This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regcomp.c: Change char used to force reading in fold swashes
[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 *
cdad3b53 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 20 * as is the custom in the West, if you wish to be answered?'
cdad3b53 21 * --Gandalf, addressing Théoden's door wardens
4ac71550
TC
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
ST
296#ifdef EBCDIC
297 u = NATIVE_TO_UTF(u);
298#endif
646ca15d
JH
299 u &= UTF_START_MASK(len);
300 uv = u;
301 ouv = uv;
302 while (slen--) {
303 if (!UTF8_IS_CONTINUATION(*s))
304 return 0;
305 uv = UTF8_ACCUMULATE(uv, *s);
48ef279e 306 if (uv < ouv)
646ca15d
JH
307 return 0;
308 ouv = uv;
309 s++;
310 }
311
312 if ((STRLEN)UNISKIP(uv) < len)
313 return 0;
314
315 return len;
316}
9041c2e3
NIS
317
318/*
87cea99e 319=for apidoc is_utf8_char
eebe1485 320
5da9da9e 321Tests if some arbitrary number of bytes begins in a valid UTF-8
2bbc8d55
SP
322character. Note that an INVARIANT (i.e. ASCII on non-EBCDIC machines)
323character is a valid UTF-8 character. The actual number of bytes in the UTF-8
324character will be returned if it is valid, otherwise 0.
9041c2e3 325
82686b01 326=cut */
067a85ef 327STRLEN
668b6d8d 328Perl_is_utf8_char(const U8 *s)
386d01d6 329{
44f8325f 330 const STRLEN len = UTF8SKIP(s);
7918f24d
NC
331
332 PERL_ARGS_ASSERT_IS_UTF8_CHAR;
3b0fc154 333#ifdef IS_UTF8_CHAR
768c67ee 334 if (IS_UTF8_CHAR_FAST(len))
3b0fc154
JH
335 return IS_UTF8_CHAR(s, len) ? len : 0;
336#endif /* #ifdef IS_UTF8_CHAR */
2c0c5f92 337 return is_utf8_char_slow(s, len);
386d01d6
GS
338}
339
eaf7a4d2 340
6662521e 341/*
87cea99e 342=for apidoc is_utf8_string
6662521e 343
c9ada85f 344Returns true if first C<len> bytes of the given string form a valid
9f7e3d64
MH
345UTF-8 string, false otherwise. If C<len> is 0, it will be calculated
346using C<strlen(s)>. Note that 'a valid UTF-8 string' does not mean 'a
347string that contains code points above 0x7F encoded in UTF-8' because a
348valid ASCII string is a valid UTF-8 string.
6662521e 349
eaf7a4d2 350See also is_ascii_string(), is_utf8_string_loclen(), and is_utf8_string_loc().
768c67ee 351
6662521e
GS
352=cut
353*/
354
8e84507e 355bool
668b6d8d 356Perl_is_utf8_string(const U8 *s, STRLEN len)
6662521e 357{
35da51f7 358 const U8* const send = s + (len ? len : strlen((const char *)s));
7fc63493 359 const U8* x = s;
067a85ef 360
7918f24d 361 PERL_ARGS_ASSERT_IS_UTF8_STRING;
1aa99e6b 362
6662521e 363 while (x < send) {
a3b680e6 364 STRLEN c;
1acdb0da
JH
365 /* Inline the easy bits of is_utf8_char() here for speed... */
366 if (UTF8_IS_INVARIANT(*x))
367 c = 1;
368 else if (!UTF8_IS_START(*x))
768c67ee 369 goto out;
1acdb0da
JH
370 else {
371 /* ... and call is_utf8_char() only if really needed. */
646ca15d
JH
372#ifdef IS_UTF8_CHAR
373 c = UTF8SKIP(x);
768c67ee
JH
374 if (IS_UTF8_CHAR_FAST(c)) {
375 if (!IS_UTF8_CHAR(x, c))
3c614e38
AD
376 c = 0;
377 }
378 else
379 c = is_utf8_char_slow(x, c);
646ca15d
JH
380#else
381 c = is_utf8_char(x);
382#endif /* #ifdef IS_UTF8_CHAR */
1acdb0da 383 if (!c)
768c67ee 384 goto out;
1acdb0da 385 }
6662521e 386 x += c;
6662521e 387 }
768c67ee
JH
388
389 out:
60006e79
JH
390 if (x != send)
391 return FALSE;
067a85ef
A
392
393 return TRUE;
6662521e
GS
394}
395
67e989fb 396/*
814fafa7
NC
397Implemented as a macro in utf8.h
398
87cea99e 399=for apidoc is_utf8_string_loc
814fafa7
NC
400
401Like is_utf8_string() but stores the location of the failure (in the
402case of "utf8ness failure") or the location s+len (in the case of
403"utf8ness success") in the C<ep>.
404
405See also is_utf8_string_loclen() and is_utf8_string().
406
87cea99e 407=for apidoc is_utf8_string_loclen
81cd54e3 408
e3e4599f 409Like is_utf8_string() but stores the location of the failure (in the
768c67ee
JH
410case of "utf8ness failure") or the location s+len (in the case of
411"utf8ness success") in the C<ep>, and the number of UTF-8
412encoded characters in the C<el>.
413
414See also is_utf8_string_loc() and is_utf8_string().
81cd54e3
JH
415
416=cut
417*/
418
419bool
668b6d8d 420Perl_is_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
81cd54e3 421{
35da51f7 422 const U8* const send = s + (len ? len : strlen((const char *)s));
7fc63493 423 const U8* x = s;
81cd54e3 424 STRLEN c;
3ebfea28 425 STRLEN outlen = 0;
7918f24d
NC
426
427 PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN;
81cd54e3 428
81cd54e3
JH
429 while (x < send) {
430 /* Inline the easy bits of is_utf8_char() here for speed... */
431 if (UTF8_IS_INVARIANT(*x))
768c67ee
JH
432 c = 1;
433 else if (!UTF8_IS_START(*x))
434 goto out;
81cd54e3 435 else {
768c67ee
JH
436 /* ... and call is_utf8_char() only if really needed. */
437#ifdef IS_UTF8_CHAR
438 c = UTF8SKIP(x);
439 if (IS_UTF8_CHAR_FAST(c)) {
440 if (!IS_UTF8_CHAR(x, c))
441 c = 0;
442 } else
443 c = is_utf8_char_slow(x, c);
444#else
445 c = is_utf8_char(x);
446#endif /* #ifdef IS_UTF8_CHAR */
447 if (!c)
448 goto out;
81cd54e3 449 }
768c67ee 450 x += c;
3ebfea28 451 outlen++;
81cd54e3 452 }
768c67ee
JH
453
454 out:
3ebfea28
AL
455 if (el)
456 *el = outlen;
457
768c67ee
JH
458 if (ep)
459 *ep = x;
3ebfea28 460 return (x == send);
81cd54e3
JH
461}
462
463/*
768c67ee 464
87cea99e 465=for apidoc utf8n_to_uvuni
67e989fb 466
9041c2e3 467Bottom level UTF-8 decode routine.
949cf498
KW
468Returns the code point value of the first character in the string C<s>
469which is assumed to be in UTF-8 (or UTF-EBCDIC) encoding and no longer than
470C<curlen> bytes; C<retlen> will be set to the length, in bytes, of that
471character.
472
473The value of C<flags> determines the behavior when C<s> does not point to a
474well-formed UTF-8 character. If C<flags> is 0, when a malformation is found,
475C<retlen> is set to the expected length of the UTF-8 character in bytes, zero
476is returned, and if UTF-8 warnings haven't been lexically disabled, a warning
477is raised.
478
479Various ALLOW flags can be set in C<flags> to allow (and not warn on)
480individual types of malformations, such as the sequence being overlong (that
481is, when there is a shorter sequence that can express the same code point;
482overlong sequences are expressly forbidden in the UTF-8 standard due to
483potential security issues). Another malformation example is the first byte of
484a character not being a legal first byte. See F<utf8.h> for the list of such
485flags. Of course, the value returned by this function under such conditions is
486not reliable.
487
488The UTF8_CHECK_ONLY flag overrides the behavior when a non-allowed (by other
489flags) malformation is found. If this flag is set, the routine assumes that
490the caller will raise a warning, and this function will silently just set
491C<retlen> to C<-1> and return zero.
492
493Certain code points are considered problematic. These are Unicode surrogates,
494Unicode non-characters, and code points above the Unicode maximum of 0x10FFF.
495By default these are considered regular code points, but certain situations
496warrant special handling for them. if C<flags> contains
497UTF8_DISALLOW_ILLEGAL_INTERCHANGE, all three classes are treated as
498malformations and handled as such. The flags UTF8_DISALLOW_SURROGATE,
499UTF8_DISALLOW_NONCHAR, and UTF8_DISALLOW_SUPER (meaning above the legal Unicode
500maximum) can be set to disallow these categories individually.
501
502The flags UTF8_WARN_ILLEGAL_INTERCHANGE, UTF8_WARN_SURROGATE,
503UTF8_WARN_NONCHAR, and UTF8_WARN_SUPER will cause warning messages to be raised
504for their respective categories, but otherwise the code points are considered
505valid (not malformations). To get a category to both be treated as a
506malformation and raise a warning, specify both the WARN and DISALLOW flags.
507(But note that warnings are not raised if lexically disabled nor if
508UTF8_CHECK_ONLY is also specified.)
509
510Very large code points (above 0x7FFF_FFFF) are considered more problematic than
511the others that are above the Unicode legal maximum. There are several
512reasons, one of which is that the original UTF-8 specification never went above
513this number (the current 0x10FFF limit was imposed later). The UTF-8 encoding
514on ASCII platforms for these large code point begins with a byte containing
5150xFE or 0xFF. The UTF8_DISALLOW_FE_FF flag will cause them to be treated as
516malformations, while allowing smaller above-Unicode code points. (Of course
517UTF8_DISALLOW_SUPER will treat all above-Unicode code points, including these,
518as malformations.) Similarly, UTF8_WARN_FE_FF acts just like the other WARN
519flags, but applies just to these code points.
520
521All other code points corresponding to Unicode characters, including private
522use and those yet to be assigned, are never considered malformed and never
523warn.
67e989fb 524
9041c2e3
NIS
525Most code should use utf8_to_uvchr() rather than call this directly.
526
37607a96
PK
527=cut
528*/
67e989fb 529
a0ed51b3 530UV
7fc63493 531Perl_utf8n_to_uvuni(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
a0ed51b3 532{
97aff369 533 dVAR;
d4c19fe8 534 const U8 * const s0 = s;
9c5ffd7c 535 UV uv = *s, ouv = 0;
ba210ebe 536 STRLEN len = 1;
949cf498 537 bool dowarn = ckWARN_d(WARN_UTF8);
7fc63493 538 const UV startbyte = *s;
ba210ebe 539 STRLEN expectlen = 0;
a0dbb045 540 U32 warning = 0;
949cf498 541 SV* sv = NULL;
a0dbb045 542
7918f24d
NC
543 PERL_ARGS_ASSERT_UTF8N_TO_UVUNI;
544
949cf498 545/* This list is a superset of the UTF8_ALLOW_XXX. */
a0dbb045
JH
546
547#define UTF8_WARN_EMPTY 1
548#define UTF8_WARN_CONTINUATION 2
549#define UTF8_WARN_NON_CONTINUATION 3
f9bc8ed7
KW
550#define UTF8_WARN_SHORT 4
551#define UTF8_WARN_OVERFLOW 5
552#define UTF8_WARN_LONG 6
a0dbb045
JH
553
554 if (curlen == 0 &&
555 !(flags & UTF8_ALLOW_EMPTY)) {
556 warning = UTF8_WARN_EMPTY;
0c443dc2
JH
557 goto malformed;
558 }
559
1d72bdf6 560 if (UTF8_IS_INVARIANT(uv)) {
a0ed51b3
LW
561 if (retlen)
562 *retlen = 1;
c4d5f83a 563 return (UV) (NATIVE_TO_UTF(*s));
a0ed51b3 564 }
67e989fb 565
421a8bf2 566 if (UTF8_IS_CONTINUATION(uv) &&
fcc8fcf6 567 !(flags & UTF8_ALLOW_CONTINUATION)) {
a0dbb045 568 warning = UTF8_WARN_CONTINUATION;
ba210ebe
JH
569 goto malformed;
570 }
571
421a8bf2 572 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
fcc8fcf6 573 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
a0dbb045 574 warning = UTF8_WARN_NON_CONTINUATION;
ba210ebe
JH
575 goto malformed;
576 }
9041c2e3 577
1d72bdf6 578#ifdef EBCDIC
75383841 579 uv = NATIVE_TO_UTF(uv);
1d72bdf6 580#else
949cf498
KW
581 if (uv == 0xfe || uv == 0xff) {
582 if (flags & (UTF8_WARN_SUPER|UTF8_WARN_FE_FF)) {
111d382d 583 sv = sv_2mortal(Perl_newSVpvf(aTHX_ "Code point beginning with byte 0x%02"UVXf" is not Unicode, and not portable", uv));
949cf498
KW
584 flags &= ~UTF8_WARN_SUPER; /* Only warn once on this problem */
585 }
586 if (flags & (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_FE_FF)) {
587 goto malformed;
588 }
a0ed51b3 589 }
1d72bdf6
NIS
590#endif
591
ba210ebe
JH
592 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
593 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
594 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
595 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
1d72bdf6
NIS
596#ifdef EBCDIC
597 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
598 else { len = 7; uv &= 0x01; }
599#else
ba210ebe
JH
600 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
601 else if (!(uv & 0x01)) { len = 7; uv = 0; }
1d72bdf6
NIS
602 else { len = 13; uv = 0; } /* whoa! */
603#endif
604
a0ed51b3
LW
605 if (retlen)
606 *retlen = len;
9041c2e3 607
ba210ebe
JH
608 expectlen = len;
609
fcc8fcf6
JH
610 if ((curlen < expectlen) &&
611 !(flags & UTF8_ALLOW_SHORT)) {
a0dbb045 612 warning = UTF8_WARN_SHORT;
ba210ebe
JH
613 goto malformed;
614 }
615
616 len--;
a0ed51b3 617 s++;
949cf498 618 ouv = uv; /* ouv is the value from the previous iteration */
ba210ebe 619
a0ed51b3 620 while (len--) {
421a8bf2
JH
621 if (!UTF8_IS_CONTINUATION(*s) &&
622 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
a0dbb045
JH
623 s--;
624 warning = UTF8_WARN_NON_CONTINUATION;
ba210ebe 625 goto malformed;
a0ed51b3
LW
626 }
627 else
8850bf83 628 uv = UTF8_ACCUMULATE(uv, *s);
949cf498
KW
629 if (!(uv > ouv)) { /* If the value didn't grow from the previous
630 iteration, something is horribly wrong */
a0dbb045
JH
631 /* These cannot be allowed. */
632 if (uv == ouv) {
75dbc644 633 if (expectlen != 13 && !(flags & UTF8_ALLOW_LONG)) {
a0dbb045
JH
634 warning = UTF8_WARN_LONG;
635 goto malformed;
636 }
637 }
638 else { /* uv < ouv */
639 /* This cannot be allowed. */
640 warning = UTF8_WARN_OVERFLOW;
641 goto malformed;
642 }
ba210ebe
JH
643 }
644 s++;
645 ouv = uv;
646 }
647
949cf498 648 if ((expectlen > (STRLEN)UNISKIP(uv)) && !(flags & UTF8_ALLOW_LONG)) {
a0dbb045 649 warning = UTF8_WARN_LONG;
ba210ebe 650 goto malformed;
949cf498
KW
651 } else if (flags & (UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE)) {
652 if (UNICODE_IS_SURROGATE(uv)) {
653 if ((flags & (UTF8_WARN_SURROGATE|UTF8_CHECK_ONLY)) == UTF8_WARN_SURROGATE) {
111d382d 654 sv = sv_2mortal(Perl_newSVpvf(aTHX_ "UTF-16 surrogate U+%04"UVXf"", uv));
949cf498
KW
655 }
656 if (flags & UTF8_DISALLOW_SURROGATE) {
657 goto disallowed;
658 }
659 }
660 else if (UNICODE_IS_NONCHAR(uv)) {
661 if ((flags & (UTF8_WARN_NONCHAR|UTF8_CHECK_ONLY)) == UTF8_WARN_NONCHAR ) {
111d382d 662 sv = sv_2mortal(Perl_newSVpvf(aTHX_ "Unicode non-character U+%04"UVXf" is illegal for open interchange", uv));
949cf498
KW
663 }
664 if (flags & UTF8_DISALLOW_NONCHAR) {
665 goto disallowed;
666 }
667 }
668 else if ((uv > PERL_UNICODE_MAX)) {
669 if ((flags & (UTF8_WARN_SUPER|UTF8_CHECK_ONLY)) == UTF8_WARN_SUPER) {
111d382d 670 sv = sv_2mortal(Perl_newSVpvf(aTHX_ "Code point 0x%04"UVXf" is not Unicode, may not be portable", uv));
949cf498
KW
671 }
672 if (flags & UTF8_DISALLOW_SUPER) {
673 goto disallowed;
674 }
675 }
676
677 /* Here, this is not considered a malformed character, so drop through
678 * to return it */
a0ed51b3 679 }
ba210ebe 680
a0ed51b3 681 return uv;
ba210ebe 682
949cf498
KW
683disallowed: /* Is disallowed, but otherwise not malformed. 'sv' will have been
684 set if there is to be a warning. */
685 if (!sv) {
686 dowarn = 0;
687 }
688
ba210ebe
JH
689malformed:
690
fcc8fcf6 691 if (flags & UTF8_CHECK_ONLY) {
ba210ebe 692 if (retlen)
10edeb5d 693 *retlen = ((STRLEN) -1);
ba210ebe
JH
694 return 0;
695 }
696
a0dbb045 697 if (dowarn) {
949cf498 698 if (! sv) {
5b311467 699 sv = newSVpvs_flags("Malformed UTF-8 character ", SVs_TEMP);
949cf498 700 }
5b311467 701
5e3d7cf5
KW
702 switch (warning) {
703 case 0: /* Intentionally empty. */ break;
704 case UTF8_WARN_EMPTY:
705 sv_catpvs(sv, "(empty string)");
706 break;
707 case UTF8_WARN_CONTINUATION:
708 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
709 break;
710 case UTF8_WARN_NON_CONTINUATION:
711 if (s == s0)
712 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
713 (UV)s[1], startbyte);
714 else {
715 const int len = (int)(s-s0);
716 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
717 (UV)s[1], len, len > 1 ? "s" : "", startbyte, (int)expectlen);
718 }
719
720 break;
721 case UTF8_WARN_SHORT:
722 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
723 (int)curlen, curlen == 1 ? "" : "s", (int)expectlen, startbyte);
724 expectlen = curlen; /* distance for caller to skip */
725 break;
726 case UTF8_WARN_OVERFLOW:
727 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
728 ouv, *s, startbyte);
729 break;
730 case UTF8_WARN_LONG:
731 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
732 (int)expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
733 break;
734 default:
735 sv_catpvs(sv, "(unknown reason)");
736 break;
737 }
a0dbb045 738
949cf498 739 if (sv) {
44f8325f 740 const char * const s = SvPVX_const(sv);
a0dbb045
JH
741
742 if (PL_op)
9014280d 743 Perl_warner(aTHX_ packWARN(WARN_UTF8),
53e06cf0 744 "%s in %s", s, OP_DESC(PL_op));
a0dbb045 745 else
9014280d 746 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
a0dbb045
JH
747 }
748 }
749
ba210ebe 750 if (retlen)
28d3d195 751 *retlen = expectlen ? expectlen : len;
ba210ebe 752
28d3d195 753 return 0;
a0ed51b3
LW
754}
755
8e84507e 756/*
87cea99e 757=for apidoc utf8_to_uvchr
9041c2e3 758
6ee84de2 759Returns the native code point of the first character in the string C<s>
1e54db1a 760which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
9041c2e3
NIS
761length, in bytes, of that character.
762
1e54db1a 763If C<s> does not point to a well-formed UTF-8 character, zero is
9041c2e3
NIS
764returned and retlen is set, if possible, to -1.
765
766=cut
767*/
768
ae0e24b1 769
9041c2e3 770UV
7fc63493 771Perl_utf8_to_uvchr(pTHX_ const U8 *s, STRLEN *retlen)
9041c2e3 772{
7918f24d
NC
773 PERL_ARGS_ASSERT_UTF8_TO_UVCHR;
774
1754c1a1 775 return utf8n_to_uvchr(s, UTF8_MAXBYTES, retlen,
ae0e24b1 776 ckWARN_d(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
9041c2e3
NIS
777}
778
779/*
87cea99e 780=for apidoc utf8_to_uvuni
9041c2e3
NIS
781
782Returns the Unicode code point of the first character in the string C<s>
1e54db1a 783which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
9041c2e3
NIS
784length, in bytes, of that character.
785
2bbc8d55 786This function should only be used when the returned UV is considered
9041c2e3
NIS
787an index into the Unicode semantic tables (e.g. swashes).
788
1e54db1a 789If C<s> does not point to a well-formed UTF-8 character, zero is
ba210ebe 790returned and retlen is set, if possible, to -1.
8e84507e
NIS
791
792=cut
793*/
794
795UV
7fc63493 796Perl_utf8_to_uvuni(pTHX_ const U8 *s, STRLEN *retlen)
8e84507e 797{
7918f24d
NC
798 PERL_ARGS_ASSERT_UTF8_TO_UVUNI;
799
9041c2e3 800 /* Call the low level routine asking for checks */
89ebb4a3 801 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXBYTES, retlen,
ae0e24b1 802 ckWARN_d(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
8e84507e
NIS
803}
804
b76347f2 805/*
87cea99e 806=for apidoc utf8_length
b76347f2
JH
807
808Return the length of the UTF-8 char encoded string C<s> in characters.
02eb7b47
JH
809Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
810up past C<e>, croaks.
b76347f2
JH
811
812=cut
813*/
814
815STRLEN
35a4481c 816Perl_utf8_length(pTHX_ const U8 *s, const U8 *e)
b76347f2 817{
97aff369 818 dVAR;
b76347f2
JH
819 STRLEN len = 0;
820
7918f24d
NC
821 PERL_ARGS_ASSERT_UTF8_LENGTH;
822
8850bf83
JH
823 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
824 * the bitops (especially ~) can create illegal UTF-8.
825 * In other words: in Perl UTF-8 is not just for Unicode. */
826
a3b680e6
AL
827 if (e < s)
828 goto warn_and_return;
b76347f2 829 while (s < e) {
8e91ec7f
AV
830 if (!UTF8_IS_INVARIANT(*s))
831 s += UTF8SKIP(s);
832 else
833 s++;
834 len++;
835 }
836
837 if (e != s) {
838 len--;
839 warn_and_return:
9b387841
NC
840 if (PL_op)
841 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
842 "%s in %s", unees, OP_DESC(PL_op));
843 else
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
c682ebef
FC
1094/* This logic is duplicated in sv_catpvn_flags, so any bug fixes will
1095 likewise need duplication. */
1096
6940069f 1097U8*
35a4481c 1098Perl_bytes_to_utf8(pTHX_ const U8 *s, STRLEN *len)
6940069f 1099{
35a4481c 1100 const U8 * const send = s + (*len);
6940069f
GS
1101 U8 *d;
1102 U8 *dst;
7918f24d
NC
1103
1104 PERL_ARGS_ASSERT_BYTES_TO_UTF8;
96a5add6 1105 PERL_UNUSED_CONTEXT;
6940069f 1106
212542aa 1107 Newx(d, (*len) * 2 + 1, U8);
6940069f
GS
1108 dst = d;
1109
1110 while (s < send) {
35a4481c 1111 const UV uv = NATIVE_TO_ASCII(*s++);
c4d5f83a 1112 if (UNI_IS_INVARIANT(uv))
eb160463 1113 *d++ = (U8)UTF_TO_NATIVE(uv);
6940069f 1114 else {
eb160463
GS
1115 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
1116 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
6940069f
GS
1117 }
1118 }
1119 *d = '\0';
6662521e 1120 *len = d-dst;
6940069f
GS
1121 return dst;
1122}
1123
a0ed51b3 1124/*
dea0fc0b 1125 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
a0ed51b3
LW
1126 *
1127 * Destination must be pre-extended to 3/2 source. Do not use in-place.
1128 * We optimize for native, for obvious reasons. */
1129
1130U8*
dea0fc0b 1131Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
a0ed51b3 1132{
dea0fc0b
JH
1133 U8* pend;
1134 U8* dstart = d;
1135
7918f24d
NC
1136 PERL_ARGS_ASSERT_UTF16_TO_UTF8;
1137
dea0fc0b 1138 if (bytelen & 1)
f5992bc4 1139 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen %"UVuf, (UV)bytelen);
dea0fc0b
JH
1140
1141 pend = p + bytelen;
1142
a0ed51b3 1143 while (p < pend) {
dea0fc0b
JH
1144 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
1145 p += 2;
a0ed51b3 1146 if (uv < 0x80) {
e294cc5d
JH
1147#ifdef EBCDIC
1148 *d++ = UNI_TO_NATIVE(uv);
1149#else
eb160463 1150 *d++ = (U8)uv;
e294cc5d 1151#endif
a0ed51b3
LW
1152 continue;
1153 }
1154 if (uv < 0x800) {
eb160463
GS
1155 *d++ = (U8)(( uv >> 6) | 0xc0);
1156 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
1157 continue;
1158 }
52b9aa85 1159 if (uv >= 0xd800 && uv <= 0xdbff) { /* surrogates */
01ea242b 1160 if (p >= pend) {
dea0fc0b 1161 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
01ea242b
NC
1162 } else {
1163 UV low = (p[0] << 8) + p[1];
1164 p += 2;
52b9aa85 1165 if (low < 0xdc00 || low > 0xdfff)
01ea242b
NC
1166 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
1167 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
1168 }
dbde1951
NC
1169 } else if (uv >= 0xdc00 && uv <= 0xdfff) {
1170 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
a0ed51b3
LW
1171 }
1172 if (uv < 0x10000) {
eb160463
GS
1173 *d++ = (U8)(( uv >> 12) | 0xe0);
1174 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
1175 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
1176 continue;
1177 }
1178 else {
eb160463
GS
1179 *d++ = (U8)(( uv >> 18) | 0xf0);
1180 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
1181 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80);
1182 *d++ = (U8)(( uv & 0x3f) | 0x80);
a0ed51b3
LW
1183 continue;
1184 }
1185 }
dea0fc0b 1186 *newlen = d - dstart;
a0ed51b3
LW
1187 return d;
1188}
1189
1190/* Note: this one is slightly destructive of the source. */
1191
1192U8*
dea0fc0b 1193Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
a0ed51b3
LW
1194{
1195 U8* s = (U8*)p;
d4c19fe8 1196 U8* const send = s + bytelen;
7918f24d
NC
1197
1198 PERL_ARGS_ASSERT_UTF16_TO_UTF8_REVERSED;
1199
e0ea5e2d
NC
1200 if (bytelen & 1)
1201 Perl_croak(aTHX_ "panic: utf16_to_utf8_reversed: odd bytelen %"UVuf,
1202 (UV)bytelen);
1203
a0ed51b3 1204 while (s < send) {
d4c19fe8 1205 const U8 tmp = s[0];
a0ed51b3
LW
1206 s[0] = s[1];
1207 s[1] = tmp;
1208 s += 2;
1209 }
dea0fc0b 1210 return utf16_to_utf8(p, d, bytelen, newlen);
a0ed51b3
LW
1211}
1212
c3fd2246
KW
1213/* for now these are all defined (inefficiently) in terms of the utf8 versions.
1214 * Note that the macros in handy.h that call these short-circuit calling them
1215 * for Latin-1 range inputs */
a0ed51b3
LW
1216
1217bool
84afefe6 1218Perl_is_uni_alnum(pTHX_ UV c)
a0ed51b3 1219{
89ebb4a3 1220 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1221 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1222 return is_utf8_alnum(tmpbuf);
1223}
1224
1225bool
84afefe6 1226Perl_is_uni_idfirst(pTHX_ UV c)
a0ed51b3 1227{
89ebb4a3 1228 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1229 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1230 return is_utf8_idfirst(tmpbuf);
1231}
1232
1233bool
84afefe6 1234Perl_is_uni_alpha(pTHX_ UV c)
a0ed51b3 1235{
89ebb4a3 1236 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1237 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1238 return is_utf8_alpha(tmpbuf);
1239}
1240
1241bool
84afefe6 1242Perl_is_uni_ascii(pTHX_ UV c)
4d61ec05 1243{
bc39fe24 1244 return isASCII(c);
4d61ec05
GS
1245}
1246
1247bool
84afefe6 1248Perl_is_uni_space(pTHX_ UV c)
a0ed51b3 1249{
89ebb4a3 1250 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1251 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1252 return is_utf8_space(tmpbuf);
1253}
1254
1255bool
84afefe6 1256Perl_is_uni_digit(pTHX_ UV c)
a0ed51b3 1257{
89ebb4a3 1258 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1259 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1260 return is_utf8_digit(tmpbuf);
1261}
1262
1263bool
84afefe6 1264Perl_is_uni_upper(pTHX_ UV c)
a0ed51b3 1265{
89ebb4a3 1266 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1267 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1268 return is_utf8_upper(tmpbuf);
1269}
1270
1271bool
84afefe6 1272Perl_is_uni_lower(pTHX_ UV c)
a0ed51b3 1273{
89ebb4a3 1274 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1275 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1276 return is_utf8_lower(tmpbuf);
1277}
1278
1279bool
84afefe6 1280Perl_is_uni_cntrl(pTHX_ UV c)
b8c5462f 1281{
7b952154 1282 return isCNTRL_L1(c);
b8c5462f
JH
1283}
1284
1285bool
84afefe6 1286Perl_is_uni_graph(pTHX_ UV c)
b8c5462f 1287{
89ebb4a3 1288 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1289 uvchr_to_utf8(tmpbuf, c);
b8c5462f
JH
1290 return is_utf8_graph(tmpbuf);
1291}
1292
1293bool
84afefe6 1294Perl_is_uni_print(pTHX_ UV c)
a0ed51b3 1295{
89ebb4a3 1296 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1297 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
1298 return is_utf8_print(tmpbuf);
1299}
1300
b8c5462f 1301bool
84afefe6 1302Perl_is_uni_punct(pTHX_ UV c)
b8c5462f 1303{
89ebb4a3 1304 U8 tmpbuf[UTF8_MAXBYTES+1];
230880c1 1305 uvchr_to_utf8(tmpbuf, c);
b8c5462f
JH
1306 return is_utf8_punct(tmpbuf);
1307}
1308
4d61ec05 1309bool
84afefe6 1310Perl_is_uni_xdigit(pTHX_ UV c)
4d61ec05 1311{
89ebb4a3 1312 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
230880c1 1313 uvchr_to_utf8(tmpbuf, c);
4d61ec05
GS
1314 return is_utf8_xdigit(tmpbuf);
1315}
1316
3a4c58c9
KW
1317UV
1318Perl__to_upper_title_latin1(pTHX_ const U8 c, U8* p, STRLEN *lenp, const char S_or_s)
1319{
1320 /* We have the latin1-range values compiled into the core, so just use
1321 * those, converting the result to utf8. The only difference between upper
1322 * and title case in this range is that LATIN_SMALL_LETTER_SHARP_S is
1323 * either "SS" or "Ss". Which one to use is passed into the routine in
1324 * 'S_or_s' to avoid a test */
1325
1326 UV converted = toUPPER_LATIN1_MOD(c);
1327
1328 PERL_ARGS_ASSERT__TO_UPPER_TITLE_LATIN1;
1329
1330 assert(S_or_s == 'S' || S_or_s == 's');
1331
1332 if (UNI_IS_INVARIANT(converted)) { /* No difference between the two for
1333 characters in this range */
1334 *p = (U8) converted;
1335 *lenp = 1;
1336 return converted;
1337 }
1338
1339 /* toUPPER_LATIN1_MOD gives the correct results except for three outliers,
1340 * which it maps to one of them, so as to only have to have one check for
1341 * it in the main case */
1342 if (UNLIKELY(converted == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS)) {
1343 switch (c) {
1344 case LATIN_SMALL_LETTER_Y_WITH_DIAERESIS:
1345 converted = LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS;
1346 break;
1347 case MICRO_SIGN:
1348 converted = GREEK_CAPITAL_LETTER_MU;
1349 break;
1350 case LATIN_SMALL_LETTER_SHARP_S:
1351 *(p)++ = 'S';
1352 *p = S_or_s;
1353 *lenp = 2;
1354 return 'S';
1355 default:
1356 Perl_croak(aTHX_ "panic: to_upper_title_latin1 did not expect '%c' to map to '%c'", c, LATIN_SMALL_LETTER_Y_WITH_DIAERESIS);
1357 /* NOTREACHED */
1358 }
1359 }
1360
1361 *(p)++ = UTF8_TWO_BYTE_HI(converted);
1362 *p = UTF8_TWO_BYTE_LO(converted);
1363 *lenp = 2;
1364
1365 return converted;
1366}
1367
50bda2c3
KW
1368/* Call the function to convert a UTF-8 encoded character to the specified case.
1369 * Note that there may be more than one character in the result.
1370 * INP is a pointer to the first byte of the input character
1371 * OUTP will be set to the first byte of the string of changed characters. It
1372 * needs to have space for UTF8_MAXBYTES_CASE+1 bytes
1373 * LENP will be set to the length in bytes of the string of changed characters
1374 *
1375 * The functions return the ordinal of the first character in the string of OUTP */
1376#define CALL_UPPER_CASE(INP, OUTP, LENP) Perl_to_utf8_case(aTHX_ INP, OUTP, LENP, &PL_utf8_toupper, "ToUc", "utf8::ToSpecUpper")
1377#define CALL_TITLE_CASE(INP, OUTP, LENP) Perl_to_utf8_case(aTHX_ INP, OUTP, LENP, &PL_utf8_totitle, "ToTc", "utf8::ToSpecTitle")
1378#define CALL_LOWER_CASE(INP, OUTP, LENP) Perl_to_utf8_case(aTHX_ INP, OUTP, LENP, &PL_utf8_tolower, "ToLc", "utf8::ToSpecLower")
1379
1380/* This additionally has the input parameter SPECIALS, which if non-zero will
1381 * cause this to use the SPECIALS hash for folding (meaning get full case
1382 * folding); otherwise, when zero, this implies a simple case fold */
1383#define CALL_FOLD_CASE(INP, OUTP, LENP, SPECIALS) Perl_to_utf8_case(aTHX_ INP, OUTP, LENP, &PL_utf8_tofold, "ToCf", (SPECIALS) ? "utf8::ToSpecFold" : NULL)
c3fd2246 1384
84afefe6
JH
1385UV
1386Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 1387{
3a4c58c9
KW
1388 dVAR;
1389
c3fd2246
KW
1390 /* Convert the Unicode character whose ordinal is c to its uppercase
1391 * version and store that in UTF-8 in p and its length in bytes in lenp.
1392 * Note that the p needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1393 * the changed version may be longer than the original character.
1394 *
1395 * The ordinal of the first character of the changed version is returned
1396 * (but note, as explained above, that there may be more.) */
1397
7918f24d
NC
1398 PERL_ARGS_ASSERT_TO_UNI_UPPER;
1399
3a4c58c9
KW
1400 if (c < 256) {
1401 return _to_upper_title_latin1((U8) c, p, lenp, 'S');
1402 }
1403
0ebc6274 1404 uvchr_to_utf8(p, c);
3a4c58c9 1405 return CALL_UPPER_CASE(p, p, lenp);
a0ed51b3
LW
1406}
1407
84afefe6
JH
1408UV
1409Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 1410{
3a4c58c9
KW
1411 dVAR;
1412
7918f24d
NC
1413 PERL_ARGS_ASSERT_TO_UNI_TITLE;
1414
3a4c58c9
KW
1415 if (c < 256) {
1416 return _to_upper_title_latin1((U8) c, p, lenp, 's');
1417 }
1418
0ebc6274 1419 uvchr_to_utf8(p, c);
3a4c58c9 1420 return CALL_TITLE_CASE(p, p, lenp);
a0ed51b3
LW
1421}
1422
afc16117
KW
1423STATIC U8
1424S_to_lower_latin1(pTHX_ const U8 c, U8* p, STRLEN *lenp)
1425{
1426 /* We have the latin1-range values compiled into the core, so just use
1427 * those, converting the result to utf8. Since the result is always just
1428 * one character, we allow p to be NULL */
1429
1430 U8 converted = toLOWER_LATIN1(c);
1431
1432 if (p != NULL) {
1433 if (UNI_IS_INVARIANT(converted)) {
1434 *p = converted;
1435 *lenp = 1;
1436 }
1437 else {
1438 *p = UTF8_TWO_BYTE_HI(converted);
1439 *(p+1) = UTF8_TWO_BYTE_LO(converted);
1440 *lenp = 2;
1441 }
1442 }
1443 return converted;
1444}
1445
84afefe6
JH
1446UV
1447Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 1448{
968c5e6a
KW
1449 dVAR;
1450
7918f24d
NC
1451 PERL_ARGS_ASSERT_TO_UNI_LOWER;
1452
afc16117
KW
1453 if (c < 256) {
1454 return to_lower_latin1((U8) c, p, lenp);
bca00c02
KW
1455 }
1456
afc16117 1457 uvchr_to_utf8(p, c);
968c5e6a 1458 return CALL_LOWER_CASE(p, p, lenp);
a0ed51b3
LW
1459}
1460
84afefe6 1461UV
36bb2ab6 1462Perl__to_uni_fold_flags(pTHX_ UV c, U8* p, STRLEN *lenp, U8 flags)
84afefe6 1463{
36bb2ab6 1464 PERL_ARGS_ASSERT__TO_UNI_FOLD_FLAGS;
7918f24d 1465
0ebc6274 1466 uvchr_to_utf8(p, c);
36bb2ab6 1467 return _to_utf8_fold_flags(p, p, lenp, flags);
84afefe6
JH
1468}
1469
a0ed51b3
LW
1470/* for now these all assume no locale info available for Unicode > 255 */
1471
1472bool
84afefe6 1473Perl_is_uni_alnum_lc(pTHX_ UV c)
a0ed51b3
LW
1474{
1475 return is_uni_alnum(c); /* XXX no locale support yet */
1476}
1477
1478bool
84afefe6 1479Perl_is_uni_idfirst_lc(pTHX_ UV c)
a0ed51b3
LW
1480{
1481 return is_uni_idfirst(c); /* XXX no locale support yet */
1482}
1483
1484bool
84afefe6 1485Perl_is_uni_alpha_lc(pTHX_ UV c)
a0ed51b3
LW
1486{
1487 return is_uni_alpha(c); /* XXX no locale support yet */
1488}
1489
1490bool
84afefe6 1491Perl_is_uni_ascii_lc(pTHX_ UV c)
4d61ec05
GS
1492{
1493 return is_uni_ascii(c); /* XXX no locale support yet */
1494}
1495
1496bool
84afefe6 1497Perl_is_uni_space_lc(pTHX_ UV c)
a0ed51b3
LW
1498{
1499 return is_uni_space(c); /* XXX no locale support yet */
1500}
1501
1502bool
84afefe6 1503Perl_is_uni_digit_lc(pTHX_ UV c)
a0ed51b3
LW
1504{
1505 return is_uni_digit(c); /* XXX no locale support yet */
1506}
1507
1508bool
84afefe6 1509Perl_is_uni_upper_lc(pTHX_ UV c)
a0ed51b3
LW
1510{
1511 return is_uni_upper(c); /* XXX no locale support yet */
1512}
1513
1514bool
84afefe6 1515Perl_is_uni_lower_lc(pTHX_ UV c)
a0ed51b3
LW
1516{
1517 return is_uni_lower(c); /* XXX no locale support yet */
1518}
1519
1520bool
84afefe6 1521Perl_is_uni_cntrl_lc(pTHX_ UV c)
b8c5462f
JH
1522{
1523 return is_uni_cntrl(c); /* XXX no locale support yet */
1524}
1525
1526bool
84afefe6 1527Perl_is_uni_graph_lc(pTHX_ UV c)
b8c5462f
JH
1528{
1529 return is_uni_graph(c); /* XXX no locale support yet */
1530}
1531
1532bool
84afefe6 1533Perl_is_uni_print_lc(pTHX_ UV c)
a0ed51b3
LW
1534{
1535 return is_uni_print(c); /* XXX no locale support yet */
1536}
1537
b8c5462f 1538bool
84afefe6 1539Perl_is_uni_punct_lc(pTHX_ UV c)
b8c5462f
JH
1540{
1541 return is_uni_punct(c); /* XXX no locale support yet */
1542}
1543
4d61ec05 1544bool
84afefe6 1545Perl_is_uni_xdigit_lc(pTHX_ UV c)
4d61ec05
GS
1546{
1547 return is_uni_xdigit(c); /* XXX no locale support yet */
1548}
1549
b7ac61fa
JH
1550U32
1551Perl_to_uni_upper_lc(pTHX_ U32 c)
1552{
ee099d14
JH
1553 /* XXX returns only the first character -- do not use XXX */
1554 /* XXX no locale support yet */
1555 STRLEN len;
89ebb4a3 1556 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
ee099d14 1557 return (U32)to_uni_upper(c, tmpbuf, &len);
b7ac61fa
JH
1558}
1559
1560U32
1561Perl_to_uni_title_lc(pTHX_ U32 c)
1562{
ee099d14
JH
1563 /* XXX returns only the first character XXX -- do not use XXX */
1564 /* XXX no locale support yet */
1565 STRLEN len;
89ebb4a3 1566 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
ee099d14 1567 return (U32)to_uni_title(c, tmpbuf, &len);
b7ac61fa
JH
1568}
1569
1570U32
1571Perl_to_uni_lower_lc(pTHX_ U32 c)
1572{
ee099d14
JH
1573 /* XXX returns only the first character -- do not use XXX */
1574 /* XXX no locale support yet */
1575 STRLEN len;
89ebb4a3 1576 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
ee099d14 1577 return (U32)to_uni_lower(c, tmpbuf, &len);
b7ac61fa
JH
1578}
1579
7452cf6a 1580static bool
5141f98e 1581S_is_utf8_common(pTHX_ const U8 *const p, SV **swash,
bde6a22d
NC
1582 const char *const swashname)
1583{
97aff369 1584 dVAR;
7918f24d
NC
1585
1586 PERL_ARGS_ASSERT_IS_UTF8_COMMON;
1587
bde6a22d
NC
1588 if (!is_utf8_char(p))
1589 return FALSE;
1590 if (!*swash)
711a919c 1591 *swash = swash_init("utf8", swashname, &PL_sv_undef, 1, 0);
bde6a22d
NC
1592 return swash_fetch(*swash, p, TRUE) != 0;
1593}
1594
1595bool
7fc63493 1596Perl_is_utf8_alnum(pTHX_ const U8 *p)
a0ed51b3 1597{
97aff369 1598 dVAR;
7918f24d
NC
1599
1600 PERL_ARGS_ASSERT_IS_UTF8_ALNUM;
1601
671c33bf
NC
1602 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1603 * descendant of isalnum(3), in other words, it doesn't
1604 * contain the '_'. --jhi */
d4c19fe8 1605 return is_utf8_common(p, &PL_utf8_alnum, "IsWord");
a0ed51b3
LW
1606}
1607
1608bool
7fc63493 1609Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
a0ed51b3 1610{
97aff369 1611 dVAR;
7918f24d
NC
1612
1613 PERL_ARGS_ASSERT_IS_UTF8_IDFIRST;
1614
82686b01
JH
1615 if (*p == '_')
1616 return TRUE;
bde6a22d 1617 /* is_utf8_idstart would be more logical. */
d4c19fe8 1618 return is_utf8_common(p, &PL_utf8_idstart, "IdStart");
82686b01
JH
1619}
1620
1621bool
c11ff943
KW
1622Perl_is_utf8_xidfirst(pTHX_ const U8 *p) /* The naming is historical. */
1623{
1624 dVAR;
1625
1626 PERL_ARGS_ASSERT_IS_UTF8_XIDFIRST;
1627
1628 if (*p == '_')
1629 return TRUE;
1630 /* is_utf8_idstart would be more logical. */
1631 return is_utf8_common(p, &PL_utf8_xidstart, "XIdStart");
1632}
1633
1634bool
b6912c02
KW
1635Perl__is_utf8__perl_idstart(pTHX_ const U8 *p)
1636{
1637 dVAR;
1638
1639 PERL_ARGS_ASSERT__IS_UTF8__PERL_IDSTART;
1640
1641 return is_utf8_common(p, &PL_utf8_perl_idstart, "_Perl_IDStart");
1642}
1643
1644bool
7fc63493 1645Perl_is_utf8_idcont(pTHX_ const U8 *p)
82686b01 1646{
97aff369 1647 dVAR;
7918f24d
NC
1648
1649 PERL_ARGS_ASSERT_IS_UTF8_IDCONT;
1650
d4c19fe8 1651 return is_utf8_common(p, &PL_utf8_idcont, "IdContinue");
a0ed51b3
LW
1652}
1653
1654bool
c11ff943
KW
1655Perl_is_utf8_xidcont(pTHX_ const U8 *p)
1656{
1657 dVAR;
1658
1659 PERL_ARGS_ASSERT_IS_UTF8_XIDCONT;
1660
c11ff943
KW
1661 return is_utf8_common(p, &PL_utf8_idcont, "XIdContinue");
1662}
1663
1664bool
7fc63493 1665Perl_is_utf8_alpha(pTHX_ const U8 *p)
a0ed51b3 1666{
97aff369 1667 dVAR;
7918f24d
NC
1668
1669 PERL_ARGS_ASSERT_IS_UTF8_ALPHA;
1670
d4c19fe8 1671 return is_utf8_common(p, &PL_utf8_alpha, "IsAlpha");
a0ed51b3
LW
1672}
1673
1674bool
7fc63493 1675Perl_is_utf8_ascii(pTHX_ const U8 *p)
b8c5462f 1676{
97aff369 1677 dVAR;
7918f24d
NC
1678
1679 PERL_ARGS_ASSERT_IS_UTF8_ASCII;
1680
bc39fe24
KW
1681 /* ASCII characters are the same whether in utf8 or not. So the macro
1682 * works on both utf8 and non-utf8 representations. */
1683 return isASCII(*p);
b8c5462f
JH
1684}
1685
1686bool
7fc63493 1687Perl_is_utf8_space(pTHX_ const U8 *p)
a0ed51b3 1688{
97aff369 1689 dVAR;
7918f24d
NC
1690
1691 PERL_ARGS_ASSERT_IS_UTF8_SPACE;
1692
a34094a9 1693 return is_utf8_common(p, &PL_utf8_space, "IsXPerlSpace");
a0ed51b3
LW
1694}
1695
1696bool
d1eb3177
YO
1697Perl_is_utf8_perl_space(pTHX_ const U8 *p)
1698{
1699 dVAR;
1700
1701 PERL_ARGS_ASSERT_IS_UTF8_PERL_SPACE;
1702
c4428693
KW
1703 /* Only true if is an ASCII space-like character, and ASCII is invariant
1704 * under utf8, so can just use the macro */
1705 return isSPACE_A(*p);
d1eb3177
YO
1706}
1707
1708bool
1709Perl_is_utf8_perl_word(pTHX_ const U8 *p)
1710{
1711 dVAR;
1712
1713 PERL_ARGS_ASSERT_IS_UTF8_PERL_WORD;
1714
c4428693
KW
1715 /* Only true if is an ASCII word character, and ASCII is invariant
1716 * under utf8, so can just use the macro */
1717 return isWORDCHAR_A(*p);
d1eb3177
YO
1718}
1719
1720bool
7fc63493 1721Perl_is_utf8_digit(pTHX_ const U8 *p)
a0ed51b3 1722{
97aff369 1723 dVAR;
7918f24d
NC
1724
1725 PERL_ARGS_ASSERT_IS_UTF8_DIGIT;
1726
d4c19fe8 1727 return is_utf8_common(p, &PL_utf8_digit, "IsDigit");
a0ed51b3
LW
1728}
1729
1730bool
d1eb3177
YO
1731Perl_is_utf8_posix_digit(pTHX_ const U8 *p)
1732{
1733 dVAR;
1734
1735 PERL_ARGS_ASSERT_IS_UTF8_POSIX_DIGIT;
1736
c4428693
KW
1737 /* Only true if is an ASCII digit character, and ASCII is invariant
1738 * under utf8, so can just use the macro */
1739 return isDIGIT_A(*p);
d1eb3177
YO
1740}
1741
1742bool
7fc63493 1743Perl_is_utf8_upper(pTHX_ const U8 *p)
a0ed51b3 1744{
97aff369 1745 dVAR;
7918f24d
NC
1746
1747 PERL_ARGS_ASSERT_IS_UTF8_UPPER;
1748
d4c19fe8 1749 return is_utf8_common(p, &PL_utf8_upper, "IsUppercase");
a0ed51b3
LW
1750}
1751
1752bool
7fc63493 1753Perl_is_utf8_lower(pTHX_ const U8 *p)
a0ed51b3 1754{
97aff369 1755 dVAR;
7918f24d
NC
1756
1757 PERL_ARGS_ASSERT_IS_UTF8_LOWER;
1758
d4c19fe8 1759 return is_utf8_common(p, &PL_utf8_lower, "IsLowercase");
a0ed51b3
LW
1760}
1761
1762bool
7fc63493 1763Perl_is_utf8_cntrl(pTHX_ const U8 *p)
b8c5462f 1764{
97aff369 1765 dVAR;
7918f24d
NC
1766
1767 PERL_ARGS_ASSERT_IS_UTF8_CNTRL;
1768
7b952154
KW
1769 if (isASCII(*p)) {
1770 return isCNTRL_A(*p);
1771 }
1772
1773 /* All controls are in Latin1 */
1774 if (! UTF8_IS_DOWNGRADEABLE_START(*p)) {
1775 return 0;
1776 }
1777 return isCNTRL_L1(TWO_BYTE_UTF8_TO_UNI(*p, *(p+1)));
b8c5462f
JH
1778}
1779
1780bool
7fc63493 1781Perl_is_utf8_graph(pTHX_ const U8 *p)
b8c5462f 1782{
97aff369 1783 dVAR;
7918f24d
NC
1784
1785 PERL_ARGS_ASSERT_IS_UTF8_GRAPH;
1786
d4c19fe8 1787 return is_utf8_common(p, &PL_utf8_graph, "IsGraph");
b8c5462f
JH
1788}
1789
1790bool
7fc63493 1791Perl_is_utf8_print(pTHX_ const U8 *p)
a0ed51b3 1792{
97aff369 1793 dVAR;
7918f24d
NC
1794
1795 PERL_ARGS_ASSERT_IS_UTF8_PRINT;
1796
d4c19fe8 1797 return is_utf8_common(p, &PL_utf8_print, "IsPrint");
a0ed51b3
LW
1798}
1799
1800bool
7fc63493 1801Perl_is_utf8_punct(pTHX_ const U8 *p)
b8c5462f 1802{
97aff369 1803 dVAR;
7918f24d
NC
1804
1805 PERL_ARGS_ASSERT_IS_UTF8_PUNCT;
1806
d4c19fe8 1807 return is_utf8_common(p, &PL_utf8_punct, "IsPunct");
b8c5462f
JH
1808}
1809
1810bool
7fc63493 1811Perl_is_utf8_xdigit(pTHX_ const U8 *p)
b8c5462f 1812{
97aff369 1813 dVAR;
7918f24d
NC
1814
1815 PERL_ARGS_ASSERT_IS_UTF8_XDIGIT;
1816
d1eb3177 1817 return is_utf8_common(p, &PL_utf8_xdigit, "IsXDigit");
b8c5462f
JH
1818}
1819
1820bool
7fc63493 1821Perl_is_utf8_mark(pTHX_ const U8 *p)
a0ed51b3 1822{
97aff369 1823 dVAR;
7918f24d
NC
1824
1825 PERL_ARGS_ASSERT_IS_UTF8_MARK;
1826
d4c19fe8 1827 return is_utf8_common(p, &PL_utf8_mark, "IsM");
a0ed51b3
LW
1828}
1829
37e2e78e
KW
1830bool
1831Perl_is_utf8_X_begin(pTHX_ const U8 *p)
1832{
1833 dVAR;
1834
1835 PERL_ARGS_ASSERT_IS_UTF8_X_BEGIN;
1836
1837 return is_utf8_common(p, &PL_utf8_X_begin, "_X_Begin");
1838}
1839
1840bool
1841Perl_is_utf8_X_extend(pTHX_ const U8 *p)
1842{
1843 dVAR;
1844
1845 PERL_ARGS_ASSERT_IS_UTF8_X_EXTEND;
1846
1847 return is_utf8_common(p, &PL_utf8_X_extend, "_X_Extend");
1848}
1849
1850bool
1851Perl_is_utf8_X_prepend(pTHX_ const U8 *p)
1852{
1853 dVAR;
1854
1855 PERL_ARGS_ASSERT_IS_UTF8_X_PREPEND;
1856
1857 return is_utf8_common(p, &PL_utf8_X_prepend, "GCB=Prepend");
1858}
1859
1860bool
1861Perl_is_utf8_X_non_hangul(pTHX_ const U8 *p)
1862{
1863 dVAR;
1864
1865 PERL_ARGS_ASSERT_IS_UTF8_X_NON_HANGUL;
1866
1867 return is_utf8_common(p, &PL_utf8_X_non_hangul, "HST=Not_Applicable");
1868}
1869
1870bool
1871Perl_is_utf8_X_L(pTHX_ const U8 *p)
1872{
1873 dVAR;
1874
1875 PERL_ARGS_ASSERT_IS_UTF8_X_L;
1876
1877 return is_utf8_common(p, &PL_utf8_X_L, "GCB=L");
1878}
1879
1880bool
1881Perl_is_utf8_X_LV(pTHX_ const U8 *p)
1882{
1883 dVAR;
1884
1885 PERL_ARGS_ASSERT_IS_UTF8_X_LV;
1886
1887 return is_utf8_common(p, &PL_utf8_X_LV, "GCB=LV");
1888}
1889
1890bool
1891Perl_is_utf8_X_LVT(pTHX_ const U8 *p)
1892{
1893 dVAR;
1894
1895 PERL_ARGS_ASSERT_IS_UTF8_X_LVT;
1896
1897 return is_utf8_common(p, &PL_utf8_X_LVT, "GCB=LVT");
1898}
1899
1900bool
1901Perl_is_utf8_X_T(pTHX_ const U8 *p)
1902{
1903 dVAR;
1904
1905 PERL_ARGS_ASSERT_IS_UTF8_X_T;
1906
1907 return is_utf8_common(p, &PL_utf8_X_T, "GCB=T");
1908}
1909
1910bool
1911Perl_is_utf8_X_V(pTHX_ const U8 *p)
1912{
1913 dVAR;
1914
1915 PERL_ARGS_ASSERT_IS_UTF8_X_V;
1916
1917 return is_utf8_common(p, &PL_utf8_X_V, "GCB=V");
1918}
1919
1920bool
1921Perl_is_utf8_X_LV_LVT_V(pTHX_ const U8 *p)
1922{
1923 dVAR;
1924
1925 PERL_ARGS_ASSERT_IS_UTF8_X_LV_LVT_V;
1926
1927 return is_utf8_common(p, &PL_utf8_X_LV_LVT_V, "_X_LV_LVT_V");
1928}
1929
6b5c0936 1930/*
87cea99e 1931=for apidoc to_utf8_case
6b5c0936
JH
1932
1933The "p" contains the pointer to the UTF-8 string encoding
1934the character that is being converted.
1935
1936The "ustrp" is a pointer to the character buffer to put the
1937conversion result to. The "lenp" is a pointer to the length
1938of the result.
1939
0134edef 1940The "swashp" is a pointer to the swash to use.
6b5c0936 1941
36bb2ab6 1942Both the special and normal mappings are stored in lib/unicore/To/Foo.pl,
8fe4d5b2 1943and loaded by SWASHNEW, using lib/utf8_heavy.pl. The special (usually,
0134edef 1944but not always, a multicharacter mapping), is tried first.
6b5c0936 1945
0134edef
JH
1946The "special" is a string like "utf8::ToSpecLower", which means the
1947hash %utf8::ToSpecLower. The access to the hash is through
1948Perl_to_utf8_case().
6b5c0936 1949
0134edef
JH
1950The "normal" is a string like "ToLower" which means the swash
1951%utf8::ToLower.
1952
1953=cut */
6b5c0936 1954
2104c8d9 1955UV
9a957fbc
AL
1956Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp,
1957 SV **swashp, const char *normal, const char *special)
a0ed51b3 1958{
97aff369 1959 dVAR;
89ebb4a3 1960 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
0134edef 1961 STRLEN len = 0;
aec46f14 1962 const UV uv0 = utf8_to_uvchr(p, NULL);
1feea2c7
JH
1963 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1964 * are necessary in EBCDIC, they are redundant no-ops
1965 * in ASCII-ish platforms, and hopefully optimized away. */
f54cb97a 1966 const UV uv1 = NATIVE_TO_UNI(uv0);
7918f24d
NC
1967
1968 PERL_ARGS_ASSERT_TO_UTF8_CASE;
1969
9ae3ac1a
KW
1970 /* Note that swash_fetch() doesn't output warnings for these because it
1971 * assumes we will */
8457b38f 1972 if (uv1 >= UNICODE_SURROGATE_FIRST) {
9ae3ac1a 1973 if (uv1 <= UNICODE_SURROGATE_LAST) {
8457b38f
KW
1974 if (ckWARN_d(WARN_SURROGATE)) {
1975 const char* desc = (PL_op) ? OP_DESC(PL_op) : normal;
1976 Perl_warner(aTHX_ packWARN(WARN_SURROGATE),
1977 "Operation \"%s\" returns its argument for UTF-16 surrogate U+%04"UVXf"", desc, uv1);
1978 }
9ae3ac1a
KW
1979 }
1980 else if (UNICODE_IS_SUPER(uv1)) {
8457b38f
KW
1981 if (ckWARN_d(WARN_NON_UNICODE)) {
1982 const char* desc = (PL_op) ? OP_DESC(PL_op) : normal;
1983 Perl_warner(aTHX_ packWARN(WARN_NON_UNICODE),
1984 "Operation \"%s\" returns its argument for non-Unicode code point 0x%04"UVXf"", desc, uv1);
1985 }
9ae3ac1a
KW
1986 }
1987
1988 /* Note that non-characters are perfectly legal, so no warning should
1989 * be given */
1990 }
1991
1feea2c7 1992 uvuni_to_utf8(tmpbuf, uv1);
0134edef
JH
1993
1994 if (!*swashp) /* load on-demand */
1995 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1996
a6f87d8c 1997 if (special) {
0134edef 1998 /* It might be "special" (sometimes, but not always,
2a37f04d 1999 * a multicharacter mapping) */
6673a63c 2000 HV * const hv = get_hv(special, 0);
b08cf34e
JH
2001 SV **svp;
2002
35da51f7 2003 if (hv &&
b08cf34e
JH
2004 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
2005 (*svp)) {
cfd0369c 2006 const char *s;
47654450 2007
cfd0369c 2008 s = SvPV_const(*svp, len);
47654450
JH
2009 if (len == 1)
2010 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
2a37f04d 2011 else {
2f9475ad
JH
2012#ifdef EBCDIC
2013 /* If we have EBCDIC we need to remap the characters
2014 * since any characters in the low 256 are Unicode
2015 * code points, not EBCDIC. */
7cda7a3d 2016 U8 *t = (U8*)s, *tend = t + len, *d;
2f9475ad
JH
2017
2018 d = tmpbuf;
b08cf34e 2019 if (SvUTF8(*svp)) {
2f9475ad
JH
2020 STRLEN tlen = 0;
2021
2022 while (t < tend) {
d4c19fe8 2023 const UV c = utf8_to_uvchr(t, &tlen);
2f9475ad
JH
2024 if (tlen > 0) {
2025 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
2026 t += tlen;
2027 }
2028 else
2029 break;
2030 }
2031 }
2032 else {
36fec512
JH
2033 while (t < tend) {
2034 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
2035 t++;
2036 }
2f9475ad
JH
2037 }
2038 len = d - tmpbuf;
2039 Copy(tmpbuf, ustrp, len, U8);
2040#else
d2dcd0fb 2041 Copy(s, ustrp, len, U8);
2f9475ad 2042#endif
29e98929 2043 }
983ffd37 2044 }
0134edef
JH
2045 }
2046
2047 if (!len && *swashp) {
d4c19fe8
AL
2048 const UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
2049
0134edef
JH
2050 if (uv2) {
2051 /* It was "normal" (a single character mapping). */
d4c19fe8 2052 const UV uv3 = UNI_TO_NATIVE(uv2);
e9101d72 2053 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
2a37f04d
JH
2054 }
2055 }
1feea2c7 2056
37e2e78e
KW
2057 if (!len) /* Neither: just copy. In other words, there was no mapping
2058 defined, which means that the code point maps to itself */
0134edef
JH
2059 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
2060
2a37f04d
JH
2061 if (lenp)
2062 *lenp = len;
2063
0134edef 2064 return len ? utf8_to_uvchr(ustrp, 0) : 0;
a0ed51b3
LW
2065}
2066
d3e79532 2067/*
87cea99e 2068=for apidoc to_utf8_upper
d3e79532
JH
2069
2070Convert the UTF-8 encoded character at p to its uppercase version and
2071store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
2072that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
2073the uppercase version may be longer than the original character.
d3e79532
JH
2074
2075The first character of the uppercased version is returned
2076(but note, as explained above, that there may be more.)
2077
2078=cut */
2079
2104c8d9 2080UV
7fc63493 2081Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
a0ed51b3 2082{
97aff369 2083 dVAR;
7918f24d
NC
2084
2085 PERL_ARGS_ASSERT_TO_UTF8_UPPER;
2086
3a4c58c9
KW
2087 if (UTF8_IS_INVARIANT(*p)) {
2088 return _to_upper_title_latin1(*p, ustrp, lenp, 'S');
2089 }
2090 else if UTF8_IS_DOWNGRADEABLE_START(*p) {
2091 return _to_upper_title_latin1(TWO_BYTE_UTF8_TO_UNI(*p, *(p+1)),
2092 ustrp, lenp, 'S');
2093 }
2094
50bda2c3 2095 return CALL_UPPER_CASE(p, ustrp, lenp);
983ffd37 2096}
a0ed51b3 2097
d3e79532 2098/*
87cea99e 2099=for apidoc to_utf8_title
d3e79532
JH
2100
2101Convert the UTF-8 encoded character at p to its titlecase version and
2102store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
2103that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
2104titlecase version may be longer than the original character.
d3e79532
JH
2105
2106The first character of the titlecased version is returned
2107(but note, as explained above, that there may be more.)
2108
2109=cut */
2110
983ffd37 2111UV
7fc63493 2112Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
983ffd37 2113{
97aff369 2114 dVAR;
7918f24d
NC
2115
2116 PERL_ARGS_ASSERT_TO_UTF8_TITLE;
2117
3a4c58c9
KW
2118 if (UTF8_IS_INVARIANT(*p)) {
2119 return _to_upper_title_latin1(*p, ustrp, lenp, 's');
2120 }
2121 else if UTF8_IS_DOWNGRADEABLE_START(*p) {
2122 return _to_upper_title_latin1(TWO_BYTE_UTF8_TO_UNI(*p, *(p+1)),
2123 ustrp, lenp, 's');
2124 }
2125
50bda2c3 2126 return CALL_TITLE_CASE(p, ustrp, lenp);
a0ed51b3
LW
2127}
2128
d3e79532 2129/*
87cea99e 2130=for apidoc to_utf8_lower
d3e79532
JH
2131
2132Convert the UTF-8 encoded character at p to its lowercase version and
2133store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
2134that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
2135lowercase version may be longer than the original character.
d3e79532
JH
2136
2137The first character of the lowercased version is returned
2138(but note, as explained above, that there may be more.)
2139
2140=cut */
2141
2104c8d9 2142UV
7fc63493 2143Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
a0ed51b3 2144{
97aff369 2145 dVAR;
7918f24d
NC
2146
2147 PERL_ARGS_ASSERT_TO_UTF8_LOWER;
2148
968c5e6a
KW
2149 if (UTF8_IS_INVARIANT(*p)) {
2150 return to_lower_latin1(*p, ustrp, lenp);
2151 }
2152 else if UTF8_IS_DOWNGRADEABLE_START(*p) {
2153 return to_lower_latin1(TWO_BYTE_UTF8_TO_UNI(*p, *(p+1)), ustrp, lenp);
2154 }
2155
50bda2c3 2156 return CALL_LOWER_CASE(p, ustrp, lenp);
b4e400f9
JH
2157}
2158
d3e79532 2159/*
87cea99e 2160=for apidoc to_utf8_fold
d3e79532
JH
2161
2162Convert the UTF-8 encoded character at p to its foldcase version and
2163store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3 2164that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
d3e79532
JH
2165foldcase version may be longer than the original character (up to
2166three characters).
2167
2168The first character of the foldcased version is returned
2169(but note, as explained above, that there may be more.)
2170
2171=cut */
2172
36bb2ab6
KW
2173/* Not currently externally documented is 'flags', which currently is non-zero
2174 * if full case folds are to be used; otherwise simple folds */
2175
b4e400f9 2176UV
36bb2ab6 2177Perl__to_utf8_fold_flags(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp, U8 flags)
b4e400f9 2178{
97aff369 2179 dVAR;
7918f24d 2180
36bb2ab6 2181 PERL_ARGS_ASSERT__TO_UTF8_FOLD_FLAGS;
7918f24d 2182
50bda2c3 2183 return CALL_FOLD_CASE(p, ustrp, lenp, flags);
a0ed51b3
LW
2184}
2185
711a919c
ST
2186/* Note:
2187 * A "swash" is a swatch hash.
2188 * A "swatch" is a bit vector generated by utf8.c:S_swash_get().
2189 * C<pkg> is a pointer to a package name for SWASHNEW, should be "utf8".
2190 * For other parameters, see utf8::SWASHNEW in lib/utf8_heavy.pl.
2191 */
a0ed51b3 2192SV*
7fc63493 2193Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
a0ed51b3 2194{
27da23d5 2195 dVAR;
a0ed51b3 2196 SV* retval;
8e84507e 2197 dSP;
7fc63493
AL
2198 const size_t pkg_len = strlen(pkg);
2199 const size_t name_len = strlen(name);
da51bb9b 2200 HV * const stash = gv_stashpvn(pkg, pkg_len, 0);
f8be5cf0 2201 SV* errsv_save;
88c57d91 2202 GV *method;
ce3b816e 2203
7918f24d
NC
2204 PERL_ARGS_ASSERT_SWASH_INIT;
2205
96ca9f55
DM
2206 PUSHSTACKi(PERLSI_MAGIC);
2207 ENTER;
ec34a119 2208 SAVEHINTS();
96ca9f55 2209 save_re_context();
ba6ff154
FC
2210 if (PL_parser && PL_parser->error_count)
2211 SAVEI8(PL_parser->error_count), PL_parser->error_count = 0;
88c57d91
NC
2212 method = gv_fetchmeth(stash, "SWASHNEW", 8, -1);
2213 if (!method) { /* demand load utf8 */
ce3b816e 2214 ENTER;
f8be5cf0 2215 errsv_save = newSVsv(ERRSV);
dc0c6abb
NC
2216 /* It is assumed that callers of this routine are not passing in any
2217 user derived data. */
2218 /* Need to do this after save_re_context() as it will set PL_tainted to
2219 1 while saving $1 etc (see the code after getrx: in Perl_magic_get).
2220 Even line to create errsv_save can turn on PL_tainted. */
2221 SAVEBOOL(PL_tainted);
2222 PL_tainted = 0;
71bed85a 2223 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
a0714e2c 2224 NULL);
f8be5cf0
JH
2225 if (!SvTRUE(ERRSV))
2226 sv_setsv(ERRSV, errsv_save);
2227 SvREFCNT_dec(errsv_save);
ce3b816e
GS
2228 LEAVE;
2229 }
2230 SPAGAIN;
a0ed51b3
LW
2231 PUSHMARK(SP);
2232 EXTEND(SP,5);
6e449a3a
MHM
2233 mPUSHp(pkg, pkg_len);
2234 mPUSHp(name, name_len);
a0ed51b3 2235 PUSHs(listsv);
6e449a3a
MHM
2236 mPUSHi(minbits);
2237 mPUSHi(none);
a0ed51b3 2238 PUTBACK;
f8be5cf0 2239 errsv_save = newSVsv(ERRSV);
88c57d91
NC
2240 /* If we already have a pointer to the method, no need to use call_method()
2241 to repeat the lookup. */
2242 if (method ? call_sv(MUTABLE_SV(method), G_SCALAR)
8b563eef 2243 : call_sv(newSVpvs_flags("SWASHNEW", SVs_TEMP), G_SCALAR | G_METHOD))
8e84507e 2244 retval = newSVsv(*PL_stack_sp--);
a0ed51b3 2245 else
e24b16f9 2246 retval = &PL_sv_undef;
f8be5cf0
JH
2247 if (!SvTRUE(ERRSV))
2248 sv_setsv(ERRSV, errsv_save);
2249 SvREFCNT_dec(errsv_save);
a0ed51b3
LW
2250 LEAVE;
2251 POPSTACK;
923e4eb5 2252 if (IN_PERL_COMPILETIME) {
623e6609 2253 CopHINTS_set(PL_curcop, PL_hints);
a0ed51b3 2254 }
bc45ce41
JH
2255 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
2256 if (SvPOK(retval))
35c1215d 2257 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
be2597df 2258 SVfARG(retval));
cea2e8a9 2259 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
bc45ce41 2260 }
a0ed51b3
LW
2261 return retval;
2262}
2263
035d37be
JH
2264
2265/* This API is wrong for special case conversions since we may need to
2266 * return several Unicode characters for a single Unicode character
2267 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
2268 * the lower-level routine, and it is similarly broken for returning
38684baa
KW
2269 * multiple values. --jhi
2270 * For those, you should use to_utf8_case() instead */
979f2922 2271/* Now SWASHGET is recasted into S_swash_get in this file. */
680c470c
ST
2272
2273/* Note:
2274 * Returns the value of property/mapping C<swash> for the first character
2275 * of the string C<ptr>. If C<do_utf8> is true, the string C<ptr> is
2276 * assumed to be in utf8. If C<do_utf8> is false, the string C<ptr> is
2277 * assumed to be in native 8-bit encoding. Caches the swatch in C<swash>.
2278 */
a0ed51b3 2279UV
680c470c 2280Perl_swash_fetch(pTHX_ SV *swash, const U8 *ptr, bool do_utf8)
a0ed51b3 2281{
27da23d5 2282 dVAR;
ef8f7699 2283 HV *const hv = MUTABLE_HV(SvRV(swash));
3568d838
JH
2284 U32 klen;
2285 U32 off;
a0ed51b3 2286 STRLEN slen;
7d85a32c 2287 STRLEN needents;
cfd0369c 2288 const U8 *tmps = NULL;
a0ed51b3 2289 U32 bit;
979f2922 2290 SV *swatch;
3568d838 2291 U8 tmputf8[2];
35da51f7 2292 const UV c = NATIVE_TO_ASCII(*ptr);
3568d838 2293
7918f24d
NC
2294 PERL_ARGS_ASSERT_SWASH_FETCH;
2295
3568d838 2296 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
979f2922
ST
2297 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
2298 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
2299 ptr = tmputf8;
3568d838
JH
2300 }
2301 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
37e2e78e 2302 * then the "swatch" is a vec() for all the chars which start
3568d838
JH
2303 * with 0xAA..0xYY
2304 * So the key in the hash (klen) is length of encoded char -1
2305 */
2306 klen = UTF8SKIP(ptr) - 1;
2307 off = ptr[klen];
a0ed51b3 2308
979f2922 2309 if (klen == 0) {
37e2e78e 2310 /* If char is invariant then swatch is for all the invariant chars
1e54db1a 2311 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
7d85a32c 2312 */
979f2922
ST
2313 needents = UTF_CONTINUATION_MARK;
2314 off = NATIVE_TO_UTF(ptr[klen]);
2315 }
2316 else {
7d85a32c 2317 /* If char is encoded then swatch is for the prefix */
979f2922
ST
2318 needents = (1 << UTF_ACCUMULATION_SHIFT);
2319 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
8457b38f 2320 if (UTF8_IS_SUPER(ptr) && ckWARN_d(WARN_NON_UNICODE)) {
9ae3ac1a
KW
2321 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0, 0);
2322
2323 /* This outputs warnings for binary properties only, assuming that
9479a769
KW
2324 * to_utf8_case() will output any for non-binary. Also, surrogates
2325 * aren't checked for, as that would warn on things like
2326 * /\p{Gc=Cs}/ */
9ae3ac1a
KW
2327 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2328 if (SvUV(*bitssvp) == 1) {
8457b38f 2329 Perl_warner(aTHX_ packWARN(WARN_NON_UNICODE),
c634fdd3 2330 "Code point 0x%04"UVXf" is not Unicode, all \\p{} matches fail; all \\P{} matches succeed", code_point);
9ae3ac1a
KW
2331 }
2332 }
979f2922 2333 }
7d85a32c 2334
a0ed51b3
LW
2335 /*
2336 * This single-entry cache saves about 1/3 of the utf8 overhead in test
2337 * suite. (That is, only 7-8% overall over just a hash cache. Still,
2338 * it's nothing to sniff at.) Pity we usually come through at least
2339 * two function calls to get here...
2340 *
2341 * NB: this code assumes that swatches are never modified, once generated!
2342 */
2343
3568d838 2344 if (hv == PL_last_swash_hv &&
a0ed51b3 2345 klen == PL_last_swash_klen &&
27da23d5 2346 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
a0ed51b3
LW
2347 {
2348 tmps = PL_last_swash_tmps;
2349 slen = PL_last_swash_slen;
2350 }
2351 else {
2352 /* Try our second-level swatch cache, kept in a hash. */
e1ec3a88 2353 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
a0ed51b3 2354
979f2922
ST
2355 /* If not cached, generate it via swash_get */
2356 if (!svp || !SvPOK(*svp)
2357 || !(tmps = (const U8*)SvPV_const(*svp, slen))) {
2b9d42f0
NIS
2358 /* We use utf8n_to_uvuni() as we want an index into
2359 Unicode tables, not a native character number.
2360 */
aec46f14 2361 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
872c91ae
JH
2362 ckWARN(WARN_UTF8) ?
2363 0 : UTF8_ALLOW_ANY);
680c470c 2364 swatch = swash_get(swash,
979f2922
ST
2365 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
2366 (klen) ? (code_point & ~(needents - 1)) : 0,
2367 needents);
2368
923e4eb5 2369 if (IN_PERL_COMPILETIME)
623e6609 2370 CopHINTS_set(PL_curcop, PL_hints);
a0ed51b3 2371
979f2922 2372 svp = hv_store(hv, (const char *)ptr, klen, swatch, 0);
a0ed51b3 2373
979f2922
ST
2374 if (!svp || !(tmps = (U8*)SvPV(*svp, slen))
2375 || (slen << 3) < needents)
660a4616 2376 Perl_croak(aTHX_ "panic: swash_fetch got improper swatch");
a0ed51b3
LW
2377 }
2378
2379 PL_last_swash_hv = hv;
16d8f38a 2380 assert(klen <= sizeof(PL_last_swash_key));
eac04b2e 2381 PL_last_swash_klen = (U8)klen;
cfd0369c
NC
2382 /* FIXME change interpvar.h? */
2383 PL_last_swash_tmps = (U8 *) tmps;
a0ed51b3
LW
2384 PL_last_swash_slen = slen;
2385 if (klen)
2386 Copy(ptr, PL_last_swash_key, klen, U8);
2387 }
2388
9faf8d75 2389 switch ((int)((slen << 3) / needents)) {
a0ed51b3
LW
2390 case 1:
2391 bit = 1 << (off & 7);
2392 off >>= 3;
2393 return (tmps[off] & bit) != 0;
2394 case 8:
2395 return tmps[off];
2396 case 16:
2397 off <<= 1;
2398 return (tmps[off] << 8) + tmps[off + 1] ;
2399 case 32:
2400 off <<= 2;
2401 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
2402 }
660a4616 2403 Perl_croak(aTHX_ "panic: swash_fetch got swatch of unexpected bit width");
670f1322 2404 NORETURN_FUNCTION_END;
a0ed51b3 2405}
2b9d42f0 2406
319009ee
KW
2407/* Read a single line of the main body of the swash input text. These are of
2408 * the form:
2409 * 0053 0056 0073
2410 * where each number is hex. The first two numbers form the minimum and
2411 * maximum of a range, and the third is the value associated with the range.
2412 * Not all swashes should have a third number
2413 *
2414 * On input: l points to the beginning of the line to be examined; it points
2415 * to somewhere in the string of the whole input text, and is
2416 * terminated by a \n or the null string terminator.
2417 * lend points to the null terminator of that string
2418 * wants_value is non-zero if the swash expects a third number
2419 * typestr is the name of the swash's mapping, like 'ToLower'
2420 * On output: *min, *max, and *val are set to the values read from the line.
2421 * returns a pointer just beyond the line examined. If there was no
2422 * valid min number on the line, returns lend+1
2423 */
2424
2425STATIC U8*
2426S_swash_scan_list_line(pTHX_ U8* l, U8* const lend, UV* min, UV* max, UV* val,
2427 const bool wants_value, const U8* const typestr)
2428{
2429 const int typeto = typestr[0] == 'T' && typestr[1] == 'o';
2430 STRLEN numlen; /* Length of the number */
02470786
KW
2431 I32 flags = PERL_SCAN_SILENT_ILLDIGIT
2432 | PERL_SCAN_DISALLOW_PREFIX
2433 | PERL_SCAN_SILENT_NON_PORTABLE;
319009ee
KW
2434
2435 /* nl points to the next \n in the scan */
2436 U8* const nl = (U8*)memchr(l, '\n', lend - l);
2437
2438 /* Get the first number on the line: the range minimum */
2439 numlen = lend - l;
2440 *min = grok_hex((char *)l, &numlen, &flags, NULL);
2441 if (numlen) /* If found a hex number, position past it */
2442 l += numlen;
2443 else if (nl) { /* Else, go handle next line, if any */
2444 return nl + 1; /* 1 is length of "\n" */
2445 }
2446 else { /* Else, no next line */
2447 return lend + 1; /* to LIST's end at which \n is not found */
2448 }
2449
2450 /* The max range value follows, separated by a BLANK */
2451 if (isBLANK(*l)) {
2452 ++l;
02470786
KW
2453 flags = PERL_SCAN_SILENT_ILLDIGIT
2454 | PERL_SCAN_DISALLOW_PREFIX
2455 | PERL_SCAN_SILENT_NON_PORTABLE;
319009ee
KW
2456 numlen = lend - l;
2457 *max = grok_hex((char *)l, &numlen, &flags, NULL);
2458 if (numlen)
2459 l += numlen;
2460 else /* If no value here, it is a single element range */
2461 *max = *min;
2462
2463 /* Non-binary tables have a third entry: what the first element of the
2464 * range maps to */
2465 if (wants_value) {
2466 if (isBLANK(*l)) {
2467 ++l;
02470786
KW
2468 flags = PERL_SCAN_SILENT_ILLDIGIT
2469 | PERL_SCAN_DISALLOW_PREFIX
2470 | PERL_SCAN_SILENT_NON_PORTABLE;
319009ee
KW
2471 numlen = lend - l;
2472 *val = grok_hex((char *)l, &numlen, &flags, NULL);
2473 if (numlen)
2474 l += numlen;
2475 else
2476 *val = 0;
2477 }
2478 else {
2479 *val = 0;
2480 if (typeto) {
2481 Perl_croak(aTHX_ "%s: illegal mapping '%s'",
2482 typestr, l);
2483 }
2484 }
2485 }
2486 else
2487 *val = 0; /* bits == 1, then any val should be ignored */
2488 }
2489 else { /* Nothing following range min, should be single element with no
2490 mapping expected */
2491 *max = *min;
2492 if (wants_value) {
2493 *val = 0;
2494 if (typeto) {
2495 Perl_croak(aTHX_ "%s: illegal mapping '%s'", typestr, l);
2496 }
2497 }
2498 else
2499 *val = 0; /* bits == 1, then val should be ignored */
2500 }
2501
2502 /* Position to next line if any, or EOF */
2503 if (nl)
2504 l = nl + 1;
2505 else
2506 l = lend;
2507
2508 return l;
2509}
2510
979f2922
ST
2511/* Note:
2512 * Returns a swatch (a bit vector string) for a code point sequence
2513 * that starts from the value C<start> and comprises the number C<span>.
2514 * A C<swash> must be an object created by SWASHNEW (see lib/utf8_heavy.pl).
2515 * Should be used via swash_fetch, which will cache the swatch in C<swash>.
2516 */
2517STATIC SV*
2518S_swash_get(pTHX_ SV* swash, UV start, UV span)
2519{
2520 SV *swatch;
77f9f126 2521 U8 *l, *lend, *x, *xend, *s, *send;
979f2922 2522 STRLEN lcur, xcur, scur;
ef8f7699 2523 HV *const hv = MUTABLE_HV(SvRV(swash));
972dd592
KW
2524
2525 /* The string containing the main body of the table */
017a3ce5 2526 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
972dd592 2527
017a3ce5
GA
2528 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2529 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2530 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2531 SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);
77f9f126 2532 SV** const invert_it_svp = hv_fetchs(hv, "INVERT_IT", FALSE);
0bd48802 2533 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
0bd48802
AL
2534 const STRLEN bits = SvUV(*bitssvp);
2535 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2536 const UV none = SvUV(*nonesvp);
2537 const UV end = start + span;
979f2922 2538
7918f24d
NC
2539 PERL_ARGS_ASSERT_SWASH_GET;
2540
979f2922 2541 if (bits != 1 && bits != 8 && bits != 16 && bits != 32) {
660a4616
ST
2542 Perl_croak(aTHX_ "panic: swash_get doesn't expect bits %"UVuf,
2543 (UV)bits);
979f2922
ST
2544 }
2545
2546 /* create and initialize $swatch */
979f2922 2547 scur = octets ? (span * octets) : (span + 7) / 8;
e524fe40
NC
2548 swatch = newSV(scur);
2549 SvPOK_on(swatch);
979f2922
ST
2550 s = (U8*)SvPVX(swatch);
2551 if (octets && none) {
0bd48802 2552 const U8* const e = s + scur;
979f2922
ST
2553 while (s < e) {
2554 if (bits == 8)
2555 *s++ = (U8)(none & 0xff);
2556 else if (bits == 16) {
2557 *s++ = (U8)((none >> 8) & 0xff);
2558 *s++ = (U8)( none & 0xff);
2559 }
2560 else if (bits == 32) {
2561 *s++ = (U8)((none >> 24) & 0xff);
2562 *s++ = (U8)((none >> 16) & 0xff);
2563 *s++ = (U8)((none >> 8) & 0xff);
2564 *s++ = (U8)( none & 0xff);
2565 }
2566 }
2567 *s = '\0';
2568 }
2569 else {
2570 (void)memzero((U8*)s, scur + 1);
2571 }
2572 SvCUR_set(swatch, scur);
2573 s = (U8*)SvPVX(swatch);
2574
2575 /* read $swash->{LIST} */
2576 l = (U8*)SvPV(*listsvp, lcur);
2577 lend = l + lcur;
2578 while (l < lend) {
35da51f7 2579 UV min, max, val;
319009ee
KW
2580 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2581 cBOOL(octets), typestr);
2582 if (l > lend) {
979f2922
ST
2583 break;
2584 }
2585
972dd592 2586 /* If looking for something beyond this range, go try the next one */
979f2922
ST
2587 if (max < start)
2588 continue;
2589
2590 if (octets) {
35da51f7 2591 UV key;
979f2922
ST
2592 if (min < start) {
2593 if (!none || val < none) {
2594 val += start - min;
2595 }
2596 min = start;
2597 }
2598 for (key = min; key <= max; key++) {
2599 STRLEN offset;
2600 if (key >= end)
2601 goto go_out_list;
2602 /* offset must be non-negative (start <= min <= key < end) */
2603 offset = octets * (key - start);
2604 if (bits == 8)
2605 s[offset] = (U8)(val & 0xff);
2606 else if (bits == 16) {
2607 s[offset ] = (U8)((val >> 8) & 0xff);
2608 s[offset + 1] = (U8)( val & 0xff);
2609 }
2610 else if (bits == 32) {
2611 s[offset ] = (U8)((val >> 24) & 0xff);
2612 s[offset + 1] = (U8)((val >> 16) & 0xff);
2613 s[offset + 2] = (U8)((val >> 8) & 0xff);
2614 s[offset + 3] = (U8)( val & 0xff);
2615 }
2616
2617 if (!none || val < none)
2618 ++val;
2619 }
2620 }
711a919c 2621 else { /* bits == 1, then val should be ignored */
35da51f7 2622 UV key;
979f2922
ST
2623 if (min < start)
2624 min = start;
2625 for (key = min; key <= max; key++) {
0bd48802 2626 const STRLEN offset = (STRLEN)(key - start);
979f2922
ST
2627 if (key >= end)
2628 goto go_out_list;
2629 s[offset >> 3] |= 1 << (offset & 7);
2630 }
2631 }
2632 } /* while */
2633 go_out_list:
2634
9479a769 2635 /* Invert if the data says it should be. Assumes that bits == 1 */
77f9f126 2636 if (invert_it_svp && SvUV(*invert_it_svp)) {
0bda3001
KW
2637
2638 /* Unicode properties should come with all bits above PERL_UNICODE_MAX
2639 * be 0, and their inversion should also be 0, as we don't succeed any
2640 * Unicode property matches for non-Unicode code points */
2641 if (start <= PERL_UNICODE_MAX) {
2642
2643 /* The code below assumes that we never cross the
2644 * Unicode/above-Unicode boundary in a range, as otherwise we would
2645 * have to figure out where to stop flipping the bits. Since this
2646 * boundary is divisible by a large power of 2, and swatches comes
2647 * in small powers of 2, this should be a valid assumption */
2648 assert(start + span - 1 <= PERL_UNICODE_MAX);
2649
507a8485
KW
2650 send = s + scur;
2651 while (s < send) {
2652 *s = ~(*s);
2653 s++;
2654 }
0bda3001 2655 }
77f9f126
KW
2656 }
2657
d73c39c5
KW
2658 /* read $swash->{EXTRAS}
2659 * This code also copied to swash_to_invlist() below */
979f2922
ST
2660 x = (U8*)SvPV(*extssvp, xcur);
2661 xend = x + xcur;
2662 while (x < xend) {
2663 STRLEN namelen;
2664 U8 *namestr;
2665 SV** othersvp;
2666 HV* otherhv;
2667 STRLEN otherbits;
2668 SV **otherbitssvp, *other;
711a919c 2669 U8 *s, *o, *nl;
979f2922
ST
2670 STRLEN slen, olen;
2671
35da51f7 2672 const U8 opc = *x++;
979f2922
ST
2673 if (opc == '\n')
2674 continue;
2675
2676 nl = (U8*)memchr(x, '\n', xend - x);
2677
2678 if (opc != '-' && opc != '+' && opc != '!' && opc != '&') {
2679 if (nl) {
2680 x = nl + 1; /* 1 is length of "\n" */
2681 continue;
2682 }
2683 else {
2684 x = xend; /* to EXTRAS' end at which \n is not found */
2685 break;
2686 }
2687 }
2688
2689 namestr = x;
2690 if (nl) {
2691 namelen = nl - namestr;
2692 x = nl + 1;
2693 }
2694 else {
2695 namelen = xend - namestr;
2696 x = xend;
2697 }
2698
2699 othersvp = hv_fetch(hv, (char *)namestr, namelen, FALSE);
ef8f7699 2700 otherhv = MUTABLE_HV(SvRV(*othersvp));
017a3ce5 2701 otherbitssvp = hv_fetchs(otherhv, "BITS", FALSE);
979f2922
ST
2702 otherbits = (STRLEN)SvUV(*otherbitssvp);
2703 if (bits < otherbits)
660a4616 2704 Perl_croak(aTHX_ "panic: swash_get found swatch size mismatch");
979f2922
ST
2705
2706 /* The "other" swatch must be destroyed after. */
2707 other = swash_get(*othersvp, start, span);
2708 o = (U8*)SvPV(other, olen);
2709
2710 if (!olen)
660a4616 2711 Perl_croak(aTHX_ "panic: swash_get got improper swatch");
979f2922
ST
2712
2713 s = (U8*)SvPV(swatch, slen);
2714 if (bits == 1 && otherbits == 1) {
2715 if (slen != olen)
660a4616 2716 Perl_croak(aTHX_ "panic: swash_get found swatch length mismatch");
979f2922
ST
2717
2718 switch (opc) {
2719 case '+':
2720 while (slen--)
2721 *s++ |= *o++;
2722 break;
2723 case '!':
2724 while (slen--)
2725 *s++ |= ~*o++;
2726 break;
2727 case '-':
2728 while (slen--)
2729 *s++ &= ~*o++;
2730 break;
2731 case '&':
2732 while (slen--)
2733 *s++ &= *o++;
2734 break;
2735 default:
2736 break;
2737 }
2738 }
711a919c 2739 else {
979f2922
ST
2740 STRLEN otheroctets = otherbits >> 3;
2741 STRLEN offset = 0;
35da51f7 2742 U8* const send = s + slen;
979f2922
ST
2743
2744 while (s < send) {
2745 UV otherval = 0;
2746
2747 if (otherbits == 1) {
2748 otherval = (o[offset >> 3] >> (offset & 7)) & 1;
2749 ++offset;
2750 }
2751 else {
2752 STRLEN vlen = otheroctets;
2753 otherval = *o++;
2754 while (--vlen) {
2755 otherval <<= 8;
2756 otherval |= *o++;
2757 }
2758 }
2759
711a919c 2760 if (opc == '+' && otherval)
6f207bd3 2761 NOOP; /* replace with otherval */
979f2922
ST
2762 else if (opc == '!' && !otherval)
2763 otherval = 1;
2764 else if (opc == '-' && otherval)
2765 otherval = 0;
2766 else if (opc == '&' && !otherval)
2767 otherval = 0;
2768 else {
711a919c 2769 s += octets; /* no replacement */
979f2922
ST
2770 continue;
2771 }
2772
2773 if (bits == 8)
2774 *s++ = (U8)( otherval & 0xff);
2775 else if (bits == 16) {
2776 *s++ = (U8)((otherval >> 8) & 0xff);
2777 *s++ = (U8)( otherval & 0xff);
2778 }
2779 else if (bits == 32) {
2780 *s++ = (U8)((otherval >> 24) & 0xff);
2781 *s++ = (U8)((otherval >> 16) & 0xff);
2782 *s++ = (U8)((otherval >> 8) & 0xff);
2783 *s++ = (U8)( otherval & 0xff);
2784 }
2785 }
2786 }
2787 sv_free(other); /* through with it! */
2788 } /* while */
2789 return swatch;
2790}
2791
064c021d 2792HV*
4c2e1131 2793Perl__swash_inversion_hash(pTHX_ SV* const swash)
064c021d
KW
2794{
2795
5662e334
KW
2796 /* Subject to change or removal. For use only in one place in regcomp.c.
2797 * Can't be used on a property that is subject to user override, as it
2798 * relies on the value of SPECIALS in the swash which would be set by
2799 * utf8_heavy.pl to the hash in the non-overriden file, and hence is not set
2800 * for overridden properties
064c021d
KW
2801 *
2802 * Returns a hash which is the inversion and closure of a swash mapping.
2803 * For example, consider the input lines:
2804 * 004B 006B
2805 * 004C 006C
2806 * 212A 006B
2807 *
2808 * The returned hash would have two keys, the utf8 for 006B and the utf8 for
2809 * 006C. The value for each key is an array. For 006C, the array would
2810 * have a two elements, the utf8 for itself, and for 004C. For 006B, there
2811 * would be three elements in its array, the utf8 for 006B, 004B and 212A.
2812 *
2813 * Essentially, for any code point, it gives all the code points that map to
2814 * it, or the list of 'froms' for that point.
2815 *
5662e334
KW
2816 * Currently it ignores any additions or deletions from other swashes,
2817 * looking at just the main body of the swash, and if there are SPECIALS
2818 * in the swash, at that hash
2819 *
2820 * The specials hash can be extra code points, and most likely consists of
2821 * maps from single code points to multiple ones (each expressed as a string
2822 * of utf8 characters). This function currently returns only 1-1 mappings.
2823 * However consider this possible input in the specials hash:
2824 * "\xEF\xAC\x85" => "\x{0073}\x{0074}", # U+FB05 => 0073 0074
2825 * "\xEF\xAC\x86" => "\x{0073}\x{0074}", # U+FB06 => 0073 0074
2826 *
2827 * Both FB05 and FB06 map to the same multi-char sequence, which we don't
2828 * currently handle. But it also means that FB05 and FB06 are equivalent in
2829 * a 1-1 mapping which we should handle, and this relationship may not be in
2830 * the main table. Therefore this function examines all the multi-char
2831 * sequences and adds the 1-1 mappings that come out of that. */
064c021d
KW
2832
2833 U8 *l, *lend;
2834 STRLEN lcur;
2835 HV *const hv = MUTABLE_HV(SvRV(swash));
2836
2837 /* The string containing the main body of the table */
2838 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2839
2840 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2841 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2842 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2843 /*SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);*/
2844 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
2845 const STRLEN bits = SvUV(*bitssvp);
2846 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2847 const UV none = SvUV(*nonesvp);
5662e334 2848 SV **specials_p = hv_fetchs(hv, "SPECIALS", 0);
064c021d
KW
2849
2850 HV* ret = newHV();
2851
2852 PERL_ARGS_ASSERT__SWASH_INVERSION_HASH;
2853
2854 /* Must have at least 8 bits to get the mappings */
2855 if (bits != 8 && bits != 16 && bits != 32) {
2856 Perl_croak(aTHX_ "panic: swash_inversion_hash doesn't expect bits %"UVuf,
2857 (UV)bits);
2858 }
2859
5662e334
KW
2860 if (specials_p) { /* It might be "special" (sometimes, but not always, a
2861 mapping to more than one character */
2862
2863 /* Construct an inverse mapping hash for the specials */
2864 HV * const specials_hv = MUTABLE_HV(SvRV(*specials_p));
2865 HV * specials_inverse = newHV();
2866 char *char_from; /* the lhs of the map */
2867 I32 from_len; /* its byte length */
2868 char *char_to; /* the rhs of the map */
2869 I32 to_len; /* its byte length */
2870 SV *sv_to; /* and in a sv */
2871 AV* from_list; /* list of things that map to each 'to' */
2872
2873 hv_iterinit(specials_hv);
2874
2875 /* The keys are the characters (in utf8) that map to the corresponding
2876 * utf8 string value. Iterate through the list creating the inverse
2877 * list. */
2878 while ((sv_to = hv_iternextsv(specials_hv, &char_from, &from_len))) {
2879 SV** listp;
2880 if (! SvPOK(sv_to)) {
2881 Perl_croak(aTHX_ "panic: value returned from hv_iternextsv() unexpectedly is not a string");
2882 }
2883 /*DEBUG_U(PerlIO_printf(Perl_debug_log, "Found mapping from %"UVXf", First char of to is %"UVXf"\n", utf8_to_uvchr((U8*) char_from, 0), utf8_to_uvchr((U8*) SvPVX(sv_to), 0)));*/
2884
2885 /* Each key in the inverse list is a mapped-to value, and the key's
2886 * hash value is a list of the strings (each in utf8) that map to
2887 * it. Those strings are all one character long */
2888 if ((listp = hv_fetch(specials_inverse,
2889 SvPVX(sv_to),
2890 SvCUR(sv_to), 0)))
2891 {
2892 from_list = (AV*) *listp;
2893 }
2894 else { /* No entry yet for it: create one */
2895 from_list = newAV();
2896 if (! hv_store(specials_inverse,
2897 SvPVX(sv_to),
2898 SvCUR(sv_to),
2899 (SV*) from_list, 0))
2900 {
2901 Perl_croak(aTHX_ "panic: hv_store() unexpectedly failed");
2902 }
2903 }
2904
2905 /* Here have the list associated with this 'to' (perhaps newly
2906 * created and empty). Just add to it. Note that we ASSUME that
2907 * the input is guaranteed to not have duplications, so we don't
2908 * check for that. Duplications just slow down execution time. */
2909 av_push(from_list, newSVpvn_utf8(char_from, from_len, TRUE));
2910 }
2911
2912 /* Here, 'specials_inverse' contains the inverse mapping. Go through
2913 * it looking for cases like the FB05/FB06 examples above. There would
2914 * be an entry in the hash like
2915 * 'st' => [ FB05, FB06 ]
2916 * In this example we will create two lists that get stored in the
2917 * returned hash, 'ret':
2918 * FB05 => [ FB05, FB06 ]
2919 * FB06 => [ FB05, FB06 ]
2920 *
2921 * Note that there is nothing to do if the array only has one element.
2922 * (In the normal 1-1 case handled below, we don't have to worry about
2923 * two lists, as everything gets tied to the single list that is
2924 * generated for the single character 'to'. But here, we are omitting
2925 * that list, ('st' in the example), so must have multiple lists.) */
2926 while ((from_list = (AV *) hv_iternextsv(specials_inverse,
2927 &char_to, &to_len)))
2928 {
2929 if (av_len(from_list) > 0) {
2930 int i;
2931
2932 /* We iterate over all combinations of i,j to place each code
2933 * point on each list */
2934 for (i = 0; i <= av_len(from_list); i++) {
2935 int j;
2936 AV* i_list = newAV();
2937 SV** entryp = av_fetch(from_list, i, FALSE);
2938 if (entryp == NULL) {
2939 Perl_croak(aTHX_ "panic: av_fetch() unexpectedly failed");
2940 }
2941 if (hv_fetch(ret, SvPVX(*entryp), SvCUR(*entryp), FALSE)) {
2942 Perl_croak(aTHX_ "panic: unexpected entry for %s", SvPVX(*entryp));
2943 }
2944 if (! hv_store(ret, SvPVX(*entryp), SvCUR(*entryp),
2945 (SV*) i_list, FALSE))
2946 {
2947 Perl_croak(aTHX_ "panic: hv_store() unexpectedly failed");
2948 }
2949
2950 /* For debugging: UV u = utf8_to_uvchr((U8*) SvPVX(*entryp), 0);*/
2951 for (j = 0; j <= av_len(from_list); j++) {
2952 entryp = av_fetch(from_list, j, FALSE);
2953 if (entryp == NULL) {
2954 Perl_croak(aTHX_ "panic: av_fetch() unexpectedly failed");
2955 }
2956
2957 /* When i==j this adds itself to the list */
2958 av_push(i_list, newSVuv(utf8_to_uvchr(
2959 (U8*) SvPVX(*entryp), 0)));
2960 /*DEBUG_U(PerlIO_printf(Perl_debug_log, "Adding %"UVXf" to list for %"UVXf"\n", utf8_to_uvchr((U8*) SvPVX(*entryp), 0), u));*/
2961 }
2962 }
2963 }
2964 }
2965 SvREFCNT_dec(specials_inverse); /* done with it */
2966 } /* End of specials */
2967
064c021d
KW
2968 /* read $swash->{LIST} */
2969 l = (U8*)SvPV(*listsvp, lcur);
2970 lend = l + lcur;
2971
2972 /* Go through each input line */
2973 while (l < lend) {
2974 UV min, max, val;
2975 UV inverse;
2976 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2977 cBOOL(octets), typestr);
2978 if (l > lend) {
2979 break;
2980 }
2981
2982 /* Each element in the range is to be inverted */
2983 for (inverse = min; inverse <= max; inverse++) {
2984 AV* list;
064c021d
KW
2985 SV** listp;
2986 IV i;
2987 bool found_key = FALSE;
5662e334 2988 bool found_inverse = FALSE;
064c021d
KW
2989
2990 /* The key is the inverse mapping */
2991 char key[UTF8_MAXBYTES+1];
2992 char* key_end = (char *) uvuni_to_utf8((U8*) key, val);
2993 STRLEN key_len = key_end - key;
2994
064c021d
KW
2995 /* Get the list for the map */
2996 if ((listp = hv_fetch(ret, key, key_len, FALSE))) {
2997 list = (AV*) *listp;
2998 }
2999 else { /* No entry yet for it: create one */
3000 list = newAV();
3001 if (! hv_store(ret, key, key_len, (SV*) list, FALSE)) {
3002 Perl_croak(aTHX_ "panic: hv_store() unexpectedly failed");
3003 }
3004 }
3005
5662e334
KW
3006 /* Look through list to see if this inverse mapping already is
3007 * listed, or if there is a mapping to itself already */
508f7cfa 3008 for (i = 0; i <= av_len(list); i++) {
064c021d
KW
3009 SV** entryp = av_fetch(list, i, FALSE);
3010 SV* entry;
3011 if (entryp == NULL) {
3012 Perl_croak(aTHX_ "panic: av_fetch() unexpectedly failed");
3013 }
3014 entry = *entryp;
5662e334 3015 /*DEBUG_U(PerlIO_printf(Perl_debug_log, "list for %"UVXf" contains %"UVXf"\n", val, SvUV(entry)));*/
56ca34ca 3016 if (SvUV(entry) == val) {
064c021d 3017 found_key = TRUE;
5662e334
KW
3018 }
3019 if (SvUV(entry) == inverse) {
3020 found_inverse = TRUE;
3021 }
3022
3023 /* No need to continue searching if found everything we are
3024 * looking for */
3025 if (found_key && found_inverse) {
064c021d
KW
3026 break;
3027 }
3028 }
56ca34ca
KW
3029
3030 /* Make sure there is a mapping to itself on the list */
064c021d 3031 if (! found_key) {
d397ff6a 3032 av_push(list, newSVuv(val));
5662e334 3033 /*DEBUG_U(PerlIO_printf(Perl_debug_log, "Adding %"UVXf" to list for %"UVXf"\n", val, val));*/
064c021d
KW
3034 }
3035
3036
3037 /* Simply add the value to the list */
5662e334
KW
3038 if (! found_inverse) {
3039 av_push(list, newSVuv(inverse));
3040 /*DEBUG_U(PerlIO_printf(Perl_debug_log, "Adding %"UVXf" to list for %"UVXf"\n", inverse, val));*/
3041 }
064c021d
KW
3042
3043 /* swash_get() increments the value of val for each element in the
3044 * range. That makes more compact tables possible. You can
3045 * express the capitalization, for example, of all consecutive
3046 * letters with a single line: 0061\t007A\t0041 This maps 0061 to
3047 * 0041, 0062 to 0042, etc. I (khw) have never understood 'none',
bd3f2f94
KW
3048 * and it's not documented; it appears to be used only in
3049 * implementing tr//; I copied the semantics from swash_get(), just
3050 * in case */
064c021d
KW
3051 if (!none || val < none) {
3052 ++val;
3053 }
3054 }
3055 }
3056
3057 return ret;
3058}
3059
a25abddc 3060SV*
d764b54e
KW
3061Perl__swash_to_invlist(pTHX_ SV* const swash)
3062{
3063
3064 /* Subject to change or removal. For use only in one place in regcomp.c */
3065
3066 U8 *l, *lend;
3067 char *loc;
3068 STRLEN lcur;
3069 HV *const hv = MUTABLE_HV(SvRV(swash));
3070 UV elements = 0; /* Number of elements in the inversion list */
b443038a 3071 U8 empty[] = "";
d764b54e
KW
3072
3073 /* The string containing the main body of the table */
3074 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
3075 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
3076 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
d73c39c5 3077 SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);
77f9f126 3078 SV** const invert_it_svp = hv_fetchs(hv, "INVERT_IT", FALSE);
d764b54e
KW
3079
3080 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
3081 const STRLEN bits = SvUV(*bitssvp);
3082 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
d73c39c5
KW
3083 U8 *x, *xend;
3084 STRLEN xcur;
d764b54e 3085
a25abddc 3086 SV* invlist;
d764b54e
KW
3087
3088 PERL_ARGS_ASSERT__SWASH_TO_INVLIST;
3089
3090 /* read $swash->{LIST} */
b443038a
KW
3091 if (SvPOK(*listsvp)) {
3092 l = (U8*)SvPV(*listsvp, lcur);
3093 }
3094 else {
3095 /* LIST legitimately doesn't contain a string during compilation phases
3096 * of Perl itself, before the Unicode tables are generated. In this
3097 * case, just fake things up by creating an empty list */
3098 l = empty;
3099 lcur = 0;
3100 }
d764b54e
KW
3101 loc = (char *) l;
3102 lend = l + lcur;
3103
3104 /* Scan the input to count the number of lines to preallocate array size
3105 * based on worst possible case, which is each line in the input creates 2
3106 * elements in the inversion list: 1) the beginning of a range in the list;
3107 * 2) the beginning of a range not in the list. */
3108 while ((loc = (strchr(loc, '\n'))) != NULL) {
3109 elements += 2;
3110 loc++;
3111 }
3112
3113 /* If the ending is somehow corrupt and isn't a new line, add another
3114 * element for the final range that isn't in the inversion list */
3115 if (! (*lend == '\n' || (*lend == '\0' && *(lend - 1) == '\n'))) {
3116 elements++;
3117 }
3118
3119 invlist = _new_invlist(elements);
3120
3121 /* Now go through the input again, adding each range to the list */
3122 while (l < lend) {
3123 UV start, end;
3124 UV val; /* Not used by this function */
3125
3126 l = S_swash_scan_list_line(aTHX_ l, lend, &start, &end, &val,
3127 cBOOL(octets), typestr);
3128
3129 if (l > lend) {
3130 break;
3131 }
3132
3133 _append_range_to_invlist(invlist, start, end);
3134 }
3135
77f9f126
KW
3136 /* Invert if the data says it should be */
3137 if (invert_it_svp && SvUV(*invert_it_svp)) {
45bb2768 3138 _invlist_invert_prop(invlist);
77f9f126
KW
3139 }
3140
d73c39c5
KW
3141 /* This code is copied from swash_get()
3142 * read $swash->{EXTRAS} */
3143 x = (U8*)SvPV(*extssvp, xcur);
3144 xend = x + xcur;
3145 while (x < xend) {
3146 STRLEN namelen;
3147 U8 *namestr;
3148 SV** othersvp;
3149 HV* otherhv;
3150 STRLEN otherbits;
3151 SV **otherbitssvp, *other;
3152 U8 *nl;
3153
3154 const U8 opc = *x++;
3155 if (opc == '\n')
3156 continue;
3157
3158 nl = (U8*)memchr(x, '\n', xend - x);
3159
3160 if (opc != '-' && opc != '+' && opc != '!' && opc != '&') {
3161 if (nl) {
3162 x = nl + 1; /* 1 is length of "\n" */
3163 continue;
3164 }
3165 else {
3166 x = xend; /* to EXTRAS' end at which \n is not found */
3167 break;
3168 }
3169 }
3170
3171 namestr = x;
3172 if (nl) {
3173 namelen = nl - namestr;
3174 x = nl + 1;
3175 }
3176 else {
3177 namelen = xend - namestr;
3178 x = xend;
3179 }
3180
3181 othersvp = hv_fetch(hv, (char *)namestr, namelen, FALSE);
3182 otherhv = MUTABLE_HV(SvRV(*othersvp));
3183 otherbitssvp = hv_fetchs(otherhv, "BITS", FALSE);
3184 otherbits = (STRLEN)SvUV(*otherbitssvp);
3185
3186 if (bits != otherbits || bits != 1) {
3187 Perl_croak(aTHX_ "panic: _swash_to_invlist only operates on boolean properties");
3188 }
3189
3190 /* The "other" swatch must be destroyed after. */
3191 other = _swash_to_invlist((SV *)*othersvp);
3192
3193 /* End of code copied from swash_get() */
3194 switch (opc) {
3195 case '+':
3196 _invlist_union(invlist, other, &invlist);
3197 break;
3198 case '!':
3199 _invlist_invert(other);
3200 _invlist_union(invlist, other, &invlist);
3201 break;
3202 case '-':
3203 _invlist_subtract(invlist, other, &invlist);
3204 break;
3205 case '&':
3206 _invlist_intersection(invlist, other, &invlist);
3207 break;
3208 default:
3209 break;
3210 }
3211 sv_free(other); /* through with it! */
3212 }
3213
d764b54e
KW
3214 return invlist;
3215}
3216
0f830e0b 3217/*
87cea99e 3218=for apidoc uvchr_to_utf8
0f830e0b 3219
6ee84de2 3220Adds the UTF-8 representation of the Native code point C<uv> to the end
0f830e0b
NC
3221of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
3222bytes available. The return value is the pointer to the byte after the
3223end of the new character. In other words,
3224
3225 d = uvchr_to_utf8(d, uv);
3226
3227is the recommended wide native character-aware way of saying
3228
3229 *(d++) = uv;
3230
3231=cut
3232*/
3233
3234/* On ASCII machines this is normally a macro but we want a
3235 real function in case XS code wants it
3236*/
0f830e0b
NC
3237U8 *
3238Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
3239{
7918f24d
NC
3240 PERL_ARGS_ASSERT_UVCHR_TO_UTF8;
3241
0f830e0b
NC
3242 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
3243}
3244
b851fbc1
JH
3245U8 *
3246Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
3247{
7918f24d
NC
3248 PERL_ARGS_ASSERT_UVCHR_TO_UTF8_FLAGS;
3249
b851fbc1
JH
3250 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
3251}
2b9d42f0
NIS
3252
3253/*
87cea99e 3254=for apidoc utf8n_to_uvchr
0f830e0b 3255
48ef279e 3256Returns the native character value of the first character in the string
0f830e0b
NC
3257C<s>
3258which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
3259length, in bytes, of that character.
3260
6ee84de2 3261length and flags are the same as utf8n_to_uvuni().
0f830e0b
NC
3262
3263=cut
3264*/
3265/* On ASCII machines this is normally a macro but we want
3266 a real function in case XS code wants it
3267*/
0f830e0b 3268UV
48ef279e 3269Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen,
0f830e0b
NC
3270U32 flags)
3271{
3272 const UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
7918f24d
NC
3273
3274 PERL_ARGS_ASSERT_UTF8N_TO_UVCHR;
3275
0f830e0b
NC
3276 return UNI_TO_NATIVE(uv);
3277}
3278
0876b9a0
KW
3279bool
3280Perl_check_utf8_print(pTHX_ register const U8* s, const STRLEN len)
3281{
3282 /* May change: warns if surrogates, non-character code points, or
3283 * non-Unicode code points are in s which has length len. Returns TRUE if
3284 * none found; FALSE otherwise. The only other validity check is to make
3285 * sure that this won't exceed the string's length */
3286
3287 const U8* const e = s + len;
3288 bool ok = TRUE;
3289
3290 PERL_ARGS_ASSERT_CHECK_UTF8_PRINT;
3291
3292 while (s < e) {
3293 if (UTF8SKIP(s) > len) {
3294 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
3295 "%s in %s", unees, PL_op ? OP_DESC(PL_op) : "print");
3296 return FALSE;
3297 }
3298 if (*s >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE) {
3299 STRLEN char_len;
3300 if (UTF8_IS_SUPER(s)) {
8457b38f
KW
3301 if (ckWARN_d(WARN_NON_UNICODE)) {
3302 UV uv = utf8_to_uvchr(s, &char_len);
3303 Perl_warner(aTHX_ packWARN(WARN_NON_UNICODE),
3304 "Code point 0x%04"UVXf" is not Unicode, may not be portable", uv);
3305 ok = FALSE;
3306 }
0876b9a0
KW
3307 }
3308 else if (UTF8_IS_SURROGATE(s)) {
8457b38f
KW
3309 if (ckWARN_d(WARN_SURROGATE)) {
3310 UV uv = utf8_to_uvchr(s, &char_len);
3311 Perl_warner(aTHX_ packWARN(WARN_SURROGATE),
3312 "Unicode surrogate U+%04"UVXf" is illegal in UTF-8", uv);
3313 ok = FALSE;
3314 }
0876b9a0
KW
3315 }
3316 else if
8457b38f
KW
3317 ((UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s))
3318 && (ckWARN_d(WARN_NONCHAR)))
0876b9a0
KW
3319 {
3320 UV uv = utf8_to_uvchr(s, &char_len);
8457b38f 3321 Perl_warner(aTHX_ packWARN(WARN_NONCHAR),
0876b9a0
KW
3322 "Unicode non-character U+%04"UVXf" is illegal for open interchange", uv);
3323 ok = FALSE;
3324 }
3325 }
3326 s += UTF8SKIP(s);
3327 }
3328
3329 return ok;
3330}
3331
0f830e0b 3332/*
87cea99e 3333=for apidoc pv_uni_display
d2cc3551
JH
3334
3335Build to the scalar dsv a displayable version of the string spv,
3336length len, the displayable version being at most pvlim bytes long
3337(if longer, the rest is truncated and "..." will be appended).
0a2ef054 3338
9e55ce06 3339The flags argument can have UNI_DISPLAY_ISPRINT set to display
00e86452 3340isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
0a2ef054
JH
3341to display the \\[nrfta\\] as the backslashed versions (like '\n')
3342(UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
3343UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
3344UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
3345
d2cc3551
JH
3346The pointer to the PV of the dsv is returned.
3347
3348=cut */
e6b2e755 3349char *
e1ec3a88 3350Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
e6b2e755
JH
3351{
3352 int truncated = 0;
e1ec3a88 3353 const char *s, *e;
e6b2e755 3354
7918f24d
NC
3355 PERL_ARGS_ASSERT_PV_UNI_DISPLAY;
3356
76f68e9b 3357 sv_setpvs(dsv, "");
7fddd944 3358 SvUTF8_off(dsv);
e1ec3a88 3359 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
e6b2e755 3360 UV u;
a49f32c6
NC
3361 /* This serves double duty as a flag and a character to print after
3362 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
3363 */
3364 char ok = 0;
c728cb41 3365
e6b2e755
JH
3366 if (pvlim && SvCUR(dsv) >= pvlim) {
3367 truncated++;
3368 break;
3369 }
3370 u = utf8_to_uvchr((U8*)s, 0);
c728cb41 3371 if (u < 256) {
a3b680e6 3372 const unsigned char c = (unsigned char)u & 0xFF;
0bd48802 3373 if (flags & UNI_DISPLAY_BACKSLASH) {
a49f32c6 3374 switch (c) {
c728cb41 3375 case '\n':
a49f32c6 3376 ok = 'n'; break;
c728cb41 3377 case '\r':
a49f32c6 3378 ok = 'r'; break;
c728cb41 3379 case '\t':
a49f32c6 3380 ok = 't'; break;
c728cb41 3381 case '\f':
a49f32c6 3382 ok = 'f'; break;
c728cb41 3383 case '\a':
a49f32c6 3384 ok = 'a'; break;
c728cb41 3385 case '\\':
a49f32c6 3386 ok = '\\'; break;
c728cb41
JH
3387 default: break;
3388 }
a49f32c6 3389 if (ok) {
88c9ea1e 3390 const char string = ok;
76f68e9b 3391 sv_catpvs(dsv, "\\");
5e7aa789 3392 sv_catpvn(dsv, &string, 1);
a49f32c6 3393 }
c728cb41 3394 }
00e86452 3395 /* isPRINT() is the locale-blind version. */
a49f32c6 3396 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
88c9ea1e 3397 const char string = c;
5e7aa789 3398 sv_catpvn(dsv, &string, 1);
a49f32c6 3399 ok = 1;
0a2ef054 3400 }
c728cb41
JH
3401 }
3402 if (!ok)
9e55ce06 3403 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
e6b2e755
JH
3404 }
3405 if (truncated)
396482e1 3406 sv_catpvs(dsv, "...");
48ef279e 3407
e6b2e755
JH
3408 return SvPVX(dsv);
3409}
2b9d42f0 3410
d2cc3551 3411/*
87cea99e 3412=for apidoc sv_uni_display
d2cc3551
JH
3413
3414Build to the scalar dsv a displayable version of the scalar sv,
0a2ef054 3415the displayable version being at most pvlim bytes long
d2cc3551 3416(if longer, the rest is truncated and "..." will be appended).
0a2ef054
JH
3417
3418The flags argument is as in pv_uni_display().
3419
d2cc3551
JH
3420The pointer to the PV of the dsv is returned.
3421
d4c19fe8
AL
3422=cut
3423*/
e6b2e755
JH
3424char *
3425Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
3426{
7918f24d
NC
3427 PERL_ARGS_ASSERT_SV_UNI_DISPLAY;
3428
cfd0369c
NC
3429 return Perl_pv_uni_display(aTHX_ dsv, (const U8*)SvPVX_const(ssv),
3430 SvCUR(ssv), pvlim, flags);
701a277b
JH
3431}
3432
d2cc3551 3433/*
e6226b18 3434=for apidoc foldEQ_utf8
d2cc3551 3435
d51c1b21 3436Returns true if the leading portions of the strings s1 and s2 (either or both
e6226b18 3437of which may be in UTF-8) are the same case-insensitively; false otherwise.
d51c1b21 3438How far into the strings to compare is determined by other input parameters.
8b35872c
KW
3439
3440If u1 is true, the string s1 is assumed to be in UTF-8-encoded Unicode;
3441otherwise it is assumed to be in native 8-bit encoding. Correspondingly for u2
3442with respect to s2.
3443
d51c1b21
KW
3444If the byte length l1 is non-zero, it says how far into s1 to check for fold
3445equality. In other words, s1+l1 will be used as a goal to reach. The
8b35872c
KW
3446scan will not be considered to be a match unless the goal is reached, and
3447scanning won't continue past that goal. Correspondingly for l2 with respect to
3448s2.
3449
3450If pe1 is non-NULL and the pointer it points to is not NULL, that pointer is
3451considered an end pointer beyond which scanning of s1 will not continue under
3452any circumstances. This means that if both l1 and pe1 are specified, and pe1
3453is less than s1+l1, the match will never be successful because it can never
d51c1b21
KW
3454get as far as its goal (and in fact is asserted against). Correspondingly for
3455pe2 with respect to s2.
8b35872c 3456
d51c1b21
KW
3457At least one of s1 and s2 must have a goal (at least one of l1 and l2 must be
3458non-zero), and if both do, both have to be
8b35872c
KW
3459reached for a successful match. Also, if the fold of a character is multiple
3460characters, all of them must be matched (see tr21 reference below for
3461'folding').
3462
e6226b18 3463Upon a successful match, if pe1 is non-NULL,
8b35872c
KW
3464it will be set to point to the beginning of the I<next> character of s1 beyond
3465what was matched. Correspondingly for pe2 and s2.
d2cc3551
JH
3466
3467For case-insensitiveness, the "casefolding" of Unicode is used
3468instead of upper/lowercasing both the characters, see
3469http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
3470
3471=cut */
a33c29bc
KW
3472
3473/* A flags parameter has been added which may change, and hence isn't
3474 * externally documented. Currently it is:
3475 * 0 for as-documented above
3476 * FOLDEQ_UTF8_NOMIX_ASCII meaning that if a non-ASCII character folds to an
3477 ASCII one, to not match
5e64d0fa
KW
3478 * FOLDEQ_UTF8_LOCALE meaning that locale rules are to be used for code
3479 * points below 256; unicode rules for above 255; and
3480 * folds that cross those boundaries are disallowed,
3481 * like the NOMIX_ASCII option
18f762c3
KW
3482 * FOLDEQ_S1_ALREADY_FOLDED s1 has already been folded before calling this
3483 * routine. This allows that step to be skipped.
3484 * FOLDEQ_S2_ALREADY_FOLDED Similarly.
a33c29bc 3485 */
701a277b 3486I32
eda9cac1 3487Perl_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 3488{
8b35872c
KW
3489 dVAR;
3490 register const U8 *p1 = (const U8*)s1; /* Point to current char */
3491 register const U8 *p2 = (const U8*)s2;
48ef279e 3492 register const U8 *g1 = NULL; /* goal for s1 */
8b35872c 3493 register const U8 *g2 = NULL;
48ef279e
KW
3494 register const U8 *e1 = NULL; /* Don't scan s1 past this */
3495 register U8 *f1 = NULL; /* Point to current folded */
8b35872c
KW
3496 register const U8 *e2 = NULL;
3497 register U8 *f2 = NULL;
48ef279e 3498 STRLEN n1 = 0, n2 = 0; /* Number of bytes in current char */
8b35872c
KW
3499 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
3500 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
48ef279e
KW
3501 U8 natbuf[2]; /* Holds native 8-bit char converted to utf8;
3502 these always fit in 2 bytes */
8b35872c 3503
eda9cac1 3504 PERL_ARGS_ASSERT_FOLDEQ_UTF8_FLAGS;
8b35872c 3505
18f762c3
KW
3506 /* The algorithm requires that input with the flags on the first line of
3507 * the assert not be pre-folded. */
3508 assert( ! ((flags & (FOLDEQ_UTF8_NOMIX_ASCII | FOLDEQ_UTF8_LOCALE))
3509 && (flags & (FOLDEQ_S1_ALREADY_FOLDED | FOLDEQ_S2_ALREADY_FOLDED))));
3510
8b35872c 3511 if (pe1) {
48ef279e 3512 e1 = *(U8**)pe1;
8b35872c
KW
3513 }
3514
3515 if (l1) {
48ef279e 3516 g1 = (const U8*)s1 + l1;
8b35872c
KW
3517 }
3518
3519 if (pe2) {
48ef279e 3520 e2 = *(U8**)pe2;
8b35872c
KW
3521 }
3522
3523 if (l2) {
48ef279e 3524 g2 = (const U8*)s2 + l2;
8b35872c
KW
3525 }
3526
3527 /* Must have at least one goal */
3528 assert(g1 || g2);
3529
3530 if (g1) {
3531
48ef279e
KW
3532 /* Will never match if goal is out-of-bounds */
3533 assert(! e1 || e1 >= g1);
8b35872c 3534
48ef279e
KW
3535 /* Here, there isn't an end pointer, or it is beyond the goal. We
3536 * only go as far as the goal */
3537 e1 = g1;
8b35872c 3538 }
313b38e5
NC
3539 else {
3540 assert(e1); /* Must have an end for looking at s1 */
3541 }
8b35872c
KW
3542
3543 /* Same for goal for s2 */
3544 if (g2) {
48ef279e
KW
3545 assert(! e2 || e2 >= g2);
3546 e2 = g2;
8b35872c 3547 }
313b38e5
NC
3548 else {
3549 assert(e2);
3550 }
8b35872c 3551
18f762c3
KW
3552 /* If both operands are already folded, we could just do a memEQ on the
3553 * whole strings at once, but it would be better if the caller realized
3554 * this and didn't even call us */
3555
8b35872c
KW
3556 /* Look through both strings, a character at a time */
3557 while (p1 < e1 && p2 < e2) {
3558
d51c1b21 3559 /* If at the beginning of a new character in s1, get its fold to use
5e64d0fa
KW
3560 * and the length of the fold. (exception: locale rules just get the
3561 * character to a single byte) */
48ef279e 3562 if (n1 == 0) {
18f762c3
KW
3563 if (flags & FOLDEQ_S1_ALREADY_FOLDED) {
3564 f1 = (U8 *) p1;
3565 n1 = UTF8SKIP(f1);
a33c29bc 3566
5e64d0fa
KW
3567 /* If in locale matching, we use two sets of rules, depending on if
3568 * the code point is above or below 255. Here, we test for and
3569 * handle locale rules */
18f762c3
KW
3570 }
3571 else {
1b4059d5
KW
3572 if ((flags & FOLDEQ_UTF8_LOCALE)
3573 && (! u1 || UTF8_IS_INVARIANT(*p1)
3574 || UTF8_IS_DOWNGRADEABLE_START(*p1)))
5e64d0fa 3575 {
1b4059d5
KW
3576 /* There is no mixing of code points above and below 255. */
3577 if (u2 && (! UTF8_IS_INVARIANT(*p2)
3578 && ! UTF8_IS_DOWNGRADEABLE_START(*p2)))
3579 {
3580 return 0;
3581 }
5e64d0fa 3582
1b4059d5
KW
3583 /* We handle locale rules by converting, if necessary, the
3584 * code point to a single byte. */
3585 if (! u1 || UTF8_IS_INVARIANT(*p1)) {
3586 *foldbuf1 = *p1;
3587 }
3588 else {
3589 *foldbuf1 = TWO_BYTE_UTF8_TO_UNI(*p1, *(p1 + 1));
3590 }
3591 n1 = 1;
5e64d0fa 3592 }
1b4059d5
KW
3593 else if (isASCII(*p1)) { /* Note, that here won't be
3594 both ASCII and using locale
3595 rules */
3596
3597 /* If trying to mix non- with ASCII, and not supposed to,
3598 * fail */
3599 if ((flags & FOLDEQ_UTF8_NOMIX_ASCII) && ! isASCII(*p2)) {
3600 return 0;
3601 }
3602 n1 = 1;
3603 *foldbuf1 = toLOWER(*p1); /* Folds in the ASCII range are
3604 just lowercased */
5e64d0fa 3605 }
1b4059d5
KW
3606 else if (u1) {
3607 to_utf8_fold(p1, foldbuf1, &n1);
a33c29bc 3608 }
1b4059d5
KW
3609 else { /* Not utf8, convert to it first and then get fold */
3610 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p1)));
3611 to_utf8_fold(natbuf, foldbuf1, &n1);
3612 }
3613 f1 = foldbuf1;
18f762c3 3614 }
48ef279e 3615 }
8b35872c 3616
48ef279e 3617 if (n2 == 0) { /* Same for s2 */
18f762c3
KW
3618 if (flags & FOLDEQ_S2_ALREADY_FOLDED) {
3619 f2 = (U8 *) p2;
3620 n2 = UTF8SKIP(f2);
3621 }
3622 else {
227968da
KW
3623 if ((flags & FOLDEQ_UTF8_LOCALE)
3624 && (! u2 || UTF8_IS_INVARIANT(*p2) || UTF8_IS_DOWNGRADEABLE_START(*p2)))
5e64d0fa 3625 {
227968da
KW
3626 /* Here, the next char in s2 is < 256. We've already
3627 * worked on s1, and if it isn't also < 256, can't match */
3628 if (u1 && (! UTF8_IS_INVARIANT(*p1)
3629 && ! UTF8_IS_DOWNGRADEABLE_START(*p1)))
3630 {
3631 return 0;
3632 }
3633 if (! u2 || UTF8_IS_INVARIANT(*p2)) {
3634 *foldbuf2 = *p2;
3635 }
3636 else {
3637 *foldbuf2 = TWO_BYTE_UTF8_TO_UNI(*p2, *(p2 + 1));
3638 }
3639
3640 /* Use another function to handle locale rules. We've made
3641 * sure that both characters to compare are single bytes */
3642 if (! foldEQ_locale((char *) f1, (char *) foldbuf2, 1)) {
3643 return 0;
3644 }
3645 n1 = n2 = 0;
5e64d0fa 3646 }
227968da
KW
3647 else if (isASCII(*p2)) {
3648 if (flags && ! isASCII(*p1)) {
3649 return 0;
3650 }
3651 n2 = 1;
3652 *foldbuf2 = toLOWER(*p2);
5e64d0fa 3653 }
227968da
KW
3654 else if (u2) {
3655 to_utf8_fold(p2, foldbuf2, &n2);
5001101e 3656 }
227968da
KW
3657 else {
3658 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p2)));
3659 to_utf8_fold(natbuf, foldbuf2, &n2);
a33c29bc 3660 }
227968da 3661 f2 = foldbuf2;
18f762c3 3662 }
48ef279e 3663 }
8b35872c 3664
5001101e 3665 /* Here f1 and f2 point to the beginning of the strings to compare.
227968da
KW
3666 * These strings are the folds of the next character from each input
3667 * string, stored in utf8. */
5e64d0fa 3668
48ef279e
KW
3669 /* While there is more to look for in both folds, see if they
3670 * continue to match */
3671 while (n1 && n2) {
3672 U8 fold_length = UTF8SKIP(f1);
3673 if (fold_length != UTF8SKIP(f2)
3674 || (fold_length == 1 && *f1 != *f2) /* Short circuit memNE
3675 function call for single
3676 character */
3677 || memNE((char*)f1, (char*)f2, fold_length))
3678 {
e6226b18 3679 return 0; /* mismatch */
48ef279e
KW
3680 }
3681
3682 /* Here, they matched, advance past them */
3683 n1 -= fold_length;
3684 f1 += fold_length;
3685 n2 -= fold_length;
3686 f2 += fold_length;
3687 }
8b35872c 3688
48ef279e
KW
3689 /* When reach the end of any fold, advance the input past it */
3690 if (n1 == 0) {
3691 p1 += u1 ? UTF8SKIP(p1) : 1;
3692 }
3693 if (n2 == 0) {
3694 p2 += u2 ? UTF8SKIP(p2) : 1;
3695 }
8b35872c
KW
3696 } /* End of loop through both strings */
3697
3698 /* A match is defined by each scan that specified an explicit length
3699 * reaching its final goal, and the other not having matched a partial
3700 * character (which can happen when the fold of a character is more than one
3701 * character). */
3702 if (! ((g1 == 0 || p1 == g1) && (g2 == 0 || p2 == g2)) || n1 || n2) {
e6226b18 3703 return 0;
8b35872c
KW
3704 }
3705
3706 /* Successful match. Set output pointers */
3707 if (pe1) {
48ef279e 3708 *pe1 = (char*)p1;
8b35872c
KW
3709 }
3710 if (pe2) {
48ef279e 3711 *pe2 = (char*)p2;
8b35872c 3712 }
e6226b18 3713 return 1;
e6b2e755 3714}
701a277b 3715
a49f32c6
NC
3716/*
3717 * Local variables:
3718 * c-indentation-style: bsd
3719 * c-basic-offset: 4
3720 * indent-tabs-mode: t
3721 * End:
3722 *
37442d52
RGS
3723 * ex: set ts=8 sts=4 sw=4 noet:
3724 */