b
NC
2630C<s>
2631which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
2632length, in bytes, of that character.
2633
2634Allows length and flags to be passed to low level routine.
2635
2636=cut
2637*/
2638/* On ASCII machines this is normally a macro but we want
2639 a real function in case XS code wants it
2640*/
0f830e0b 2641UV
48ef279e 2642Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen,
0f830e0b
NC
2643U32 flags)
2644{
2645 const UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
7918f24d
NC
2646
2647 PERL_ARGS_ASSERT_UTF8N_TO_UVCHR;
2648
0f830e0b
NC
2649 return UNI_TO_NATIVE(uv);
2650}
2651
2652/*
87cea99e 2653=for apidoc pv_uni_display
d2cc3551
JH
2654
2655Build to the scalar dsv a displayable version of the string spv,
2656length len, the displayable version being at most pvlim bytes long
2657(if longer, the rest is truncated and "..." will be appended).
0a2ef054 2658
9e55ce06 2659The flags argument can have UNI_DISPLAY_ISPRINT set to display
00e86452 2660isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
0a2ef054
JH
2661to display the \\[nrfta\\] as the backslashed versions (like '\n')
2662(UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
2663UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
2664UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
2665
d2cc3551
JH
2666The pointer to the PV of the dsv is returned.
2667
2668=cut */
e6b2e755 2669char *
e1ec3a88 2670Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
e6b2e755
JH
2671{
2672 int truncated = 0;
e1ec3a88 2673 const char *s, *e;
e6b2e755 2674
7918f24d
NC
2675 PERL_ARGS_ASSERT_PV_UNI_DISPLAY;
2676
76f68e9b 2677 sv_setpvs(dsv, "");
7fddd944 2678 SvUTF8_off(dsv);
e1ec3a88 2679 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
e6b2e755 2680 UV u;
a49f32c6
NC
2681 /* This serves double duty as a flag and a character to print after
2682 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
2683 */
2684 char ok = 0;
c728cb41 2685
e6b2e755
JH
2686 if (pvlim && SvCUR(dsv) >= pvlim) {
2687 truncated++;
2688 break;
2689 }
2690 u = utf8_to_uvchr((U8*)s, 0);
c728cb41 2691 if (u < 256) {
a3b680e6 2692 const unsigned char c = (unsigned char)u & 0xFF;
0bd48802 2693 if (flags & UNI_DISPLAY_BACKSLASH) {
a49f32c6 2694 switch (c) {
c728cb41 2695 case '\n':
a49f32c6 2696 ok = 'n'; break;
c728cb41 2697 case '\r':
a49f32c6 2698 ok = 'r'; break;
c728cb41 2699 case '\t':
a49f32c6 2700 ok = 't'; break;
c728cb41 2701 case '\f':
a49f32c6 2702 ok = 'f'; break;
c728cb41 2703 case '\a':
a49f32c6 2704 ok = 'a'; break;
c728cb41 2705 case '\\':
a49f32c6 2706 ok = '\\'; break;
c728cb41
JH
2707 default: break;
2708 }
a49f32c6 2709 if (ok) {
88c9ea1e 2710 const char string = ok;
76f68e9b 2711 sv_catpvs(dsv, "\\");
5e7aa789 2712 sv_catpvn(dsv, &string, 1);
a49f32c6 2713 }
c728cb41 2714 }
00e86452 2715 /* isPRINT() is the locale-blind version. */
a49f32c6 2716 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
88c9ea1e 2717 const char string = c;
5e7aa789 2718 sv_catpvn(dsv, &string, 1);
a49f32c6 2719 ok = 1;
0a2ef054 2720 }
c728cb41
JH
2721 }
2722 if (!ok)
9e55ce06 2723 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
e6b2e755
JH
2724 }
2725 if (truncated)
396482e1 2726 sv_catpvs(dsv, "...");
48ef279e 2727
e6b2e755
JH
2728 return SvPVX(dsv);
2729}
2b9d42f0 2730
d2cc3551 2731/*
87cea99e 2732=for apidoc sv_uni_display
d2cc3551
JH
2733
2734Build to the scalar dsv a displayable version of the scalar sv,
0a2ef054 2735the displayable version being at most pvlim bytes long
d2cc3551 2736(if longer, the rest is truncated and "..." will be appended).
0a2ef054
JH
2737
2738The flags argument is as in pv_uni_display().
2739
d2cc3551
JH
2740The pointer to the PV of the dsv is returned.
2741
d4c19fe8
AL
2742=cut
2743*/
e6b2e755
JH
2744char *
2745Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
2746{
7918f24d
NC
2747 PERL_ARGS_ASSERT_SV_UNI_DISPLAY;
2748
cfd0369c
NC
2749 return Perl_pv_uni_display(aTHX_ dsv, (const U8*)SvPVX_const(ssv),
2750 SvCUR(ssv), pvlim, flags);
701a277b
JH
2751}
2752
d2cc3551 2753/*
e6226b18 2754=for apidoc foldEQ_utf8
d2cc3551 2755
d51c1b21 2756Returns true if the leading portions of the strings s1 and s2 (either or both
e6226b18 2757of which may be in UTF-8) are the same case-insensitively; false otherwise.
d51c1b21 2758How far into the strings to compare is determined by other input parameters.
8b35872c
KW
2759
2760If u1 is true, the string s1 is assumed to be in UTF-8-encoded Unicode;
2761otherwise it is assumed to be in native 8-bit encoding. Correspondingly for u2
2762with respect to s2.
2763
d51c1b21
KW
2764If the byte length l1 is non-zero, it says how far into s1 to check for fold
2765equality. In other words, s1+l1 will be used as a goal to reach. The
8b35872c
KW
2766scan will not be considered to be a match unless the goal is reached, and
2767scanning won't continue past that goal. Correspondingly for l2 with respect to
2768s2.
2769
2770If pe1 is non-NULL and the pointer it points to is not NULL, that pointer is
2771considered an end pointer beyond which scanning of s1 will not continue under
2772any circumstances. This means that if both l1 and pe1 are specified, and pe1
2773is less than s1+l1, the match will never be successful because it can never
d51c1b21
KW
2774get as far as its goal (and in fact is asserted against). Correspondingly for
2775pe2 with respect to s2.
8b35872c 2776
d51c1b21
KW
2777At least one of s1 and s2 must have a goal (at least one of l1 and l2 must be
2778non-zero), and if both do, both have to be
8b35872c
KW
2779reached for a successful match. Also, if the fold of a character is multiple
2780characters, all of them must be matched (see tr21 reference below for
2781'folding').
2782
e6226b18 2783Upon a successful match, if pe1 is non-NULL,
8b35872c
KW
2784it will be set to point to the beginning of the I<next> character of s1 beyond
2785what was matched. Correspondingly for pe2 and s2.
d2cc3551
JH
2786
2787For case-insensitiveness, the "casefolding" of Unicode is used
2788instead of upper/lowercasing both the characters, see
2789http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
2790
2791=cut */
701a277b 2792I32
e6226b18 2793Perl_foldEQ_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
332ddc25 2794{
8b35872c
KW
2795 dVAR;
2796 register const U8 *p1 = (const U8*)s1; /* Point to current char */
2797 register const U8 *p2 = (const U8*)s2;
48ef279e 2798 register const U8 *g1 = NULL; /* goal for s1 */
8b35872c 2799 register const U8 *g2 = NULL;
48ef279e
KW
2800 register const U8 *e1 = NULL; /* Don't scan s1 past this */
2801 register U8 *f1 = NULL; /* Point to current folded */
8b35872c
KW
2802 register const U8 *e2 = NULL;
2803 register U8 *f2 = NULL;
48ef279e 2804 STRLEN n1 = 0, n2 = 0; /* Number of bytes in current char */
8b35872c
KW
2805 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
2806 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
48ef279e
KW
2807 U8 natbuf[2]; /* Holds native 8-bit char converted to utf8;
2808 these always fit in 2 bytes */
8b35872c 2809
e6226b18 2810 PERL_ARGS_ASSERT_FOLDEQ_UTF8;
8b35872c
KW
2811
2812 if (pe1) {
48ef279e 2813 e1 = *(U8**)pe1;
8b35872c
KW
2814 }
2815
2816 if (l1) {
48ef279e 2817 g1 = (const U8*)s1 + l1;
8b35872c
KW
2818 }
2819
2820 if (pe2) {
48ef279e 2821 e2 = *(U8**)pe2;
8b35872c
KW
2822 }
2823
2824 if (l2) {
48ef279e 2825 g2 = (const U8*)s2 + l2;
8b35872c
KW
2826 }
2827
2828 /* Must have at least one goal */
2829 assert(g1 || g2);
2830
2831 if (g1) {
2832
48ef279e
KW
2833 /* Will never match if goal is out-of-bounds */
2834 assert(! e1 || e1 >= g1);
8b35872c 2835
48ef279e
KW
2836 /* Here, there isn't an end pointer, or it is beyond the goal. We
2837 * only go as far as the goal */
2838 e1 = g1;
8b35872c 2839 }
313b38e5
NC
2840 else {
2841 assert(e1); /* Must have an end for looking at s1 */
2842 }
8b35872c
KW
2843
2844 /* Same for goal for s2 */
2845 if (g2) {
48ef279e
KW
2846 assert(! e2 || e2 >= g2);
2847 e2 = g2;
8b35872c 2848 }
313b38e5
NC
2849 else {
2850 assert(e2);
2851 }
8b35872c
KW
2852
2853 /* Look through both strings, a character at a time */
2854 while (p1 < e1 && p2 < e2) {
2855
d51c1b21
KW
2856 /* If at the beginning of a new character in s1, get its fold to use
2857 * and the length of the fold */
48ef279e
KW
2858 if (n1 == 0) {
2859 if (u1) {
2860 to_utf8_fold(p1, foldbuf1, &n1);
2861 }
2862 else { /* Not utf8, convert to it first and then get fold */
2863 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p1)));
2864 to_utf8_fold(natbuf, foldbuf1, &n1);
2865 }
2866 f1 = foldbuf1;
2867 }
8b35872c 2868
48ef279e
KW
2869 if (n2 == 0) { /* Same for s2 */
2870 if (u2) {
2871 to_utf8_fold(p2, foldbuf2, &n2);
2872 }
2873 else {
2874 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p2)));
2875 to_utf8_fold(natbuf, foldbuf2, &n2);
2876 }
2877 f2 = foldbuf2;
2878 }
8b35872c 2879
48ef279e
KW
2880 /* While there is more to look for in both folds, see if they
2881 * continue to match */
2882 while (n1 && n2) {
2883 U8 fold_length = UTF8SKIP(f1);
2884 if (fold_length != UTF8SKIP(f2)
2885 || (fold_length == 1 && *f1 != *f2) /* Short circuit memNE
2886 function call for single
2887 character */
2888 || memNE((char*)f1, (char*)f2, fold_length))
2889 {
e6226b18 2890 return 0; /* mismatch */
48ef279e
KW
2891 }
2892
2893 /* Here, they matched, advance past them */
2894 n1 -= fold_length;
2895 f1 += fold_length;
2896 n2 -= fold_length;
2897 f2 += fold_length;
2898 }
8b35872c 2899
48ef279e
KW
2900 /* When reach the end of any fold, advance the input past it */
2901 if (n1 == 0) {
2902 p1 += u1 ? UTF8SKIP(p1) : 1;
2903 }
2904 if (n2 == 0) {
2905 p2 += u2 ? UTF8SKIP(p2) : 1;
2906 }
8b35872c
KW
2907 } /* End of loop through both strings */
2908
2909 /* A match is defined by each scan that specified an explicit length
2910 * reaching its final goal, and the other not having matched a partial
2911 * character (which can happen when the fold of a character is more than one
2912 * character). */
2913 if (! ((g1 == 0 || p1 == g1) && (g2 == 0 || p2 == g2)) || n1 || n2) {
e6226b18 2914 return 0;
8b35872c
KW
2915 }
2916
2917 /* Successful match. Set output pointers */
2918 if (pe1) {
48ef279e 2919 *pe1 = (char*)p1;
8b35872c
KW
2920 }
2921 if (pe2) {
48ef279e 2922 *pe2 = (char*)p2;
8b35872c 2923 }
e6226b18 2924 return 1;
e6b2e755 2925}
701a277b 2926
a49f32c6
NC
2927/*
2928 * Local variables:
2929 * c-indentation-style: bsd
2930 * c-basic-offset: 4
2931 * indent-tabs-mode: t
2932 * End:
2933 *
37442d52
RGS
2934 * ex: set ts=8 sts=4 sw=4 noet:
2935 */