This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regcomp.c: Generate different property for /i matching
[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
6ee84de2
KW
91Adds the UTF-8 representation of the code point C<uv> to the end
92of the string C<d>; C<d> should 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
949cf498 106This is the recommended Unicode-aware way of saying
eebe1485
SC
107
108 *(d++) = uv;
109
949cf498
KW
110This function will convert to UTF-8 (and not warn) even code points that aren't
111legal Unicode or are problematic, unless C<flags> contains one or more of the
112following flags.
113If C<uv> is a Unicode surrogate code point and UNICODE_WARN_SURROGATE is set,
114the function will raise a warning, provided UTF8 warnings are enabled. If instead
115UNICODE_DISALLOW_SURROGATE is set, the function will fail and return NULL.
116If both flags are set, the function will both warn and return NULL.
117
118The UNICODE_WARN_NONCHAR and UNICODE_DISALLOW_NONCHAR flags correspondingly
119affect how the function handles a Unicode non-character. And, likewise for the
120UNICODE_WARN_SUPER and UNICODE_DISALLOW_SUPER flags, and code points that are
121above the Unicode maximum of 0x10FFFF. Code points above 0x7FFF_FFFF (which are
122even less portable) can be warned and/or disallowed even if other above-Unicode
123code points are accepted by the UNICODE_WARN_FE_FF and UNICODE_DISALLOW_FE_FF
124flags.
125
126And finally, the flag UNICODE_WARN_ILLEGAL_INTERCHANGE selects all four of the
127above WARN flags; and UNICODE_DISALLOW_ILLEGAL_INTERCHANGE selects all four
128DISALLOW flags.
129
130
eebe1485
SC
131=cut
132*/
133
dfe13c55 134U8 *
b851fbc1 135Perl_uvuni_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
a0ed51b3 136{
7918f24d
NC
137 PERL_ARGS_ASSERT_UVUNI_TO_UTF8_FLAGS;
138
949cf498
KW
139 if (ckWARN_d(WARN_UTF8)) {
140 if (UNICODE_IS_SURROGATE(uv)) {
141 if (flags & UNICODE_WARN_SURROGATE) {
142 Perl_warner(aTHX_ packWARN(WARN_UTF8),
143 "UTF-16 surrogate U+%04"UVXf, uv);
144 }
145 if (flags & UNICODE_DISALLOW_SURROGATE) {
146 return NULL;
147 }
148 }
149 else if (UNICODE_IS_SUPER(uv)) {
150 if (flags & UNICODE_WARN_SUPER
151 || (UNICODE_IS_FE_FF(uv) && (flags & UNICODE_WARN_FE_FF)))
152 {
153 Perl_warner(aTHX_ packWARN(WARN_UTF8),
154 "Code point 0x%04"UVXf" is not Unicode, may not be portable", uv);
155 }
156 if (flags & UNICODE_DISALLOW_SUPER
157 || (UNICODE_IS_FE_FF(uv) && (flags & UNICODE_DISALLOW_FE_FF)))
158 {
159 return NULL;
160 }
161 }
162 else if (UNICODE_IS_NONCHAR(uv)) {
163 if (flags & UNICODE_WARN_NONCHAR) {
164 Perl_warner(aTHX_ packWARN(WARN_UTF8),
165 "Unicode non-character U+%04"UVXf" is illegal for open interchange",
166 uv);
167 }
168 if (flags & UNICODE_DISALLOW_NONCHAR) {
169 return NULL;
170 }
171 }
507b9800 172 }
c4d5f83a 173 if (UNI_IS_INVARIANT(uv)) {
eb160463 174 *d++ = (U8)UTF_TO_NATIVE(uv);
a0ed51b3
LW
175 return d;
176 }
2d331972 177#if defined(EBCDIC)
1d72bdf6
NIS
178 else {
179 STRLEN len = UNISKIP(uv);
180 U8 *p = d+len-1;
181 while (p > d) {
eb160463 182 *p-- = (U8)UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
1d72bdf6
NIS
183 uv >>= UTF_ACCUMULATION_SHIFT;
184 }
eb160463 185 *p = (U8)UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
1d72bdf6
NIS
186 return d+len;
187 }
188#else /* Non loop style */
a0ed51b3 189 if (uv < 0x800) {
eb160463
GS
190 *d++ = (U8)(( uv >> 6) | 0xc0);
191 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
192 return d;
193 }
194 if (uv < 0x10000) {
eb160463
GS
195 *d++ = (U8)(( uv >> 12) | 0xe0);
196 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
197 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
198 return d;
199 }
200 if (uv < 0x200000) {
eb160463
GS
201 *d++ = (U8)(( uv >> 18) | 0xf0);
202 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
203 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
204 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
205 return d;
206 }
207 if (uv < 0x4000000) {
eb160463
GS
208 *d++ = (U8)(( uv >> 24) | 0xf8);
209 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
210 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
211 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
212 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
213 return d;
214 }
215 if (uv < 0x80000000) {
eb160463
GS
216 *d++ = (U8)(( uv >> 30) | 0xfc);
217 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
218 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
219 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
220 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
221 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
222 return d;
223 }
6b8eaf93 224#ifdef HAS_QUAD
d7578b48 225 if (uv < UTF8_QUAD_MAX)
a0ed51b3
LW
226#endif
227 {
eb160463
GS
228 *d++ = 0xfe; /* Can't match U+FEFF! */
229 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
230 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
231 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
232 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
233 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
234 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
235 return d;
236 }
6b8eaf93 237#ifdef HAS_QUAD
a0ed51b3 238 {
eb160463
GS
239 *d++ = 0xff; /* Can't match U+FFFE! */
240 *d++ = 0x80; /* 6 Reserved bits */
241 *d++ = (U8)(((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
242 *d++ = (U8)(((uv >> 54) & 0x3f) | 0x80);
243 *d++ = (U8)(((uv >> 48) & 0x3f) | 0x80);
244 *d++ = (U8)(((uv >> 42) & 0x3f) | 0x80);
245 *d++ = (U8)(((uv >> 36) & 0x3f) | 0x80);
246 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
247 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
248 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
249 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
250 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
251 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
252 return d;
253 }
254#endif
1d72bdf6 255#endif /* Loop style */
a0ed51b3 256}
9041c2e3 257
646ca15d
JH
258/*
259
260Tests if some arbitrary number of bytes begins in a valid UTF-8
261character. Note that an INVARIANT (i.e. ASCII) character is a valid
262UTF-8 character. The actual number of bytes in the UTF-8 character
263will be returned if it is valid, otherwise 0.
264
265This is the "slow" version as opposed to the "fast" version which is
266the "unrolled" IS_UTF8_CHAR(). E.g. for t/uni/class.t the speed
267difference is a factor of 2 to 3. For lengths (UTF8SKIP(s)) of four
268or less you should use the IS_UTF8_CHAR(), for lengths of five or more
269you should use the _slow(). In practice this means that the _slow()
270will be used very rarely, since the maximum Unicode code point (as of
271Unicode 4.1) is U+10FFFF, which encodes in UTF-8 to four bytes. Only
272the "Perl extended UTF-8" (the infamous 'v-strings') will encode into
273five bytes or more.
274
275=cut */
c053b435 276STATIC STRLEN
5f66b61c 277S_is_utf8_char_slow(const U8 *s, const STRLEN len)
646ca15d
JH
278{
279 U8 u = *s;
280 STRLEN slen;
281 UV uv, ouv;
282
7918f24d
NC
283 PERL_ARGS_ASSERT_IS_UTF8_CHAR_SLOW;
284
646ca15d
JH
285 if (UTF8_IS_INVARIANT(u))
286 return 1;
287
288 if (!UTF8_IS_START(u))
289 return 0;
290
291 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
292 return 0;
293
294 slen = len - 1;
295 s++;
77263263
ST
296#ifdef EBCDIC
297 u = NATIVE_TO_UTF(u);
298#endif
646ca15d
JH
299 u &= UTF_START_MASK(len);
300 uv = u;
301 ouv = uv;
302 while (slen--) {
303 if (!UTF8_IS_CONTINUATION(*s))
304 return 0;
305 uv = UTF8_ACCUMULATE(uv, *s);
48ef279e 306 if (uv < ouv)
646ca15d
JH
307 return 0;
308 ouv = uv;
309 s++;
310 }
311
312 if ((STRLEN)UNISKIP(uv) < len)
313 return 0;
314
315 return len;
316}
9041c2e3
NIS
317
318/*
87cea99e 319=for apidoc is_utf8_char
eebe1485 320
5da9da9e 321Tests if some arbitrary number of bytes begins in a valid UTF-8
2bbc8d55
SP
322character. Note that an INVARIANT (i.e. ASCII on non-EBCDIC machines)
323character is a valid UTF-8 character. The actual number of bytes in the UTF-8
324character will be returned if it is valid, otherwise 0.
9041c2e3 325
82686b01 326=cut */
067a85ef 327STRLEN
668b6d8d 328Perl_is_utf8_char(const U8 *s)
386d01d6 329{
44f8325f 330 const STRLEN len = UTF8SKIP(s);
7918f24d
NC
331
332 PERL_ARGS_ASSERT_IS_UTF8_CHAR;
3b0fc154 333#ifdef IS_UTF8_CHAR
768c67ee 334 if (IS_UTF8_CHAR_FAST(len))
3b0fc154
JH
335 return IS_UTF8_CHAR(s, len) ? len : 0;
336#endif /* #ifdef IS_UTF8_CHAR */
2c0c5f92 337 return is_utf8_char_slow(s, len);
386d01d6
GS
338}
339
eaf7a4d2 340
6662521e 341/*
87cea99e 342=for apidoc is_utf8_string
6662521e 343
c9ada85f 344Returns true if first C<len> bytes of the given string form a valid
9f7e3d64
MH
345UTF-8 string, false otherwise. If C<len> is 0, it will be calculated
346using C<strlen(s)>. Note that 'a valid UTF-8 string' does not mean 'a
347string that contains code points above 0x7F encoded in UTF-8' because a
348valid ASCII string is a valid UTF-8 string.
6662521e 349
eaf7a4d2 350See also is_ascii_string(), is_utf8_string_loclen(), and is_utf8_string_loc().
768c67ee 351
6662521e
GS
352=cut
353*/
354
8e84507e 355bool
668b6d8d 356Perl_is_utf8_string(const U8 *s, STRLEN len)
6662521e 357{
35da51f7 358 const U8* const send = s + (len ? len : strlen((const char *)s));
7fc63493 359 const U8* x = s;
067a85ef 360
7918f24d 361 PERL_ARGS_ASSERT_IS_UTF8_STRING;
1aa99e6b 362
6662521e 363 while (x < send) {
a3b680e6 364 STRLEN c;
1acdb0da
JH
365 /* Inline the easy bits of is_utf8_char() here for speed... */
366 if (UTF8_IS_INVARIANT(*x))
367 c = 1;
368 else if (!UTF8_IS_START(*x))
768c67ee 369 goto out;
1acdb0da
JH
370 else {
371 /* ... and call is_utf8_char() only if really needed. */
646ca15d
JH
372#ifdef IS_UTF8_CHAR
373 c = UTF8SKIP(x);
768c67ee
JH
374 if (IS_UTF8_CHAR_FAST(c)) {
375 if (!IS_UTF8_CHAR(x, c))
3c614e38
AD
376 c = 0;
377 }
378 else
379 c = is_utf8_char_slow(x, c);
646ca15d
JH
380#else
381 c = is_utf8_char(x);
382#endif /* #ifdef IS_UTF8_CHAR */
1acdb0da 383 if (!c)
768c67ee 384 goto out;
1acdb0da 385 }
6662521e 386 x += c;
6662521e 387 }
768c67ee
JH
388
389 out:
60006e79
JH
390 if (x != send)
391 return FALSE;
067a85ef
A
392
393 return TRUE;
6662521e
GS
394}
395
67e989fb 396/*
814fafa7
NC
397Implemented as a macro in utf8.h
398
87cea99e 399=for apidoc is_utf8_string_loc
814fafa7
NC
400
401Like is_utf8_string() but stores the location of the failure (in the
402case of "utf8ness failure") or the location s+len (in the case of
403"utf8ness success") in the C<ep>.
404
405See also is_utf8_string_loclen() and is_utf8_string().
406
87cea99e 407=for apidoc is_utf8_string_loclen
81cd54e3 408
e3e4599f 409Like is_utf8_string() but stores the location of the failure (in the
768c67ee
JH
410case of "utf8ness failure") or the location s+len (in the case of
411"utf8ness success") in the C<ep>, and the number of UTF-8
412encoded characters in the C<el>.
413
414See also is_utf8_string_loc() and is_utf8_string().
81cd54e3
JH
415
416=cut
417*/
418
419bool
668b6d8d 420Perl_is_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
81cd54e3 421{
35da51f7 422 const U8* const send = s + (len ? len : strlen((const char *)s));
7fc63493 423 const U8* x = s;
81cd54e3 424 STRLEN c;
3ebfea28 425 STRLEN outlen = 0;
7918f24d
NC
426
427 PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN;
81cd54e3 428
81cd54e3
JH
429 while (x < send) {
430 /* Inline the easy bits of is_utf8_char() here for speed... */
431 if (UTF8_IS_INVARIANT(*x))
768c67ee
JH
432 c = 1;
433 else if (!UTF8_IS_START(*x))
434 goto out;
81cd54e3 435 else {
768c67ee
JH
436 /* ... and call is_utf8_char() only if really needed. */
437#ifdef IS_UTF8_CHAR
438 c = UTF8SKIP(x);
439 if (IS_UTF8_CHAR_FAST(c)) {
440 if (!IS_UTF8_CHAR(x, c))
441 c = 0;
442 } else
443 c = is_utf8_char_slow(x, c);
444#else
445 c = is_utf8_char(x);
446#endif /* #ifdef IS_UTF8_CHAR */
447 if (!c)
448 goto out;
81cd54e3 449 }
768c67ee 450 x += c;
3ebfea28 451 outlen++;
81cd54e3 452 }
768c67ee
JH
453
454 out:
3ebfea28
AL
455 if (el)
456 *el = outlen;
457
768c67ee
JH
458 if (ep)
459 *ep = x;
3ebfea28 460 return (x == send);
81cd54e3
JH
461}
462
463/*
768c67ee 464
87cea99e 465=for apidoc utf8n_to_uvuni
67e989fb 466
9041c2e3 467Bottom level UTF-8 decode routine.
949cf498
KW
468Returns the code point value of the first character in the string C<s>
469which is assumed to be in UTF-8 (or UTF-EBCDIC) encoding and no longer than
470C<curlen> bytes; C<retlen> will be set to the length, in bytes, of that
471character.
472
473The value of C<flags> determines the behavior when C<s> does not point to a
474well-formed UTF-8 character. If C<flags> is 0, when a malformation is found,
475C<retlen> is set to the expected length of the UTF-8 character in bytes, zero
476is returned, and if UTF-8 warnings haven't been lexically disabled, a warning
477is raised.
478
479Various ALLOW flags can be set in C<flags> to allow (and not warn on)
480individual types of malformations, such as the sequence being overlong (that
481is, when there is a shorter sequence that can express the same code point;
482overlong sequences are expressly forbidden in the UTF-8 standard due to
483potential security issues). Another malformation example is the first byte of
484a character not being a legal first byte. See F<utf8.h> for the list of such
485flags. Of course, the value returned by this function under such conditions is
486not reliable.
487
488The UTF8_CHECK_ONLY flag overrides the behavior when a non-allowed (by other
489flags) malformation is found. If this flag is set, the routine assumes that
490the caller will raise a warning, and this function will silently just set
491C<retlen> to C<-1> and return zero.
492
493Certain code points are considered problematic. These are Unicode surrogates,
494Unicode non-characters, and code points above the Unicode maximum of 0x10FFF.
495By default these are considered regular code points, but certain situations
496warrant special handling for them. if C<flags> contains
497UTF8_DISALLOW_ILLEGAL_INTERCHANGE, all three classes are treated as
498malformations and handled as such. The flags UTF8_DISALLOW_SURROGATE,
499UTF8_DISALLOW_NONCHAR, and UTF8_DISALLOW_SUPER (meaning above the legal Unicode
500maximum) can be set to disallow these categories individually.
501
502The flags UTF8_WARN_ILLEGAL_INTERCHANGE, UTF8_WARN_SURROGATE,
503UTF8_WARN_NONCHAR, and UTF8_WARN_SUPER will cause warning messages to be raised
504for their respective categories, but otherwise the code points are considered
505valid (not malformations). To get a category to both be treated as a
506malformation and raise a warning, specify both the WARN and DISALLOW flags.
507(But note that warnings are not raised if lexically disabled nor if
508UTF8_CHECK_ONLY is also specified.)
509
510Very large code points (above 0x7FFF_FFFF) are considered more problematic than
511the others that are above the Unicode legal maximum. There are several
512reasons, one of which is that the original UTF-8 specification never went above
513this number (the current 0x10FFF limit was imposed later). The UTF-8 encoding
514on ASCII platforms for these large code point begins with a byte containing
5150xFE or 0xFF. The UTF8_DISALLOW_FE_FF flag will cause them to be treated as
516malformations, while allowing smaller above-Unicode code points. (Of course
517UTF8_DISALLOW_SUPER will treat all above-Unicode code points, including these,
518as malformations.) Similarly, UTF8_WARN_FE_FF acts just like the other WARN
519flags, but applies just to these code points.
520
521All other code points corresponding to Unicode characters, including private
522use and those yet to be assigned, are never considered malformed and never
523warn.
67e989fb 524
9041c2e3
NIS
525Most code should use utf8_to_uvchr() rather than call this directly.
526
37607a96
PK
527=cut
528*/
67e989fb 529
a0ed51b3 530UV
7fc63493 531Perl_utf8n_to_uvuni(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
a0ed51b3 532{
97aff369 533 dVAR;
d4c19fe8 534 const U8 * const s0 = s;
9c5ffd7c 535 UV uv = *s, ouv = 0;
ba210ebe 536 STRLEN len = 1;
949cf498 537 bool dowarn = ckWARN_d(WARN_UTF8);
7fc63493 538 const UV startbyte = *s;
ba210ebe 539 STRLEN expectlen = 0;
a0dbb045 540 U32 warning = 0;
949cf498 541 SV* sv = NULL;
a0dbb045 542
7918f24d
NC
543 PERL_ARGS_ASSERT_UTF8N_TO_UVUNI;
544
949cf498 545/* This list is a superset of the UTF8_ALLOW_XXX. */
a0dbb045
JH
546
547#define UTF8_WARN_EMPTY 1
548#define UTF8_WARN_CONTINUATION 2
549#define UTF8_WARN_NON_CONTINUATION 3
f9bc8ed7
KW
550#define UTF8_WARN_SHORT 4
551#define UTF8_WARN_OVERFLOW 5
552#define UTF8_WARN_LONG 6
a0dbb045
JH
553
554 if (curlen == 0 &&
555 !(flags & UTF8_ALLOW_EMPTY)) {
556 warning = UTF8_WARN_EMPTY;
0c443dc2
JH
557 goto malformed;
558 }
559
1d72bdf6 560 if (UTF8_IS_INVARIANT(uv)) {
a0ed51b3
LW
561 if (retlen)
562 *retlen = 1;
c4d5f83a 563 return (UV) (NATIVE_TO_UTF(*s));
a0ed51b3 564 }
67e989fb 565
421a8bf2 566 if (UTF8_IS_CONTINUATION(uv) &&
fcc8fcf6 567 !(flags & UTF8_ALLOW_CONTINUATION)) {
a0dbb045 568 warning = UTF8_WARN_CONTINUATION;
ba210ebe
JH
569 goto malformed;
570 }
571
421a8bf2 572 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
fcc8fcf6 573 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
a0dbb045 574 warning = UTF8_WARN_NON_CONTINUATION;
ba210ebe
JH
575 goto malformed;
576 }
9041c2e3 577
1d72bdf6 578#ifdef EBCDIC
75383841 579 uv = NATIVE_TO_UTF(uv);
1d72bdf6 580#else
949cf498
KW
581 if (uv == 0xfe || uv == 0xff) {
582 if (flags & (UTF8_WARN_SUPER|UTF8_WARN_FE_FF)) {
111d382d 583 sv = sv_2mortal(Perl_newSVpvf(aTHX_ "Code point beginning with byte 0x%02"UVXf" is not Unicode, and not portable", uv));
949cf498
KW
584 flags &= ~UTF8_WARN_SUPER; /* Only warn once on this problem */
585 }
586 if (flags & (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_FE_FF)) {
587 goto malformed;
588 }
a0ed51b3 589 }
1d72bdf6
NIS
590#endif
591
ba210ebe
JH
592 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
593 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
594 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
595 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
1d72bdf6
NIS
596#ifdef EBCDIC
597 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
598 else { len = 7; uv &= 0x01; }
599#else
ba210ebe
JH
600 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
601 else if (!(uv & 0x01)) { len = 7; uv = 0; }
1d72bdf6
NIS
602 else { len = 13; uv = 0; } /* whoa! */
603#endif
604
a0ed51b3
LW
605 if (retlen)
606 *retlen = len;
9041c2e3 607
ba210ebe
JH
608 expectlen = len;
609
fcc8fcf6
JH
610 if ((curlen < expectlen) &&
611 !(flags & UTF8_ALLOW_SHORT)) {
a0dbb045 612 warning = UTF8_WARN_SHORT;
ba210ebe
JH
613 goto malformed;
614 }
615
616 len--;
a0ed51b3 617 s++;
949cf498 618 ouv = uv; /* ouv is the value from the previous iteration */
ba210ebe 619
a0ed51b3 620 while (len--) {
421a8bf2
JH
621 if (!UTF8_IS_CONTINUATION(*s) &&
622 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
a0dbb045
JH
623 s--;
624 warning = UTF8_WARN_NON_CONTINUATION;
ba210ebe 625 goto malformed;
a0ed51b3
LW
626 }
627 else
8850bf83 628 uv = UTF8_ACCUMULATE(uv, *s);
949cf498
KW
629 if (!(uv > ouv)) { /* If the value didn't grow from the previous
630 iteration, something is horribly wrong */
a0dbb045
JH
631 /* These cannot be allowed. */
632 if (uv == ouv) {
75dbc644 633 if (expectlen != 13 && !(flags & UTF8_ALLOW_LONG)) {
a0dbb045
JH
634 warning = UTF8_WARN_LONG;
635 goto malformed;
636 }
637 }
638 else { /* uv < ouv */
639 /* This cannot be allowed. */
640 warning = UTF8_WARN_OVERFLOW;
641 goto malformed;
642 }
ba210ebe
JH
643 }
644 s++;
645 ouv = uv;
646 }
647
949cf498 648 if ((expectlen > (STRLEN)UNISKIP(uv)) && !(flags & UTF8_ALLOW_LONG)) {
a0dbb045 649 warning = UTF8_WARN_LONG;
ba210ebe 650 goto malformed;
949cf498
KW
651 } else if (flags & (UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE)) {
652 if (UNICODE_IS_SURROGATE(uv)) {
653 if ((flags & (UTF8_WARN_SURROGATE|UTF8_CHECK_ONLY)) == UTF8_WARN_SURROGATE) {
111d382d 654 sv = sv_2mortal(Perl_newSVpvf(aTHX_ "UTF-16 surrogate U+%04"UVXf"", uv));
949cf498
KW
655 }
656 if (flags & UTF8_DISALLOW_SURROGATE) {
657 goto disallowed;
658 }
659 }
660 else if (UNICODE_IS_NONCHAR(uv)) {
661 if ((flags & (UTF8_WARN_NONCHAR|UTF8_CHECK_ONLY)) == UTF8_WARN_NONCHAR ) {
111d382d 662 sv = sv_2mortal(Perl_newSVpvf(aTHX_ "Unicode non-character U+%04"UVXf" is illegal for open interchange", uv));
949cf498
KW
663 }
664 if (flags & UTF8_DISALLOW_NONCHAR) {
665 goto disallowed;
666 }
667 }
668 else if ((uv > PERL_UNICODE_MAX)) {
669 if ((flags & (UTF8_WARN_SUPER|UTF8_CHECK_ONLY)) == UTF8_WARN_SUPER) {
111d382d 670 sv = sv_2mortal(Perl_newSVpvf(aTHX_ "Code point 0x%04"UVXf" is not Unicode, may not be portable", uv));
949cf498
KW
671 }
672 if (flags & UTF8_DISALLOW_SUPER) {
673 goto disallowed;
674 }
675 }
676
677 /* Here, this is not considered a malformed character, so drop through
678 * to return it */
a0ed51b3 679 }
ba210ebe 680
a0ed51b3 681 return uv;
ba210ebe 682
949cf498
KW
683disallowed: /* Is disallowed, but otherwise not malformed. 'sv' will have been
684 set if there is to be a warning. */
685 if (!sv) {
686 dowarn = 0;
687 }
688
ba210ebe
JH
689malformed:
690
fcc8fcf6 691 if (flags & UTF8_CHECK_ONLY) {
ba210ebe 692 if (retlen)
10edeb5d 693 *retlen = ((STRLEN) -1);
ba210ebe
JH
694 return 0;
695 }
696
a0dbb045 697 if (dowarn) {
949cf498 698 if (! sv) {
5b311467 699 sv = newSVpvs_flags("Malformed UTF-8 character ", SVs_TEMP);
949cf498 700 }
5b311467 701
5e3d7cf5
KW
702 switch (warning) {
703 case 0: /* Intentionally empty. */ break;
704 case UTF8_WARN_EMPTY:
705 sv_catpvs(sv, "(empty string)");
706 break;
707 case UTF8_WARN_CONTINUATION:
708 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
709 break;
710 case UTF8_WARN_NON_CONTINUATION:
711 if (s == s0)
712 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
713 (UV)s[1], startbyte);
714 else {
715 const int len = (int)(s-s0);
716 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
717 (UV)s[1], len, len > 1 ? "s" : "", startbyte, (int)expectlen);
718 }
719
720 break;
721 case UTF8_WARN_SHORT:
722 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
723 (int)curlen, curlen == 1 ? "" : "s", (int)expectlen, startbyte);
724 expectlen = curlen; /* distance for caller to skip */
725 break;
726 case UTF8_WARN_OVERFLOW:
727 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
728 ouv, *s, startbyte);
729 break;
730 case UTF8_WARN_LONG:
731 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
732 (int)expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
733 break;
734 default:
735 sv_catpvs(sv, "(unknown reason)");
736 break;
737 }
a0dbb045 738
949cf498 739 if (sv) {
44f8325f 740 const char * const s = SvPVX_const(sv);
a0dbb045
JH
741
742 if (PL_op)
9014280d 743 Perl_warner(aTHX_ packWARN(WARN_UTF8),
53e06cf0 744 "%s in %s", s, OP_DESC(PL_op));
a0dbb045 745 else
9014280d 746 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
a0dbb045
JH
747 }
748 }
749
ba210ebe 750 if (retlen)
28d3d195 751 *retlen = expectlen ? expectlen : len;
ba210ebe 752
28d3d195 753 return 0;
a0ed51b3
LW
754}
755
8e84507e 756/*
87cea99e 757=for apidoc utf8_to_uvchr
9041c2e3 758
6ee84de2 759Returns the native code point of the first character in the string C<s>
1e54db1a 760which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
9041c2e3
NIS
761length, in bytes, of that character.
762
1e54db1a 763If C<s> does not point to a well-formed UTF-8 character, zero is
9041c2e3
NIS
764returned and retlen is set, if possible, to -1.
765
766=cut
767*/
768
ae0e24b1 769
9041c2e3 770UV
7fc63493 771Perl_utf8_to_uvchr(pTHX_ const U8 *s, STRLEN *retlen)
9041c2e3 772{
7918f24d
NC
773 PERL_ARGS_ASSERT_UTF8_TO_UVCHR;
774
1754c1a1 775 return utf8n_to_uvchr(s, UTF8_MAXBYTES, retlen,
ae0e24b1 776 ckWARN_d(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
9041c2e3
NIS
777}
778
779/*
87cea99e 780=for apidoc utf8_to_uvuni
9041c2e3
NIS
781
782Returns the Unicode code point of the first character in the string C<s>
1e54db1a 783which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
9041c2e3
NIS
784length, in bytes, of that character.
785
2bbc8d55 786This function should only be used when the returned UV is considered
9041c2e3
NIS
787an index into the Unicode semantic tables (e.g. swashes).
788
1e54db1a 789If C<s> does not point to a well-formed UTF-8 character, zero is
ba210ebe 790returned and retlen is set, if possible, to -1.
8e84507e
NIS
791
792=cut
793*/
794
795UV
7fc63493 796Perl_utf8_to_uvuni(pTHX_ const U8 *s, STRLEN *retlen)
8e84507e 797{
7918f24d
NC
798 PERL_ARGS_ASSERT_UTF8_TO_UVUNI;
799
9041c2e3 800 /* Call the low level routine asking for checks */
89ebb4a3 801 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXBYTES, retlen,
ae0e24b1 802 ckWARN_d(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
8e84507e
NIS
803}
804
b76347f2 805/*
87cea99e 806=for apidoc utf8_length
b76347f2
JH
807
808Return the length of the UTF-8 char encoded string C<s> in characters.
02eb7b47
JH
809Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
810up past C<e>, croaks.
b76347f2
JH
811
812=cut
813*/
814
815STRLEN
35a4481c 816Perl_utf8_length(pTHX_ const U8 *s, const U8 *e)
b76347f2 817{
97aff369 818 dVAR;
b76347f2
JH
819 STRLEN len = 0;
820
7918f24d
NC
821 PERL_ARGS_ASSERT_UTF8_LENGTH;
822
8850bf83
JH
823 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
824 * the bitops (especially ~) can create illegal UTF-8.
825 * In other words: in Perl UTF-8 is not just for Unicode. */
826
a3b680e6
AL
827 if (e < s)
828 goto warn_and_return;
b76347f2 829 while (s < e) {
8e91ec7f
AV
830 if (!UTF8_IS_INVARIANT(*s))
831 s += UTF8SKIP(s);
832 else
833 s++;
834 len++;
835 }
836
837 if (e != s) {
838 len--;
839 warn_and_return:
9b387841
NC
840 if (PL_op)
841 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
842 "%s in %s", unees, OP_DESC(PL_op));
843 else
844 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8), unees);
b76347f2
JH
845 }
846
847 return len;
848}
849
b06226ff 850/*
87cea99e 851=for apidoc utf8_distance
b06226ff 852
1e54db1a 853Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
b06226ff
JH
854and C<b>.
855
856WARNING: use only if you *know* that the pointers point inside the
857same UTF-8 buffer.
858
37607a96
PK
859=cut
860*/
a0ed51b3 861
02eb7b47 862IV
35a4481c 863Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
a0ed51b3 864{
7918f24d
NC
865 PERL_ARGS_ASSERT_UTF8_DISTANCE;
866
bf1665bc 867 return (a < b) ? -1 * (IV) utf8_length(a, b) : (IV) utf8_length(b, a);
a0ed51b3
LW
868}
869
b06226ff 870/*
87cea99e 871=for apidoc utf8_hop
b06226ff 872
8850bf83
JH
873Return the UTF-8 pointer C<s> displaced by C<off> characters, either
874forward or backward.
b06226ff
JH
875
876WARNING: do not use the following unless you *know* C<off> is within
8850bf83
JH
877the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
878on the first byte of character or just after the last byte of a character.
b06226ff 879
37607a96
PK
880=cut
881*/
a0ed51b3
LW
882
883U8 *
4373e329 884Perl_utf8_hop(pTHX_ const U8 *s, I32 off)
a0ed51b3 885{
7918f24d
NC
886 PERL_ARGS_ASSERT_UTF8_HOP;
887
96a5add6 888 PERL_UNUSED_CONTEXT;
8850bf83
JH
889 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
890 * the bitops (especially ~) can create illegal UTF-8.
891 * In other words: in Perl UTF-8 is not just for Unicode. */
892
a0ed51b3
LW
893 if (off >= 0) {
894 while (off--)
895 s += UTF8SKIP(s);
896 }
897 else {
898 while (off++) {
899 s--;
8850bf83
JH
900 while (UTF8_IS_CONTINUATION(*s))
901 s--;
a0ed51b3
LW
902 }
903 }
4373e329 904 return (U8 *)s;
a0ed51b3
LW
905}
906
6940069f 907/*
fed3ba5d
NC
908=for apidoc bytes_cmp_utf8
909
910Compares the sequence of characters (stored as octets) in b, blen with the
911sequence of characters (stored as UTF-8) in u, ulen. Returns 0 if they are
912equal, -1 or -2 if the first string is less than the second string, +1 or +2
913if the first string is greater than the second string.
914
915-1 or +1 is returned if the shorter string was identical to the start of the
916longer string. -2 or +2 is returned if the was a difference between characters
917within the strings.
918
919=cut
920*/
921
922int
923Perl_bytes_cmp_utf8(pTHX_ const U8 *b, STRLEN blen, const U8 *u, STRLEN ulen)
924{
925 const U8 *const bend = b + blen;
926 const U8 *const uend = u + ulen;
927
928 PERL_ARGS_ASSERT_BYTES_CMP_UTF8;
929
930 PERL_UNUSED_CONTEXT;
931
932 while (b < bend && u < uend) {
933 U8 c = *u++;
934 if (!UTF8_IS_INVARIANT(c)) {
935 if (UTF8_IS_DOWNGRADEABLE_START(c)) {
936 if (u < uend) {
937 U8 c1 = *u++;
938 if (UTF8_IS_CONTINUATION(c1)) {
356979f4 939 c = UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(c, c1));
fed3ba5d
NC
940 } else {
941 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
942 "Malformed UTF-8 character "
943 "(unexpected non-continuation byte 0x%02x"
944 ", immediately after start byte 0x%02x)"
945 /* Dear diag.t, it's in the pod. */
946 "%s%s", c1, c,
947 PL_op ? " in " : "",
948 PL_op ? OP_DESC(PL_op) : "");
949 return -2;
950 }
951 } else {
952 if (PL_op)
953 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
954 "%s in %s", unees, OP_DESC(PL_op));
955 else
956 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8), unees);
957 return -2; /* Really want to return undef :-) */
958 }
959 } else {
960 return -2;
961 }
962 }
963 if (*b != c) {
964 return *b < c ? -2 : +2;
965 }
966 ++b;
967 }
968
969 if (b == bend && u == uend)
970 return 0;
971
972 return b < bend ? +1 : -1;
973}
974
975/*
87cea99e 976=for apidoc utf8_to_bytes
6940069f 977
2bbc8d55 978Converts a string C<s> of length C<len> from UTF-8 into native byte encoding.
246fae53
MG
979Unlike C<bytes_to_utf8>, this over-writes the original string, and
980updates len to contain the new length.
67e989fb 981Returns zero on failure, setting C<len> to -1.
6940069f 982
95be277c
NC
983If you need a copy of the string, see C<bytes_from_utf8>.
984
6940069f
GS
985=cut
986*/
987
988U8 *
37607a96 989Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
6940069f 990{
d4c19fe8
AL
991 U8 * const save = s;
992 U8 * const send = s + *len;
6940069f 993 U8 *d;
246fae53 994
7918f24d
NC
995 PERL_ARGS_ASSERT_UTF8_TO_BYTES;
996
1e54db1a 997 /* ensure valid UTF-8 and chars < 256 before updating string */
d4c19fe8 998 while (s < send) {
dcad2880
JH
999 U8 c = *s++;
1000
1d72bdf6
NIS
1001 if (!UTF8_IS_INVARIANT(c) &&
1002 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
1003 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
10edeb5d 1004 *len = ((STRLEN) -1);
dcad2880
JH
1005 return 0;
1006 }
246fae53 1007 }
dcad2880
JH
1008
1009 d = s = save;
6940069f 1010 while (s < send) {
ed646e6e 1011 STRLEN ulen;
9041c2e3 1012 *d++ = (U8)utf8_to_uvchr(s, &ulen);
ed646e6e 1013 s += ulen;
6940069f
GS
1014 }
1015 *d = '\0';
246fae53 1016 *len = d - save;
6940069f
GS
1017 return save;
1018}
1019
1020/*
87cea99e 1021=for apidoc bytes_from_utf8
f9a63242 1022
2bbc8d55 1023Converts a string C<s> of length C<len> from UTF-8 into native byte encoding.
35a4481c 1024Unlike C<utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
ef9edfd0
JH
1025the newly-created string, and updates C<len> to contain the new
1026length. Returns the original string if no conversion occurs, C<len>
1027is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
2bbc8d55
SP
10280 if C<s> is converted or consisted entirely of characters that are invariant
1029in utf8 (i.e., US-ASCII on non-EBCDIC machines).
f9a63242 1030
37607a96
PK
1031=cut
1032*/
f9a63242
JH
1033
1034U8 *
e1ec3a88 1035Perl_bytes_from_utf8(pTHX_ const U8 *s, STRLEN *len, bool *is_utf8)
f9a63242 1036{
f9a63242 1037 U8 *d;
e1ec3a88
AL
1038 const U8 *start = s;
1039 const U8 *send;
f9a63242
JH
1040 I32 count = 0;
1041
7918f24d
NC
1042 PERL_ARGS_ASSERT_BYTES_FROM_UTF8;
1043
96a5add6 1044 PERL_UNUSED_CONTEXT;
f9a63242 1045 if (!*is_utf8)
73d840c0 1046 return (U8 *)start;
f9a63242 1047
1e54db1a 1048 /* ensure valid UTF-8 and chars < 256 before converting string */
f9a63242 1049 for (send = s + *len; s < send;) {
e1ec3a88 1050 U8 c = *s++;
1d72bdf6 1051 if (!UTF8_IS_INVARIANT(c)) {
db42d148
NIS
1052 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
1053 (c = *s++) && UTF8_IS_CONTINUATION(c))
1054 count++;
1055 else
73d840c0 1056 return (U8 *)start;
db42d148 1057 }
f9a63242
JH
1058 }
1059
35da51f7 1060 *is_utf8 = FALSE;
f9a63242 1061
212542aa 1062 Newx(d, (*len) - count + 1, U8);
ef9edfd0 1063 s = start; start = d;
f9a63242
JH
1064 while (s < send) {
1065 U8 c = *s++;
c4d5f83a
NIS
1066 if (!UTF8_IS_INVARIANT(c)) {
1067 /* Then it is two-byte encoded */
356979f4 1068 c = UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(c, *s++));
c4d5f83a
NIS
1069 }
1070 *d++ = c;
f9a63242
JH
1071 }
1072 *d = '\0';
1073 *len = d - start;
73d840c0 1074 return (U8 *)start;
f9a63242
JH
1075}
1076
1077/*
87cea99e 1078=for apidoc bytes_to_utf8
6940069f 1079
ff97e5cf
KW
1080Converts a string C<s> of length C<len> bytes from the native encoding into
1081UTF-8.
6662521e 1082Returns a pointer to the newly-created string, and sets C<len> to
ff97e5cf 1083reflect the new length in bytes.
6940069f 1084
2bbc8d55
SP
1085A NUL character will be written after the end of the string.
1086
1087If you want to convert to UTF-8 from encodings other than
1088the native (Latin1 or EBCDIC),
c9ada85f
JH
1089see sv_recode_to_utf8().
1090
497711e7 1091=cut
6940069f
GS
1092*/
1093
1094U8*
35a4481c 1095Perl_bytes_to_utf8(pTHX_ const U8 *s, STRLEN *len)
6940069f 1096{
35a4481c 1097 const U8 * const send = s + (*len);
6940069f
GS
1098 U8 *d;
1099 U8 *dst;
7918f24d
NC
1100
1101 PERL_ARGS_ASSERT_BYTES_TO_UTF8;
96a5add6 1102 PERL_UNUSED_CONTEXT;
6940069f 1103
212542aa 1104 Newx(d, (*len) * 2 + 1, U8);
6940069f
GS
1105 dst = d;
1106
1107 while (s < send) {
35a4481c 1108 const UV uv = NATIVE_TO_ASCII(*s++);
c4d5f83a 1109 if (UNI_IS_INVARIANT(uv))
eb160463 1110 *d++ = (U8)UTF_TO_NATIVE(uv);
6940069f 1111 else {
eb160463
GS
1112 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
1113 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
6940069f
GS
1114 }
1115 }
1116 *d = '\0';
6662521e 1117 *len = d-dst;
6940069f
GS
1118 return dst;
1119}
1120
a0ed51b3 1121/*
dea0fc0b 1122 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
a0ed51b3
LW
1123 *
1124 * Destination must be pre-extended to 3/2 source. Do not use in-place.
1125 * We optimize for native, for obvious reasons. */
1126
1127U8*
dea0fc0b 1128Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
a0ed51b3 1129{
dea0fc0b
JH
1130 U8* pend;
1131 U8* dstart = d;
1132
7918f24d
NC
1133 PERL_ARGS_ASSERT_UTF16_TO_UTF8;
1134
dea0fc0b 1135 if (bytelen & 1)
f5992bc4 1136 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %"UVuf, (UV)bytelen);
dea0fc0b
JH
1137
1138 pend = p + bytelen;
1139
a0ed51b3 1140 while (p < pend) {
dea0fc0b
JH
1141 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
1142 p += 2;
a0ed51b3 1143 if (uv < 0x80) {
e294cc5d
JH
1144#ifdef EBCDIC
1145 *d++ = UNI_TO_NATIVE(uv);
1146#else
eb160463 1147 *d++ = (U8)uv;
e294cc5d 1148#endif
a0ed51b3
LW
1149 continue;
1150 }
1151 if (uv < 0x800) {
eb160463
GS
1152 *d++ = (U8)(( uv >> 6) | 0xc0);
1153 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
1154 continue;
1155 }
52b9aa85 1156 if (uv >= 0xd800 && uv <= 0xdbff) { /* surrogates */
01ea242b 1157 if (p >= pend) {
dea0fc0b 1158 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
01ea242b
NC
1159 } else {
1160 UV low = (p[0] << 8) + p[1];
1161 p += 2;
52b9aa85 1162 if (low < 0xdc00 || low > 0xdfff)
01ea242b
NC
1163 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
1164 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
1165 }
dbde1951
NC
1166 } else if (uv >= 0xdc00 && uv <= 0xdfff) {
1167 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
a0ed51b3
LW
1168 }
1169 if (uv < 0x10000) {
eb160463
GS
1170 *d++ = (U8)(( uv >> 12) | 0xe0);
1171 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
1172 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
1173 continue;
1174 }
1175 else {
eb160463
GS
1176 *d++ = (U8)(( uv >> 18) | 0xf0);
1177 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
1178 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
1179 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
1180 continue;
1181 }
1182 }
dea0fc0b 1183 *newlen = d - dstart;
a0ed51b3
LW
1184 return d;
1185}
1186
1187/* Note: this one is slightly destructive of the source. */
1188
1189U8*
dea0fc0b 1190Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
a0ed51b3
LW
1191{
1192 U8* s = (U8*)p;
d4c19fe8 1193 U8* const send = s + bytelen;
7918f24d
NC
1194
1195 PERL_ARGS_ASSERT_UTF16_TO_UTF8_REVERSED;
1196
e0ea5e2d
NC
1197 if (bytelen & 1)
1198 Perl_croak(aTHX_ "panic: utf16_to_utf8_reversed: odd bytelen %"UVuf,
1199 (UV)bytelen);
1200
a0ed51b3 1201 while (s < send) {
d4c19fe8 1202 const U8 tmp = s[0];
a0ed51b3
LW
1203 s[0] = s[1];
1204 s[1] = tmp;
1205 s += 2;
1206 }
dea0fc0b 1207 return utf16_to_utf8(p, d, bytelen, newlen);
a0ed51b3
LW
1208}
1209
1210/* for now these are all defined (inefficiently) in terms of the utf8 versions */
1211
1212bool
84afefe6 1213Perl_is_uni_alnum(pTHX_ UV c)
a0ed51b3 1214{
89ebb4a3 1215 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1216 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1217 return is_utf8_alnum(tmpbuf);
1218}
1219
1220bool
84afefe6 1221Perl_is_uni_idfirst(pTHX_ UV c)
a0ed51b3 1222{
89ebb4a3 1223 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1224 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1225 return is_utf8_idfirst(tmpbuf);
1226}
1227
1228bool
84afefe6 1229Perl_is_uni_alpha(pTHX_ UV c)
a0ed51b3 1230{
89ebb4a3 1231 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1232 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1233 return is_utf8_alpha(tmpbuf);
1234}
1235
1236bool
84afefe6 1237Perl_is_uni_ascii(pTHX_ UV c)
4d61ec05 1238{
89ebb4a3 1239 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1240 uvchr_to_utf8(tmpbuf, c);
4d61ec05
GS
1241 return is_utf8_ascii(tmpbuf);
1242}
1243
1244bool
84afefe6 1245Perl_is_uni_space(pTHX_ UV c)
a0ed51b3 1246{
89ebb4a3 1247 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1248 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1249 return is_utf8_space(tmpbuf);
1250}
1251
1252bool
84afefe6 1253Perl_is_uni_digit(pTHX_ UV c)
a0ed51b3 1254{
89ebb4a3 1255 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1256 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1257 return is_utf8_digit(tmpbuf);
1258}
1259
1260bool
84afefe6 1261Perl_is_uni_upper(pTHX_ UV c)
a0ed51b3 1262{
89ebb4a3 1263 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1264 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1265 return is_utf8_upper(tmpbuf);
1266}
1267
1268bool
84afefe6 1269Perl_is_uni_lower(pTHX_ UV c)
a0ed51b3 1270{
89ebb4a3 1271 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1272 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1273 return is_utf8_lower(tmpbuf);
1274}
1275
1276bool
84afefe6 1277Perl_is_uni_cntrl(pTHX_ UV c)
b8c5462f 1278{
89ebb4a3 1279 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1280 uvchr_to_utf8(tmpbuf, c);
b8c5462f
JH
1281 return is_utf8_cntrl(tmpbuf);
1282}
1283
1284bool
84afefe6 1285Perl_is_uni_graph(pTHX_ UV c)
b8c5462f 1286{
89ebb4a3 1287 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1288 uvchr_to_utf8(tmpbuf, c);
b8c5462f
JH
1289 return is_utf8_graph(tmpbuf);
1290}
1291
1292bool
84afefe6 1293Perl_is_uni_print(pTHX_ UV c)
a0ed51b3 1294{
89ebb4a3 1295 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1296 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1297 return is_utf8_print(tmpbuf);
1298}
1299
b8c5462f 1300bool
84afefe6 1301Perl_is_uni_punct(pTHX_ UV c)
b8c5462f 1302{
89ebb4a3 1303 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1304 uvchr_to_utf8(tmpbuf, c);
b8c5462f
JH
1305 return is_utf8_punct(tmpbuf);
1306}
1307
4d61ec05 1308bool
84afefe6 1309Perl_is_uni_xdigit(pTHX_ UV c)
4d61ec05 1310{
89ebb4a3 1311 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
230880c1 1312 uvchr_to_utf8(tmpbuf, c);
4d61ec05
GS
1313 return is_utf8_xdigit(tmpbuf);
1314}
1315
84afefe6
JH
1316UV
1317Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 1318{
7918f24d
NC
1319 PERL_ARGS_ASSERT_TO_UNI_UPPER;
1320
0ebc6274
JH
1321 uvchr_to_utf8(p, c);
1322 return to_utf8_upper(p, p, lenp);
a0ed51b3
LW
1323}
1324
84afefe6
JH
1325UV
1326Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 1327{
7918f24d
NC
1328 PERL_ARGS_ASSERT_TO_UNI_TITLE;
1329
0ebc6274
JH
1330 uvchr_to_utf8(p, c);
1331 return to_utf8_title(p, p, lenp);
a0ed51b3
LW
1332}
1333
84afefe6
JH
1334UV
1335Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 1336{
7918f24d
NC
1337 PERL_ARGS_ASSERT_TO_UNI_LOWER;
1338
0ebc6274
JH
1339 uvchr_to_utf8(p, c);
1340 return to_utf8_lower(p, p, lenp);
a0ed51b3
LW
1341}
1342
84afefe6
JH
1343UV
1344Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1345{
7918f24d
NC
1346 PERL_ARGS_ASSERT_TO_UNI_FOLD;
1347
0ebc6274
JH
1348 uvchr_to_utf8(p, c);
1349 return to_utf8_fold(p, p, lenp);
84afefe6
JH
1350}
1351
a0ed51b3
LW
1352/* for now these all assume no locale info available for Unicode > 255 */
1353
1354bool
84afefe6 1355Perl_is_uni_alnum_lc(pTHX_ UV c)
a0ed51b3
LW
1356{
1357 return is_uni_alnum(c); /* XXX no locale support yet */
1358}
1359
1360bool
84afefe6 1361Perl_is_uni_idfirst_lc(pTHX_ UV c)
a0ed51b3
LW
1362{
1363 return is_uni_idfirst(c); /* XXX no locale support yet */
1364}
1365
1366bool
84afefe6 1367Perl_is_uni_alpha_lc(pTHX_ UV c)
a0ed51b3
LW
1368{
1369 return is_uni_alpha(c); /* XXX no locale support yet */
1370}
1371
1372bool
84afefe6 1373Perl_is_uni_ascii_lc(pTHX_ UV c)
4d61ec05
GS
1374{
1375 return is_uni_ascii(c); /* XXX no locale support yet */
1376}
1377
1378bool
84afefe6 1379Perl_is_uni_space_lc(pTHX_ UV c)
a0ed51b3
LW
1380{
1381 return is_uni_space(c); /* XXX no locale support yet */
1382}
1383
1384bool
84afefe6 1385Perl_is_uni_digit_lc(pTHX_ UV c)
a0ed51b3
LW
1386{
1387 return is_uni_digit(c); /* XXX no locale support yet */
1388}
1389
1390bool
84afefe6 1391Perl_is_uni_upper_lc(pTHX_ UV c)
a0ed51b3
LW
1392{
1393 return is_uni_upper(c); /* XXX no locale support yet */
1394}
1395
1396bool
84afefe6 1397Perl_is_uni_lower_lc(pTHX_ UV c)
a0ed51b3
LW
1398{
1399 return is_uni_lower(c); /* XXX no locale support yet */
1400}
1401
1402bool
84afefe6 1403Perl_is_uni_cntrl_lc(pTHX_ UV c)
b8c5462f
JH
1404{
1405 return is_uni_cntrl(c); /* XXX no locale support yet */
1406}
1407
1408bool
84afefe6 1409Perl_is_uni_graph_lc(pTHX_ UV c)
b8c5462f
JH
1410{
1411 return is_uni_graph(c); /* XXX no locale support yet */
1412}
1413
1414bool
84afefe6 1415Perl_is_uni_print_lc(pTHX_ UV c)
a0ed51b3
LW
1416{
1417 return is_uni_print(c); /* XXX no locale support yet */
1418}
1419
b8c5462f 1420bool
84afefe6 1421Perl_is_uni_punct_lc(pTHX_ UV c)
b8c5462f
JH
1422{
1423 return is_uni_punct(c); /* XXX no locale support yet */
1424}
1425
4d61ec05 1426bool
84afefe6 1427Perl_is_uni_xdigit_lc(pTHX_ UV c)
4d61ec05
GS
1428{
1429 return is_uni_xdigit(c); /* XXX no locale support yet */
1430}
1431
b7ac61fa
JH
1432U32
1433Perl_to_uni_upper_lc(pTHX_ U32 c)
1434{
ee099d14
JH
1435 /* XXX returns only the first character -- do not use XXX */
1436 /* XXX no locale support yet */
1437 STRLEN len;
89ebb4a3 1438 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
ee099d14 1439 return (U32)to_uni_upper(c, tmpbuf, &len);
b7ac61fa
JH
1440}
1441
1442U32
1443Perl_to_uni_title_lc(pTHX_ U32 c)
1444{
ee099d14
JH
1445 /* XXX returns only the first character XXX -- do not use XXX */
1446 /* XXX no locale support yet */
1447 STRLEN len;
89ebb4a3 1448 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
ee099d14 1449 return (U32)to_uni_title(c, tmpbuf, &len);
b7ac61fa
JH
1450}
1451
1452U32
1453Perl_to_uni_lower_lc(pTHX_ U32 c)
1454{
ee099d14
JH
1455 /* XXX returns only the first character -- do not use XXX */
1456 /* XXX no locale support yet */
1457 STRLEN len;
89ebb4a3 1458 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
ee099d14 1459 return (U32)to_uni_lower(c, tmpbuf, &len);
b7ac61fa
JH
1460}
1461
7452cf6a 1462static bool
5141f98e 1463S_is_utf8_common(pTHX_ const U8 *const p, SV **swash,
bde6a22d
NC
1464 const char *const swashname)
1465{
97aff369 1466 dVAR;
7918f24d
NC
1467
1468 PERL_ARGS_ASSERT_IS_UTF8_COMMON;
1469
bde6a22d
NC
1470 if (!is_utf8_char(p))
1471 return FALSE;
1472 if (!*swash)
711a919c 1473 *swash = swash_init("utf8", swashname, &PL_sv_undef, 1, 0);
bde6a22d
NC
1474 return swash_fetch(*swash, p, TRUE) != 0;
1475}
1476
1477bool
7fc63493 1478Perl_is_utf8_alnum(pTHX_ const U8 *p)
a0ed51b3 1479{
97aff369 1480 dVAR;
7918f24d
NC
1481
1482 PERL_ARGS_ASSERT_IS_UTF8_ALNUM;
1483
671c33bf
NC
1484 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1485 * descendant of isalnum(3), in other words, it doesn't
1486 * contain the '_'. --jhi */
d4c19fe8 1487 return is_utf8_common(p, &PL_utf8_alnum, "IsWord");
a0ed51b3
LW
1488}
1489
1490bool
7fc63493 1491Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
a0ed51b3 1492{
97aff369 1493 dVAR;
7918f24d
NC
1494
1495 PERL_ARGS_ASSERT_IS_UTF8_IDFIRST;
1496
82686b01
JH
1497 if (*p == '_')
1498 return TRUE;
bde6a22d 1499 /* is_utf8_idstart would be more logical. */
d4c19fe8 1500 return is_utf8_common(p, &PL_utf8_idstart, "IdStart");
82686b01
JH
1501}
1502
1503bool
7fc63493 1504Perl_is_utf8_idcont(pTHX_ const U8 *p)
82686b01 1505{
97aff369 1506 dVAR;
7918f24d
NC
1507
1508 PERL_ARGS_ASSERT_IS_UTF8_IDCONT;
1509
82686b01
JH
1510 if (*p == '_')
1511 return TRUE;
d4c19fe8 1512 return is_utf8_common(p, &PL_utf8_idcont, "IdContinue");
a0ed51b3
LW
1513}
1514
1515bool
7fc63493 1516Perl_is_utf8_alpha(pTHX_ const U8 *p)
a0ed51b3 1517{
97aff369 1518 dVAR;
7918f24d
NC
1519
1520 PERL_ARGS_ASSERT_IS_UTF8_ALPHA;
1521
d4c19fe8 1522 return is_utf8_common(p, &PL_utf8_alpha, "IsAlpha");
a0ed51b3
LW
1523}
1524
1525bool
7fc63493 1526Perl_is_utf8_ascii(pTHX_ const U8 *p)
b8c5462f 1527{
97aff369 1528 dVAR;
7918f24d
NC
1529
1530 PERL_ARGS_ASSERT_IS_UTF8_ASCII;
1531
d4c19fe8 1532 return is_utf8_common(p, &PL_utf8_ascii, "IsAscii");
b8c5462f
JH
1533}
1534
1535bool
7fc63493 1536Perl_is_utf8_space(pTHX_ const U8 *p)
a0ed51b3 1537{
97aff369 1538 dVAR;
7918f24d
NC
1539
1540 PERL_ARGS_ASSERT_IS_UTF8_SPACE;
1541
d4c19fe8 1542 return is_utf8_common(p, &PL_utf8_space, "IsSpacePerl");
a0ed51b3
LW
1543}
1544
1545bool
d1eb3177
YO
1546Perl_is_utf8_perl_space(pTHX_ const U8 *p)
1547{
1548 dVAR;
1549
1550 PERL_ARGS_ASSERT_IS_UTF8_PERL_SPACE;
1551
1552 return is_utf8_common(p, &PL_utf8_perl_space, "IsPerlSpace");
1553}
1554
1555bool
1556Perl_is_utf8_perl_word(pTHX_ const U8 *p)
1557{
1558 dVAR;
1559
1560 PERL_ARGS_ASSERT_IS_UTF8_PERL_WORD;
1561
1562 return is_utf8_common(p, &PL_utf8_perl_word, "IsPerlWord");
1563}
1564
1565bool
7fc63493 1566Perl_is_utf8_digit(pTHX_ const U8 *p)
a0ed51b3 1567{
97aff369 1568 dVAR;
7918f24d
NC
1569
1570 PERL_ARGS_ASSERT_IS_UTF8_DIGIT;
1571
d4c19fe8 1572 return is_utf8_common(p, &PL_utf8_digit, "IsDigit");
a0ed51b3
LW
1573}
1574
1575bool
d1eb3177
YO
1576Perl_is_utf8_posix_digit(pTHX_ const U8 *p)
1577{
1578 dVAR;
1579
1580 PERL_ARGS_ASSERT_IS_UTF8_POSIX_DIGIT;
1581
1582 return is_utf8_common(p, &PL_utf8_posix_digit, "IsPosixDigit");
1583}
1584
1585bool
7fc63493 1586Perl_is_utf8_upper(pTHX_ const U8 *p)
a0ed51b3 1587{
97aff369 1588 dVAR;
7918f24d
NC
1589
1590 PERL_ARGS_ASSERT_IS_UTF8_UPPER;
1591
d4c19fe8 1592 return is_utf8_common(p, &PL_utf8_upper, "IsUppercase");
a0ed51b3
LW
1593}
1594
1595bool
7fc63493 1596Perl_is_utf8_lower(pTHX_ const U8 *p)
a0ed51b3 1597{
97aff369 1598 dVAR;
7918f24d
NC
1599
1600 PERL_ARGS_ASSERT_IS_UTF8_LOWER;
1601
d4c19fe8 1602 return is_utf8_common(p, &PL_utf8_lower, "IsLowercase");
a0ed51b3
LW
1603}
1604
1605bool
7fc63493 1606Perl_is_utf8_cntrl(pTHX_ const U8 *p)
b8c5462f 1607{
97aff369 1608 dVAR;
7918f24d
NC
1609
1610 PERL_ARGS_ASSERT_IS_UTF8_CNTRL;
1611
d4c19fe8 1612 return is_utf8_common(p, &PL_utf8_cntrl, "IsCntrl");
b8c5462f
JH
1613}
1614
1615bool
7fc63493 1616Perl_is_utf8_graph(pTHX_ const U8 *p)
b8c5462f 1617{
97aff369 1618 dVAR;
7918f24d
NC
1619
1620 PERL_ARGS_ASSERT_IS_UTF8_GRAPH;
1621
d4c19fe8 1622 return is_utf8_common(p, &PL_utf8_graph, "IsGraph");
b8c5462f
JH
1623}
1624
1625bool
7fc63493 1626Perl_is_utf8_print(pTHX_ const U8 *p)
a0ed51b3 1627{
97aff369 1628 dVAR;
7918f24d
NC
1629
1630 PERL_ARGS_ASSERT_IS_UTF8_PRINT;
1631
d4c19fe8 1632 return is_utf8_common(p, &PL_utf8_print, "IsPrint");
a0ed51b3
LW
1633}
1634
1635bool
7fc63493 1636Perl_is_utf8_punct(pTHX_ const U8 *p)
b8c5462f 1637{
97aff369 1638 dVAR;
7918f24d
NC
1639
1640 PERL_ARGS_ASSERT_IS_UTF8_PUNCT;
1641
d4c19fe8 1642 return is_utf8_common(p, &PL_utf8_punct, "IsPunct");
b8c5462f
JH
1643}
1644
1645bool
7fc63493 1646Perl_is_utf8_xdigit(pTHX_ const U8 *p)
b8c5462f 1647{
97aff369 1648 dVAR;
7918f24d
NC
1649
1650 PERL_ARGS_ASSERT_IS_UTF8_XDIGIT;
1651
d1eb3177 1652 return is_utf8_common(p, &PL_utf8_xdigit, "IsXDigit");
b8c5462f
JH
1653}
1654
1655bool
7fc63493 1656Perl_is_utf8_mark(pTHX_ const U8 *p)
a0ed51b3 1657{
97aff369 1658 dVAR;
7918f24d
NC
1659
1660 PERL_ARGS_ASSERT_IS_UTF8_MARK;
1661
d4c19fe8 1662 return is_utf8_common(p, &PL_utf8_mark, "IsM");
a0ed51b3
LW
1663}
1664
37e2e78e
KW
1665bool
1666Perl_is_utf8_X_begin(pTHX_ const U8 *p)
1667{
1668 dVAR;
1669
1670 PERL_ARGS_ASSERT_IS_UTF8_X_BEGIN;
1671
1672 return is_utf8_common(p, &PL_utf8_X_begin, "_X_Begin");
1673}
1674
1675bool
1676Perl_is_utf8_X_extend(pTHX_ const U8 *p)
1677{
1678 dVAR;
1679
1680 PERL_ARGS_ASSERT_IS_UTF8_X_EXTEND;
1681
1682 return is_utf8_common(p, &PL_utf8_X_extend, "_X_Extend");
1683}
1684
1685bool
1686Perl_is_utf8_X_prepend(pTHX_ const U8 *p)
1687{
1688 dVAR;
1689
1690 PERL_ARGS_ASSERT_IS_UTF8_X_PREPEND;
1691
1692 return is_utf8_common(p, &PL_utf8_X_prepend, "GCB=Prepend");
1693}
1694
1695bool
1696Perl_is_utf8_X_non_hangul(pTHX_ const U8 *p)
1697{
1698 dVAR;
1699
1700 PERL_ARGS_ASSERT_IS_UTF8_X_NON_HANGUL;
1701
1702 return is_utf8_common(p, &PL_utf8_X_non_hangul, "HST=Not_Applicable");
1703}
1704
1705bool
1706Perl_is_utf8_X_L(pTHX_ const U8 *p)
1707{
1708 dVAR;
1709
1710 PERL_ARGS_ASSERT_IS_UTF8_X_L;
1711
1712 return is_utf8_common(p, &PL_utf8_X_L, "GCB=L");
1713}
1714
1715bool
1716Perl_is_utf8_X_LV(pTHX_ const U8 *p)
1717{
1718 dVAR;
1719
1720 PERL_ARGS_ASSERT_IS_UTF8_X_LV;
1721
1722 return is_utf8_common(p, &PL_utf8_X_LV, "GCB=LV");
1723}
1724
1725bool
1726Perl_is_utf8_X_LVT(pTHX_ const U8 *p)
1727{
1728 dVAR;
1729
1730 PERL_ARGS_ASSERT_IS_UTF8_X_LVT;
1731
1732 return is_utf8_common(p, &PL_utf8_X_LVT, "GCB=LVT");
1733}
1734
1735bool
1736Perl_is_utf8_X_T(pTHX_ const U8 *p)
1737{
1738 dVAR;
1739
1740 PERL_ARGS_ASSERT_IS_UTF8_X_T;
1741
1742 return is_utf8_common(p, &PL_utf8_X_T, "GCB=T");
1743}
1744
1745bool
1746Perl_is_utf8_X_V(pTHX_ const U8 *p)
1747{
1748 dVAR;
1749
1750 PERL_ARGS_ASSERT_IS_UTF8_X_V;
1751
1752 return is_utf8_common(p, &PL_utf8_X_V, "GCB=V");
1753}
1754
1755bool
1756Perl_is_utf8_X_LV_LVT_V(pTHX_ const U8 *p)
1757{
1758 dVAR;
1759
1760 PERL_ARGS_ASSERT_IS_UTF8_X_LV_LVT_V;
1761
1762 return is_utf8_common(p, &PL_utf8_X_LV_LVT_V, "_X_LV_LVT_V");
1763}
1764
6b5c0936 1765/*
87cea99e 1766=for apidoc to_utf8_case
6b5c0936
JH
1767
1768The "p" contains the pointer to the UTF-8 string encoding
1769the character that is being converted.
1770
1771The "ustrp" is a pointer to the character buffer to put the
1772conversion result to. The "lenp" is a pointer to the length
1773of the result.
1774
0134edef 1775The "swashp" is a pointer to the swash to use.
6b5c0936 1776
0134edef 1777Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
8fe4d5b2 1778and loaded by SWASHNEW, using lib/utf8_heavy.pl. The special (usually,
0134edef 1779but not always, a multicharacter mapping), is tried first.
6b5c0936 1780
0134edef
JH
1781The "special" is a string like "utf8::ToSpecLower", which means the
1782hash %utf8::ToSpecLower. The access to the hash is through
1783Perl_to_utf8_case().
6b5c0936 1784
0134edef
JH
1785The "normal" is a string like "ToLower" which means the swash
1786%utf8::ToLower.
1787
1788=cut */
6b5c0936 1789
2104c8d9 1790UV
9a957fbc
AL
1791Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp,
1792 SV **swashp, const char *normal, const char *special)
a0ed51b3 1793{
97aff369 1794 dVAR;
89ebb4a3 1795 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
0134edef 1796 STRLEN len = 0;
aec46f14 1797 const UV uv0 = utf8_to_uvchr(p, NULL);
1feea2c7
JH
1798 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1799 * are necessary in EBCDIC, they are redundant no-ops
1800 * in ASCII-ish platforms, and hopefully optimized away. */
f54cb97a 1801 const UV uv1 = NATIVE_TO_UNI(uv0);
7918f24d
NC
1802
1803 PERL_ARGS_ASSERT_TO_UTF8_CASE;
1804
9ae3ac1a
KW
1805 /* Note that swash_fetch() doesn't output warnings for these because it
1806 * assumes we will */
1807 if (uv1 >= UNICODE_SURROGATE_FIRST && ckWARN_d(WARN_UTF8)) {
1808 if (uv1 <= UNICODE_SURROGATE_LAST) {
1809 const char* desc = (PL_op) ? OP_DESC(PL_op) : normal;
1810 Perl_warner(aTHX_ packWARN(WARN_UTF8),
1811 "Operation \"%s\" returns its argument for UTF-16 surrogate U+%04"UVXf"", desc, uv1);
1812 }
1813 else if (UNICODE_IS_SUPER(uv1)) {
1814 const char* desc = (PL_op) ? OP_DESC(PL_op) : normal;
1815 Perl_warner(aTHX_ packWARN(WARN_UTF8),
1816 "Operation \"%s\" returns its argument for non-Unicode code point 0x%04"UVXf"", desc, uv1);
1817 }
1818
1819 /* Note that non-characters are perfectly legal, so no warning should
1820 * be given */
1821 }
1822
1feea2c7 1823 uvuni_to_utf8(tmpbuf, uv1);
0134edef
JH
1824
1825 if (!*swashp) /* load on-demand */
1826 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
37e2e78e
KW
1827 /* This is the beginnings of a skeleton of code to read the info section
1828 * that is in all the swashes in case we ever want to do that, so one can
1829 * read things whose maps aren't code points, and whose default if missing
1830 * is not to the code point itself. This was just to see if it actually
1831 * worked. Details on what the possibilities are are in perluniprops.pod
1832 HV * const hv = get_hv("utf8::SwashInfo", 0);
1833 if (hv) {
1834 SV **svp;
1835 svp = hv_fetch(hv, (const char*)normal, strlen(normal), FALSE);
1836 const char *s;
1837
1838 HV * const this_hash = SvRV(*svp);
1839 svp = hv_fetch(this_hash, "type", strlen("type"), FALSE);
1840 s = SvPV_const(*svp, len);
1841 }
1842 }*/
0134edef 1843
a6f87d8c 1844 if (special) {
0134edef 1845 /* It might be "special" (sometimes, but not always,
2a37f04d 1846 * a multicharacter mapping) */
6673a63c 1847 HV * const hv = get_hv(special, 0);
b08cf34e
JH
1848 SV **svp;
1849
35da51f7 1850 if (hv &&
b08cf34e
JH
1851 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1852 (*svp)) {
cfd0369c 1853 const char *s;
47654450 1854
cfd0369c 1855 s = SvPV_const(*svp, len);
47654450
JH
1856 if (len == 1)
1857 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
2a37f04d 1858 else {
2f9475ad
JH
1859#ifdef EBCDIC
1860 /* If we have EBCDIC we need to remap the characters
1861 * since any characters in the low 256 are Unicode
1862 * code points, not EBCDIC. */
7cda7a3d 1863 U8 *t = (U8*)s, *tend = t + len, *d;
2f9475ad
JH
1864
1865 d = tmpbuf;
b08cf34e 1866 if (SvUTF8(*svp)) {
2f9475ad
JH
1867 STRLEN tlen = 0;
1868
1869 while (t < tend) {
d4c19fe8 1870 const UV c = utf8_to_uvchr(t, &tlen);
2f9475ad
JH
1871 if (tlen > 0) {
1872 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1873 t += tlen;
1874 }
1875 else
1876 break;
1877 }
1878 }
1879 else {
36fec512
JH
1880 while (t < tend) {
1881 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1882 t++;
1883 }
2f9475ad
JH
1884 }
1885 len = d - tmpbuf;
1886 Copy(tmpbuf, ustrp, len, U8);
1887#else
d2dcd0fb 1888 Copy(s, ustrp, len, U8);
2f9475ad 1889#endif
29e98929 1890 }
983ffd37 1891 }
0134edef
JH
1892 }
1893
1894 if (!len && *swashp) {
d4c19fe8
AL
1895 const UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1896
0134edef
JH
1897 if (uv2) {
1898 /* It was "normal" (a single character mapping). */
d4c19fe8 1899 const UV uv3 = UNI_TO_NATIVE(uv2);
e9101d72 1900 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
2a37f04d
JH
1901 }
1902 }
1feea2c7 1903
37e2e78e
KW
1904 if (!len) /* Neither: just copy. In other words, there was no mapping
1905 defined, which means that the code point maps to itself */
0134edef
JH
1906 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1907
2a37f04d
JH
1908 if (lenp)
1909 *lenp = len;
1910
0134edef 1911 return len ? utf8_to_uvchr(ustrp, 0) : 0;
a0ed51b3
LW
1912}
1913
d3e79532 1914/*
87cea99e 1915=for apidoc to_utf8_upper
d3e79532
JH
1916
1917Convert the UTF-8 encoded character at p to its uppercase version and
1918store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
1919that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1920the uppercase version may be longer than the original character.
d3e79532
JH
1921
1922The first character of the uppercased version is returned
1923(but note, as explained above, that there may be more.)
1924
1925=cut */
1926
2104c8d9 1927UV
7fc63493 1928Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
a0ed51b3 1929{
97aff369 1930 dVAR;
7918f24d
NC
1931
1932 PERL_ARGS_ASSERT_TO_UTF8_UPPER;
1933
983ffd37 1934 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
b4e400f9 1935 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
983ffd37 1936}
a0ed51b3 1937
d3e79532 1938/*
87cea99e 1939=for apidoc to_utf8_title
d3e79532
JH
1940
1941Convert the UTF-8 encoded character at p to its titlecase version and
1942store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
1943that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1944titlecase version may be longer than the original character.
d3e79532
JH
1945
1946The first character of the titlecased version is returned
1947(but note, as explained above, that there may be more.)
1948
1949=cut */
1950
983ffd37 1951UV
7fc63493 1952Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
983ffd37 1953{
97aff369 1954 dVAR;
7918f24d
NC
1955
1956 PERL_ARGS_ASSERT_TO_UTF8_TITLE;
1957
983ffd37 1958 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
b4e400f9 1959 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
a0ed51b3
LW
1960}
1961
d3e79532 1962/*
87cea99e 1963=for apidoc to_utf8_lower
d3e79532
JH
1964
1965Convert the UTF-8 encoded character at p to its lowercase version and
1966store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
1967that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1968lowercase version may be longer than the original character.
d3e79532
JH
1969
1970The first character of the lowercased version is returned
1971(but note, as explained above, that there may be more.)
1972
1973=cut */
1974
2104c8d9 1975UV
7fc63493 1976Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
a0ed51b3 1977{
97aff369 1978 dVAR;
7918f24d
NC
1979
1980 PERL_ARGS_ASSERT_TO_UTF8_LOWER;
1981
983ffd37 1982 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
b4e400f9
JH
1983 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1984}
1985
d3e79532 1986/*
87cea99e 1987=for apidoc to_utf8_fold
d3e79532
JH
1988
1989Convert the UTF-8 encoded character at p to its foldcase version and
1990store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3 1991that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
d3e79532
JH
1992foldcase version may be longer than the original character (up to
1993three characters).
1994
1995The first character of the foldcased version is returned
1996(but note, as explained above, that there may be more.)
1997
1998=cut */
1999
b4e400f9 2000UV
7fc63493 2001Perl_to_utf8_fold(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
b4e400f9 2002{
97aff369 2003 dVAR;
7918f24d
NC
2004
2005 PERL_ARGS_ASSERT_TO_UTF8_FOLD;
2006
b4e400f9
JH
2007 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
2008 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
a0ed51b3
LW
2009}
2010
711a919c
ST
2011/* Note:
2012 * A "swash" is a swatch hash.
2013 * A "swatch" is a bit vector generated by utf8.c:S_swash_get().
2014 * C<pkg> is a pointer to a package name for SWASHNEW, should be "utf8".
2015 * For other parameters, see utf8::SWASHNEW in lib/utf8_heavy.pl.
2016 */
a0ed51b3 2017SV*
7fc63493 2018Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
a0ed51b3 2019{
27da23d5 2020 dVAR;
a0ed51b3 2021 SV* retval;
8e84507e 2022 dSP;
7fc63493
AL
2023 const size_t pkg_len = strlen(pkg);
2024 const size_t name_len = strlen(name);
da51bb9b 2025 HV * const stash = gv_stashpvn(pkg, pkg_len, 0);
f8be5cf0 2026 SV* errsv_save;
88c57d91 2027 GV *method;
ce3b816e 2028
7918f24d
NC
2029 PERL_ARGS_ASSERT_SWASH_INIT;
2030
96ca9f55
DM
2031 PUSHSTACKi(PERLSI_MAGIC);
2032 ENTER;
ec34a119 2033 SAVEHINTS();
96ca9f55 2034 save_re_context();
88c57d91
NC
2035 method = gv_fetchmeth(stash, "SWASHNEW", 8, -1);
2036 if (!method) { /* demand load utf8 */
ce3b816e 2037 ENTER;
f8be5cf0 2038 errsv_save = newSVsv(ERRSV);
dc0c6abb
NC
2039 /* It is assumed that callers of this routine are not passing in any
2040 user derived data. */
2041 /* Need to do this after save_re_context() as it will set PL_tainted to
2042 1 while saving $1 etc (see the code after getrx: in Perl_magic_get).
2043 Even line to create errsv_save can turn on PL_tainted. */
2044 SAVEBOOL(PL_tainted);
2045 PL_tainted = 0;
71bed85a 2046 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
a0714e2c 2047 NULL);
f8be5cf0
JH
2048 if (!SvTRUE(ERRSV))
2049 sv_setsv(ERRSV, errsv_save);
2050 SvREFCNT_dec(errsv_save);
ce3b816e
GS
2051 LEAVE;
2052 }
2053 SPAGAIN;
a0ed51b3
LW
2054 PUSHMARK(SP);
2055 EXTEND(SP,5);
6e449a3a
MHM
2056 mPUSHp(pkg, pkg_len);
2057 mPUSHp(name, name_len);
a0ed51b3 2058 PUSHs(listsv);
6e449a3a
MHM
2059 mPUSHi(minbits);
2060 mPUSHi(none);
a0ed51b3 2061 PUTBACK;
f8be5cf0 2062 errsv_save = newSVsv(ERRSV);
88c57d91
NC
2063 /* If we already have a pointer to the method, no need to use call_method()
2064 to repeat the lookup. */
2065 if (method ? call_sv(MUTABLE_SV(method), G_SCALAR)
8b563eef 2066 : call_sv(newSVpvs_flags("SWASHNEW", SVs_TEMP), G_SCALAR | G_METHOD))
8e84507e 2067 retval = newSVsv(*PL_stack_sp--);
a0ed51b3 2068 else
e24b16f9 2069 retval = &PL_sv_undef;
f8be5cf0
JH
2070 if (!SvTRUE(ERRSV))
2071 sv_setsv(ERRSV, errsv_save);
2072 SvREFCNT_dec(errsv_save);
a0ed51b3
LW
2073 LEAVE;
2074 POPSTACK;
923e4eb5 2075 if (IN_PERL_COMPILETIME) {
623e6609 2076 CopHINTS_set(PL_curcop, PL_hints);
a0ed51b3 2077 }
bc45ce41
JH
2078 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
2079 if (SvPOK(retval))
35c1215d 2080 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
be2597df 2081 SVfARG(retval));
cea2e8a9 2082 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
bc45ce41 2083 }
a0ed51b3
LW
2084 return retval;
2085}
2086
035d37be
JH
2087
2088/* This API is wrong for special case conversions since we may need to
2089 * return several Unicode characters for a single Unicode character
2090 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
2091 * the lower-level routine, and it is similarly broken for returning
38684baa
KW
2092 * multiple values. --jhi
2093 * For those, you should use to_utf8_case() instead */
979f2922 2094/* Now SWASHGET is recasted into S_swash_get in this file. */
680c470c
ST
2095
2096/* Note:
2097 * Returns the value of property/mapping C<swash> for the first character
2098 * of the string C<ptr>. If C<do_utf8> is true, the string C<ptr> is
2099 * assumed to be in utf8. If C<do_utf8> is false, the string C<ptr> is
2100 * assumed to be in native 8-bit encoding. Caches the swatch in C<swash>.
2101 */
a0ed51b3 2102UV
680c470c 2103Perl_swash_fetch(pTHX_ SV *swash, const U8 *ptr, bool do_utf8)
a0ed51b3 2104{
27da23d5 2105 dVAR;
ef8f7699 2106 HV *const hv = MUTABLE_HV(SvRV(swash));
3568d838
JH
2107 U32 klen;
2108 U32 off;
a0ed51b3 2109 STRLEN slen;
7d85a32c 2110 STRLEN needents;
cfd0369c 2111 const U8 *tmps = NULL;
a0ed51b3 2112 U32 bit;
979f2922 2113 SV *swatch;
3568d838 2114 U8 tmputf8[2];
35da51f7 2115 const UV c = NATIVE_TO_ASCII(*ptr);
3568d838 2116
7918f24d
NC
2117 PERL_ARGS_ASSERT_SWASH_FETCH;
2118
3568d838 2119 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
979f2922
ST
2120 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
2121 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
2122 ptr = tmputf8;
3568d838
JH
2123 }
2124 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
37e2e78e 2125 * then the "swatch" is a vec() for all the chars which start
3568d838
JH
2126 * with 0xAA..0xYY
2127 * So the key in the hash (klen) is length of encoded char -1
2128 */
2129 klen = UTF8SKIP(ptr) - 1;
2130 off = ptr[klen];
a0ed51b3 2131
979f2922 2132 if (klen == 0) {
37e2e78e 2133 /* If char is invariant then swatch is for all the invariant chars
1e54db1a 2134 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
7d85a32c 2135 */
979f2922
ST
2136 needents = UTF_CONTINUATION_MARK;
2137 off = NATIVE_TO_UTF(ptr[klen]);
2138 }
2139 else {
7d85a32c 2140 /* If char is encoded then swatch is for the prefix */
979f2922
ST
2141 needents = (1 << UTF_ACCUMULATION_SHIFT);
2142 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
9ae3ac1a
KW
2143 if (UTF8_IS_SUPER(ptr) && ckWARN_d(WARN_UTF8)) {
2144 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0, 0);
2145
2146 /* This outputs warnings for binary properties only, assuming that
2147 * to_utf8_case() will output any. Also, surrogates aren't checked
2148 * for, as that would warn on things like /\p{Gc=Cs}/ */
2149 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2150 if (SvUV(*bitssvp) == 1) {
2151 Perl_warner(aTHX_ packWARN(WARN_UTF8),
2152 "Code point 0x%04"UVXf" is not Unicode, no properties match it; all inverse properties do", code_point);
2153 }
2154 }
979f2922 2155 }
7d85a32c 2156
a0ed51b3
LW
2157 /*
2158 * This single-entry cache saves about 1/3 of the utf8 overhead in test
2159 * suite. (That is, only 7-8% overall over just a hash cache. Still,
2160 * it's nothing to sniff at.) Pity we usually come through at least
2161 * two function calls to get here...
2162 *
2163 * NB: this code assumes that swatches are never modified, once generated!
2164 */
2165
3568d838 2166 if (hv == PL_last_swash_hv &&
a0ed51b3 2167 klen == PL_last_swash_klen &&
27da23d5 2168 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
a0ed51b3
LW
2169 {
2170 tmps = PL_last_swash_tmps;
2171 slen = PL_last_swash_slen;
2172 }
2173 else {
2174 /* Try our second-level swatch cache, kept in a hash. */
e1ec3a88 2175 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
a0ed51b3 2176
979f2922
ST
2177 /* If not cached, generate it via swash_get */
2178 if (!svp || !SvPOK(*svp)
2179 || !(tmps = (const U8*)SvPV_const(*svp, slen))) {
2b9d42f0
NIS
2180 /* We use utf8n_to_uvuni() as we want an index into
2181 Unicode tables, not a native character number.
2182 */
aec46f14 2183 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
872c91ae
JH
2184 ckWARN(WARN_UTF8) ?
2185 0 : UTF8_ALLOW_ANY);
680c470c 2186 swatch = swash_get(swash,
979f2922
ST
2187 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
2188 (klen) ? (code_point & ~(needents - 1)) : 0,
2189 needents);
2190
923e4eb5 2191 if (IN_PERL_COMPILETIME)
623e6609 2192 CopHINTS_set(PL_curcop, PL_hints);
a0ed51b3 2193
979f2922 2194 svp = hv_store(hv, (const char *)ptr, klen, swatch, 0);
a0ed51b3 2195
979f2922
ST
2196 if (!svp || !(tmps = (U8*)SvPV(*svp, slen))
2197 || (slen << 3) < needents)
660a4616 2198 Perl_croak(aTHX_ "panic: swash_fetch got improper swatch");
a0ed51b3
LW
2199 }
2200
2201 PL_last_swash_hv = hv;
16d8f38a 2202 assert(klen <= sizeof(PL_last_swash_key));
eac04b2e 2203 PL_last_swash_klen = (U8)klen;
cfd0369c
NC
2204 /* FIXME change interpvar.h? */
2205 PL_last_swash_tmps = (U8 *) tmps;
a0ed51b3
LW
2206 PL_last_swash_slen = slen;
2207 if (klen)
2208 Copy(ptr, PL_last_swash_key, klen, U8);
2209 }
2210
9faf8d75 2211 switch ((int)((slen << 3) / needents)) {
a0ed51b3
LW
2212 case 1:
2213 bit = 1 << (off & 7);
2214 off >>= 3;
2215 return (tmps[off] & bit) != 0;
2216 case 8:
2217 return tmps[off];
2218 case 16:
2219 off <<= 1;
2220 return (tmps[off] << 8) + tmps[off + 1] ;
2221 case 32:
2222 off <<= 2;
2223 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
2224 }
660a4616 2225 Perl_croak(aTHX_ "panic: swash_fetch got swatch of unexpected bit width");
670f1322 2226 NORETURN_FUNCTION_END;
a0ed51b3 2227}
2b9d42f0 2228
319009ee
KW
2229/* Read a single line of the main body of the swash input text. These are of
2230 * the form:
2231 * 0053 0056 0073
2232 * where each number is hex. The first two numbers form the minimum and
2233 * maximum of a range, and the third is the value associated with the range.
2234 * Not all swashes should have a third number
2235 *
2236 * On input: l points to the beginning of the line to be examined; it points
2237 * to somewhere in the string of the whole input text, and is
2238 * terminated by a \n or the null string terminator.
2239 * lend points to the null terminator of that string
2240 * wants_value is non-zero if the swash expects a third number
2241 * typestr is the name of the swash's mapping, like 'ToLower'
2242 * On output: *min, *max, and *val are set to the values read from the line.
2243 * returns a pointer just beyond the line examined. If there was no
2244 * valid min number on the line, returns lend+1
2245 */
2246
2247STATIC U8*
2248S_swash_scan_list_line(pTHX_ U8* l, U8* const lend, UV* min, UV* max, UV* val,
2249 const bool wants_value, const U8* const typestr)
2250{
2251 const int typeto = typestr[0] == 'T' && typestr[1] == 'o';
2252 STRLEN numlen; /* Length of the number */
2253 I32 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
2254
2255 /* nl points to the next \n in the scan */
2256 U8* const nl = (U8*)memchr(l, '\n', lend - l);
2257
2258 /* Get the first number on the line: the range minimum */
2259 numlen = lend - l;
2260 *min = grok_hex((char *)l, &numlen, &flags, NULL);
2261 if (numlen) /* If found a hex number, position past it */
2262 l += numlen;
2263 else if (nl) { /* Else, go handle next line, if any */
2264 return nl + 1; /* 1 is length of "\n" */
2265 }
2266 else { /* Else, no next line */
2267 return lend + 1; /* to LIST's end at which \n is not found */
2268 }
2269
2270 /* The max range value follows, separated by a BLANK */
2271 if (isBLANK(*l)) {
2272 ++l;
2273 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
2274 numlen = lend - l;
2275 *max = grok_hex((char *)l, &numlen, &flags, NULL);
2276 if (numlen)
2277 l += numlen;
2278 else /* If no value here, it is a single element range */
2279 *max = *min;
2280
2281 /* Non-binary tables have a third entry: what the first element of the
2282 * range maps to */
2283 if (wants_value) {
2284 if (isBLANK(*l)) {
2285 ++l;
2286 flags = PERL_SCAN_SILENT_ILLDIGIT |
2287 PERL_SCAN_DISALLOW_PREFIX;
2288 numlen = lend - l;
2289 *val = grok_hex((char *)l, &numlen, &flags, NULL);
2290 if (numlen)
2291 l += numlen;
2292 else
2293 *val = 0;
2294 }
2295 else {
2296 *val = 0;
2297 if (typeto) {
2298 Perl_croak(aTHX_ "%s: illegal mapping '%s'",
2299 typestr, l);
2300 }
2301 }
2302 }
2303 else
2304 *val = 0; /* bits == 1, then any val should be ignored */
2305 }
2306 else { /* Nothing following range min, should be single element with no
2307 mapping expected */
2308 *max = *min;
2309 if (wants_value) {
2310 *val = 0;
2311 if (typeto) {
2312 Perl_croak(aTHX_ "%s: illegal mapping '%s'", typestr, l);
2313 }
2314 }
2315 else
2316 *val = 0; /* bits == 1, then val should be ignored */
2317 }
2318
2319 /* Position to next line if any, or EOF */
2320 if (nl)
2321 l = nl + 1;
2322 else
2323 l = lend;
2324
2325 return l;
2326}
2327
979f2922
ST
2328/* Note:
2329 * Returns a swatch (a bit vector string) for a code point sequence
2330 * that starts from the value C<start> and comprises the number C<span>.
2331 * A C<swash> must be an object created by SWASHNEW (see lib/utf8_heavy.pl).
2332 * Should be used via swash_fetch, which will cache the swatch in C<swash>.
2333 */
2334STATIC SV*
2335S_swash_get(pTHX_ SV* swash, UV start, UV span)
2336{
2337 SV *swatch;
711a919c 2338 U8 *l, *lend, *x, *xend, *s;
979f2922 2339 STRLEN lcur, xcur, scur;
ef8f7699 2340 HV *const hv = MUTABLE_HV(SvRV(swash));
972dd592
KW
2341
2342 /* The string containing the main body of the table */
017a3ce5 2343 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
972dd592 2344
017a3ce5
GA
2345 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2346 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2347 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2348 SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);
0bd48802 2349 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
0bd48802
AL
2350 const STRLEN bits = SvUV(*bitssvp);
2351 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2352 const UV none = SvUV(*nonesvp);
2353 const UV end = start + span;
979f2922 2354
7918f24d
NC
2355 PERL_ARGS_ASSERT_SWASH_GET;
2356
979f2922 2357 if (bits != 1 && bits != 8 && bits != 16 && bits != 32) {
660a4616
ST
2358 Perl_croak(aTHX_ "panic: swash_get doesn't expect bits %"UVuf,
2359 (UV)bits);
979f2922
ST
2360 }
2361
2362 /* create and initialize $swatch */
979f2922 2363 scur = octets ? (span * octets) : (span + 7) / 8;
e524fe40
NC
2364 swatch = newSV(scur);
2365 SvPOK_on(swatch);
979f2922
ST
2366 s = (U8*)SvPVX(swatch);
2367 if (octets && none) {
0bd48802 2368 const U8* const e = s + scur;
979f2922
ST
2369 while (s < e) {
2370 if (bits == 8)
2371 *s++ = (U8)(none & 0xff);
2372 else if (bits == 16) {
2373 *s++ = (U8)((none >> 8) & 0xff);
2374 *s++ = (U8)( none & 0xff);
2375 }
2376 else if (bits == 32) {
2377 *s++ = (U8)((none >> 24) & 0xff);
2378 *s++ = (U8)((none >> 16) & 0xff);
2379 *s++ = (U8)((none >> 8) & 0xff);
2380 *s++ = (U8)( none & 0xff);
2381 }
2382 }
2383 *s = '\0';
2384 }
2385 else {
2386 (void)memzero((U8*)s, scur + 1);
2387 }
2388 SvCUR_set(swatch, scur);
2389 s = (U8*)SvPVX(swatch);
2390
2391 /* read $swash->{LIST} */
2392 l = (U8*)SvPV(*listsvp, lcur);
2393 lend = l + lcur;
2394 while (l < lend) {
35da51f7 2395 UV min, max, val;
319009ee
KW
2396 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2397 cBOOL(octets), typestr);
2398 if (l > lend) {
979f2922
ST
2399 break;
2400 }
2401
972dd592 2402 /* If looking for something beyond this range, go try the next one */
979f2922
ST
2403 if (max < start)
2404 continue;
2405
2406 if (octets) {
35da51f7 2407 UV key;
979f2922
ST
2408 if (min < start) {
2409 if (!none || val < none) {
2410 val += start - min;
2411 }
2412 min = start;
2413 }
2414 for (key = min; key <= max; key++) {
2415 STRLEN offset;
2416 if (key >= end)
2417 goto go_out_list;
2418 /* offset must be non-negative (start <= min <= key < end) */
2419 offset = octets * (key - start);
2420 if (bits == 8)
2421 s[offset] = (U8)(val & 0xff);
2422 else if (bits == 16) {
2423 s[offset ] = (U8)((val >> 8) & 0xff);
2424 s[offset + 1] = (U8)( val & 0xff);
2425 }
2426 else if (bits == 32) {
2427 s[offset ] = (U8)((val >> 24) & 0xff);
2428 s[offset + 1] = (U8)((val >> 16) & 0xff);
2429 s[offset + 2] = (U8)((val >> 8) & 0xff);
2430 s[offset + 3] = (U8)( val & 0xff);
2431 }
2432
2433 if (!none || val < none)
2434 ++val;
2435 }
2436 }
711a919c 2437 else { /* bits == 1, then val should be ignored */
35da51f7 2438 UV key;
979f2922
ST
2439 if (min < start)
2440 min = start;
2441 for (key = min; key <= max; key++) {
0bd48802 2442 const STRLEN offset = (STRLEN)(key - start);
979f2922
ST
2443 if (key >= end)
2444 goto go_out_list;
2445 s[offset >> 3] |= 1 << (offset & 7);
2446 }
2447 }
2448 } /* while */
2449 go_out_list:
2450
2451 /* read $swash->{EXTRAS} */
2452 x = (U8*)SvPV(*extssvp, xcur);
2453 xend = x + xcur;
2454 while (x < xend) {
2455 STRLEN namelen;
2456 U8 *namestr;
2457 SV** othersvp;
2458 HV* otherhv;
2459 STRLEN otherbits;
2460 SV **otherbitssvp, *other;
711a919c 2461 U8 *s, *o, *nl;
979f2922
ST
2462 STRLEN slen, olen;
2463
35da51f7 2464 const U8 opc = *x++;
979f2922
ST
2465 if (opc == '\n')
2466 continue;
2467
2468 nl = (U8*)memchr(x, '\n', xend - x);
2469
2470 if (opc != '-' && opc != '+' && opc != '!' && opc != '&') {
2471 if (nl) {
2472 x = nl + 1; /* 1 is length of "\n" */
2473 continue;
2474 }
2475 else {
2476 x = xend; /* to EXTRAS' end at which \n is not found */
2477 break;
2478 }
2479 }
2480
2481 namestr = x;
2482 if (nl) {
2483 namelen = nl - namestr;
2484 x = nl + 1;
2485 }
2486 else {
2487 namelen = xend - namestr;
2488 x = xend;
2489 }
2490
2491 othersvp = hv_fetch(hv, (char *)namestr, namelen, FALSE);
ef8f7699 2492 otherhv = MUTABLE_HV(SvRV(*othersvp));
017a3ce5 2493 otherbitssvp = hv_fetchs(otherhv, "BITS", FALSE);
979f2922
ST
2494 otherbits = (STRLEN)SvUV(*otherbitssvp);
2495 if (bits < otherbits)
660a4616 2496 Perl_croak(aTHX_ "panic: swash_get found swatch size mismatch");
979f2922
ST
2497
2498 /* The "other" swatch must be destroyed after. */
2499 other = swash_get(*othersvp, start, span);
2500 o = (U8*)SvPV(other, olen);
2501
2502 if (!olen)
660a4616 2503 Perl_croak(aTHX_ "panic: swash_get got improper swatch");
979f2922
ST
2504
2505 s = (U8*)SvPV(swatch, slen);
2506 if (bits == 1 && otherbits == 1) {
2507 if (slen != olen)
660a4616 2508 Perl_croak(aTHX_ "panic: swash_get found swatch length mismatch");
979f2922
ST
2509
2510 switch (opc) {
2511 case '+':
2512 while (slen--)
2513 *s++ |= *o++;
2514 break;
2515 case '!':
2516 while (slen--)
2517 *s++ |= ~*o++;
2518 break;
2519 case '-':
2520 while (slen--)
2521 *s++ &= ~*o++;
2522 break;
2523 case '&':
2524 while (slen--)
2525 *s++ &= *o++;
2526 break;
2527 default:
2528 break;
2529 }
2530 }
711a919c 2531 else {
979f2922
ST
2532 STRLEN otheroctets = otherbits >> 3;
2533 STRLEN offset = 0;
35da51f7 2534 U8* const send = s + slen;
979f2922
ST
2535
2536 while (s < send) {
2537 UV otherval = 0;
2538
2539 if (otherbits == 1) {
2540 otherval = (o[offset >> 3] >> (offset & 7)) & 1;
2541 ++offset;
2542 }
2543 else {
2544 STRLEN vlen = otheroctets;
2545 otherval = *o++;
2546 while (--vlen) {
2547 otherval <<= 8;
2548 otherval |= *o++;
2549 }
2550 }
2551
711a919c 2552 if (opc == '+' && otherval)
6f207bd3 2553 NOOP; /* replace with otherval */
979f2922
ST
2554 else if (opc == '!' && !otherval)
2555 otherval = 1;
2556 else if (opc == '-' && otherval)
2557 otherval = 0;
2558 else if (opc == '&' && !otherval)
2559 otherval = 0;
2560 else {
711a919c 2561 s += octets; /* no replacement */
979f2922
ST
2562 continue;
2563 }
2564
2565 if (bits == 8)
2566 *s++ = (U8)( otherval & 0xff);
2567 else if (bits == 16) {
2568 *s++ = (U8)((otherval >> 8) & 0xff);
2569 *s++ = (U8)( otherval & 0xff);
2570 }
2571 else if (bits == 32) {
2572 *s++ = (U8)((otherval >> 24) & 0xff);
2573 *s++ = (U8)((otherval >> 16) & 0xff);
2574 *s++ = (U8)((otherval >> 8) & 0xff);
2575 *s++ = (U8)( otherval & 0xff);
2576 }
2577 }
2578 }
2579 sv_free(other); /* through with it! */
2580 } /* while */
2581 return swatch;
2582}
2583
064c021d
KW
2584HV*
2585Perl__swash_inversion_hash(pTHX_ SV* swash)
2586{
2587
2588 /* Subject to change or removal. For use only in one place in regexec.c
2589 *
2590 * Returns a hash which is the inversion and closure of a swash mapping.
2591 * For example, consider the input lines:
2592 * 004B 006B
2593 * 004C 006C
2594 * 212A 006B
2595 *
2596 * The returned hash would have two keys, the utf8 for 006B and the utf8 for
2597 * 006C. The value for each key is an array. For 006C, the array would
2598 * have a two elements, the utf8 for itself, and for 004C. For 006B, there
2599 * would be three elements in its array, the utf8 for 006B, 004B and 212A.
2600 *
2601 * Essentially, for any code point, it gives all the code points that map to
2602 * it, or the list of 'froms' for that point.
2603 *
2604 * Currently it only looks at the main body of the swash, and ignores any
2605 * additions or deletions from other swashes */
2606
2607 U8 *l, *lend;
2608 STRLEN lcur;
2609 HV *const hv = MUTABLE_HV(SvRV(swash));
2610
2611 /* The string containing the main body of the table */
2612 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2613
2614 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2615 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2616 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2617 /*SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);*/
2618 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
2619 const STRLEN bits = SvUV(*bitssvp);
2620 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2621 const UV none = SvUV(*nonesvp);
2622
2623 HV* ret = newHV();
2624
2625 PERL_ARGS_ASSERT__SWASH_INVERSION_HASH;
2626
2627 /* Must have at least 8 bits to get the mappings */
2628 if (bits != 8 && bits != 16 && bits != 32) {
2629 Perl_croak(aTHX_ "panic: swash_inversion_hash doesn't expect bits %"UVuf,
2630 (UV)bits);
2631 }
2632
2633 /* read $swash->{LIST} */
2634 l = (U8*)SvPV(*listsvp, lcur);
2635 lend = l + lcur;
2636
2637 /* Go through each input line */
2638 while (l < lend) {
2639 UV min, max, val;
2640 UV inverse;
2641 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2642 cBOOL(octets), typestr);
2643 if (l > lend) {
2644 break;
2645 }
2646
2647 /* Each element in the range is to be inverted */
2648 for (inverse = min; inverse <= max; inverse++) {
2649 AV* list;
2650 SV* element;
2651 SV** listp;
2652 IV i;
2653 bool found_key = FALSE;
2654
2655 /* The key is the inverse mapping */
2656 char key[UTF8_MAXBYTES+1];
2657 char* key_end = (char *) uvuni_to_utf8((U8*) key, val);
2658 STRLEN key_len = key_end - key;
2659
2660 /* And the value is what the forward mapping is from. */
2661 char utf8_inverse[UTF8_MAXBYTES+1];
2662 char *utf8_inverse_end = (char *) uvuni_to_utf8((U8*) utf8_inverse, inverse);
2663
2664 /* Get the list for the map */
2665 if ((listp = hv_fetch(ret, key, key_len, FALSE))) {
2666 list = (AV*) *listp;
2667 }
2668 else { /* No entry yet for it: create one */
2669 list = newAV();
2670 if (! hv_store(ret, key, key_len, (SV*) list, FALSE)) {
2671 Perl_croak(aTHX_ "panic: hv_store() unexpectedly failed");
2672 }
2673 }
2674
2675 for (i = 0; i < av_len(list); i++) {
2676 SV** entryp = av_fetch(list, i, FALSE);
2677 SV* entry;
2678 if (entryp == NULL) {
2679 Perl_croak(aTHX_ "panic: av_fetch() unexpectedly failed");
2680 }
2681 entry = *entryp;
2682 if (SvCUR(entry) != key_len) {
2683 continue;
2684 }
2685 if (memEQ(key, SvPVX(entry), key_len)) {
2686 found_key = TRUE;
2687 break;
2688 }
2689 }
2690 if (! found_key) {
2691 element = newSVpvn_flags(key, key_len, SVf_UTF8);
2692 av_push(list, element);
2693 }
2694
2695
2696 /* Simply add the value to the list */
2697 element = newSVpvn_flags(utf8_inverse, utf8_inverse_end - utf8_inverse, SVf_UTF8);
2698 av_push(list, element);
2699
2700 /* swash_get() increments the value of val for each element in the
2701 * range. That makes more compact tables possible. You can
2702 * express the capitalization, for example, of all consecutive
2703 * letters with a single line: 0061\t007A\t0041 This maps 0061 to
2704 * 0041, 0062 to 0042, etc. I (khw) have never understood 'none',
2705 * and it's not documented, and perhaps not even currently used,
2706 * but I copied the semantics from swash_get(), just in case */
2707 if (!none || val < none) {
2708 ++val;
2709 }
2710 }
2711 }
2712
2713 return ret;
2714}
2715
0f830e0b 2716/*
87cea99e 2717=for apidoc uvchr_to_utf8
0f830e0b 2718
6ee84de2 2719Adds the UTF-8 representation of the Native code point C<uv> to the end
0f830e0b
NC
2720of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
2721bytes available. The return value is the pointer to the byte after the
2722end of the new character. In other words,
2723
2724 d = uvchr_to_utf8(d, uv);
2725
2726is the recommended wide native character-aware way of saying
2727
2728 *(d++) = uv;
2729
2730=cut
2731*/
2732
2733/* On ASCII machines this is normally a macro but we want a
2734 real function in case XS code wants it
2735*/
0f830e0b
NC
2736U8 *
2737Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
2738{
7918f24d
NC
2739 PERL_ARGS_ASSERT_UVCHR_TO_UTF8;
2740
0f830e0b
NC
2741 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
2742}
2743
b851fbc1
JH
2744U8 *
2745Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
2746{
7918f24d
NC
2747 PERL_ARGS_ASSERT_UVCHR_TO_UTF8_FLAGS;
2748
b851fbc1
JH
2749 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
2750}
2b9d42f0
NIS
2751
2752/*
87cea99e 2753=for apidoc utf8n_to_uvchr
0f830e0b 2754
48ef279e 2755Returns the native character value of the first character in the string
0f830e0b
NC
2756C<s>
2757which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
2758length, in bytes, of that character.
2759
6ee84de2 2760length and flags are the same as utf8n_to_uvuni().
0f830e0b
NC
2761
2762=cut
2763*/
2764/* On ASCII machines this is normally a macro but we want
2765 a real function in case XS code wants it
2766*/
0f830e0b 2767UV
48ef279e 2768Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen,
0f830e0b
NC
2769U32 flags)
2770{
2771 const UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
7918f24d
NC
2772
2773 PERL_ARGS_ASSERT_UTF8N_TO_UVCHR;
2774
0f830e0b
NC
2775 return UNI_TO_NATIVE(uv);
2776}
2777
0876b9a0
KW
2778bool
2779Perl_check_utf8_print(pTHX_ register const U8* s, const STRLEN len)
2780{
2781 /* May change: warns if surrogates, non-character code points, or
2782 * non-Unicode code points are in s which has length len. Returns TRUE if
2783 * none found; FALSE otherwise. The only other validity check is to make
2784 * sure that this won't exceed the string's length */
2785
2786 const U8* const e = s + len;
2787 bool ok = TRUE;
2788
2789 PERL_ARGS_ASSERT_CHECK_UTF8_PRINT;
2790
2791 while (s < e) {
2792 if (UTF8SKIP(s) > len) {
2793 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
2794 "%s in %s", unees, PL_op ? OP_DESC(PL_op) : "print");
2795 return FALSE;
2796 }
2797 if (*s >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE) {
2798 STRLEN char_len;
2799 if (UTF8_IS_SUPER(s)) {
2800 UV uv = utf8_to_uvchr(s, &char_len);
2801 Perl_warner(aTHX_ packWARN(WARN_UTF8),
2802 "Code point 0x%04"UVXf" is not Unicode, may not be portable", uv);
2803 ok = FALSE;
2804 }
2805 else if (UTF8_IS_SURROGATE(s)) {
2806 UV uv = utf8_to_uvchr(s, &char_len);
2807 Perl_warner(aTHX_ packWARN(WARN_UTF8),
2808 "Unicode surrogate U+%04"UVXf" is illegal in UTF-8", uv);
2809 ok = FALSE;
2810 }
2811 else if
2812 (UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s))
2813 {
2814 UV uv = utf8_to_uvchr(s, &char_len);
2815 Perl_warner(aTHX_ packWARN(WARN_UTF8),
2816 "Unicode non-character U+%04"UVXf" is illegal for open interchange", uv);
2817 ok = FALSE;
2818 }
2819 }
2820 s += UTF8SKIP(s);
2821 }
2822
2823 return ok;
2824}
2825
0f830e0b 2826/*
87cea99e 2827=for apidoc pv_uni_display
d2cc3551
JH
2828
2829Build to the scalar dsv a displayable version of the string spv,
2830length len, the displayable version being at most pvlim bytes long
2831(if longer, the rest is truncated and "..." will be appended).
0a2ef054 2832
9e55ce06 2833The flags argument can have UNI_DISPLAY_ISPRINT set to display
00e86452 2834isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
0a2ef054
JH
2835to display the \\[nrfta\\] as the backslashed versions (like '\n')
2836(UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
2837UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
2838UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
2839
d2cc3551
JH
2840The pointer to the PV of the dsv is returned.
2841
2842=cut */
e6b2e755 2843char *
e1ec3a88 2844Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
e6b2e755
JH
2845{
2846 int truncated = 0;
e1ec3a88 2847 const char *s, *e;
e6b2e755 2848
7918f24d
NC
2849 PERL_ARGS_ASSERT_PV_UNI_DISPLAY;
2850
76f68e9b 2851 sv_setpvs(dsv, "");
7fddd944 2852 SvUTF8_off(dsv);
e1ec3a88 2853 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
e6b2e755 2854 UV u;
a49f32c6
NC
2855 /* This serves double duty as a flag and a character to print after
2856 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
2857 */
2858 char ok = 0;
c728cb41 2859
e6b2e755
JH
2860 if (pvlim && SvCUR(dsv) >= pvlim) {
2861 truncated++;
2862 break;
2863 }
2864 u = utf8_to_uvchr((U8*)s, 0);
c728cb41 2865 if (u < 256) {
a3b680e6 2866 const unsigned char c = (unsigned char)u & 0xFF;
0bd48802 2867 if (flags & UNI_DISPLAY_BACKSLASH) {
a49f32c6 2868 switch (c) {
c728cb41 2869 case '\n':
a49f32c6 2870 ok = 'n'; break;
c728cb41 2871 case '\r':
a49f32c6 2872 ok = 'r'; break;
c728cb41 2873 case '\t':
a49f32c6 2874 ok = 't'; break;
c728cb41 2875 case '\f':
a49f32c6 2876 ok = 'f'; break;
c728cb41 2877 case '\a':
a49f32c6 2878 ok = 'a'; break;
c728cb41 2879 case '\\':
a49f32c6 2880 ok = '\\'; break;
c728cb41
JH
2881 default: break;
2882 }
a49f32c6 2883 if (ok) {
88c9ea1e 2884 const char string = ok;
76f68e9b 2885 sv_catpvs(dsv, "\\");
5e7aa789 2886 sv_catpvn(dsv, &string, 1);
a49f32c6 2887 }
c728cb41 2888 }
00e86452 2889 /* isPRINT() is the locale-blind version. */
a49f32c6 2890 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
88c9ea1e 2891 const char string = c;
5e7aa789 2892 sv_catpvn(dsv, &string, 1);
a49f32c6 2893 ok = 1;
0a2ef054 2894 }
c728cb41
JH
2895 }
2896 if (!ok)
9e55ce06 2897 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
e6b2e755
JH
2898 }
2899 if (truncated)
396482e1 2900 sv_catpvs(dsv, "...");
48ef279e 2901
e6b2e755
JH
2902 return SvPVX(dsv);
2903}
2b9d42f0 2904
d2cc3551 2905/*
87cea99e 2906=for apidoc sv_uni_display
d2cc3551
JH
2907
2908Build to the scalar dsv a displayable version of the scalar sv,
0a2ef054 2909the displayable version being at most pvlim bytes long
d2cc3551 2910(if longer, the rest is truncated and "..." will be appended).
0a2ef054
JH
2911
2912The flags argument is as in pv_uni_display().
2913
d2cc3551
JH
2914The pointer to the PV of the dsv is returned.
2915
d4c19fe8
AL
2916=cut
2917*/
e6b2e755
JH
2918char *
2919Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
2920{
7918f24d
NC
2921 PERL_ARGS_ASSERT_SV_UNI_DISPLAY;
2922
cfd0369c
NC
2923 return Perl_pv_uni_display(aTHX_ dsv, (const U8*)SvPVX_const(ssv),
2924 SvCUR(ssv), pvlim, flags);
701a277b
JH
2925}
2926
d2cc3551 2927/*
e6226b18 2928=for apidoc foldEQ_utf8
d2cc3551 2929
d51c1b21 2930Returns true if the leading portions of the strings s1 and s2 (either or both
e6226b18 2931of which may be in UTF-8) are the same case-insensitively; false otherwise.
d51c1b21 2932How far into the strings to compare is determined by other input parameters.
8b35872c
KW
2933
2934If u1 is true, the string s1 is assumed to be in UTF-8-encoded Unicode;
2935otherwise it is assumed to be in native 8-bit encoding. Correspondingly for u2
2936with respect to s2.
2937
d51c1b21
KW
2938If the byte length l1 is non-zero, it says how far into s1 to check for fold
2939equality. In other words, s1+l1 will be used as a goal to reach. The
8b35872c
KW
2940scan will not be considered to be a match unless the goal is reached, and
2941scanning won't continue past that goal. Correspondingly for l2 with respect to
2942s2.
2943
2944If pe1 is non-NULL and the pointer it points to is not NULL, that pointer is
2945considered an end pointer beyond which scanning of s1 will not continue under
2946any circumstances. This means that if both l1 and pe1 are specified, and pe1
2947is less than s1+l1, the match will never be successful because it can never
d51c1b21
KW
2948get as far as its goal (and in fact is asserted against). Correspondingly for
2949pe2 with respect to s2.
8b35872c 2950
d51c1b21
KW
2951At least one of s1 and s2 must have a goal (at least one of l1 and l2 must be
2952non-zero), and if both do, both have to be
8b35872c
KW
2953reached for a successful match. Also, if the fold of a character is multiple
2954characters, all of them must be matched (see tr21 reference below for
2955'folding').
2956
e6226b18 2957Upon a successful match, if pe1 is non-NULL,
8b35872c
KW
2958it will be set to point to the beginning of the I<next> character of s1 beyond
2959what was matched. Correspondingly for pe2 and s2.
d2cc3551
JH
2960
2961For case-insensitiveness, the "casefolding" of Unicode is used
2962instead of upper/lowercasing both the characters, see
2963http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
2964
2965=cut */
701a277b 2966I32
e6226b18 2967Perl_foldEQ_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
332ddc25 2968{
8b35872c
KW
2969 dVAR;
2970 register const U8 *p1 = (const U8*)s1; /* Point to current char */
2971 register const U8 *p2 = (const U8*)s2;
48ef279e 2972 register const U8 *g1 = NULL; /* goal for s1 */
8b35872c 2973 register const U8 *g2 = NULL;
48ef279e
KW
2974 register const U8 *e1 = NULL; /* Don't scan s1 past this */
2975 register U8 *f1 = NULL; /* Point to current folded */
8b35872c
KW
2976 register const U8 *e2 = NULL;
2977 register U8 *f2 = NULL;
48ef279e 2978 STRLEN n1 = 0, n2 = 0; /* Number of bytes in current char */
8b35872c
KW
2979 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
2980 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
48ef279e
KW
2981 U8 natbuf[2]; /* Holds native 8-bit char converted to utf8;
2982 these always fit in 2 bytes */
8b35872c 2983
e6226b18 2984 PERL_ARGS_ASSERT_FOLDEQ_UTF8;
8b35872c
KW
2985
2986 if (pe1) {
48ef279e 2987 e1 = *(U8**)pe1;
8b35872c
KW
2988 }
2989
2990 if (l1) {
48ef279e 2991 g1 = (const U8*)s1 + l1;
8b35872c
KW
2992 }
2993
2994 if (pe2) {
48ef279e 2995 e2 = *(U8**)pe2;
8b35872c
KW
2996 }
2997
2998 if (l2) {
48ef279e 2999 g2 = (const U8*)s2 + l2;
8b35872c
KW
3000 }
3001
3002 /* Must have at least one goal */
3003 assert(g1 || g2);
3004
3005 if (g1) {
3006
48ef279e
KW
3007 /* Will never match if goal is out-of-bounds */
3008 assert(! e1 || e1 >= g1);
8b35872c 3009
48ef279e
KW
3010 /* Here, there isn't an end pointer, or it is beyond the goal. We
3011 * only go as far as the goal */
3012 e1 = g1;
8b35872c 3013 }
313b38e5
NC
3014 else {
3015 assert(e1); /* Must have an end for looking at s1 */
3016 }
8b35872c
KW
3017
3018 /* Same for goal for s2 */
3019 if (g2) {
48ef279e
KW
3020 assert(! e2 || e2 >= g2);
3021 e2 = g2;
8b35872c 3022 }
313b38e5
NC
3023 else {
3024 assert(e2);
3025 }
8b35872c
KW
3026
3027 /* Look through both strings, a character at a time */
3028 while (p1 < e1 && p2 < e2) {
3029
d51c1b21
KW
3030 /* If at the beginning of a new character in s1, get its fold to use
3031 * and the length of the fold */
48ef279e
KW
3032 if (n1 == 0) {
3033 if (u1) {
3034 to_utf8_fold(p1, foldbuf1, &n1);
3035 }
3036 else { /* Not utf8, convert to it first and then get fold */
3037 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p1)));
3038 to_utf8_fold(natbuf, foldbuf1, &n1);
3039 }
3040 f1 = foldbuf1;
3041 }
8b35872c 3042
48ef279e
KW
3043 if (n2 == 0) { /* Same for s2 */
3044 if (u2) {
3045 to_utf8_fold(p2, foldbuf2, &n2);
3046 }
3047 else {
3048 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p2)));
3049 to_utf8_fold(natbuf, foldbuf2, &n2);
3050 }
3051 f2 = foldbuf2;
3052 }
8b35872c 3053
48ef279e
KW
3054 /* While there is more to look for in both folds, see if they
3055 * continue to match */
3056 while (n1 && n2) {
3057 U8 fold_length = UTF8SKIP(f1);
3058 if (fold_length != UTF8SKIP(f2)
3059 || (fold_length == 1 && *f1 != *f2) /* Short circuit memNE
3060 function call for single
3061 character */
3062 || memNE((char*)f1, (char*)f2, fold_length))
3063 {
e6226b18 3064 return 0; /* mismatch */
48ef279e
KW
3065 }
3066
3067 /* Here, they matched, advance past them */
3068 n1 -= fold_length;
3069 f1 += fold_length;
3070 n2 -= fold_length;
3071 f2 += fold_length;
3072 }
8b35872c 3073
48ef279e
KW
3074 /* When reach the end of any fold, advance the input past it */
3075 if (n1 == 0) {
3076 p1 += u1 ? UTF8SKIP(p1) : 1;
3077 }
3078 if (n2 == 0) {
3079 p2 += u2 ? UTF8SKIP(p2) : 1;
3080 }
8b35872c
KW
3081 } /* End of loop through both strings */
3082
3083 /* A match is defined by each scan that specified an explicit length
3084 * reaching its final goal, and the other not having matched a partial
3085 * character (which can happen when the fold of a character is more than one
3086 * character). */
3087 if (! ((g1 == 0 || p1 == g1) && (g2 == 0 || p2 == g2)) || n1 || n2) {
e6226b18 3088 return 0;
8b35872c
KW
3089 }
3090
3091 /* Successful match. Set output pointers */
3092 if (pe1) {
48ef279e 3093 *pe1 = (char*)p1;
8b35872c
KW
3094 }
3095 if (pe2) {
48ef279e 3096 *pe2 = (char*)p2;
8b35872c 3097 }
e6226b18 3098 return 1;
e6b2e755 3099}
701a277b 3100
a49f32c6
NC
3101/*
3102 * Local variables:
3103 * c-indentation-style: bsd
3104 * c-basic-offset: 4
3105 * indent-tabs-mode: t
3106 * End:
3107 *
37442d52
RGS
3108 * ex: set ts=8 sts=4 sw=4 noet:
3109 */