This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
utf8.c: "<" should be "<="
[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) {
8457b38f 142 Perl_ck_warner_d(aTHX_ packWARN(WARN_SURROGATE),
949cf498
KW
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 {
8457b38f 153 Perl_ck_warner_d(aTHX_ packWARN(WARN_NON_UNICODE),
949cf498
KW
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) {
8457b38f 164 Perl_ck_warner_d(aTHX_ packWARN(WARN_NONCHAR),
949cf498
KW
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
TS
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
61a12c31 844 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8), "%s", 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
61a12c31 956 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8), "%s", unees);
fed3ba5d
NC
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 1343UV
36bb2ab6 1344Perl__to_uni_fold_flags(pTHX_ UV c, U8* p, STRLEN *lenp, U8 flags)
84afefe6 1345{
36bb2ab6 1346 PERL_ARGS_ASSERT__TO_UNI_FOLD_FLAGS;
7918f24d 1347
0ebc6274 1348 uvchr_to_utf8(p, c);
36bb2ab6 1349 return _to_utf8_fold_flags(p, p, lenp, flags);
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
c11ff943
KW
1504Perl_is_utf8_xidfirst(pTHX_ const U8 *p) /* The naming is historical. */
1505{
1506 dVAR;
1507
1508 PERL_ARGS_ASSERT_IS_UTF8_XIDFIRST;
1509
1510 if (*p == '_')
1511 return TRUE;
1512 /* is_utf8_idstart would be more logical. */
1513 return is_utf8_common(p, &PL_utf8_xidstart, "XIdStart");
1514}
1515
1516bool
7fc63493 1517Perl_is_utf8_idcont(pTHX_ const U8 *p)
82686b01 1518{
97aff369 1519 dVAR;
7918f24d
NC
1520
1521 PERL_ARGS_ASSERT_IS_UTF8_IDCONT;
1522
82686b01
JH
1523 if (*p == '_')
1524 return TRUE;
d4c19fe8 1525 return is_utf8_common(p, &PL_utf8_idcont, "IdContinue");
a0ed51b3
LW
1526}
1527
1528bool
c11ff943
KW
1529Perl_is_utf8_xidcont(pTHX_ const U8 *p)
1530{
1531 dVAR;
1532
1533 PERL_ARGS_ASSERT_IS_UTF8_XIDCONT;
1534
1535 if (*p == '_')
1536 return TRUE;
1537 return is_utf8_common(p, &PL_utf8_idcont, "XIdContinue");
1538}
1539
1540bool
7fc63493 1541Perl_is_utf8_alpha(pTHX_ const U8 *p)
a0ed51b3 1542{
97aff369 1543 dVAR;
7918f24d
NC
1544
1545 PERL_ARGS_ASSERT_IS_UTF8_ALPHA;
1546
d4c19fe8 1547 return is_utf8_common(p, &PL_utf8_alpha, "IsAlpha");
a0ed51b3
LW
1548}
1549
1550bool
7fc63493 1551Perl_is_utf8_ascii(pTHX_ const U8 *p)
b8c5462f 1552{
97aff369 1553 dVAR;
7918f24d
NC
1554
1555 PERL_ARGS_ASSERT_IS_UTF8_ASCII;
1556
d4c19fe8 1557 return is_utf8_common(p, &PL_utf8_ascii, "IsAscii");
b8c5462f
JH
1558}
1559
1560bool
7fc63493 1561Perl_is_utf8_space(pTHX_ const U8 *p)
a0ed51b3 1562{
97aff369 1563 dVAR;
7918f24d
NC
1564
1565 PERL_ARGS_ASSERT_IS_UTF8_SPACE;
1566
d4c19fe8 1567 return is_utf8_common(p, &PL_utf8_space, "IsSpacePerl");
a0ed51b3
LW
1568}
1569
1570bool
d1eb3177
YO
1571Perl_is_utf8_perl_space(pTHX_ const U8 *p)
1572{
1573 dVAR;
1574
1575 PERL_ARGS_ASSERT_IS_UTF8_PERL_SPACE;
1576
1577 return is_utf8_common(p, &PL_utf8_perl_space, "IsPerlSpace");
1578}
1579
1580bool
1581Perl_is_utf8_perl_word(pTHX_ const U8 *p)
1582{
1583 dVAR;
1584
1585 PERL_ARGS_ASSERT_IS_UTF8_PERL_WORD;
1586
1587 return is_utf8_common(p, &PL_utf8_perl_word, "IsPerlWord");
1588}
1589
1590bool
7fc63493 1591Perl_is_utf8_digit(pTHX_ const U8 *p)
a0ed51b3 1592{
97aff369 1593 dVAR;
7918f24d
NC
1594
1595 PERL_ARGS_ASSERT_IS_UTF8_DIGIT;
1596
d4c19fe8 1597 return is_utf8_common(p, &PL_utf8_digit, "IsDigit");
a0ed51b3
LW
1598}
1599
1600bool
d1eb3177
YO
1601Perl_is_utf8_posix_digit(pTHX_ const U8 *p)
1602{
1603 dVAR;
1604
1605 PERL_ARGS_ASSERT_IS_UTF8_POSIX_DIGIT;
1606
1607 return is_utf8_common(p, &PL_utf8_posix_digit, "IsPosixDigit");
1608}
1609
1610bool
7fc63493 1611Perl_is_utf8_upper(pTHX_ const U8 *p)
a0ed51b3 1612{
97aff369 1613 dVAR;
7918f24d
NC
1614
1615 PERL_ARGS_ASSERT_IS_UTF8_UPPER;
1616
d4c19fe8 1617 return is_utf8_common(p, &PL_utf8_upper, "IsUppercase");
a0ed51b3
LW
1618}
1619
1620bool
7fc63493 1621Perl_is_utf8_lower(pTHX_ const U8 *p)
a0ed51b3 1622{
97aff369 1623 dVAR;
7918f24d
NC
1624
1625 PERL_ARGS_ASSERT_IS_UTF8_LOWER;
1626
d4c19fe8 1627 return is_utf8_common(p, &PL_utf8_lower, "IsLowercase");
a0ed51b3
LW
1628}
1629
1630bool
7fc63493 1631Perl_is_utf8_cntrl(pTHX_ const U8 *p)
b8c5462f 1632{
97aff369 1633 dVAR;
7918f24d
NC
1634
1635 PERL_ARGS_ASSERT_IS_UTF8_CNTRL;
1636
d4c19fe8 1637 return is_utf8_common(p, &PL_utf8_cntrl, "IsCntrl");
b8c5462f
JH
1638}
1639
1640bool
7fc63493 1641Perl_is_utf8_graph(pTHX_ const U8 *p)
b8c5462f 1642{
97aff369 1643 dVAR;
7918f24d
NC
1644
1645 PERL_ARGS_ASSERT_IS_UTF8_GRAPH;
1646
d4c19fe8 1647 return is_utf8_common(p, &PL_utf8_graph, "IsGraph");
b8c5462f
JH
1648}
1649
1650bool
7fc63493 1651Perl_is_utf8_print(pTHX_ const U8 *p)
a0ed51b3 1652{
97aff369 1653 dVAR;
7918f24d
NC
1654
1655 PERL_ARGS_ASSERT_IS_UTF8_PRINT;
1656
d4c19fe8 1657 return is_utf8_common(p, &PL_utf8_print, "IsPrint");
a0ed51b3
LW
1658}
1659
1660bool
7fc63493 1661Perl_is_utf8_punct(pTHX_ const U8 *p)
b8c5462f 1662{
97aff369 1663 dVAR;
7918f24d
NC
1664
1665 PERL_ARGS_ASSERT_IS_UTF8_PUNCT;
1666
d4c19fe8 1667 return is_utf8_common(p, &PL_utf8_punct, "IsPunct");
b8c5462f
JH
1668}
1669
1670bool
7fc63493 1671Perl_is_utf8_xdigit(pTHX_ const U8 *p)
b8c5462f 1672{
97aff369 1673 dVAR;
7918f24d
NC
1674
1675 PERL_ARGS_ASSERT_IS_UTF8_XDIGIT;
1676
d1eb3177 1677 return is_utf8_common(p, &PL_utf8_xdigit, "IsXDigit");
b8c5462f
JH
1678}
1679
1680bool
7fc63493 1681Perl_is_utf8_mark(pTHX_ const U8 *p)
a0ed51b3 1682{
97aff369 1683 dVAR;
7918f24d
NC
1684
1685 PERL_ARGS_ASSERT_IS_UTF8_MARK;
1686
d4c19fe8 1687 return is_utf8_common(p, &PL_utf8_mark, "IsM");
a0ed51b3
LW
1688}
1689
37e2e78e
KW
1690bool
1691Perl_is_utf8_X_begin(pTHX_ const U8 *p)
1692{
1693 dVAR;
1694
1695 PERL_ARGS_ASSERT_IS_UTF8_X_BEGIN;
1696
1697 return is_utf8_common(p, &PL_utf8_X_begin, "_X_Begin");
1698}
1699
1700bool
1701Perl_is_utf8_X_extend(pTHX_ const U8 *p)
1702{
1703 dVAR;
1704
1705 PERL_ARGS_ASSERT_IS_UTF8_X_EXTEND;
1706
1707 return is_utf8_common(p, &PL_utf8_X_extend, "_X_Extend");
1708}
1709
1710bool
1711Perl_is_utf8_X_prepend(pTHX_ const U8 *p)
1712{
1713 dVAR;
1714
1715 PERL_ARGS_ASSERT_IS_UTF8_X_PREPEND;
1716
1717 return is_utf8_common(p, &PL_utf8_X_prepend, "GCB=Prepend");
1718}
1719
1720bool
1721Perl_is_utf8_X_non_hangul(pTHX_ const U8 *p)
1722{
1723 dVAR;
1724
1725 PERL_ARGS_ASSERT_IS_UTF8_X_NON_HANGUL;
1726
1727 return is_utf8_common(p, &PL_utf8_X_non_hangul, "HST=Not_Applicable");
1728}
1729
1730bool
1731Perl_is_utf8_X_L(pTHX_ const U8 *p)
1732{
1733 dVAR;
1734
1735 PERL_ARGS_ASSERT_IS_UTF8_X_L;
1736
1737 return is_utf8_common(p, &PL_utf8_X_L, "GCB=L");
1738}
1739
1740bool
1741Perl_is_utf8_X_LV(pTHX_ const U8 *p)
1742{
1743 dVAR;
1744
1745 PERL_ARGS_ASSERT_IS_UTF8_X_LV;
1746
1747 return is_utf8_common(p, &PL_utf8_X_LV, "GCB=LV");
1748}
1749
1750bool
1751Perl_is_utf8_X_LVT(pTHX_ const U8 *p)
1752{
1753 dVAR;
1754
1755 PERL_ARGS_ASSERT_IS_UTF8_X_LVT;
1756
1757 return is_utf8_common(p, &PL_utf8_X_LVT, "GCB=LVT");
1758}
1759
1760bool
1761Perl_is_utf8_X_T(pTHX_ const U8 *p)
1762{
1763 dVAR;
1764
1765 PERL_ARGS_ASSERT_IS_UTF8_X_T;
1766
1767 return is_utf8_common(p, &PL_utf8_X_T, "GCB=T");
1768}
1769
1770bool
1771Perl_is_utf8_X_V(pTHX_ const U8 *p)
1772{
1773 dVAR;
1774
1775 PERL_ARGS_ASSERT_IS_UTF8_X_V;
1776
1777 return is_utf8_common(p, &PL_utf8_X_V, "GCB=V");
1778}
1779
1780bool
1781Perl_is_utf8_X_LV_LVT_V(pTHX_ const U8 *p)
1782{
1783 dVAR;
1784
1785 PERL_ARGS_ASSERT_IS_UTF8_X_LV_LVT_V;
1786
1787 return is_utf8_common(p, &PL_utf8_X_LV_LVT_V, "_X_LV_LVT_V");
1788}
1789
6b5c0936 1790/*
87cea99e 1791=for apidoc to_utf8_case
6b5c0936
JH
1792
1793The "p" contains the pointer to the UTF-8 string encoding
1794the character that is being converted.
1795
1796The "ustrp" is a pointer to the character buffer to put the
1797conversion result to. The "lenp" is a pointer to the length
1798of the result.
1799
0134edef 1800The "swashp" is a pointer to the swash to use.
6b5c0936 1801
36bb2ab6 1802Both the special and normal mappings are stored in lib/unicore/To/Foo.pl,
8fe4d5b2 1803and loaded by SWASHNEW, using lib/utf8_heavy.pl. The special (usually,
0134edef 1804but not always, a multicharacter mapping), is tried first.
6b5c0936 1805
0134edef
JH
1806The "special" is a string like "utf8::ToSpecLower", which means the
1807hash %utf8::ToSpecLower. The access to the hash is through
1808Perl_to_utf8_case().
6b5c0936 1809
0134edef
JH
1810The "normal" is a string like "ToLower" which means the swash
1811%utf8::ToLower.
1812
1813=cut */
6b5c0936 1814
2104c8d9 1815UV
9a957fbc
AL
1816Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp,
1817 SV **swashp, const char *normal, const char *special)
a0ed51b3 1818{
97aff369 1819 dVAR;
89ebb4a3 1820 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
0134edef 1821 STRLEN len = 0;
aec46f14 1822 const UV uv0 = utf8_to_uvchr(p, NULL);
1feea2c7
JH
1823 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1824 * are necessary in EBCDIC, they are redundant no-ops
1825 * in ASCII-ish platforms, and hopefully optimized away. */
f54cb97a 1826 const UV uv1 = NATIVE_TO_UNI(uv0);
7918f24d
NC
1827
1828 PERL_ARGS_ASSERT_TO_UTF8_CASE;
1829
9ae3ac1a
KW
1830 /* Note that swash_fetch() doesn't output warnings for these because it
1831 * assumes we will */
8457b38f 1832 if (uv1 >= UNICODE_SURROGATE_FIRST) {
9ae3ac1a 1833 if (uv1 <= UNICODE_SURROGATE_LAST) {
8457b38f
KW
1834 if (ckWARN_d(WARN_SURROGATE)) {
1835 const char* desc = (PL_op) ? OP_DESC(PL_op) : normal;
1836 Perl_warner(aTHX_ packWARN(WARN_SURROGATE),
1837 "Operation \"%s\" returns its argument for UTF-16 surrogate U+%04"UVXf"", desc, uv1);
1838 }
9ae3ac1a
KW
1839 }
1840 else if (UNICODE_IS_SUPER(uv1)) {
8457b38f
KW
1841 if (ckWARN_d(WARN_NON_UNICODE)) {
1842 const char* desc = (PL_op) ? OP_DESC(PL_op) : normal;
1843 Perl_warner(aTHX_ packWARN(WARN_NON_UNICODE),
1844 "Operation \"%s\" returns its argument for non-Unicode code point 0x%04"UVXf"", desc, uv1);
1845 }
9ae3ac1a
KW
1846 }
1847
1848 /* Note that non-characters are perfectly legal, so no warning should
1849 * be given */
1850 }
1851
1feea2c7 1852 uvuni_to_utf8(tmpbuf, uv1);
0134edef
JH
1853
1854 if (!*swashp) /* load on-demand */
1855 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
37e2e78e
KW
1856 /* This is the beginnings of a skeleton of code to read the info section
1857 * that is in all the swashes in case we ever want to do that, so one can
1858 * read things whose maps aren't code points, and whose default if missing
1859 * is not to the code point itself. This was just to see if it actually
1860 * worked. Details on what the possibilities are are in perluniprops.pod
1861 HV * const hv = get_hv("utf8::SwashInfo", 0);
1862 if (hv) {
1863 SV **svp;
1864 svp = hv_fetch(hv, (const char*)normal, strlen(normal), FALSE);
1865 const char *s;
1866
1867 HV * const this_hash = SvRV(*svp);
1868 svp = hv_fetch(this_hash, "type", strlen("type"), FALSE);
1869 s = SvPV_const(*svp, len);
1870 }
1871 }*/
0134edef 1872
a6f87d8c 1873 if (special) {
0134edef 1874 /* It might be "special" (sometimes, but not always,
2a37f04d 1875 * a multicharacter mapping) */
6673a63c 1876 HV * const hv = get_hv(special, 0);
b08cf34e
JH
1877 SV **svp;
1878
35da51f7 1879 if (hv &&
b08cf34e
JH
1880 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1881 (*svp)) {
cfd0369c 1882 const char *s;
47654450 1883
cfd0369c 1884 s = SvPV_const(*svp, len);
47654450
JH
1885 if (len == 1)
1886 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
2a37f04d 1887 else {
2f9475ad
JH
1888#ifdef EBCDIC
1889 /* If we have EBCDIC we need to remap the characters
1890 * since any characters in the low 256 are Unicode
1891 * code points, not EBCDIC. */
7cda7a3d 1892 U8 *t = (U8*)s, *tend = t + len, *d;
2f9475ad
JH
1893
1894 d = tmpbuf;
b08cf34e 1895 if (SvUTF8(*svp)) {
2f9475ad
JH
1896 STRLEN tlen = 0;
1897
1898 while (t < tend) {
d4c19fe8 1899 const UV c = utf8_to_uvchr(t, &tlen);
2f9475ad
JH
1900 if (tlen > 0) {
1901 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1902 t += tlen;
1903 }
1904 else
1905 break;
1906 }
1907 }
1908 else {
36fec512
JH
1909 while (t < tend) {
1910 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1911 t++;
1912 }
2f9475ad
JH
1913 }
1914 len = d - tmpbuf;
1915 Copy(tmpbuf, ustrp, len, U8);
1916#else
d2dcd0fb 1917 Copy(s, ustrp, len, U8);
2f9475ad 1918#endif
29e98929 1919 }
983ffd37 1920 }
0134edef
JH
1921 }
1922
1923 if (!len && *swashp) {
d4c19fe8
AL
1924 const UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1925
0134edef
JH
1926 if (uv2) {
1927 /* It was "normal" (a single character mapping). */
d4c19fe8 1928 const UV uv3 = UNI_TO_NATIVE(uv2);
e9101d72 1929 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
2a37f04d
JH
1930 }
1931 }
1feea2c7 1932
37e2e78e
KW
1933 if (!len) /* Neither: just copy. In other words, there was no mapping
1934 defined, which means that the code point maps to itself */
0134edef
JH
1935 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1936
2a37f04d
JH
1937 if (lenp)
1938 *lenp = len;
1939
0134edef 1940 return len ? utf8_to_uvchr(ustrp, 0) : 0;
a0ed51b3
LW
1941}
1942
d3e79532 1943/*
87cea99e 1944=for apidoc to_utf8_upper
d3e79532
JH
1945
1946Convert the UTF-8 encoded character at p to its uppercase version and
1947store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
1948that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1949the uppercase version may be longer than the original character.
d3e79532
JH
1950
1951The first character of the uppercased version is returned
1952(but note, as explained above, that there may be more.)
1953
1954=cut */
1955
2104c8d9 1956UV
7fc63493 1957Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
a0ed51b3 1958{
97aff369 1959 dVAR;
7918f24d
NC
1960
1961 PERL_ARGS_ASSERT_TO_UTF8_UPPER;
1962
983ffd37 1963 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
b4e400f9 1964 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
983ffd37 1965}
a0ed51b3 1966
d3e79532 1967/*
87cea99e 1968=for apidoc to_utf8_title
d3e79532
JH
1969
1970Convert the UTF-8 encoded character at p to its titlecase version and
1971store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
1972that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1973titlecase version may be longer than the original character.
d3e79532
JH
1974
1975The first character of the titlecased version is returned
1976(but note, as explained above, that there may be more.)
1977
1978=cut */
1979
983ffd37 1980UV
7fc63493 1981Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
983ffd37 1982{
97aff369 1983 dVAR;
7918f24d
NC
1984
1985 PERL_ARGS_ASSERT_TO_UTF8_TITLE;
1986
983ffd37 1987 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
b4e400f9 1988 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
a0ed51b3
LW
1989}
1990
d3e79532 1991/*
87cea99e 1992=for apidoc to_utf8_lower
d3e79532
JH
1993
1994Convert the UTF-8 encoded character at p to its lowercase version and
1995store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
1996that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
1997lowercase version may be longer than the original character.
d3e79532
JH
1998
1999The first character of the lowercased version is returned
2000(but note, as explained above, that there may be more.)
2001
2002=cut */
2003
2104c8d9 2004UV
7fc63493 2005Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
a0ed51b3 2006{
97aff369 2007 dVAR;
7918f24d
NC
2008
2009 PERL_ARGS_ASSERT_TO_UTF8_LOWER;
2010
983ffd37 2011 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
b4e400f9
JH
2012 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
2013}
2014
d3e79532 2015/*
87cea99e 2016=for apidoc to_utf8_fold
d3e79532
JH
2017
2018Convert the UTF-8 encoded character at p to its foldcase version and
2019store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3 2020that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
d3e79532
JH
2021foldcase version may be longer than the original character (up to
2022three characters).
2023
2024The first character of the foldcased version is returned
2025(but note, as explained above, that there may be more.)
2026
2027=cut */
2028
36bb2ab6
KW
2029/* Not currently externally documented is 'flags', which currently is non-zero
2030 * if full case folds are to be used; otherwise simple folds */
2031
b4e400f9 2032UV
36bb2ab6 2033Perl__to_utf8_fold_flags(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp, U8 flags)
b4e400f9 2034{
36bb2ab6
KW
2035 const char *specials = (flags) ? "utf8::ToSpecFold" : NULL;
2036
97aff369 2037 dVAR;
7918f24d 2038
36bb2ab6 2039 PERL_ARGS_ASSERT__TO_UTF8_FOLD_FLAGS;
7918f24d 2040
b4e400f9 2041 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
36bb2ab6 2042 &PL_utf8_tofold, "ToFold", specials);
a0ed51b3
LW
2043}
2044
711a919c
TS
2045/* Note:
2046 * A "swash" is a swatch hash.
2047 * A "swatch" is a bit vector generated by utf8.c:S_swash_get().
2048 * C<pkg> is a pointer to a package name for SWASHNEW, should be "utf8".
2049 * For other parameters, see utf8::SWASHNEW in lib/utf8_heavy.pl.
2050 */
a0ed51b3 2051SV*
7fc63493 2052Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
a0ed51b3 2053{
27da23d5 2054 dVAR;
a0ed51b3 2055 SV* retval;
8e84507e 2056 dSP;
7fc63493
AL
2057 const size_t pkg_len = strlen(pkg);
2058 const size_t name_len = strlen(name);
da51bb9b 2059 HV * const stash = gv_stashpvn(pkg, pkg_len, 0);
f8be5cf0 2060 SV* errsv_save;
88c57d91 2061 GV *method;
ce3b816e 2062
7918f24d
NC
2063 PERL_ARGS_ASSERT_SWASH_INIT;
2064
96ca9f55
DM
2065 PUSHSTACKi(PERLSI_MAGIC);
2066 ENTER;
ec34a119 2067 SAVEHINTS();
96ca9f55 2068 save_re_context();
88c57d91
NC
2069 method = gv_fetchmeth(stash, "SWASHNEW", 8, -1);
2070 if (!method) { /* demand load utf8 */
ce3b816e 2071 ENTER;
f8be5cf0 2072 errsv_save = newSVsv(ERRSV);
dc0c6abb
NC
2073 /* It is assumed that callers of this routine are not passing in any
2074 user derived data. */
2075 /* Need to do this after save_re_context() as it will set PL_tainted to
2076 1 while saving $1 etc (see the code after getrx: in Perl_magic_get).
2077 Even line to create errsv_save can turn on PL_tainted. */
2078 SAVEBOOL(PL_tainted);
2079 PL_tainted = 0;
71bed85a 2080 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
a0714e2c 2081 NULL);
f8be5cf0
JH
2082 if (!SvTRUE(ERRSV))
2083 sv_setsv(ERRSV, errsv_save);
2084 SvREFCNT_dec(errsv_save);
ce3b816e
GS
2085 LEAVE;
2086 }
2087 SPAGAIN;
a0ed51b3
LW
2088 PUSHMARK(SP);
2089 EXTEND(SP,5);
6e449a3a
MHM
2090 mPUSHp(pkg, pkg_len);
2091 mPUSHp(name, name_len);
a0ed51b3 2092 PUSHs(listsv);
6e449a3a
MHM
2093 mPUSHi(minbits);
2094 mPUSHi(none);
a0ed51b3 2095 PUTBACK;
f8be5cf0 2096 errsv_save = newSVsv(ERRSV);
88c57d91
NC
2097 /* If we already have a pointer to the method, no need to use call_method()
2098 to repeat the lookup. */
2099 if (method ? call_sv(MUTABLE_SV(method), G_SCALAR)
8b563eef 2100 : call_sv(newSVpvs_flags("SWASHNEW", SVs_TEMP), G_SCALAR | G_METHOD))
8e84507e 2101 retval = newSVsv(*PL_stack_sp--);
a0ed51b3 2102 else
e24b16f9 2103 retval = &PL_sv_undef;
f8be5cf0
JH
2104 if (!SvTRUE(ERRSV))
2105 sv_setsv(ERRSV, errsv_save);
2106 SvREFCNT_dec(errsv_save);
a0ed51b3
LW
2107 LEAVE;
2108 POPSTACK;
923e4eb5 2109 if (IN_PERL_COMPILETIME) {
623e6609 2110 CopHINTS_set(PL_curcop, PL_hints);
a0ed51b3 2111 }
bc45ce41
JH
2112 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
2113 if (SvPOK(retval))
35c1215d 2114 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
be2597df 2115 SVfARG(retval));
cea2e8a9 2116 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
bc45ce41 2117 }
a0ed51b3
LW
2118 return retval;
2119}
2120
035d37be
JH
2121
2122/* This API is wrong for special case conversions since we may need to
2123 * return several Unicode characters for a single Unicode character
2124 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
2125 * the lower-level routine, and it is similarly broken for returning
38684baa
KW
2126 * multiple values. --jhi
2127 * For those, you should use to_utf8_case() instead */
979f2922 2128/* Now SWASHGET is recasted into S_swash_get in this file. */
680c470c
TS
2129
2130/* Note:
2131 * Returns the value of property/mapping C<swash> for the first character
2132 * of the string C<ptr>. If C<do_utf8> is true, the string C<ptr> is
2133 * assumed to be in utf8. If C<do_utf8> is false, the string C<ptr> is
2134 * assumed to be in native 8-bit encoding. Caches the swatch in C<swash>.
2135 */
a0ed51b3 2136UV
680c470c 2137Perl_swash_fetch(pTHX_ SV *swash, const U8 *ptr, bool do_utf8)
a0ed51b3 2138{
27da23d5 2139 dVAR;
ef8f7699 2140 HV *const hv = MUTABLE_HV(SvRV(swash));
3568d838
JH
2141 U32 klen;
2142 U32 off;
a0ed51b3 2143 STRLEN slen;
7d85a32c 2144 STRLEN needents;
cfd0369c 2145 const U8 *tmps = NULL;
a0ed51b3 2146 U32 bit;
979f2922 2147 SV *swatch;
3568d838 2148 U8 tmputf8[2];
35da51f7 2149 const UV c = NATIVE_TO_ASCII(*ptr);
3568d838 2150
7918f24d
NC
2151 PERL_ARGS_ASSERT_SWASH_FETCH;
2152
3568d838 2153 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
979f2922
TS
2154 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
2155 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
2156 ptr = tmputf8;
3568d838
JH
2157 }
2158 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
37e2e78e 2159 * then the "swatch" is a vec() for all the chars which start
3568d838
JH
2160 * with 0xAA..0xYY
2161 * So the key in the hash (klen) is length of encoded char -1
2162 */
2163 klen = UTF8SKIP(ptr) - 1;
2164 off = ptr[klen];
a0ed51b3 2165
979f2922 2166 if (klen == 0) {
37e2e78e 2167 /* If char is invariant then swatch is for all the invariant chars
1e54db1a 2168 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
7d85a32c 2169 */
979f2922
TS
2170 needents = UTF_CONTINUATION_MARK;
2171 off = NATIVE_TO_UTF(ptr[klen]);
2172 }
2173 else {
7d85a32c 2174 /* If char is encoded then swatch is for the prefix */
979f2922
TS
2175 needents = (1 << UTF_ACCUMULATION_SHIFT);
2176 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
8457b38f 2177 if (UTF8_IS_SUPER(ptr) && ckWARN_d(WARN_NON_UNICODE)) {
9ae3ac1a
KW
2178 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0, 0);
2179
2180 /* This outputs warnings for binary properties only, assuming that
2181 * to_utf8_case() will output any. Also, surrogates aren't checked
2182 * for, as that would warn on things like /\p{Gc=Cs}/ */
2183 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2184 if (SvUV(*bitssvp) == 1) {
8457b38f 2185 Perl_warner(aTHX_ packWARN(WARN_NON_UNICODE),
9ae3ac1a
KW
2186 "Code point 0x%04"UVXf" is not Unicode, no properties match it; all inverse properties do", code_point);
2187 }
2188 }
979f2922 2189 }
7d85a32c 2190
a0ed51b3
LW
2191 /*
2192 * This single-entry cache saves about 1/3 of the utf8 overhead in test
2193 * suite. (That is, only 7-8% overall over just a hash cache. Still,
2194 * it's nothing to sniff at.) Pity we usually come through at least
2195 * two function calls to get here...
2196 *
2197 * NB: this code assumes that swatches are never modified, once generated!
2198 */
2199
3568d838 2200 if (hv == PL_last_swash_hv &&
a0ed51b3 2201 klen == PL_last_swash_klen &&
27da23d5 2202 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
a0ed51b3
LW
2203 {
2204 tmps = PL_last_swash_tmps;
2205 slen = PL_last_swash_slen;
2206 }
2207 else {
2208 /* Try our second-level swatch cache, kept in a hash. */
e1ec3a88 2209 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
a0ed51b3 2210
979f2922
TS
2211 /* If not cached, generate it via swash_get */
2212 if (!svp || !SvPOK(*svp)
2213 || !(tmps = (const U8*)SvPV_const(*svp, slen))) {
2b9d42f0
NIS
2214 /* We use utf8n_to_uvuni() as we want an index into
2215 Unicode tables, not a native character number.
2216 */
aec46f14 2217 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
872c91ae
JH
2218 ckWARN(WARN_UTF8) ?
2219 0 : UTF8_ALLOW_ANY);
680c470c 2220 swatch = swash_get(swash,
979f2922
TS
2221 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
2222 (klen) ? (code_point & ~(needents - 1)) : 0,
2223 needents);
2224
923e4eb5 2225 if (IN_PERL_COMPILETIME)
623e6609 2226 CopHINTS_set(PL_curcop, PL_hints);
a0ed51b3 2227
979f2922 2228 svp = hv_store(hv, (const char *)ptr, klen, swatch, 0);
a0ed51b3 2229
979f2922
TS
2230 if (!svp || !(tmps = (U8*)SvPV(*svp, slen))
2231 || (slen << 3) < needents)
660a4616 2232 Perl_croak(aTHX_ "panic: swash_fetch got improper swatch");
a0ed51b3
LW
2233 }
2234
2235 PL_last_swash_hv = hv;
16d8f38a 2236 assert(klen <= sizeof(PL_last_swash_key));
eac04b2e 2237 PL_last_swash_klen = (U8)klen;
cfd0369c
NC
2238 /* FIXME change interpvar.h? */
2239 PL_last_swash_tmps = (U8 *) tmps;
a0ed51b3
LW
2240 PL_last_swash_slen = slen;
2241 if (klen)
2242 Copy(ptr, PL_last_swash_key, klen, U8);
2243 }
2244
9faf8d75 2245 switch ((int)((slen << 3) / needents)) {
a0ed51b3
LW
2246 case 1:
2247 bit = 1 << (off & 7);
2248 off >>= 3;
2249 return (tmps[off] & bit) != 0;
2250 case 8:
2251 return tmps[off];
2252 case 16:
2253 off <<= 1;
2254 return (tmps[off] << 8) + tmps[off + 1] ;
2255 case 32:
2256 off <<= 2;
2257 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
2258 }
660a4616 2259 Perl_croak(aTHX_ "panic: swash_fetch got swatch of unexpected bit width");
670f1322 2260 NORETURN_FUNCTION_END;
a0ed51b3 2261}
2b9d42f0 2262
319009ee
KW
2263/* Read a single line of the main body of the swash input text. These are of
2264 * the form:
2265 * 0053 0056 0073
2266 * where each number is hex. The first two numbers form the minimum and
2267 * maximum of a range, and the third is the value associated with the range.
2268 * Not all swashes should have a third number
2269 *
2270 * On input: l points to the beginning of the line to be examined; it points
2271 * to somewhere in the string of the whole input text, and is
2272 * terminated by a \n or the null string terminator.
2273 * lend points to the null terminator of that string
2274 * wants_value is non-zero if the swash expects a third number
2275 * typestr is the name of the swash's mapping, like 'ToLower'
2276 * On output: *min, *max, and *val are set to the values read from the line.
2277 * returns a pointer just beyond the line examined. If there was no
2278 * valid min number on the line, returns lend+1
2279 */
2280
2281STATIC U8*
2282S_swash_scan_list_line(pTHX_ U8* l, U8* const lend, UV* min, UV* max, UV* val,
2283 const bool wants_value, const U8* const typestr)
2284{
2285 const int typeto = typestr[0] == 'T' && typestr[1] == 'o';
2286 STRLEN numlen; /* Length of the number */
2287 I32 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
2288
2289 /* nl points to the next \n in the scan */
2290 U8* const nl = (U8*)memchr(l, '\n', lend - l);
2291
2292 /* Get the first number on the line: the range minimum */
2293 numlen = lend - l;
2294 *min = grok_hex((char *)l, &numlen, &flags, NULL);
2295 if (numlen) /* If found a hex number, position past it */
2296 l += numlen;
2297 else if (nl) { /* Else, go handle next line, if any */
2298 return nl + 1; /* 1 is length of "\n" */
2299 }
2300 else { /* Else, no next line */
2301 return lend + 1; /* to LIST's end at which \n is not found */
2302 }
2303
2304 /* The max range value follows, separated by a BLANK */
2305 if (isBLANK(*l)) {
2306 ++l;
2307 flags = PERL_SCAN_SILENT_ILLDIGIT | PERL_SCAN_DISALLOW_PREFIX;
2308 numlen = lend - l;
2309 *max = grok_hex((char *)l, &numlen, &flags, NULL);
2310 if (numlen)
2311 l += numlen;
2312 else /* If no value here, it is a single element range */
2313 *max = *min;
2314
2315 /* Non-binary tables have a third entry: what the first element of the
2316 * range maps to */
2317 if (wants_value) {
2318 if (isBLANK(*l)) {
2319 ++l;
2320 flags = PERL_SCAN_SILENT_ILLDIGIT |
2321 PERL_SCAN_DISALLOW_PREFIX;
2322 numlen = lend - l;
2323 *val = grok_hex((char *)l, &numlen, &flags, NULL);
2324 if (numlen)
2325 l += numlen;
2326 else
2327 *val = 0;
2328 }
2329 else {
2330 *val = 0;
2331 if (typeto) {
2332 Perl_croak(aTHX_ "%s: illegal mapping '%s'",
2333 typestr, l);
2334 }
2335 }
2336 }
2337 else
2338 *val = 0; /* bits == 1, then any val should be ignored */
2339 }
2340 else { /* Nothing following range min, should be single element with no
2341 mapping expected */
2342 *max = *min;
2343 if (wants_value) {
2344 *val = 0;
2345 if (typeto) {
2346 Perl_croak(aTHX_ "%s: illegal mapping '%s'", typestr, l);
2347 }
2348 }
2349 else
2350 *val = 0; /* bits == 1, then val should be ignored */
2351 }
2352
2353 /* Position to next line if any, or EOF */
2354 if (nl)
2355 l = nl + 1;
2356 else
2357 l = lend;
2358
2359 return l;
2360}
2361
979f2922
TS
2362/* Note:
2363 * Returns a swatch (a bit vector string) for a code point sequence
2364 * that starts from the value C<start> and comprises the number C<span>.
2365 * A C<swash> must be an object created by SWASHNEW (see lib/utf8_heavy.pl).
2366 * Should be used via swash_fetch, which will cache the swatch in C<swash>.
2367 */
2368STATIC SV*
2369S_swash_get(pTHX_ SV* swash, UV start, UV span)
2370{
2371 SV *swatch;
711a919c 2372 U8 *l, *lend, *x, *xend, *s;
979f2922 2373 STRLEN lcur, xcur, scur;
ef8f7699 2374 HV *const hv = MUTABLE_HV(SvRV(swash));
972dd592
KW
2375
2376 /* The string containing the main body of the table */
017a3ce5 2377 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
972dd592 2378
017a3ce5
GA
2379 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2380 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2381 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2382 SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);
0bd48802 2383 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
0bd48802
AL
2384 const STRLEN bits = SvUV(*bitssvp);
2385 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2386 const UV none = SvUV(*nonesvp);
2387 const UV end = start + span;
979f2922 2388
7918f24d
NC
2389 PERL_ARGS_ASSERT_SWASH_GET;
2390
979f2922 2391 if (bits != 1 && bits != 8 && bits != 16 && bits != 32) {
660a4616
TS
2392 Perl_croak(aTHX_ "panic: swash_get doesn't expect bits %"UVuf,
2393 (UV)bits);
979f2922
TS
2394 }
2395
2396 /* create and initialize $swatch */
979f2922 2397 scur = octets ? (span * octets) : (span + 7) / 8;
e524fe40
NC
2398 swatch = newSV(scur);
2399 SvPOK_on(swatch);
979f2922
TS
2400 s = (U8*)SvPVX(swatch);
2401 if (octets && none) {
0bd48802 2402 const U8* const e = s + scur;
979f2922
TS
2403 while (s < e) {
2404 if (bits == 8)
2405 *s++ = (U8)(none & 0xff);
2406 else if (bits == 16) {
2407 *s++ = (U8)((none >> 8) & 0xff);
2408 *s++ = (U8)( none & 0xff);
2409 }
2410 else if (bits == 32) {
2411 *s++ = (U8)((none >> 24) & 0xff);
2412 *s++ = (U8)((none >> 16) & 0xff);
2413 *s++ = (U8)((none >> 8) & 0xff);
2414 *s++ = (U8)( none & 0xff);
2415 }
2416 }
2417 *s = '\0';
2418 }
2419 else {
2420 (void)memzero((U8*)s, scur + 1);
2421 }
2422 SvCUR_set(swatch, scur);
2423 s = (U8*)SvPVX(swatch);
2424
2425 /* read $swash->{LIST} */
2426 l = (U8*)SvPV(*listsvp, lcur);
2427 lend = l + lcur;
2428 while (l < lend) {
35da51f7 2429 UV min, max, val;
319009ee
KW
2430 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2431 cBOOL(octets), typestr);
2432 if (l > lend) {
979f2922
TS
2433 break;
2434 }
2435
972dd592 2436 /* If looking for something beyond this range, go try the next one */
979f2922
TS
2437 if (max < start)
2438 continue;
2439
2440 if (octets) {
35da51f7 2441 UV key;
979f2922
TS
2442 if (min < start) {
2443 if (!none || val < none) {
2444 val += start - min;
2445 }
2446 min = start;
2447 }
2448 for (key = min; key <= max; key++) {
2449 STRLEN offset;
2450 if (key >= end)
2451 goto go_out_list;
2452 /* offset must be non-negative (start <= min <= key < end) */
2453 offset = octets * (key - start);
2454 if (bits == 8)
2455 s[offset] = (U8)(val & 0xff);
2456 else if (bits == 16) {
2457 s[offset ] = (U8)((val >> 8) & 0xff);
2458 s[offset + 1] = (U8)( val & 0xff);
2459 }
2460 else if (bits == 32) {
2461 s[offset ] = (U8)((val >> 24) & 0xff);
2462 s[offset + 1] = (U8)((val >> 16) & 0xff);
2463 s[offset + 2] = (U8)((val >> 8) & 0xff);
2464 s[offset + 3] = (U8)( val & 0xff);
2465 }
2466
2467 if (!none || val < none)
2468 ++val;
2469 }
2470 }
711a919c 2471 else { /* bits == 1, then val should be ignored */
35da51f7 2472 UV key;
979f2922
TS
2473 if (min < start)
2474 min = start;
2475 for (key = min; key <= max; key++) {
0bd48802 2476 const STRLEN offset = (STRLEN)(key - start);
979f2922
TS
2477 if (key >= end)
2478 goto go_out_list;
2479 s[offset >> 3] |= 1 << (offset & 7);
2480 }
2481 }
2482 } /* while */
2483 go_out_list:
2484
2485 /* read $swash->{EXTRAS} */
2486 x = (U8*)SvPV(*extssvp, xcur);
2487 xend = x + xcur;
2488 while (x < xend) {
2489 STRLEN namelen;
2490 U8 *namestr;
2491 SV** othersvp;
2492 HV* otherhv;
2493 STRLEN otherbits;
2494 SV **otherbitssvp, *other;
711a919c 2495 U8 *s, *o, *nl;
979f2922
TS
2496 STRLEN slen, olen;
2497
35da51f7 2498 const U8 opc = *x++;
979f2922
TS
2499 if (opc == '\n')
2500 continue;
2501
2502 nl = (U8*)memchr(x, '\n', xend - x);
2503
2504 if (opc != '-' && opc != '+' && opc != '!' && opc != '&') {
2505 if (nl) {
2506 x = nl + 1; /* 1 is length of "\n" */
2507 continue;
2508 }
2509 else {
2510 x = xend; /* to EXTRAS' end at which \n is not found */
2511 break;
2512 }
2513 }
2514
2515 namestr = x;
2516 if (nl) {
2517 namelen = nl - namestr;
2518 x = nl + 1;
2519 }
2520 else {
2521 namelen = xend - namestr;
2522 x = xend;
2523 }
2524
2525 othersvp = hv_fetch(hv, (char *)namestr, namelen, FALSE);
ef8f7699 2526 otherhv = MUTABLE_HV(SvRV(*othersvp));
017a3ce5 2527 otherbitssvp = hv_fetchs(otherhv, "BITS", FALSE);
979f2922
TS
2528 otherbits = (STRLEN)SvUV(*otherbitssvp);
2529 if (bits < otherbits)
660a4616 2530 Perl_croak(aTHX_ "panic: swash_get found swatch size mismatch");
979f2922
TS
2531
2532 /* The "other" swatch must be destroyed after. */
2533 other = swash_get(*othersvp, start, span);
2534 o = (U8*)SvPV(other, olen);
2535
2536 if (!olen)
660a4616 2537 Perl_croak(aTHX_ "panic: swash_get got improper swatch");
979f2922
TS
2538
2539 s = (U8*)SvPV(swatch, slen);
2540 if (bits == 1 && otherbits == 1) {
2541 if (slen != olen)
660a4616 2542 Perl_croak(aTHX_ "panic: swash_get found swatch length mismatch");
979f2922
TS
2543
2544 switch (opc) {
2545 case '+':
2546 while (slen--)
2547 *s++ |= *o++;
2548 break;
2549 case '!':
2550 while (slen--)
2551 *s++ |= ~*o++;
2552 break;
2553 case '-':
2554 while (slen--)
2555 *s++ &= ~*o++;
2556 break;
2557 case '&':
2558 while (slen--)
2559 *s++ &= *o++;
2560 break;
2561 default:
2562 break;
2563 }
2564 }
711a919c 2565 else {
979f2922
TS
2566 STRLEN otheroctets = otherbits >> 3;
2567 STRLEN offset = 0;
35da51f7 2568 U8* const send = s + slen;
979f2922
TS
2569
2570 while (s < send) {
2571 UV otherval = 0;
2572
2573 if (otherbits == 1) {
2574 otherval = (o[offset >> 3] >> (offset & 7)) & 1;
2575 ++offset;
2576 }
2577 else {
2578 STRLEN vlen = otheroctets;
2579 otherval = *o++;
2580 while (--vlen) {
2581 otherval <<= 8;
2582 otherval |= *o++;
2583 }
2584 }
2585
711a919c 2586 if (opc == '+' && otherval)
6f207bd3 2587 NOOP; /* replace with otherval */
979f2922
TS
2588 else if (opc == '!' && !otherval)
2589 otherval = 1;
2590 else if (opc == '-' && otherval)
2591 otherval = 0;
2592 else if (opc == '&' && !otherval)
2593 otherval = 0;
2594 else {
711a919c 2595 s += octets; /* no replacement */
979f2922
TS
2596 continue;
2597 }
2598
2599 if (bits == 8)
2600 *s++ = (U8)( otherval & 0xff);
2601 else if (bits == 16) {
2602 *s++ = (U8)((otherval >> 8) & 0xff);
2603 *s++ = (U8)( otherval & 0xff);
2604 }
2605 else if (bits == 32) {
2606 *s++ = (U8)((otherval >> 24) & 0xff);
2607 *s++ = (U8)((otherval >> 16) & 0xff);
2608 *s++ = (U8)((otherval >> 8) & 0xff);
2609 *s++ = (U8)( otherval & 0xff);
2610 }
2611 }
2612 }
2613 sv_free(other); /* through with it! */
2614 } /* while */
2615 return swatch;
2616}
2617
064c021d 2618HV*
4c2e1131 2619Perl__swash_inversion_hash(pTHX_ SV* const swash)
064c021d
KW
2620{
2621
2622 /* Subject to change or removal. For use only in one place in regexec.c
2623 *
2624 * Returns a hash which is the inversion and closure of a swash mapping.
2625 * For example, consider the input lines:
2626 * 004B 006B
2627 * 004C 006C
2628 * 212A 006B
2629 *
2630 * The returned hash would have two keys, the utf8 for 006B and the utf8 for
2631 * 006C. The value for each key is an array. For 006C, the array would
2632 * have a two elements, the utf8 for itself, and for 004C. For 006B, there
2633 * would be three elements in its array, the utf8 for 006B, 004B and 212A.
2634 *
2635 * Essentially, for any code point, it gives all the code points that map to
2636 * it, or the list of 'froms' for that point.
2637 *
2638 * Currently it only looks at the main body of the swash, and ignores any
2639 * additions or deletions from other swashes */
2640
2641 U8 *l, *lend;
2642 STRLEN lcur;
2643 HV *const hv = MUTABLE_HV(SvRV(swash));
2644
2645 /* The string containing the main body of the table */
2646 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2647
2648 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2649 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2650 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2651 /*SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);*/
2652 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
2653 const STRLEN bits = SvUV(*bitssvp);
2654 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2655 const UV none = SvUV(*nonesvp);
2656
2657 HV* ret = newHV();
2658
2659 PERL_ARGS_ASSERT__SWASH_INVERSION_HASH;
2660
2661 /* Must have at least 8 bits to get the mappings */
2662 if (bits != 8 && bits != 16 && bits != 32) {
2663 Perl_croak(aTHX_ "panic: swash_inversion_hash doesn't expect bits %"UVuf,
2664 (UV)bits);
2665 }
2666
2667 /* read $swash->{LIST} */
2668 l = (U8*)SvPV(*listsvp, lcur);
2669 lend = l + lcur;
2670
2671 /* Go through each input line */
2672 while (l < lend) {
2673 UV min, max, val;
2674 UV inverse;
2675 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2676 cBOOL(octets), typestr);
2677 if (l > lend) {
2678 break;
2679 }
2680
2681 /* Each element in the range is to be inverted */
2682 for (inverse = min; inverse <= max; inverse++) {
2683 AV* list;
2684 SV* element;
2685 SV** listp;
2686 IV i;
2687 bool found_key = FALSE;
2688
2689 /* The key is the inverse mapping */
2690 char key[UTF8_MAXBYTES+1];
2691 char* key_end = (char *) uvuni_to_utf8((U8*) key, val);
2692 STRLEN key_len = key_end - key;
2693
064c021d
KW
2694 /* Get the list for the map */
2695 if ((listp = hv_fetch(ret, key, key_len, FALSE))) {
2696 list = (AV*) *listp;
2697 }
2698 else { /* No entry yet for it: create one */
2699 list = newAV();
2700 if (! hv_store(ret, key, key_len, (SV*) list, FALSE)) {
2701 Perl_croak(aTHX_ "panic: hv_store() unexpectedly failed");
2702 }
2703 }
2704
508f7cfa 2705 for (i = 0; i <= av_len(list); i++) {
064c021d
KW
2706 SV** entryp = av_fetch(list, i, FALSE);
2707 SV* entry;
2708 if (entryp == NULL) {
2709 Perl_croak(aTHX_ "panic: av_fetch() unexpectedly failed");
2710 }
2711 entry = *entryp;
56ca34ca 2712 if (SvUV(entry) == val) {
064c021d
KW
2713 found_key = TRUE;
2714 break;
2715 }
2716 }
56ca34ca
KW
2717
2718 /* Make sure there is a mapping to itself on the list */
064c021d 2719 if (! found_key) {
56ca34ca 2720 element = newSVuv(val);
064c021d
KW
2721 av_push(list, element);
2722 }
2723
2724
2725 /* Simply add the value to the list */
56ca34ca 2726 element = newSVuv(inverse);
064c021d
KW
2727 av_push(list, element);
2728
2729 /* swash_get() increments the value of val for each element in the
2730 * range. That makes more compact tables possible. You can
2731 * express the capitalization, for example, of all consecutive
2732 * letters with a single line: 0061\t007A\t0041 This maps 0061 to
2733 * 0041, 0062 to 0042, etc. I (khw) have never understood 'none',
2734 * and it's not documented, and perhaps not even currently used,
2735 * but I copied the semantics from swash_get(), just in case */
2736 if (!none || val < none) {
2737 ++val;
2738 }
2739 }
2740 }
2741
2742 return ret;
2743}
2744
d764b54e
KW
2745HV*
2746Perl__swash_to_invlist(pTHX_ SV* const swash)
2747{
2748
2749 /* Subject to change or removal. For use only in one place in regcomp.c */
2750
2751 U8 *l, *lend;
2752 char *loc;
2753 STRLEN lcur;
2754 HV *const hv = MUTABLE_HV(SvRV(swash));
2755 UV elements = 0; /* Number of elements in the inversion list */
b443038a 2756 U8 empty[] = "";
d764b54e
KW
2757
2758 /* The string containing the main body of the table */
2759 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2760 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2761 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2762
2763 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
2764 const STRLEN bits = SvUV(*bitssvp);
2765 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2766
2767 HV* invlist;
2768
2769 PERL_ARGS_ASSERT__SWASH_TO_INVLIST;
2770
2771 /* read $swash->{LIST} */
b443038a
KW
2772 if (SvPOK(*listsvp)) {
2773 l = (U8*)SvPV(*listsvp, lcur);
2774 }
2775 else {
2776 /* LIST legitimately doesn't contain a string during compilation phases
2777 * of Perl itself, before the Unicode tables are generated. In this
2778 * case, just fake things up by creating an empty list */
2779 l = empty;
2780 lcur = 0;
2781 }
d764b54e
KW
2782 loc = (char *) l;
2783 lend = l + lcur;
2784
2785 /* Scan the input to count the number of lines to preallocate array size
2786 * based on worst possible case, which is each line in the input creates 2
2787 * elements in the inversion list: 1) the beginning of a range in the list;
2788 * 2) the beginning of a range not in the list. */
2789 while ((loc = (strchr(loc, '\n'))) != NULL) {
2790 elements += 2;
2791 loc++;
2792 }
2793
2794 /* If the ending is somehow corrupt and isn't a new line, add another
2795 * element for the final range that isn't in the inversion list */
2796 if (! (*lend == '\n' || (*lend == '\0' && *(lend - 1) == '\n'))) {
2797 elements++;
2798 }
2799
2800 invlist = _new_invlist(elements);
2801
2802 /* Now go through the input again, adding each range to the list */
2803 while (l < lend) {
2804 UV start, end;
2805 UV val; /* Not used by this function */
2806
2807 l = S_swash_scan_list_line(aTHX_ l, lend, &start, &end, &val,
2808 cBOOL(octets), typestr);
2809
2810 if (l > lend) {
2811 break;
2812 }
2813
2814 _append_range_to_invlist(invlist, start, end);
2815 }
2816
2817 return invlist;
2818}
2819
0f830e0b 2820/*
87cea99e 2821=for apidoc uvchr_to_utf8
0f830e0b 2822
6ee84de2 2823Adds the UTF-8 representation of the Native code point C<uv> to the end
0f830e0b
NC
2824of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
2825bytes available. The return value is the pointer to the byte after the
2826end of the new character. In other words,
2827
2828 d = uvchr_to_utf8(d, uv);
2829
2830is the recommended wide native character-aware way of saying
2831
2832 *(d++) = uv;
2833
2834=cut
2835*/
2836
2837/* On ASCII machines this is normally a macro but we want a
2838 real function in case XS code wants it
2839*/
0f830e0b
NC
2840U8 *
2841Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
2842{
7918f24d
NC
2843 PERL_ARGS_ASSERT_UVCHR_TO_UTF8;
2844
0f830e0b
NC
2845 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
2846}
2847
b851fbc1
JH
2848U8 *
2849Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
2850{
7918f24d
NC
2851 PERL_ARGS_ASSERT_UVCHR_TO_UTF8_FLAGS;
2852
b851fbc1
JH
2853 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
2854}
2b9d42f0
NIS
2855
2856/*
87cea99e 2857=for apidoc utf8n_to_uvchr
0f830e0b 2858
48ef279e 2859Returns the native character value of the first character in the string
0f830e0b
NC
2860C<s>
2861which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
2862length, in bytes, of that character.
2863
6ee84de2 2864length and flags are the same as utf8n_to_uvuni().
0f830e0b
NC
2865
2866=cut
2867*/
2868/* On ASCII machines this is normally a macro but we want
2869 a real function in case XS code wants it
2870*/
0f830e0b 2871UV
48ef279e 2872Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen,
0f830e0b
NC
2873U32 flags)
2874{
2875 const UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
7918f24d
NC
2876
2877 PERL_ARGS_ASSERT_UTF8N_TO_UVCHR;
2878
0f830e0b
NC
2879 return UNI_TO_NATIVE(uv);
2880}
2881
0876b9a0
KW
2882bool
2883Perl_check_utf8_print(pTHX_ register const U8* s, const STRLEN len)
2884{
2885 /* May change: warns if surrogates, non-character code points, or
2886 * non-Unicode code points are in s which has length len. Returns TRUE if
2887 * none found; FALSE otherwise. The only other validity check is to make
2888 * sure that this won't exceed the string's length */
2889
2890 const U8* const e = s + len;
2891 bool ok = TRUE;
2892
2893 PERL_ARGS_ASSERT_CHECK_UTF8_PRINT;
2894
2895 while (s < e) {
2896 if (UTF8SKIP(s) > len) {
2897 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
2898 "%s in %s", unees, PL_op ? OP_DESC(PL_op) : "print");
2899 return FALSE;
2900 }
2901 if (*s >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE) {
2902 STRLEN char_len;
2903 if (UTF8_IS_SUPER(s)) {
8457b38f
KW
2904 if (ckWARN_d(WARN_NON_UNICODE)) {
2905 UV uv = utf8_to_uvchr(s, &char_len);
2906 Perl_warner(aTHX_ packWARN(WARN_NON_UNICODE),
2907 "Code point 0x%04"UVXf" is not Unicode, may not be portable", uv);
2908 ok = FALSE;
2909 }
0876b9a0
KW
2910 }
2911 else if (UTF8_IS_SURROGATE(s)) {
8457b38f
KW
2912 if (ckWARN_d(WARN_SURROGATE)) {
2913 UV uv = utf8_to_uvchr(s, &char_len);
2914 Perl_warner(aTHX_ packWARN(WARN_SURROGATE),
2915 "Unicode surrogate U+%04"UVXf" is illegal in UTF-8", uv);
2916 ok = FALSE;
2917 }
0876b9a0
KW
2918 }
2919 else if
8457b38f
KW
2920 ((UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s))
2921 && (ckWARN_d(WARN_NONCHAR)))
0876b9a0
KW
2922 {
2923 UV uv = utf8_to_uvchr(s, &char_len);
8457b38f 2924 Perl_warner(aTHX_ packWARN(WARN_NONCHAR),
0876b9a0
KW
2925 "Unicode non-character U+%04"UVXf" is illegal for open interchange", uv);
2926 ok = FALSE;
2927 }
2928 }
2929 s += UTF8SKIP(s);
2930 }
2931
2932 return ok;
2933}
2934
0f830e0b 2935/*
87cea99e 2936=for apidoc pv_uni_display
d2cc3551
JH
2937
2938Build to the scalar dsv a displayable version of the string spv,
2939length len, the displayable version being at most pvlim bytes long
2940(if longer, the rest is truncated and "..." will be appended).
0a2ef054 2941
9e55ce06 2942The flags argument can have UNI_DISPLAY_ISPRINT set to display
00e86452 2943isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
0a2ef054
JH
2944to display the \\[nrfta\\] as the backslashed versions (like '\n')
2945(UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
2946UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
2947UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
2948
d2cc3551
JH
2949The pointer to the PV of the dsv is returned.
2950
2951=cut */
e6b2e755 2952char *
e1ec3a88 2953Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
e6b2e755
JH
2954{
2955 int truncated = 0;
e1ec3a88 2956 const char *s, *e;
e6b2e755 2957
7918f24d
NC
2958 PERL_ARGS_ASSERT_PV_UNI_DISPLAY;
2959
76f68e9b 2960 sv_setpvs(dsv, "");
7fddd944 2961 SvUTF8_off(dsv);
e1ec3a88 2962 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
e6b2e755 2963 UV u;
a49f32c6
NC
2964 /* This serves double duty as a flag and a character to print after
2965 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
2966 */
2967 char ok = 0;
c728cb41 2968
e6b2e755
JH
2969 if (pvlim && SvCUR(dsv) >= pvlim) {
2970 truncated++;
2971 break;
2972 }
2973 u = utf8_to_uvchr((U8*)s, 0);
c728cb41 2974 if (u < 256) {
a3b680e6 2975 const unsigned char c = (unsigned char)u & 0xFF;
0bd48802 2976 if (flags & UNI_DISPLAY_BACKSLASH) {
a49f32c6 2977 switch (c) {
c728cb41 2978 case '\n':
a49f32c6 2979 ok = 'n'; break;
c728cb41 2980 case '\r':
a49f32c6 2981 ok = 'r'; break;
c728cb41 2982 case '\t':
a49f32c6 2983 ok = 't'; break;
c728cb41 2984 case '\f':
a49f32c6 2985 ok = 'f'; break;
c728cb41 2986 case '\a':
a49f32c6 2987 ok = 'a'; break;
c728cb41 2988 case '\\':
a49f32c6 2989 ok = '\\'; break;
c728cb41
JH
2990 default: break;
2991 }
a49f32c6 2992 if (ok) {
88c9ea1e 2993 const char string = ok;
76f68e9b 2994 sv_catpvs(dsv, "\\");
5e7aa789 2995 sv_catpvn(dsv, &string, 1);
a49f32c6 2996 }
c728cb41 2997 }
00e86452 2998 /* isPRINT() is the locale-blind version. */
a49f32c6 2999 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
88c9ea1e 3000 const char string = c;
5e7aa789 3001 sv_catpvn(dsv, &string, 1);
a49f32c6 3002 ok = 1;
0a2ef054 3003 }
c728cb41
JH
3004 }
3005 if (!ok)
9e55ce06 3006 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
e6b2e755
JH
3007 }
3008 if (truncated)
396482e1 3009 sv_catpvs(dsv, "...");
48ef279e 3010
e6b2e755
JH
3011 return SvPVX(dsv);
3012}
2b9d42f0 3013
d2cc3551 3014/*
87cea99e 3015=for apidoc sv_uni_display
d2cc3551
JH
3016
3017Build to the scalar dsv a displayable version of the scalar sv,
0a2ef054 3018the displayable version being at most pvlim bytes long
d2cc3551 3019(if longer, the rest is truncated and "..." will be appended).
0a2ef054
JH
3020
3021The flags argument is as in pv_uni_display().
3022
d2cc3551
JH
3023The pointer to the PV of the dsv is returned.
3024
d4c19fe8
AL
3025=cut
3026*/
e6b2e755
JH
3027char *
3028Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
3029{
7918f24d
NC
3030 PERL_ARGS_ASSERT_SV_UNI_DISPLAY;
3031
cfd0369c
NC
3032 return Perl_pv_uni_display(aTHX_ dsv, (const U8*)SvPVX_const(ssv),
3033 SvCUR(ssv), pvlim, flags);
701a277b
JH
3034}
3035
d2cc3551 3036/*
e6226b18 3037=for apidoc foldEQ_utf8
d2cc3551 3038
d51c1b21 3039Returns true if the leading portions of the strings s1 and s2 (either or both
e6226b18 3040of which may be in UTF-8) are the same case-insensitively; false otherwise.
d51c1b21 3041How far into the strings to compare is determined by other input parameters.
8b35872c
KW
3042
3043If u1 is true, the string s1 is assumed to be in UTF-8-encoded Unicode;
3044otherwise it is assumed to be in native 8-bit encoding. Correspondingly for u2
3045with respect to s2.
3046
d51c1b21
KW
3047If the byte length l1 is non-zero, it says how far into s1 to check for fold
3048equality. In other words, s1+l1 will be used as a goal to reach. The
8b35872c
KW
3049scan will not be considered to be a match unless the goal is reached, and
3050scanning won't continue past that goal. Correspondingly for l2 with respect to
3051s2.
3052
3053If pe1 is non-NULL and the pointer it points to is not NULL, that pointer is
3054considered an end pointer beyond which scanning of s1 will not continue under
3055any circumstances. This means that if both l1 and pe1 are specified, and pe1
3056is less than s1+l1, the match will never be successful because it can never
d51c1b21
KW
3057get as far as its goal (and in fact is asserted against). Correspondingly for
3058pe2 with respect to s2.
8b35872c 3059
d51c1b21
KW
3060At least one of s1 and s2 must have a goal (at least one of l1 and l2 must be
3061non-zero), and if both do, both have to be
8b35872c
KW
3062reached for a successful match. Also, if the fold of a character is multiple
3063characters, all of them must be matched (see tr21 reference below for
3064'folding').
3065
e6226b18 3066Upon a successful match, if pe1 is non-NULL,
8b35872c
KW
3067it will be set to point to the beginning of the I<next> character of s1 beyond
3068what was matched. Correspondingly for pe2 and s2.
d2cc3551
JH
3069
3070For case-insensitiveness, the "casefolding" of Unicode is used
3071instead of upper/lowercasing both the characters, see
3072http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
3073
3074=cut */
a33c29bc
KW
3075
3076/* A flags parameter has been added which may change, and hence isn't
3077 * externally documented. Currently it is:
3078 * 0 for as-documented above
3079 * FOLDEQ_UTF8_NOMIX_ASCII meaning that if a non-ASCII character folds to an
3080 ASCII one, to not match
5e64d0fa
KW
3081 * FOLDEQ_UTF8_LOCALE meaning that locale rules are to be used for code
3082 * points below 256; unicode rules for above 255; and
3083 * folds that cross those boundaries are disallowed,
3084 * like the NOMIX_ASCII option
a33c29bc 3085 */
701a277b 3086I32
eda9cac1 3087Perl_foldEQ_utf8_flags(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2, U32 flags)
332ddc25 3088{
8b35872c
KW
3089 dVAR;
3090 register const U8 *p1 = (const U8*)s1; /* Point to current char */
3091 register const U8 *p2 = (const U8*)s2;
48ef279e 3092 register const U8 *g1 = NULL; /* goal for s1 */
8b35872c 3093 register const U8 *g2 = NULL;
48ef279e
KW
3094 register const U8 *e1 = NULL; /* Don't scan s1 past this */
3095 register U8 *f1 = NULL; /* Point to current folded */
8b35872c
KW
3096 register const U8 *e2 = NULL;
3097 register U8 *f2 = NULL;
48ef279e 3098 STRLEN n1 = 0, n2 = 0; /* Number of bytes in current char */
8b35872c
KW
3099 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
3100 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
48ef279e
KW
3101 U8 natbuf[2]; /* Holds native 8-bit char converted to utf8;
3102 these always fit in 2 bytes */
8b35872c 3103
eda9cac1 3104 PERL_ARGS_ASSERT_FOLDEQ_UTF8_FLAGS;
8b35872c
KW
3105
3106 if (pe1) {
48ef279e 3107 e1 = *(U8**)pe1;
8b35872c
KW
3108 }
3109
3110 if (l1) {
48ef279e 3111 g1 = (const U8*)s1 + l1;
8b35872c
KW
3112 }
3113
3114 if (pe2) {
48ef279e 3115 e2 = *(U8**)pe2;
8b35872c
KW
3116 }
3117
3118 if (l2) {
48ef279e 3119 g2 = (const U8*)s2 + l2;
8b35872c
KW
3120 }
3121
3122 /* Must have at least one goal */
3123 assert(g1 || g2);
3124
3125 if (g1) {
3126
48ef279e
KW
3127 /* Will never match if goal is out-of-bounds */
3128 assert(! e1 || e1 >= g1);
8b35872c 3129
48ef279e
KW
3130 /* Here, there isn't an end pointer, or it is beyond the goal. We
3131 * only go as far as the goal */
3132 e1 = g1;
8b35872c 3133 }
313b38e5
NC
3134 else {
3135 assert(e1); /* Must have an end for looking at s1 */
3136 }
8b35872c
KW
3137
3138 /* Same for goal for s2 */
3139 if (g2) {
48ef279e
KW
3140 assert(! e2 || e2 >= g2);
3141 e2 = g2;
8b35872c 3142 }
313b38e5
NC
3143 else {
3144 assert(e2);
3145 }
8b35872c
KW
3146
3147 /* Look through both strings, a character at a time */
3148 while (p1 < e1 && p2 < e2) {
3149
d51c1b21 3150 /* If at the beginning of a new character in s1, get its fold to use
5e64d0fa
KW
3151 * and the length of the fold. (exception: locale rules just get the
3152 * character to a single byte) */
48ef279e 3153 if (n1 == 0) {
a33c29bc 3154
5e64d0fa
KW
3155 /* If in locale matching, we use two sets of rules, depending on if
3156 * the code point is above or below 255. Here, we test for and
3157 * handle locale rules */
3158 if ((flags & FOLDEQ_UTF8_LOCALE)
3159 && (! u1 || UTF8_IS_INVARIANT(*p1) || UTF8_IS_DOWNGRADEABLE_START(*p1)))
3160 {
3161 /* There is no mixing of code points above and below 255. */
3162 if (u2 && (! UTF8_IS_INVARIANT(*p2)
3163 && ! UTF8_IS_DOWNGRADEABLE_START(*p2)))
3164 {
3165 return 0;
3166 }
3167
3168 /* We handle locale rules by converting, if necessary, the code
3169 * point to a single byte. */
3170 if (! u1 || UTF8_IS_INVARIANT(*p1)) {
3171 *foldbuf1 = *p1;
3172 }
3173 else {
3174 *foldbuf1 = TWO_BYTE_UTF8_TO_UNI(*p1, *(p1 + 1));
3175 }
3176 n1 = 1;
3177 }
3178 else if (isASCII(*p1)) { /* Note, that here won't be both ASCII
3179 and using locale rules */
3180
3181 /* If trying to mix non- with ASCII, and not supposed to, fail */
a33c29bc
KW
3182 if ((flags & FOLDEQ_UTF8_NOMIX_ASCII) && ! isASCII(*p2)) {
3183 return 0;
3184 }
3185 n1 = 1;
5e64d0fa
KW
3186 *foldbuf1 = toLOWER(*p1); /* Folds in the ASCII range are
3187 just lowercased */
a33c29bc
KW
3188 }
3189 else if (u1) {
48ef279e
KW
3190 to_utf8_fold(p1, foldbuf1, &n1);
3191 }
3192 else { /* Not utf8, convert to it first and then get fold */
3193 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p1)));
3194 to_utf8_fold(natbuf, foldbuf1, &n1);
3195 }
3196 f1 = foldbuf1;
3197 }
8b35872c 3198
48ef279e 3199 if (n2 == 0) { /* Same for s2 */
5e64d0fa
KW
3200 if ((flags & FOLDEQ_UTF8_LOCALE)
3201 && (! u2 || UTF8_IS_INVARIANT(*p2) || UTF8_IS_DOWNGRADEABLE_START(*p2)))
3202 {
5001101e
KW
3203 /* Here, the next char in s2 is < 256. We've already worked on
3204 * s1, and if it isn't also < 256, can't match */
5e64d0fa
KW
3205 if (u1 && (! UTF8_IS_INVARIANT(*p1)
3206 && ! UTF8_IS_DOWNGRADEABLE_START(*p1)))
3207 {
3208 return 0;
3209 }
3210 if (! u2 || UTF8_IS_INVARIANT(*p2)) {
3211 *foldbuf2 = *p2;
3212 }
3213 else {
3214 *foldbuf2 = TWO_BYTE_UTF8_TO_UNI(*p2, *(p2 + 1));
3215 }
5001101e
KW
3216
3217 /* Use another function to handle locale rules. We've made
3218 * sure that both characters to compare are single bytes */
3219 if (! foldEQ_locale((char *) f1, (char *) foldbuf2, 1)) {
3220 return 0;
3221 }
3222 n1 = n2 = 0;
5e64d0fa
KW
3223 }
3224 else if (isASCII(*p2)) {
a33c29bc
KW
3225 if (flags && ! isASCII(*p1)) {
3226 return 0;
3227 }
3228 n2 = 1;
3229 *foldbuf2 = toLOWER(*p2);
3230 }
3231 else if (u2) {
48ef279e
KW
3232 to_utf8_fold(p2, foldbuf2, &n2);
3233 }
3234 else {
3235 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p2)));
3236 to_utf8_fold(natbuf, foldbuf2, &n2);
3237 }
3238 f2 = foldbuf2;
3239 }
8b35872c 3240
5001101e
KW
3241 /* Here f1 and f2 point to the beginning of the strings to compare.
3242 * These strings are the folds of the input characters, stored in utf8.
3243 */
5e64d0fa 3244
48ef279e
KW
3245 /* While there is more to look for in both folds, see if they
3246 * continue to match */
3247 while (n1 && n2) {
3248 U8 fold_length = UTF8SKIP(f1);
3249 if (fold_length != UTF8SKIP(f2)
3250 || (fold_length == 1 && *f1 != *f2) /* Short circuit memNE
3251 function call for single
3252 character */
3253 || memNE((char*)f1, (char*)f2, fold_length))
3254 {
e6226b18 3255 return 0; /* mismatch */
48ef279e
KW
3256 }
3257
3258 /* Here, they matched, advance past them */
3259 n1 -= fold_length;
3260 f1 += fold_length;
3261 n2 -= fold_length;
3262 f2 += fold_length;
3263 }
8b35872c 3264
48ef279e
KW
3265 /* When reach the end of any fold, advance the input past it */
3266 if (n1 == 0) {
3267 p1 += u1 ? UTF8SKIP(p1) : 1;
3268 }
3269 if (n2 == 0) {
3270 p2 += u2 ? UTF8SKIP(p2) : 1;
3271 }
8b35872c
KW
3272 } /* End of loop through both strings */
3273
3274 /* A match is defined by each scan that specified an explicit length
3275 * reaching its final goal, and the other not having matched a partial
3276 * character (which can happen when the fold of a character is more than one
3277 * character). */
3278 if (! ((g1 == 0 || p1 == g1) && (g2 == 0 || p2 == g2)) || n1 || n2) {
e6226b18 3279 return 0;
8b35872c
KW
3280 }
3281
3282 /* Successful match. Set output pointers */
3283 if (pe1) {
48ef279e 3284 *pe1 = (char*)p1;
8b35872c
KW
3285 }
3286 if (pe2) {
48ef279e 3287 *pe2 = (char*)p2;
8b35872c 3288 }
e6226b18 3289 return 1;
e6b2e755 3290}
701a277b 3291
a49f32c6
NC
3292/*
3293 * Local variables:
3294 * c-indentation-style: bsd
3295 * c-basic-offset: 4
3296 * indent-tabs-mode: t
3297 * End:
3298 *
37442d52
RGS
3299 * ex: set ts=8 sts=4 sw=4 noet:
3300 */