This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
utf8.c: Expand use of refactored to_uni_lower
[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
50bda2c3
KW
1317/* Call the function to convert a UTF-8 encoded character to the specified case.
1318 * Note that there may be more than one character in the result.
1319 * INP is a pointer to the first byte of the input character
1320 * OUTP will be set to the first byte of the string of changed characters. It
1321 * needs to have space for UTF8_MAXBYTES_CASE+1 bytes
1322 * LENP will be set to the length in bytes of the string of changed characters
1323 *
1324 * The functions return the ordinal of the first character in the string of OUTP */
1325#define CALL_UPPER_CASE(INP, OUTP, LENP) Perl_to_utf8_case(aTHX_ INP, OUTP, LENP, &PL_utf8_toupper, "ToUc", "utf8::ToSpecUpper")
1326#define CALL_TITLE_CASE(INP, OUTP, LENP) Perl_to_utf8_case(aTHX_ INP, OUTP, LENP, &PL_utf8_totitle, "ToTc", "utf8::ToSpecTitle")
1327#define CALL_LOWER_CASE(INP, OUTP, LENP) Perl_to_utf8_case(aTHX_ INP, OUTP, LENP, &PL_utf8_tolower, "ToLc", "utf8::ToSpecLower")
1328
1329/* This additionally has the input parameter SPECIALS, which if non-zero will
1330 * cause this to use the SPECIALS hash for folding (meaning get full case
1331 * folding); otherwise, when zero, this implies a simple case fold */
1332#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 1333
84afefe6
JH
1334UV
1335Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 1336{
c3fd2246
KW
1337 /* Convert the Unicode character whose ordinal is c to its uppercase
1338 * version and store that in UTF-8 in p and its length in bytes in lenp.
1339 * Note that the p needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
1340 * the changed version may be longer than the original character.
1341 *
1342 * The ordinal of the first character of the changed version is returned
1343 * (but note, as explained above, that there may be more.) */
1344
7918f24d
NC
1345 PERL_ARGS_ASSERT_TO_UNI_UPPER;
1346
0ebc6274
JH
1347 uvchr_to_utf8(p, c);
1348 return to_utf8_upper(p, p, lenp);
a0ed51b3
LW
1349}
1350
84afefe6
JH
1351UV
1352Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 1353{
7918f24d
NC
1354 PERL_ARGS_ASSERT_TO_UNI_TITLE;
1355
0ebc6274
JH
1356 uvchr_to_utf8(p, c);
1357 return to_utf8_title(p, p, lenp);
a0ed51b3
LW
1358}
1359
afc16117
KW
1360STATIC U8
1361S_to_lower_latin1(pTHX_ const U8 c, U8* p, STRLEN *lenp)
1362{
1363 /* We have the latin1-range values compiled into the core, so just use
1364 * those, converting the result to utf8. Since the result is always just
1365 * one character, we allow p to be NULL */
1366
1367 U8 converted = toLOWER_LATIN1(c);
1368
1369 if (p != NULL) {
1370 if (UNI_IS_INVARIANT(converted)) {
1371 *p = converted;
1372 *lenp = 1;
1373 }
1374 else {
1375 *p = UTF8_TWO_BYTE_HI(converted);
1376 *(p+1) = UTF8_TWO_BYTE_LO(converted);
1377 *lenp = 2;
1378 }
1379 }
1380 return converted;
1381}
1382
84afefe6
JH
1383UV
1384Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 1385{
968c5e6a
KW
1386 dVAR;
1387
7918f24d
NC
1388 PERL_ARGS_ASSERT_TO_UNI_LOWER;
1389
afc16117
KW
1390 if (c < 256) {
1391 return to_lower_latin1((U8) c, p, lenp);
bca00c02
KW
1392 }
1393
afc16117 1394 uvchr_to_utf8(p, c);
968c5e6a 1395 return CALL_LOWER_CASE(p, p, lenp);
a0ed51b3
LW
1396}
1397
84afefe6 1398UV
36bb2ab6 1399Perl__to_uni_fold_flags(pTHX_ UV c, U8* p, STRLEN *lenp, U8 flags)
84afefe6 1400{
36bb2ab6 1401 PERL_ARGS_ASSERT__TO_UNI_FOLD_FLAGS;
7918f24d 1402
0ebc6274 1403 uvchr_to_utf8(p, c);
36bb2ab6 1404 return _to_utf8_fold_flags(p, p, lenp, flags);
84afefe6
JH
1405}
1406
a0ed51b3
LW
1407/* for now these all assume no locale info available for Unicode > 255 */
1408
1409bool
84afefe6 1410Perl_is_uni_alnum_lc(pTHX_ UV c)
a0ed51b3
LW
1411{
1412 return is_uni_alnum(c); /* XXX no locale support yet */
1413}
1414
1415bool
84afefe6 1416Perl_is_uni_idfirst_lc(pTHX_ UV c)
a0ed51b3
LW
1417{
1418 return is_uni_idfirst(c); /* XXX no locale support yet */
1419}
1420
1421bool
84afefe6 1422Perl_is_uni_alpha_lc(pTHX_ UV c)
a0ed51b3
LW
1423{
1424 return is_uni_alpha(c); /* XXX no locale support yet */
1425}
1426
1427bool
84afefe6 1428Perl_is_uni_ascii_lc(pTHX_ UV c)
4d61ec05
GS
1429{
1430 return is_uni_ascii(c); /* XXX no locale support yet */
1431}
1432
1433bool
84afefe6 1434Perl_is_uni_space_lc(pTHX_ UV c)
a0ed51b3
LW
1435{
1436 return is_uni_space(c); /* XXX no locale support yet */
1437}
1438
1439bool
84afefe6 1440Perl_is_uni_digit_lc(pTHX_ UV c)
a0ed51b3
LW
1441{
1442 return is_uni_digit(c); /* XXX no locale support yet */
1443}
1444
1445bool
84afefe6 1446Perl_is_uni_upper_lc(pTHX_ UV c)
a0ed51b3
LW
1447{
1448 return is_uni_upper(c); /* XXX no locale support yet */
1449}
1450
1451bool
84afefe6 1452Perl_is_uni_lower_lc(pTHX_ UV c)
a0ed51b3
LW
1453{
1454 return is_uni_lower(c); /* XXX no locale support yet */
1455}
1456
1457bool
84afefe6 1458Perl_is_uni_cntrl_lc(pTHX_ UV c)
b8c5462f
JH
1459{
1460 return is_uni_cntrl(c); /* XXX no locale support yet */
1461}
1462
1463bool
84afefe6 1464Perl_is_uni_graph_lc(pTHX_ UV c)
b8c5462f
JH
1465{
1466 return is_uni_graph(c); /* XXX no locale support yet */
1467}
1468
1469bool
84afefe6 1470Perl_is_uni_print_lc(pTHX_ UV c)
a0ed51b3
LW
1471{
1472 return is_uni_print(c); /* XXX no locale support yet */
1473}
1474
b8c5462f 1475bool
84afefe6 1476Perl_is_uni_punct_lc(pTHX_ UV c)
b8c5462f
JH
1477{
1478 return is_uni_punct(c); /* XXX no locale support yet */
1479}
1480
4d61ec05 1481bool
84afefe6 1482Perl_is_uni_xdigit_lc(pTHX_ UV c)
4d61ec05
GS
1483{
1484 return is_uni_xdigit(c); /* XXX no locale support yet */
1485}
1486
b7ac61fa
JH
1487U32
1488Perl_to_uni_upper_lc(pTHX_ U32 c)
1489{
ee099d14
JH
1490 /* XXX returns only the first character -- do not use XXX */
1491 /* XXX no locale support yet */
1492 STRLEN len;
89ebb4a3 1493 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
ee099d14 1494 return (U32)to_uni_upper(c, tmpbuf, &len);
b7ac61fa
JH
1495}
1496
1497U32
1498Perl_to_uni_title_lc(pTHX_ U32 c)
1499{
ee099d14
JH
1500 /* XXX returns only the first character XXX -- do not use XXX */
1501 /* XXX no locale support yet */
1502 STRLEN len;
89ebb4a3 1503 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
ee099d14 1504 return (U32)to_uni_title(c, tmpbuf, &len);
b7ac61fa
JH
1505}
1506
1507U32
1508Perl_to_uni_lower_lc(pTHX_ U32 c)
1509{
ee099d14
JH
1510 /* XXX returns only the first character -- do not use XXX */
1511 /* XXX no locale support yet */
1512 STRLEN len;
89ebb4a3 1513 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
ee099d14 1514 return (U32)to_uni_lower(c, tmpbuf, &len);
b7ac61fa
JH
1515}
1516
7452cf6a 1517static bool
5141f98e 1518S_is_utf8_common(pTHX_ const U8 *const p, SV **swash,
bde6a22d
NC
1519 const char *const swashname)
1520{
97aff369 1521 dVAR;
7918f24d
NC
1522
1523 PERL_ARGS_ASSERT_IS_UTF8_COMMON;
1524
bde6a22d
NC
1525 if (!is_utf8_char(p))
1526 return FALSE;
1527 if (!*swash)
711a919c 1528 *swash = swash_init("utf8", swashname, &PL_sv_undef, 1, 0);
bde6a22d
NC
1529 return swash_fetch(*swash, p, TRUE) != 0;
1530}
1531
1532bool
7fc63493 1533Perl_is_utf8_alnum(pTHX_ const U8 *p)
a0ed51b3 1534{
97aff369 1535 dVAR;
7918f24d
NC
1536
1537 PERL_ARGS_ASSERT_IS_UTF8_ALNUM;
1538
671c33bf
NC
1539 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1540 * descendant of isalnum(3), in other words, it doesn't
1541 * contain the '_'. --jhi */
d4c19fe8 1542 return is_utf8_common(p, &PL_utf8_alnum, "IsWord");
a0ed51b3
LW
1543}
1544
1545bool
7fc63493 1546Perl_is_utf8_idfirst(pTHX_ const U8 *p) /* The naming is historical. */
a0ed51b3 1547{
97aff369 1548 dVAR;
7918f24d
NC
1549
1550 PERL_ARGS_ASSERT_IS_UTF8_IDFIRST;
1551
82686b01
JH
1552 if (*p == '_')
1553 return TRUE;
bde6a22d 1554 /* is_utf8_idstart would be more logical. */
d4c19fe8 1555 return is_utf8_common(p, &PL_utf8_idstart, "IdStart");
82686b01
JH
1556}
1557
1558bool
c11ff943
KW
1559Perl_is_utf8_xidfirst(pTHX_ const U8 *p) /* The naming is historical. */
1560{
1561 dVAR;
1562
1563 PERL_ARGS_ASSERT_IS_UTF8_XIDFIRST;
1564
1565 if (*p == '_')
1566 return TRUE;
1567 /* is_utf8_idstart would be more logical. */
1568 return is_utf8_common(p, &PL_utf8_xidstart, "XIdStart");
1569}
1570
1571bool
b6912c02
KW
1572Perl__is_utf8__perl_idstart(pTHX_ const U8 *p)
1573{
1574 dVAR;
1575
1576 PERL_ARGS_ASSERT__IS_UTF8__PERL_IDSTART;
1577
1578 return is_utf8_common(p, &PL_utf8_perl_idstart, "_Perl_IDStart");
1579}
1580
1581bool
7fc63493 1582Perl_is_utf8_idcont(pTHX_ const U8 *p)
82686b01 1583{
97aff369 1584 dVAR;
7918f24d
NC
1585
1586 PERL_ARGS_ASSERT_IS_UTF8_IDCONT;
1587
d4c19fe8 1588 return is_utf8_common(p, &PL_utf8_idcont, "IdContinue");
a0ed51b3
LW
1589}
1590
1591bool
c11ff943
KW
1592Perl_is_utf8_xidcont(pTHX_ const U8 *p)
1593{
1594 dVAR;
1595
1596 PERL_ARGS_ASSERT_IS_UTF8_XIDCONT;
1597
c11ff943
KW
1598 return is_utf8_common(p, &PL_utf8_idcont, "XIdContinue");
1599}
1600
1601bool
7fc63493 1602Perl_is_utf8_alpha(pTHX_ const U8 *p)
a0ed51b3 1603{
97aff369 1604 dVAR;
7918f24d
NC
1605
1606 PERL_ARGS_ASSERT_IS_UTF8_ALPHA;
1607
d4c19fe8 1608 return is_utf8_common(p, &PL_utf8_alpha, "IsAlpha");
a0ed51b3
LW
1609}
1610
1611bool
7fc63493 1612Perl_is_utf8_ascii(pTHX_ const U8 *p)
b8c5462f 1613{
97aff369 1614 dVAR;
7918f24d
NC
1615
1616 PERL_ARGS_ASSERT_IS_UTF8_ASCII;
1617
bc39fe24
KW
1618 /* ASCII characters are the same whether in utf8 or not. So the macro
1619 * works on both utf8 and non-utf8 representations. */
1620 return isASCII(*p);
b8c5462f
JH
1621}
1622
1623bool
7fc63493 1624Perl_is_utf8_space(pTHX_ const U8 *p)
a0ed51b3 1625{
97aff369 1626 dVAR;
7918f24d
NC
1627
1628 PERL_ARGS_ASSERT_IS_UTF8_SPACE;
1629
a34094a9 1630 return is_utf8_common(p, &PL_utf8_space, "IsXPerlSpace");
a0ed51b3
LW
1631}
1632
1633bool
d1eb3177
YO
1634Perl_is_utf8_perl_space(pTHX_ const U8 *p)
1635{
1636 dVAR;
1637
1638 PERL_ARGS_ASSERT_IS_UTF8_PERL_SPACE;
1639
c4428693
KW
1640 /* Only true if is an ASCII space-like character, and ASCII is invariant
1641 * under utf8, so can just use the macro */
1642 return isSPACE_A(*p);
d1eb3177
YO
1643}
1644
1645bool
1646Perl_is_utf8_perl_word(pTHX_ const U8 *p)
1647{
1648 dVAR;
1649
1650 PERL_ARGS_ASSERT_IS_UTF8_PERL_WORD;
1651
c4428693
KW
1652 /* Only true if is an ASCII word character, and ASCII is invariant
1653 * under utf8, so can just use the macro */
1654 return isWORDCHAR_A(*p);
d1eb3177
YO
1655}
1656
1657bool
7fc63493 1658Perl_is_utf8_digit(pTHX_ const U8 *p)
a0ed51b3 1659{
97aff369 1660 dVAR;
7918f24d
NC
1661
1662 PERL_ARGS_ASSERT_IS_UTF8_DIGIT;
1663
d4c19fe8 1664 return is_utf8_common(p, &PL_utf8_digit, "IsDigit");
a0ed51b3
LW
1665}
1666
1667bool
d1eb3177
YO
1668Perl_is_utf8_posix_digit(pTHX_ const U8 *p)
1669{
1670 dVAR;
1671
1672 PERL_ARGS_ASSERT_IS_UTF8_POSIX_DIGIT;
1673
c4428693
KW
1674 /* Only true if is an ASCII digit character, and ASCII is invariant
1675 * under utf8, so can just use the macro */
1676 return isDIGIT_A(*p);
d1eb3177
YO
1677}
1678
1679bool
7fc63493 1680Perl_is_utf8_upper(pTHX_ const U8 *p)
a0ed51b3 1681{
97aff369 1682 dVAR;
7918f24d
NC
1683
1684 PERL_ARGS_ASSERT_IS_UTF8_UPPER;
1685
d4c19fe8 1686 return is_utf8_common(p, &PL_utf8_upper, "IsUppercase");
a0ed51b3
LW
1687}
1688
1689bool
7fc63493 1690Perl_is_utf8_lower(pTHX_ const U8 *p)
a0ed51b3 1691{
97aff369 1692 dVAR;
7918f24d
NC
1693
1694 PERL_ARGS_ASSERT_IS_UTF8_LOWER;
1695
d4c19fe8 1696 return is_utf8_common(p, &PL_utf8_lower, "IsLowercase");
a0ed51b3
LW
1697}
1698
1699bool
7fc63493 1700Perl_is_utf8_cntrl(pTHX_ const U8 *p)
b8c5462f 1701{
97aff369 1702 dVAR;
7918f24d
NC
1703
1704 PERL_ARGS_ASSERT_IS_UTF8_CNTRL;
1705
7b952154
KW
1706 if (isASCII(*p)) {
1707 return isCNTRL_A(*p);
1708 }
1709
1710 /* All controls are in Latin1 */
1711 if (! UTF8_IS_DOWNGRADEABLE_START(*p)) {
1712 return 0;
1713 }
1714 return isCNTRL_L1(TWO_BYTE_UTF8_TO_UNI(*p, *(p+1)));
b8c5462f
JH
1715}
1716
1717bool
7fc63493 1718Perl_is_utf8_graph(pTHX_ const U8 *p)
b8c5462f 1719{
97aff369 1720 dVAR;
7918f24d
NC
1721
1722 PERL_ARGS_ASSERT_IS_UTF8_GRAPH;
1723
d4c19fe8 1724 return is_utf8_common(p, &PL_utf8_graph, "IsGraph");
b8c5462f
JH
1725}
1726
1727bool
7fc63493 1728Perl_is_utf8_print(pTHX_ const U8 *p)
a0ed51b3 1729{
97aff369 1730 dVAR;
7918f24d
NC
1731
1732 PERL_ARGS_ASSERT_IS_UTF8_PRINT;
1733
d4c19fe8 1734 return is_utf8_common(p, &PL_utf8_print, "IsPrint");
a0ed51b3
LW
1735}
1736
1737bool
7fc63493 1738Perl_is_utf8_punct(pTHX_ const U8 *p)
b8c5462f 1739{
97aff369 1740 dVAR;
7918f24d
NC
1741
1742 PERL_ARGS_ASSERT_IS_UTF8_PUNCT;
1743
d4c19fe8 1744 return is_utf8_common(p, &PL_utf8_punct, "IsPunct");
b8c5462f
JH
1745}
1746
1747bool
7fc63493 1748Perl_is_utf8_xdigit(pTHX_ const U8 *p)
b8c5462f 1749{
97aff369 1750 dVAR;
7918f24d
NC
1751
1752 PERL_ARGS_ASSERT_IS_UTF8_XDIGIT;
1753
d1eb3177 1754 return is_utf8_common(p, &PL_utf8_xdigit, "IsXDigit");
b8c5462f
JH
1755}
1756
1757bool
7fc63493 1758Perl_is_utf8_mark(pTHX_ const U8 *p)
a0ed51b3 1759{
97aff369 1760 dVAR;
7918f24d
NC
1761
1762 PERL_ARGS_ASSERT_IS_UTF8_MARK;
1763
d4c19fe8 1764 return is_utf8_common(p, &PL_utf8_mark, "IsM");
a0ed51b3
LW
1765}
1766
37e2e78e
KW
1767bool
1768Perl_is_utf8_X_begin(pTHX_ const U8 *p)
1769{
1770 dVAR;
1771
1772 PERL_ARGS_ASSERT_IS_UTF8_X_BEGIN;
1773
1774 return is_utf8_common(p, &PL_utf8_X_begin, "_X_Begin");
1775}
1776
1777bool
1778Perl_is_utf8_X_extend(pTHX_ const U8 *p)
1779{
1780 dVAR;
1781
1782 PERL_ARGS_ASSERT_IS_UTF8_X_EXTEND;
1783
1784 return is_utf8_common(p, &PL_utf8_X_extend, "_X_Extend");
1785}
1786
1787bool
1788Perl_is_utf8_X_prepend(pTHX_ const U8 *p)
1789{
1790 dVAR;
1791
1792 PERL_ARGS_ASSERT_IS_UTF8_X_PREPEND;
1793
1794 return is_utf8_common(p, &PL_utf8_X_prepend, "GCB=Prepend");
1795}
1796
1797bool
1798Perl_is_utf8_X_non_hangul(pTHX_ const U8 *p)
1799{
1800 dVAR;
1801
1802 PERL_ARGS_ASSERT_IS_UTF8_X_NON_HANGUL;
1803
1804 return is_utf8_common(p, &PL_utf8_X_non_hangul, "HST=Not_Applicable");
1805}
1806
1807bool
1808Perl_is_utf8_X_L(pTHX_ const U8 *p)
1809{
1810 dVAR;
1811
1812 PERL_ARGS_ASSERT_IS_UTF8_X_L;
1813
1814 return is_utf8_common(p, &PL_utf8_X_L, "GCB=L");
1815}
1816
1817bool
1818Perl_is_utf8_X_LV(pTHX_ const U8 *p)
1819{
1820 dVAR;
1821
1822 PERL_ARGS_ASSERT_IS_UTF8_X_LV;
1823
1824 return is_utf8_common(p, &PL_utf8_X_LV, "GCB=LV");
1825}
1826
1827bool
1828Perl_is_utf8_X_LVT(pTHX_ const U8 *p)
1829{
1830 dVAR;
1831
1832 PERL_ARGS_ASSERT_IS_UTF8_X_LVT;
1833
1834 return is_utf8_common(p, &PL_utf8_X_LVT, "GCB=LVT");
1835}
1836
1837bool
1838Perl_is_utf8_X_T(pTHX_ const U8 *p)
1839{
1840 dVAR;
1841
1842 PERL_ARGS_ASSERT_IS_UTF8_X_T;
1843
1844 return is_utf8_common(p, &PL_utf8_X_T, "GCB=T");
1845}
1846
1847bool
1848Perl_is_utf8_X_V(pTHX_ const U8 *p)
1849{
1850 dVAR;
1851
1852 PERL_ARGS_ASSERT_IS_UTF8_X_V;
1853
1854 return is_utf8_common(p, &PL_utf8_X_V, "GCB=V");
1855}
1856
1857bool
1858Perl_is_utf8_X_LV_LVT_V(pTHX_ const U8 *p)
1859{
1860 dVAR;
1861
1862 PERL_ARGS_ASSERT_IS_UTF8_X_LV_LVT_V;
1863
1864 return is_utf8_common(p, &PL_utf8_X_LV_LVT_V, "_X_LV_LVT_V");
1865}
1866
6b5c0936 1867/*
87cea99e 1868=for apidoc to_utf8_case
6b5c0936
JH
1869
1870The "p" contains the pointer to the UTF-8 string encoding
1871the character that is being converted.
1872
1873The "ustrp" is a pointer to the character buffer to put the
1874conversion result to. The "lenp" is a pointer to the length
1875of the result.
1876
0134edef 1877The "swashp" is a pointer to the swash to use.
6b5c0936 1878
36bb2ab6 1879Both the special and normal mappings are stored in lib/unicore/To/Foo.pl,
8fe4d5b2 1880and loaded by SWASHNEW, using lib/utf8_heavy.pl. The special (usually,
0134edef 1881but not always, a multicharacter mapping), is tried first.
6b5c0936 1882
0134edef
JH
1883The "special" is a string like "utf8::ToSpecLower", which means the
1884hash %utf8::ToSpecLower. The access to the hash is through
1885Perl_to_utf8_case().
6b5c0936 1886
0134edef
JH
1887The "normal" is a string like "ToLower" which means the swash
1888%utf8::ToLower.
1889
1890=cut */
6b5c0936 1891
2104c8d9 1892UV
9a957fbc
AL
1893Perl_to_utf8_case(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp,
1894 SV **swashp, const char *normal, const char *special)
a0ed51b3 1895{
97aff369 1896 dVAR;
89ebb4a3 1897 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
0134edef 1898 STRLEN len = 0;
aec46f14 1899 const UV uv0 = utf8_to_uvchr(p, NULL);
1feea2c7
JH
1900 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1901 * are necessary in EBCDIC, they are redundant no-ops
1902 * in ASCII-ish platforms, and hopefully optimized away. */
f54cb97a 1903 const UV uv1 = NATIVE_TO_UNI(uv0);
7918f24d
NC
1904
1905 PERL_ARGS_ASSERT_TO_UTF8_CASE;
1906
9ae3ac1a
KW
1907 /* Note that swash_fetch() doesn't output warnings for these because it
1908 * assumes we will */
8457b38f 1909 if (uv1 >= UNICODE_SURROGATE_FIRST) {
9ae3ac1a 1910 if (uv1 <= UNICODE_SURROGATE_LAST) {
8457b38f
KW
1911 if (ckWARN_d(WARN_SURROGATE)) {
1912 const char* desc = (PL_op) ? OP_DESC(PL_op) : normal;
1913 Perl_warner(aTHX_ packWARN(WARN_SURROGATE),
1914 "Operation \"%s\" returns its argument for UTF-16 surrogate U+%04"UVXf"", desc, uv1);
1915 }
9ae3ac1a
KW
1916 }
1917 else if (UNICODE_IS_SUPER(uv1)) {
8457b38f
KW
1918 if (ckWARN_d(WARN_NON_UNICODE)) {
1919 const char* desc = (PL_op) ? OP_DESC(PL_op) : normal;
1920 Perl_warner(aTHX_ packWARN(WARN_NON_UNICODE),
1921 "Operation \"%s\" returns its argument for non-Unicode code point 0x%04"UVXf"", desc, uv1);
1922 }
9ae3ac1a
KW
1923 }
1924
1925 /* Note that non-characters are perfectly legal, so no warning should
1926 * be given */
1927 }
1928
1feea2c7 1929 uvuni_to_utf8(tmpbuf, uv1);
0134edef
JH
1930
1931 if (!*swashp) /* load on-demand */
1932 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1933
a6f87d8c 1934 if (special) {
0134edef 1935 /* It might be "special" (sometimes, but not always,
2a37f04d 1936 * a multicharacter mapping) */
6673a63c 1937 HV * const hv = get_hv(special, 0);
b08cf34e
JH
1938 SV **svp;
1939
35da51f7 1940 if (hv &&
b08cf34e
JH
1941 (svp = hv_fetch(hv, (const char*)tmpbuf, UNISKIP(uv1), FALSE)) &&
1942 (*svp)) {
cfd0369c 1943 const char *s;
47654450 1944
cfd0369c 1945 s = SvPV_const(*svp, len);
47654450
JH
1946 if (len == 1)
1947 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
2a37f04d 1948 else {
2f9475ad
JH
1949#ifdef EBCDIC
1950 /* If we have EBCDIC we need to remap the characters
1951 * since any characters in the low 256 are Unicode
1952 * code points, not EBCDIC. */
7cda7a3d 1953 U8 *t = (U8*)s, *tend = t + len, *d;
2f9475ad
JH
1954
1955 d = tmpbuf;
b08cf34e 1956 if (SvUTF8(*svp)) {
2f9475ad
JH
1957 STRLEN tlen = 0;
1958
1959 while (t < tend) {
d4c19fe8 1960 const UV c = utf8_to_uvchr(t, &tlen);
2f9475ad
JH
1961 if (tlen > 0) {
1962 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1963 t += tlen;
1964 }
1965 else
1966 break;
1967 }
1968 }
1969 else {
36fec512
JH
1970 while (t < tend) {
1971 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1972 t++;
1973 }
2f9475ad
JH
1974 }
1975 len = d - tmpbuf;
1976 Copy(tmpbuf, ustrp, len, U8);
1977#else
d2dcd0fb 1978 Copy(s, ustrp, len, U8);
2f9475ad 1979#endif
29e98929 1980 }
983ffd37 1981 }
0134edef
JH
1982 }
1983
1984 if (!len && *swashp) {
d4c19fe8
AL
1985 const UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1986
0134edef
JH
1987 if (uv2) {
1988 /* It was "normal" (a single character mapping). */
d4c19fe8 1989 const UV uv3 = UNI_TO_NATIVE(uv2);
e9101d72 1990 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
2a37f04d
JH
1991 }
1992 }
1feea2c7 1993
37e2e78e
KW
1994 if (!len) /* Neither: just copy. In other words, there was no mapping
1995 defined, which means that the code point maps to itself */
0134edef
JH
1996 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1997
2a37f04d
JH
1998 if (lenp)
1999 *lenp = len;
2000
0134edef 2001 return len ? utf8_to_uvchr(ustrp, 0) : 0;
a0ed51b3
LW
2002}
2003
d3e79532 2004/*
87cea99e 2005=for apidoc to_utf8_upper
d3e79532
JH
2006
2007Convert the UTF-8 encoded character at p to its uppercase version and
2008store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
2009that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since
2010the uppercase version may be longer than the original character.
d3e79532
JH
2011
2012The first character of the uppercased version is returned
2013(but note, as explained above, that there may be more.)
2014
2015=cut */
2016
2104c8d9 2017UV
7fc63493 2018Perl_to_utf8_upper(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
a0ed51b3 2019{
97aff369 2020 dVAR;
7918f24d
NC
2021
2022 PERL_ARGS_ASSERT_TO_UTF8_UPPER;
2023
50bda2c3 2024 return CALL_UPPER_CASE(p, ustrp, lenp);
983ffd37 2025}
a0ed51b3 2026
d3e79532 2027/*
87cea99e 2028=for apidoc to_utf8_title
d3e79532
JH
2029
2030Convert the UTF-8 encoded character at p to its titlecase version and
2031store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
2032that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
2033titlecase version may be longer than the original character.
d3e79532
JH
2034
2035The first character of the titlecased version is returned
2036(but note, as explained above, that there may be more.)
2037
2038=cut */
2039
983ffd37 2040UV
7fc63493 2041Perl_to_utf8_title(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
983ffd37 2042{
97aff369 2043 dVAR;
7918f24d
NC
2044
2045 PERL_ARGS_ASSERT_TO_UTF8_TITLE;
2046
50bda2c3 2047 return CALL_TITLE_CASE(p, ustrp, lenp);
a0ed51b3
LW
2048}
2049
d3e79532 2050/*
87cea99e 2051=for apidoc to_utf8_lower
d3e79532
JH
2052
2053Convert the UTF-8 encoded character at p to its lowercase version and
2054store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3
JH
2055that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
2056lowercase version may be longer than the original character.
d3e79532
JH
2057
2058The first character of the lowercased version is returned
2059(but note, as explained above, that there may be more.)
2060
2061=cut */
2062
2104c8d9 2063UV
7fc63493 2064Perl_to_utf8_lower(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp)
a0ed51b3 2065{
97aff369 2066 dVAR;
7918f24d
NC
2067
2068 PERL_ARGS_ASSERT_TO_UTF8_LOWER;
2069
968c5e6a
KW
2070 if (UTF8_IS_INVARIANT(*p)) {
2071 return to_lower_latin1(*p, ustrp, lenp);
2072 }
2073 else if UTF8_IS_DOWNGRADEABLE_START(*p) {
2074 return to_lower_latin1(TWO_BYTE_UTF8_TO_UNI(*p, *(p+1)), ustrp, lenp);
2075 }
2076
50bda2c3 2077 return CALL_LOWER_CASE(p, ustrp, lenp);
b4e400f9
JH
2078}
2079
d3e79532 2080/*
87cea99e 2081=for apidoc to_utf8_fold
d3e79532
JH
2082
2083Convert the UTF-8 encoded character at p to its foldcase version and
2084store that in UTF-8 in ustrp and its length in bytes in lenp. Note
89ebb4a3 2085that the ustrp needs to be at least UTF8_MAXBYTES_CASE+1 bytes since the
d3e79532
JH
2086foldcase version may be longer than the original character (up to
2087three characters).
2088
2089The first character of the foldcased version is returned
2090(but note, as explained above, that there may be more.)
2091
2092=cut */
2093
36bb2ab6
KW
2094/* Not currently externally documented is 'flags', which currently is non-zero
2095 * if full case folds are to be used; otherwise simple folds */
2096
b4e400f9 2097UV
36bb2ab6 2098Perl__to_utf8_fold_flags(pTHX_ const U8 *p, U8* ustrp, STRLEN *lenp, U8 flags)
b4e400f9 2099{
97aff369 2100 dVAR;
7918f24d 2101
36bb2ab6 2102 PERL_ARGS_ASSERT__TO_UTF8_FOLD_FLAGS;
7918f24d 2103
50bda2c3 2104 return CALL_FOLD_CASE(p, ustrp, lenp, flags);
a0ed51b3
LW
2105}
2106
711a919c
ST
2107/* Note:
2108 * A "swash" is a swatch hash.
2109 * A "swatch" is a bit vector generated by utf8.c:S_swash_get().
2110 * C<pkg> is a pointer to a package name for SWASHNEW, should be "utf8".
2111 * For other parameters, see utf8::SWASHNEW in lib/utf8_heavy.pl.
2112 */
a0ed51b3 2113SV*
7fc63493 2114Perl_swash_init(pTHX_ const char* pkg, const char* name, SV *listsv, I32 minbits, I32 none)
a0ed51b3 2115{
27da23d5 2116 dVAR;
a0ed51b3 2117 SV* retval;
8e84507e 2118 dSP;
7fc63493
AL
2119 const size_t pkg_len = strlen(pkg);
2120 const size_t name_len = strlen(name);
da51bb9b 2121 HV * const stash = gv_stashpvn(pkg, pkg_len, 0);
f8be5cf0 2122 SV* errsv_save;
88c57d91 2123 GV *method;
ce3b816e 2124
7918f24d
NC
2125 PERL_ARGS_ASSERT_SWASH_INIT;
2126
96ca9f55
DM
2127 PUSHSTACKi(PERLSI_MAGIC);
2128 ENTER;
ec34a119 2129 SAVEHINTS();
96ca9f55 2130 save_re_context();
ba6ff154
FC
2131 if (PL_parser && PL_parser->error_count)
2132 SAVEI8(PL_parser->error_count), PL_parser->error_count = 0;
88c57d91
NC
2133 method = gv_fetchmeth(stash, "SWASHNEW", 8, -1);
2134 if (!method) { /* demand load utf8 */
ce3b816e 2135 ENTER;
f8be5cf0 2136 errsv_save = newSVsv(ERRSV);
dc0c6abb
NC
2137 /* It is assumed that callers of this routine are not passing in any
2138 user derived data. */
2139 /* Need to do this after save_re_context() as it will set PL_tainted to
2140 1 while saving $1 etc (see the code after getrx: in Perl_magic_get).
2141 Even line to create errsv_save can turn on PL_tainted. */
2142 SAVEBOOL(PL_tainted);
2143 PL_tainted = 0;
71bed85a 2144 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpvn(pkg,pkg_len),
a0714e2c 2145 NULL);
f8be5cf0
JH
2146 if (!SvTRUE(ERRSV))
2147 sv_setsv(ERRSV, errsv_save);
2148 SvREFCNT_dec(errsv_save);
ce3b816e
GS
2149 LEAVE;
2150 }
2151 SPAGAIN;
a0ed51b3
LW
2152 PUSHMARK(SP);
2153 EXTEND(SP,5);
6e449a3a
MHM
2154 mPUSHp(pkg, pkg_len);
2155 mPUSHp(name, name_len);
a0ed51b3 2156 PUSHs(listsv);
6e449a3a
MHM
2157 mPUSHi(minbits);
2158 mPUSHi(none);
a0ed51b3 2159 PUTBACK;
f8be5cf0 2160 errsv_save = newSVsv(ERRSV);
88c57d91
NC
2161 /* If we already have a pointer to the method, no need to use call_method()
2162 to repeat the lookup. */
2163 if (method ? call_sv(MUTABLE_SV(method), G_SCALAR)
8b563eef 2164 : call_sv(newSVpvs_flags("SWASHNEW", SVs_TEMP), G_SCALAR | G_METHOD))
8e84507e 2165 retval = newSVsv(*PL_stack_sp--);
a0ed51b3 2166 else
e24b16f9 2167 retval = &PL_sv_undef;
f8be5cf0
JH
2168 if (!SvTRUE(ERRSV))
2169 sv_setsv(ERRSV, errsv_save);
2170 SvREFCNT_dec(errsv_save);
a0ed51b3
LW
2171 LEAVE;
2172 POPSTACK;
923e4eb5 2173 if (IN_PERL_COMPILETIME) {
623e6609 2174 CopHINTS_set(PL_curcop, PL_hints);
a0ed51b3 2175 }
bc45ce41
JH
2176 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
2177 if (SvPOK(retval))
35c1215d 2178 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"",
be2597df 2179 SVfARG(retval));
cea2e8a9 2180 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
bc45ce41 2181 }
a0ed51b3
LW
2182 return retval;
2183}
2184
035d37be
JH
2185
2186/* This API is wrong for special case conversions since we may need to
2187 * return several Unicode characters for a single Unicode character
2188 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
2189 * the lower-level routine, and it is similarly broken for returning
38684baa
KW
2190 * multiple values. --jhi
2191 * For those, you should use to_utf8_case() instead */
979f2922 2192/* Now SWASHGET is recasted into S_swash_get in this file. */
680c470c
ST
2193
2194/* Note:
2195 * Returns the value of property/mapping C<swash> for the first character
2196 * of the string C<ptr>. If C<do_utf8> is true, the string C<ptr> is
2197 * assumed to be in utf8. If C<do_utf8> is false, the string C<ptr> is
2198 * assumed to be in native 8-bit encoding. Caches the swatch in C<swash>.
2199 */
a0ed51b3 2200UV
680c470c 2201Perl_swash_fetch(pTHX_ SV *swash, const U8 *ptr, bool do_utf8)
a0ed51b3 2202{
27da23d5 2203 dVAR;
ef8f7699 2204 HV *const hv = MUTABLE_HV(SvRV(swash));
3568d838
JH
2205 U32 klen;
2206 U32 off;
a0ed51b3 2207 STRLEN slen;
7d85a32c 2208 STRLEN needents;
cfd0369c 2209 const U8 *tmps = NULL;
a0ed51b3 2210 U32 bit;
979f2922 2211 SV *swatch;
3568d838 2212 U8 tmputf8[2];
35da51f7 2213 const UV c = NATIVE_TO_ASCII(*ptr);
3568d838 2214
7918f24d
NC
2215 PERL_ARGS_ASSERT_SWASH_FETCH;
2216
3568d838 2217 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
979f2922
ST
2218 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
2219 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
2220 ptr = tmputf8;
3568d838
JH
2221 }
2222 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
37e2e78e 2223 * then the "swatch" is a vec() for all the chars which start
3568d838
JH
2224 * with 0xAA..0xYY
2225 * So the key in the hash (klen) is length of encoded char -1
2226 */
2227 klen = UTF8SKIP(ptr) - 1;
2228 off = ptr[klen];
a0ed51b3 2229
979f2922 2230 if (klen == 0) {
37e2e78e 2231 /* If char is invariant then swatch is for all the invariant chars
1e54db1a 2232 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK
7d85a32c 2233 */
979f2922
ST
2234 needents = UTF_CONTINUATION_MARK;
2235 off = NATIVE_TO_UTF(ptr[klen]);
2236 }
2237 else {
7d85a32c 2238 /* If char is encoded then swatch is for the prefix */
979f2922
ST
2239 needents = (1 << UTF_ACCUMULATION_SHIFT);
2240 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
8457b38f 2241 if (UTF8_IS_SUPER(ptr) && ckWARN_d(WARN_NON_UNICODE)) {
9ae3ac1a
KW
2242 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0, 0);
2243
2244 /* This outputs warnings for binary properties only, assuming that
9479a769
KW
2245 * to_utf8_case() will output any for non-binary. Also, surrogates
2246 * aren't checked for, as that would warn on things like
2247 * /\p{Gc=Cs}/ */
9ae3ac1a
KW
2248 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2249 if (SvUV(*bitssvp) == 1) {
8457b38f 2250 Perl_warner(aTHX_ packWARN(WARN_NON_UNICODE),
c634fdd3 2251 "Code point 0x%04"UVXf" is not Unicode, all \\p{} matches fail; all \\P{} matches succeed", code_point);
9ae3ac1a
KW
2252 }
2253 }
979f2922 2254 }
7d85a32c 2255
a0ed51b3
LW
2256 /*
2257 * This single-entry cache saves about 1/3 of the utf8 overhead in test
2258 * suite. (That is, only 7-8% overall over just a hash cache. Still,
2259 * it's nothing to sniff at.) Pity we usually come through at least
2260 * two function calls to get here...
2261 *
2262 * NB: this code assumes that swatches are never modified, once generated!
2263 */
2264
3568d838 2265 if (hv == PL_last_swash_hv &&
a0ed51b3 2266 klen == PL_last_swash_klen &&
27da23d5 2267 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
a0ed51b3
LW
2268 {
2269 tmps = PL_last_swash_tmps;
2270 slen = PL_last_swash_slen;
2271 }
2272 else {
2273 /* Try our second-level swatch cache, kept in a hash. */
e1ec3a88 2274 SV** svp = hv_fetch(hv, (const char*)ptr, klen, FALSE);
a0ed51b3 2275
979f2922
ST
2276 /* If not cached, generate it via swash_get */
2277 if (!svp || !SvPOK(*svp)
2278 || !(tmps = (const U8*)SvPV_const(*svp, slen))) {
2b9d42f0
NIS
2279 /* We use utf8n_to_uvuni() as we want an index into
2280 Unicode tables, not a native character number.
2281 */
aec46f14 2282 const UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXBYTES, 0,
872c91ae
JH
2283 ckWARN(WARN_UTF8) ?
2284 0 : UTF8_ALLOW_ANY);
680c470c 2285 swatch = swash_get(swash,
979f2922
ST
2286 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
2287 (klen) ? (code_point & ~(needents - 1)) : 0,
2288 needents);
2289
923e4eb5 2290 if (IN_PERL_COMPILETIME)
623e6609 2291 CopHINTS_set(PL_curcop, PL_hints);
a0ed51b3 2292
979f2922 2293 svp = hv_store(hv, (const char *)ptr, klen, swatch, 0);
a0ed51b3 2294
979f2922
ST
2295 if (!svp || !(tmps = (U8*)SvPV(*svp, slen))
2296 || (slen << 3) < needents)
660a4616 2297 Perl_croak(aTHX_ "panic: swash_fetch got improper swatch");
a0ed51b3
LW
2298 }
2299
2300 PL_last_swash_hv = hv;
16d8f38a 2301 assert(klen <= sizeof(PL_last_swash_key));
eac04b2e 2302 PL_last_swash_klen = (U8)klen;
cfd0369c
NC
2303 /* FIXME change interpvar.h? */
2304 PL_last_swash_tmps = (U8 *) tmps;
a0ed51b3
LW
2305 PL_last_swash_slen = slen;
2306 if (klen)
2307 Copy(ptr, PL_last_swash_key, klen, U8);
2308 }
2309
9faf8d75 2310 switch ((int)((slen << 3) / needents)) {
a0ed51b3
LW
2311 case 1:
2312 bit = 1 << (off & 7);
2313 off >>= 3;
2314 return (tmps[off] & bit) != 0;
2315 case 8:
2316 return tmps[off];
2317 case 16:
2318 off <<= 1;
2319 return (tmps[off] << 8) + tmps[off + 1] ;
2320 case 32:
2321 off <<= 2;
2322 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
2323 }
660a4616 2324 Perl_croak(aTHX_ "panic: swash_fetch got swatch of unexpected bit width");
670f1322 2325 NORETURN_FUNCTION_END;
a0ed51b3 2326}
2b9d42f0 2327
319009ee
KW
2328/* Read a single line of the main body of the swash input text. These are of
2329 * the form:
2330 * 0053 0056 0073
2331 * where each number is hex. The first two numbers form the minimum and
2332 * maximum of a range, and the third is the value associated with the range.
2333 * Not all swashes should have a third number
2334 *
2335 * On input: l points to the beginning of the line to be examined; it points
2336 * to somewhere in the string of the whole input text, and is
2337 * terminated by a \n or the null string terminator.
2338 * lend points to the null terminator of that string
2339 * wants_value is non-zero if the swash expects a third number
2340 * typestr is the name of the swash's mapping, like 'ToLower'
2341 * On output: *min, *max, and *val are set to the values read from the line.
2342 * returns a pointer just beyond the line examined. If there was no
2343 * valid min number on the line, returns lend+1
2344 */
2345
2346STATIC U8*
2347S_swash_scan_list_line(pTHX_ U8* l, U8* const lend, UV* min, UV* max, UV* val,
2348 const bool wants_value, const U8* const typestr)
2349{
2350 const int typeto = typestr[0] == 'T' && typestr[1] == 'o';
2351 STRLEN numlen; /* Length of the number */
02470786
KW
2352 I32 flags = PERL_SCAN_SILENT_ILLDIGIT
2353 | PERL_SCAN_DISALLOW_PREFIX
2354 | PERL_SCAN_SILENT_NON_PORTABLE;
319009ee
KW
2355
2356 /* nl points to the next \n in the scan */
2357 U8* const nl = (U8*)memchr(l, '\n', lend - l);
2358
2359 /* Get the first number on the line: the range minimum */
2360 numlen = lend - l;
2361 *min = grok_hex((char *)l, &numlen, &flags, NULL);
2362 if (numlen) /* If found a hex number, position past it */
2363 l += numlen;
2364 else if (nl) { /* Else, go handle next line, if any */
2365 return nl + 1; /* 1 is length of "\n" */
2366 }
2367 else { /* Else, no next line */
2368 return lend + 1; /* to LIST's end at which \n is not found */
2369 }
2370
2371 /* The max range value follows, separated by a BLANK */
2372 if (isBLANK(*l)) {
2373 ++l;
02470786
KW
2374 flags = PERL_SCAN_SILENT_ILLDIGIT
2375 | PERL_SCAN_DISALLOW_PREFIX
2376 | PERL_SCAN_SILENT_NON_PORTABLE;
319009ee
KW
2377 numlen = lend - l;
2378 *max = grok_hex((char *)l, &numlen, &flags, NULL);
2379 if (numlen)
2380 l += numlen;
2381 else /* If no value here, it is a single element range */
2382 *max = *min;
2383
2384 /* Non-binary tables have a third entry: what the first element of the
2385 * range maps to */
2386 if (wants_value) {
2387 if (isBLANK(*l)) {
2388 ++l;
02470786
KW
2389 flags = PERL_SCAN_SILENT_ILLDIGIT
2390 | PERL_SCAN_DISALLOW_PREFIX
2391 | PERL_SCAN_SILENT_NON_PORTABLE;
319009ee
KW
2392 numlen = lend - l;
2393 *val = grok_hex((char *)l, &numlen, &flags, NULL);
2394 if (numlen)
2395 l += numlen;
2396 else
2397 *val = 0;
2398 }
2399 else {
2400 *val = 0;
2401 if (typeto) {
2402 Perl_croak(aTHX_ "%s: illegal mapping '%s'",
2403 typestr, l);
2404 }
2405 }
2406 }
2407 else
2408 *val = 0; /* bits == 1, then any val should be ignored */
2409 }
2410 else { /* Nothing following range min, should be single element with no
2411 mapping expected */
2412 *max = *min;
2413 if (wants_value) {
2414 *val = 0;
2415 if (typeto) {
2416 Perl_croak(aTHX_ "%s: illegal mapping '%s'", typestr, l);
2417 }
2418 }
2419 else
2420 *val = 0; /* bits == 1, then val should be ignored */
2421 }
2422
2423 /* Position to next line if any, or EOF */
2424 if (nl)
2425 l = nl + 1;
2426 else
2427 l = lend;
2428
2429 return l;
2430}
2431
979f2922
ST
2432/* Note:
2433 * Returns a swatch (a bit vector string) for a code point sequence
2434 * that starts from the value C<start> and comprises the number C<span>.
2435 * A C<swash> must be an object created by SWASHNEW (see lib/utf8_heavy.pl).
2436 * Should be used via swash_fetch, which will cache the swatch in C<swash>.
2437 */
2438STATIC SV*
2439S_swash_get(pTHX_ SV* swash, UV start, UV span)
2440{
2441 SV *swatch;
77f9f126 2442 U8 *l, *lend, *x, *xend, *s, *send;
979f2922 2443 STRLEN lcur, xcur, scur;
ef8f7699 2444 HV *const hv = MUTABLE_HV(SvRV(swash));
972dd592
KW
2445
2446 /* The string containing the main body of the table */
017a3ce5 2447 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
972dd592 2448
017a3ce5
GA
2449 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2450 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2451 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2452 SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);
77f9f126 2453 SV** const invert_it_svp = hv_fetchs(hv, "INVERT_IT", FALSE);
0bd48802 2454 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
0bd48802
AL
2455 const STRLEN bits = SvUV(*bitssvp);
2456 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2457 const UV none = SvUV(*nonesvp);
2458 const UV end = start + span;
979f2922 2459
7918f24d
NC
2460 PERL_ARGS_ASSERT_SWASH_GET;
2461
979f2922 2462 if (bits != 1 && bits != 8 && bits != 16 && bits != 32) {
660a4616
ST
2463 Perl_croak(aTHX_ "panic: swash_get doesn't expect bits %"UVuf,
2464 (UV)bits);
979f2922
ST
2465 }
2466
2467 /* create and initialize $swatch */
979f2922 2468 scur = octets ? (span * octets) : (span + 7) / 8;
e524fe40
NC
2469 swatch = newSV(scur);
2470 SvPOK_on(swatch);
979f2922
ST
2471 s = (U8*)SvPVX(swatch);
2472 if (octets && none) {
0bd48802 2473 const U8* const e = s + scur;
979f2922
ST
2474 while (s < e) {
2475 if (bits == 8)
2476 *s++ = (U8)(none & 0xff);
2477 else if (bits == 16) {
2478 *s++ = (U8)((none >> 8) & 0xff);
2479 *s++ = (U8)( none & 0xff);
2480 }
2481 else if (bits == 32) {
2482 *s++ = (U8)((none >> 24) & 0xff);
2483 *s++ = (U8)((none >> 16) & 0xff);
2484 *s++ = (U8)((none >> 8) & 0xff);
2485 *s++ = (U8)( none & 0xff);
2486 }
2487 }
2488 *s = '\0';
2489 }
2490 else {
2491 (void)memzero((U8*)s, scur + 1);
2492 }
2493 SvCUR_set(swatch, scur);
2494 s = (U8*)SvPVX(swatch);
2495
2496 /* read $swash->{LIST} */
2497 l = (U8*)SvPV(*listsvp, lcur);
2498 lend = l + lcur;
2499 while (l < lend) {
35da51f7 2500 UV min, max, val;
319009ee
KW
2501 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2502 cBOOL(octets), typestr);
2503 if (l > lend) {
979f2922
ST
2504 break;
2505 }
2506
972dd592 2507 /* If looking for something beyond this range, go try the next one */
979f2922
ST
2508 if (max < start)
2509 continue;
2510
2511 if (octets) {
35da51f7 2512 UV key;
979f2922
ST
2513 if (min < start) {
2514 if (!none || val < none) {
2515 val += start - min;
2516 }
2517 min = start;
2518 }
2519 for (key = min; key <= max; key++) {
2520 STRLEN offset;
2521 if (key >= end)
2522 goto go_out_list;
2523 /* offset must be non-negative (start <= min <= key < end) */
2524 offset = octets * (key - start);
2525 if (bits == 8)
2526 s[offset] = (U8)(val & 0xff);
2527 else if (bits == 16) {
2528 s[offset ] = (U8)((val >> 8) & 0xff);
2529 s[offset + 1] = (U8)( val & 0xff);
2530 }
2531 else if (bits == 32) {
2532 s[offset ] = (U8)((val >> 24) & 0xff);
2533 s[offset + 1] = (U8)((val >> 16) & 0xff);
2534 s[offset + 2] = (U8)((val >> 8) & 0xff);
2535 s[offset + 3] = (U8)( val & 0xff);
2536 }
2537
2538 if (!none || val < none)
2539 ++val;
2540 }
2541 }
711a919c 2542 else { /* bits == 1, then val should be ignored */
35da51f7 2543 UV key;
979f2922
ST
2544 if (min < start)
2545 min = start;
2546 for (key = min; key <= max; key++) {
0bd48802 2547 const STRLEN offset = (STRLEN)(key - start);
979f2922
ST
2548 if (key >= end)
2549 goto go_out_list;
2550 s[offset >> 3] |= 1 << (offset & 7);
2551 }
2552 }
2553 } /* while */
2554 go_out_list:
2555
9479a769 2556 /* Invert if the data says it should be. Assumes that bits == 1 */
77f9f126 2557 if (invert_it_svp && SvUV(*invert_it_svp)) {
0bda3001
KW
2558
2559 /* Unicode properties should come with all bits above PERL_UNICODE_MAX
2560 * be 0, and their inversion should also be 0, as we don't succeed any
2561 * Unicode property matches for non-Unicode code points */
2562 if (start <= PERL_UNICODE_MAX) {
2563
2564 /* The code below assumes that we never cross the
2565 * Unicode/above-Unicode boundary in a range, as otherwise we would
2566 * have to figure out where to stop flipping the bits. Since this
2567 * boundary is divisible by a large power of 2, and swatches comes
2568 * in small powers of 2, this should be a valid assumption */
2569 assert(start + span - 1 <= PERL_UNICODE_MAX);
2570
507a8485
KW
2571 send = s + scur;
2572 while (s < send) {
2573 *s = ~(*s);
2574 s++;
2575 }
0bda3001 2576 }
77f9f126
KW
2577 }
2578
d73c39c5
KW
2579 /* read $swash->{EXTRAS}
2580 * This code also copied to swash_to_invlist() below */
979f2922
ST
2581 x = (U8*)SvPV(*extssvp, xcur);
2582 xend = x + xcur;
2583 while (x < xend) {
2584 STRLEN namelen;
2585 U8 *namestr;
2586 SV** othersvp;
2587 HV* otherhv;
2588 STRLEN otherbits;
2589 SV **otherbitssvp, *other;
711a919c 2590 U8 *s, *o, *nl;
979f2922
ST
2591 STRLEN slen, olen;
2592
35da51f7 2593 const U8 opc = *x++;
979f2922
ST
2594 if (opc == '\n')
2595 continue;
2596
2597 nl = (U8*)memchr(x, '\n', xend - x);
2598
2599 if (opc != '-' && opc != '+' && opc != '!' && opc != '&') {
2600 if (nl) {
2601 x = nl + 1; /* 1 is length of "\n" */
2602 continue;
2603 }
2604 else {
2605 x = xend; /* to EXTRAS' end at which \n is not found */
2606 break;
2607 }
2608 }
2609
2610 namestr = x;
2611 if (nl) {
2612 namelen = nl - namestr;
2613 x = nl + 1;
2614 }
2615 else {
2616 namelen = xend - namestr;
2617 x = xend;
2618 }
2619
2620 othersvp = hv_fetch(hv, (char *)namestr, namelen, FALSE);
ef8f7699 2621 otherhv = MUTABLE_HV(SvRV(*othersvp));
017a3ce5 2622 otherbitssvp = hv_fetchs(otherhv, "BITS", FALSE);
979f2922
ST
2623 otherbits = (STRLEN)SvUV(*otherbitssvp);
2624 if (bits < otherbits)
660a4616 2625 Perl_croak(aTHX_ "panic: swash_get found swatch size mismatch");
979f2922
ST
2626
2627 /* The "other" swatch must be destroyed after. */
2628 other = swash_get(*othersvp, start, span);
2629 o = (U8*)SvPV(other, olen);
2630
2631 if (!olen)
660a4616 2632 Perl_croak(aTHX_ "panic: swash_get got improper swatch");
979f2922
ST
2633
2634 s = (U8*)SvPV(swatch, slen);
2635 if (bits == 1 && otherbits == 1) {
2636 if (slen != olen)
660a4616 2637 Perl_croak(aTHX_ "panic: swash_get found swatch length mismatch");
979f2922
ST
2638
2639 switch (opc) {
2640 case '+':
2641 while (slen--)
2642 *s++ |= *o++;
2643 break;
2644 case '!':
2645 while (slen--)
2646 *s++ |= ~*o++;
2647 break;
2648 case '-':
2649 while (slen--)
2650 *s++ &= ~*o++;
2651 break;
2652 case '&':
2653 while (slen--)
2654 *s++ &= *o++;
2655 break;
2656 default:
2657 break;
2658 }
2659 }
711a919c 2660 else {
979f2922
ST
2661 STRLEN otheroctets = otherbits >> 3;
2662 STRLEN offset = 0;
35da51f7 2663 U8* const send = s + slen;
979f2922
ST
2664
2665 while (s < send) {
2666 UV otherval = 0;
2667
2668 if (otherbits == 1) {
2669 otherval = (o[offset >> 3] >> (offset & 7)) & 1;
2670 ++offset;
2671 }
2672 else {
2673 STRLEN vlen = otheroctets;
2674 otherval = *o++;
2675 while (--vlen) {
2676 otherval <<= 8;
2677 otherval |= *o++;
2678 }
2679 }
2680
711a919c 2681 if (opc == '+' && otherval)
6f207bd3 2682 NOOP; /* replace with otherval */
979f2922
ST
2683 else if (opc == '!' && !otherval)
2684 otherval = 1;
2685 else if (opc == '-' && otherval)
2686 otherval = 0;
2687 else if (opc == '&' && !otherval)
2688 otherval = 0;
2689 else {
711a919c 2690 s += octets; /* no replacement */
979f2922
ST
2691 continue;
2692 }
2693
2694 if (bits == 8)
2695 *s++ = (U8)( otherval & 0xff);
2696 else if (bits == 16) {
2697 *s++ = (U8)((otherval >> 8) & 0xff);
2698 *s++ = (U8)( otherval & 0xff);
2699 }
2700 else if (bits == 32) {
2701 *s++ = (U8)((otherval >> 24) & 0xff);
2702 *s++ = (U8)((otherval >> 16) & 0xff);
2703 *s++ = (U8)((otherval >> 8) & 0xff);
2704 *s++ = (U8)( otherval & 0xff);
2705 }
2706 }
2707 }
2708 sv_free(other); /* through with it! */
2709 } /* while */
2710 return swatch;
2711}
2712
064c021d 2713HV*
4c2e1131 2714Perl__swash_inversion_hash(pTHX_ SV* const swash)
064c021d
KW
2715{
2716
5662e334
KW
2717 /* Subject to change or removal. For use only in one place in regcomp.c.
2718 * Can't be used on a property that is subject to user override, as it
2719 * relies on the value of SPECIALS in the swash which would be set by
2720 * utf8_heavy.pl to the hash in the non-overriden file, and hence is not set
2721 * for overridden properties
064c021d
KW
2722 *
2723 * Returns a hash which is the inversion and closure of a swash mapping.
2724 * For example, consider the input lines:
2725 * 004B 006B
2726 * 004C 006C
2727 * 212A 006B
2728 *
2729 * The returned hash would have two keys, the utf8 for 006B and the utf8 for
2730 * 006C. The value for each key is an array. For 006C, the array would
2731 * have a two elements, the utf8 for itself, and for 004C. For 006B, there
2732 * would be three elements in its array, the utf8 for 006B, 004B and 212A.
2733 *
2734 * Essentially, for any code point, it gives all the code points that map to
2735 * it, or the list of 'froms' for that point.
2736 *
5662e334
KW
2737 * Currently it ignores any additions or deletions from other swashes,
2738 * looking at just the main body of the swash, and if there are SPECIALS
2739 * in the swash, at that hash
2740 *
2741 * The specials hash can be extra code points, and most likely consists of
2742 * maps from single code points to multiple ones (each expressed as a string
2743 * of utf8 characters). This function currently returns only 1-1 mappings.
2744 * However consider this possible input in the specials hash:
2745 * "\xEF\xAC\x85" => "\x{0073}\x{0074}", # U+FB05 => 0073 0074
2746 * "\xEF\xAC\x86" => "\x{0073}\x{0074}", # U+FB06 => 0073 0074
2747 *
2748 * Both FB05 and FB06 map to the same multi-char sequence, which we don't
2749 * currently handle. But it also means that FB05 and FB06 are equivalent in
2750 * a 1-1 mapping which we should handle, and this relationship may not be in
2751 * the main table. Therefore this function examines all the multi-char
2752 * sequences and adds the 1-1 mappings that come out of that. */
064c021d
KW
2753
2754 U8 *l, *lend;
2755 STRLEN lcur;
2756 HV *const hv = MUTABLE_HV(SvRV(swash));
2757
2758 /* The string containing the main body of the table */
2759 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2760
2761 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2762 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
2763 SV** const nonesvp = hv_fetchs(hv, "NONE", FALSE);
2764 /*SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);*/
2765 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
2766 const STRLEN bits = SvUV(*bitssvp);
2767 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
2768 const UV none = SvUV(*nonesvp);
5662e334 2769 SV **specials_p = hv_fetchs(hv, "SPECIALS", 0);
064c021d
KW
2770
2771 HV* ret = newHV();
2772
2773 PERL_ARGS_ASSERT__SWASH_INVERSION_HASH;
2774
2775 /* Must have at least 8 bits to get the mappings */
2776 if (bits != 8 && bits != 16 && bits != 32) {
2777 Perl_croak(aTHX_ "panic: swash_inversion_hash doesn't expect bits %"UVuf,
2778 (UV)bits);
2779 }
2780
5662e334
KW
2781 if (specials_p) { /* It might be "special" (sometimes, but not always, a
2782 mapping to more than one character */
2783
2784 /* Construct an inverse mapping hash for the specials */
2785 HV * const specials_hv = MUTABLE_HV(SvRV(*specials_p));
2786 HV * specials_inverse = newHV();
2787 char *char_from; /* the lhs of the map */
2788 I32 from_len; /* its byte length */
2789 char *char_to; /* the rhs of the map */
2790 I32 to_len; /* its byte length */
2791 SV *sv_to; /* and in a sv */
2792 AV* from_list; /* list of things that map to each 'to' */
2793
2794 hv_iterinit(specials_hv);
2795
2796 /* The keys are the characters (in utf8) that map to the corresponding
2797 * utf8 string value. Iterate through the list creating the inverse
2798 * list. */
2799 while ((sv_to = hv_iternextsv(specials_hv, &char_from, &from_len))) {
2800 SV** listp;
2801 if (! SvPOK(sv_to)) {
2802 Perl_croak(aTHX_ "panic: value returned from hv_iternextsv() unexpectedly is not a string");
2803 }
2804 /*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)));*/
2805
2806 /* Each key in the inverse list is a mapped-to value, and the key's
2807 * hash value is a list of the strings (each in utf8) that map to
2808 * it. Those strings are all one character long */
2809 if ((listp = hv_fetch(specials_inverse,
2810 SvPVX(sv_to),
2811 SvCUR(sv_to), 0)))
2812 {
2813 from_list = (AV*) *listp;
2814 }
2815 else { /* No entry yet for it: create one */
2816 from_list = newAV();
2817 if (! hv_store(specials_inverse,
2818 SvPVX(sv_to),
2819 SvCUR(sv_to),
2820 (SV*) from_list, 0))
2821 {
2822 Perl_croak(aTHX_ "panic: hv_store() unexpectedly failed");
2823 }
2824 }
2825
2826 /* Here have the list associated with this 'to' (perhaps newly
2827 * created and empty). Just add to it. Note that we ASSUME that
2828 * the input is guaranteed to not have duplications, so we don't
2829 * check for that. Duplications just slow down execution time. */
2830 av_push(from_list, newSVpvn_utf8(char_from, from_len, TRUE));
2831 }
2832
2833 /* Here, 'specials_inverse' contains the inverse mapping. Go through
2834 * it looking for cases like the FB05/FB06 examples above. There would
2835 * be an entry in the hash like
2836 * 'st' => [ FB05, FB06 ]
2837 * In this example we will create two lists that get stored in the
2838 * returned hash, 'ret':
2839 * FB05 => [ FB05, FB06 ]
2840 * FB06 => [ FB05, FB06 ]
2841 *
2842 * Note that there is nothing to do if the array only has one element.
2843 * (In the normal 1-1 case handled below, we don't have to worry about
2844 * two lists, as everything gets tied to the single list that is
2845 * generated for the single character 'to'. But here, we are omitting
2846 * that list, ('st' in the example), so must have multiple lists.) */
2847 while ((from_list = (AV *) hv_iternextsv(specials_inverse,
2848 &char_to, &to_len)))
2849 {
2850 if (av_len(from_list) > 0) {
2851 int i;
2852
2853 /* We iterate over all combinations of i,j to place each code
2854 * point on each list */
2855 for (i = 0; i <= av_len(from_list); i++) {
2856 int j;
2857 AV* i_list = newAV();
2858 SV** entryp = av_fetch(from_list, i, FALSE);
2859 if (entryp == NULL) {
2860 Perl_croak(aTHX_ "panic: av_fetch() unexpectedly failed");
2861 }
2862 if (hv_fetch(ret, SvPVX(*entryp), SvCUR(*entryp), FALSE)) {
2863 Perl_croak(aTHX_ "panic: unexpected entry for %s", SvPVX(*entryp));
2864 }
2865 if (! hv_store(ret, SvPVX(*entryp), SvCUR(*entryp),
2866 (SV*) i_list, FALSE))
2867 {
2868 Perl_croak(aTHX_ "panic: hv_store() unexpectedly failed");
2869 }
2870
2871 /* For debugging: UV u = utf8_to_uvchr((U8*) SvPVX(*entryp), 0);*/
2872 for (j = 0; j <= av_len(from_list); j++) {
2873 entryp = av_fetch(from_list, j, FALSE);
2874 if (entryp == NULL) {
2875 Perl_croak(aTHX_ "panic: av_fetch() unexpectedly failed");
2876 }
2877
2878 /* When i==j this adds itself to the list */
2879 av_push(i_list, newSVuv(utf8_to_uvchr(
2880 (U8*) SvPVX(*entryp), 0)));
2881 /*DEBUG_U(PerlIO_printf(Perl_debug_log, "Adding %"UVXf" to list for %"UVXf"\n", utf8_to_uvchr((U8*) SvPVX(*entryp), 0), u));*/
2882 }
2883 }
2884 }
2885 }
2886 SvREFCNT_dec(specials_inverse); /* done with it */
2887 } /* End of specials */
2888
064c021d
KW
2889 /* read $swash->{LIST} */
2890 l = (U8*)SvPV(*listsvp, lcur);
2891 lend = l + lcur;
2892
2893 /* Go through each input line */
2894 while (l < lend) {
2895 UV min, max, val;
2896 UV inverse;
2897 l = S_swash_scan_list_line(aTHX_ l, lend, &min, &max, &val,
2898 cBOOL(octets), typestr);
2899 if (l > lend) {
2900 break;
2901 }
2902
2903 /* Each element in the range is to be inverted */
2904 for (inverse = min; inverse <= max; inverse++) {
2905 AV* list;
064c021d
KW
2906 SV** listp;
2907 IV i;
2908 bool found_key = FALSE;
5662e334 2909 bool found_inverse = FALSE;
064c021d
KW
2910
2911 /* The key is the inverse mapping */
2912 char key[UTF8_MAXBYTES+1];
2913 char* key_end = (char *) uvuni_to_utf8((U8*) key, val);
2914 STRLEN key_len = key_end - key;
2915
064c021d
KW
2916 /* Get the list for the map */
2917 if ((listp = hv_fetch(ret, key, key_len, FALSE))) {
2918 list = (AV*) *listp;
2919 }
2920 else { /* No entry yet for it: create one */
2921 list = newAV();
2922 if (! hv_store(ret, key, key_len, (SV*) list, FALSE)) {
2923 Perl_croak(aTHX_ "panic: hv_store() unexpectedly failed");
2924 }
2925 }
2926
5662e334
KW
2927 /* Look through list to see if this inverse mapping already is
2928 * listed, or if there is a mapping to itself already */
508f7cfa 2929 for (i = 0; i <= av_len(list); i++) {
064c021d
KW
2930 SV** entryp = av_fetch(list, i, FALSE);
2931 SV* entry;
2932 if (entryp == NULL) {
2933 Perl_croak(aTHX_ "panic: av_fetch() unexpectedly failed");
2934 }
2935 entry = *entryp;
5662e334 2936 /*DEBUG_U(PerlIO_printf(Perl_debug_log, "list for %"UVXf" contains %"UVXf"\n", val, SvUV(entry)));*/
56ca34ca 2937 if (SvUV(entry) == val) {
064c021d 2938 found_key = TRUE;
5662e334
KW
2939 }
2940 if (SvUV(entry) == inverse) {
2941 found_inverse = TRUE;
2942 }
2943
2944 /* No need to continue searching if found everything we are
2945 * looking for */
2946 if (found_key && found_inverse) {
064c021d
KW
2947 break;
2948 }
2949 }
56ca34ca
KW
2950
2951 /* Make sure there is a mapping to itself on the list */
064c021d 2952 if (! found_key) {
d397ff6a 2953 av_push(list, newSVuv(val));
5662e334 2954 /*DEBUG_U(PerlIO_printf(Perl_debug_log, "Adding %"UVXf" to list for %"UVXf"\n", val, val));*/
064c021d
KW
2955 }
2956
2957
2958 /* Simply add the value to the list */
5662e334
KW
2959 if (! found_inverse) {
2960 av_push(list, newSVuv(inverse));
2961 /*DEBUG_U(PerlIO_printf(Perl_debug_log, "Adding %"UVXf" to list for %"UVXf"\n", inverse, val));*/
2962 }
064c021d
KW
2963
2964 /* swash_get() increments the value of val for each element in the
2965 * range. That makes more compact tables possible. You can
2966 * express the capitalization, for example, of all consecutive
2967 * letters with a single line: 0061\t007A\t0041 This maps 0061 to
2968 * 0041, 0062 to 0042, etc. I (khw) have never understood 'none',
bd3f2f94
KW
2969 * and it's not documented; it appears to be used only in
2970 * implementing tr//; I copied the semantics from swash_get(), just
2971 * in case */
064c021d
KW
2972 if (!none || val < none) {
2973 ++val;
2974 }
2975 }
2976 }
2977
2978 return ret;
2979}
2980
a25abddc 2981SV*
d764b54e
KW
2982Perl__swash_to_invlist(pTHX_ SV* const swash)
2983{
2984
2985 /* Subject to change or removal. For use only in one place in regcomp.c */
2986
2987 U8 *l, *lend;
2988 char *loc;
2989 STRLEN lcur;
2990 HV *const hv = MUTABLE_HV(SvRV(swash));
2991 UV elements = 0; /* Number of elements in the inversion list */
b443038a 2992 U8 empty[] = "";
d764b54e
KW
2993
2994 /* The string containing the main body of the table */
2995 SV** const listsvp = hv_fetchs(hv, "LIST", FALSE);
2996 SV** const typesvp = hv_fetchs(hv, "TYPE", FALSE);
2997 SV** const bitssvp = hv_fetchs(hv, "BITS", FALSE);
d73c39c5 2998 SV** const extssvp = hv_fetchs(hv, "EXTRAS", FALSE);
77f9f126 2999 SV** const invert_it_svp = hv_fetchs(hv, "INVERT_IT", FALSE);
d764b54e
KW
3000
3001 const U8* const typestr = (U8*)SvPV_nolen(*typesvp);
3002 const STRLEN bits = SvUV(*bitssvp);
3003 const STRLEN octets = bits >> 3; /* if bits == 1, then octets == 0 */
d73c39c5
KW
3004 U8 *x, *xend;
3005 STRLEN xcur;
d764b54e 3006
a25abddc 3007 SV* invlist;
d764b54e
KW
3008
3009 PERL_ARGS_ASSERT__SWASH_TO_INVLIST;
3010
3011 /* read $swash->{LIST} */
b443038a
KW
3012 if (SvPOK(*listsvp)) {
3013 l = (U8*)SvPV(*listsvp, lcur);
3014 }
3015 else {
3016 /* LIST legitimately doesn't contain a string during compilation phases
3017 * of Perl itself, before the Unicode tables are generated. In this
3018 * case, just fake things up by creating an empty list */
3019 l = empty;
3020 lcur = 0;
3021 }
d764b54e
KW
3022 loc = (char *) l;
3023 lend = l + lcur;
3024
3025 /* Scan the input to count the number of lines to preallocate array size
3026 * based on worst possible case, which is each line in the input creates 2
3027 * elements in the inversion list: 1) the beginning of a range in the list;
3028 * 2) the beginning of a range not in the list. */
3029 while ((loc = (strchr(loc, '\n'))) != NULL) {
3030 elements += 2;
3031 loc++;
3032 }
3033
3034 /* If the ending is somehow corrupt and isn't a new line, add another
3035 * element for the final range that isn't in the inversion list */
3036 if (! (*lend == '\n' || (*lend == '\0' && *(lend - 1) == '\n'))) {
3037 elements++;
3038 }
3039
3040 invlist = _new_invlist(elements);
3041
3042 /* Now go through the input again, adding each range to the list */
3043 while (l < lend) {
3044 UV start, end;
3045 UV val; /* Not used by this function */
3046
3047 l = S_swash_scan_list_line(aTHX_ l, lend, &start, &end, &val,
3048 cBOOL(octets), typestr);
3049
3050 if (l > lend) {
3051 break;
3052 }
3053
3054 _append_range_to_invlist(invlist, start, end);
3055 }
3056
77f9f126
KW
3057 /* Invert if the data says it should be */
3058 if (invert_it_svp && SvUV(*invert_it_svp)) {
45bb2768 3059 _invlist_invert_prop(invlist);
77f9f126
KW
3060 }
3061
d73c39c5
KW
3062 /* This code is copied from swash_get()
3063 * read $swash->{EXTRAS} */
3064 x = (U8*)SvPV(*extssvp, xcur);
3065 xend = x + xcur;
3066 while (x < xend) {
3067 STRLEN namelen;
3068 U8 *namestr;
3069 SV** othersvp;
3070 HV* otherhv;
3071 STRLEN otherbits;
3072 SV **otherbitssvp, *other;
3073 U8 *nl;
3074
3075 const U8 opc = *x++;
3076 if (opc == '\n')
3077 continue;
3078
3079 nl = (U8*)memchr(x, '\n', xend - x);
3080
3081 if (opc != '-' && opc != '+' && opc != '!' && opc != '&') {
3082 if (nl) {
3083 x = nl + 1; /* 1 is length of "\n" */
3084 continue;
3085 }
3086 else {
3087 x = xend; /* to EXTRAS' end at which \n is not found */
3088 break;
3089 }
3090 }
3091
3092 namestr = x;
3093 if (nl) {
3094 namelen = nl - namestr;
3095 x = nl + 1;
3096 }
3097 else {
3098 namelen = xend - namestr;
3099 x = xend;
3100 }
3101
3102 othersvp = hv_fetch(hv, (char *)namestr, namelen, FALSE);
3103 otherhv = MUTABLE_HV(SvRV(*othersvp));
3104 otherbitssvp = hv_fetchs(otherhv, "BITS", FALSE);
3105 otherbits = (STRLEN)SvUV(*otherbitssvp);
3106
3107 if (bits != otherbits || bits != 1) {
3108 Perl_croak(aTHX_ "panic: _swash_to_invlist only operates on boolean properties");
3109 }
3110
3111 /* The "other" swatch must be destroyed after. */
3112 other = _swash_to_invlist((SV *)*othersvp);
3113
3114 /* End of code copied from swash_get() */
3115 switch (opc) {
3116 case '+':
3117 _invlist_union(invlist, other, &invlist);
3118 break;
3119 case '!':
3120 _invlist_invert(other);
3121 _invlist_union(invlist, other, &invlist);
3122 break;
3123 case '-':
3124 _invlist_subtract(invlist, other, &invlist);
3125 break;
3126 case '&':
3127 _invlist_intersection(invlist, other, &invlist);
3128 break;
3129 default:
3130 break;
3131 }
3132 sv_free(other); /* through with it! */
3133 }
3134
d764b54e
KW
3135 return invlist;
3136}
3137
0f830e0b 3138/*
87cea99e 3139=for apidoc uvchr_to_utf8
0f830e0b 3140
6ee84de2 3141Adds the UTF-8 representation of the Native code point C<uv> to the end
0f830e0b
NC
3142of the string C<d>; C<d> should be have at least C<UTF8_MAXBYTES+1> free
3143bytes available. The return value is the pointer to the byte after the
3144end of the new character. In other words,
3145
3146 d = uvchr_to_utf8(d, uv);
3147
3148is the recommended wide native character-aware way of saying
3149
3150 *(d++) = uv;
3151
3152=cut
3153*/
3154
3155/* On ASCII machines this is normally a macro but we want a
3156 real function in case XS code wants it
3157*/
0f830e0b
NC
3158U8 *
3159Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
3160{
7918f24d
NC
3161 PERL_ARGS_ASSERT_UVCHR_TO_UTF8;
3162
0f830e0b
NC
3163 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
3164}
3165
b851fbc1
JH
3166U8 *
3167Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
3168{
7918f24d
NC
3169 PERL_ARGS_ASSERT_UVCHR_TO_UTF8_FLAGS;
3170
b851fbc1
JH
3171 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
3172}
2b9d42f0
NIS
3173
3174/*
87cea99e 3175=for apidoc utf8n_to_uvchr
0f830e0b 3176
48ef279e 3177Returns the native character value of the first character in the string
0f830e0b
NC
3178C<s>
3179which is assumed to be in UTF-8 encoding; C<retlen> will be set to the
3180length, in bytes, of that character.
3181
6ee84de2 3182length and flags are the same as utf8n_to_uvuni().
0f830e0b
NC
3183
3184=cut
3185*/
3186/* On ASCII machines this is normally a macro but we want
3187 a real function in case XS code wants it
3188*/
0f830e0b 3189UV
48ef279e 3190Perl_utf8n_to_uvchr(pTHX_ const U8 *s, STRLEN curlen, STRLEN *retlen,
0f830e0b
NC
3191U32 flags)
3192{
3193 const UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
7918f24d
NC
3194
3195 PERL_ARGS_ASSERT_UTF8N_TO_UVCHR;
3196
0f830e0b
NC
3197 return UNI_TO_NATIVE(uv);
3198}
3199
0876b9a0
KW
3200bool
3201Perl_check_utf8_print(pTHX_ register const U8* s, const STRLEN len)
3202{
3203 /* May change: warns if surrogates, non-character code points, or
3204 * non-Unicode code points are in s which has length len. Returns TRUE if
3205 * none found; FALSE otherwise. The only other validity check is to make
3206 * sure that this won't exceed the string's length */
3207
3208 const U8* const e = s + len;
3209 bool ok = TRUE;
3210
3211 PERL_ARGS_ASSERT_CHECK_UTF8_PRINT;
3212
3213 while (s < e) {
3214 if (UTF8SKIP(s) > len) {
3215 Perl_ck_warner_d(aTHX_ packWARN(WARN_UTF8),
3216 "%s in %s", unees, PL_op ? OP_DESC(PL_op) : "print");
3217 return FALSE;
3218 }
3219 if (*s >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE) {
3220 STRLEN char_len;
3221 if (UTF8_IS_SUPER(s)) {
8457b38f
KW
3222 if (ckWARN_d(WARN_NON_UNICODE)) {
3223 UV uv = utf8_to_uvchr(s, &char_len);
3224 Perl_warner(aTHX_ packWARN(WARN_NON_UNICODE),
3225 "Code point 0x%04"UVXf" is not Unicode, may not be portable", uv);
3226 ok = FALSE;
3227 }
0876b9a0
KW
3228 }
3229 else if (UTF8_IS_SURROGATE(s)) {
8457b38f
KW
3230 if (ckWARN_d(WARN_SURROGATE)) {
3231 UV uv = utf8_to_uvchr(s, &char_len);
3232 Perl_warner(aTHX_ packWARN(WARN_SURROGATE),
3233 "Unicode surrogate U+%04"UVXf" is illegal in UTF-8", uv);
3234 ok = FALSE;
3235 }
0876b9a0
KW
3236 }
3237 else if
8457b38f
KW
3238 ((UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s))
3239 && (ckWARN_d(WARN_NONCHAR)))
0876b9a0
KW
3240 {
3241 UV uv = utf8_to_uvchr(s, &char_len);
8457b38f 3242 Perl_warner(aTHX_ packWARN(WARN_NONCHAR),
0876b9a0
KW
3243 "Unicode non-character U+%04"UVXf" is illegal for open interchange", uv);
3244 ok = FALSE;
3245 }
3246 }
3247 s += UTF8SKIP(s);
3248 }
3249
3250 return ok;
3251}
3252
0f830e0b 3253/*
87cea99e 3254=for apidoc pv_uni_display
d2cc3551
JH
3255
3256Build to the scalar dsv a displayable version of the string spv,
3257length len, the displayable version being at most pvlim bytes long
3258(if longer, the rest is truncated and "..." will be appended).
0a2ef054 3259
9e55ce06 3260The flags argument can have UNI_DISPLAY_ISPRINT set to display
00e86452 3261isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
0a2ef054
JH
3262to display the \\[nrfta\\] as the backslashed versions (like '\n')
3263(UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
3264UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
3265UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
3266
d2cc3551
JH
3267The pointer to the PV of the dsv is returned.
3268
3269=cut */
e6b2e755 3270char *
e1ec3a88 3271Perl_pv_uni_display(pTHX_ SV *dsv, const U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
e6b2e755
JH
3272{
3273 int truncated = 0;
e1ec3a88 3274 const char *s, *e;
e6b2e755 3275
7918f24d
NC
3276 PERL_ARGS_ASSERT_PV_UNI_DISPLAY;
3277
76f68e9b 3278 sv_setpvs(dsv, "");
7fddd944 3279 SvUTF8_off(dsv);
e1ec3a88 3280 for (s = (const char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
e6b2e755 3281 UV u;
a49f32c6
NC
3282 /* This serves double duty as a flag and a character to print after
3283 a \ when flags & UNI_DISPLAY_BACKSLASH is true.
3284 */
3285 char ok = 0;
c728cb41 3286
e6b2e755
JH
3287 if (pvlim && SvCUR(dsv) >= pvlim) {
3288 truncated++;
3289 break;
3290 }
3291 u = utf8_to_uvchr((U8*)s, 0);
c728cb41 3292 if (u < 256) {
a3b680e6 3293 const unsigned char c = (unsigned char)u & 0xFF;
0bd48802 3294 if (flags & UNI_DISPLAY_BACKSLASH) {
a49f32c6 3295 switch (c) {
c728cb41 3296 case '\n':
a49f32c6 3297 ok = 'n'; break;
c728cb41 3298 case '\r':
a49f32c6 3299 ok = 'r'; break;
c728cb41 3300 case '\t':
a49f32c6 3301 ok = 't'; break;
c728cb41 3302 case '\f':
a49f32c6 3303 ok = 'f'; break;
c728cb41 3304 case '\a':
a49f32c6 3305 ok = 'a'; break;
c728cb41 3306 case '\\':
a49f32c6 3307 ok = '\\'; break;
c728cb41
JH
3308 default: break;
3309 }
a49f32c6 3310 if (ok) {
88c9ea1e 3311 const char string = ok;
76f68e9b 3312 sv_catpvs(dsv, "\\");
5e7aa789 3313 sv_catpvn(dsv, &string, 1);
a49f32c6 3314 }
c728cb41 3315 }
00e86452 3316 /* isPRINT() is the locale-blind version. */
a49f32c6 3317 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(c)) {
88c9ea1e 3318 const char string = c;
5e7aa789 3319 sv_catpvn(dsv, &string, 1);
a49f32c6 3320 ok = 1;
0a2ef054 3321 }
c728cb41
JH
3322 }
3323 if (!ok)
9e55ce06 3324 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
e6b2e755
JH
3325 }
3326 if (truncated)
396482e1 3327 sv_catpvs(dsv, "...");
48ef279e 3328
e6b2e755
JH
3329 return SvPVX(dsv);
3330}
2b9d42f0 3331
d2cc3551 3332/*
87cea99e 3333=for apidoc sv_uni_display
d2cc3551
JH
3334
3335Build to the scalar dsv a displayable version of the scalar sv,
0a2ef054 3336the displayable version being at most pvlim bytes long
d2cc3551 3337(if longer, the rest is truncated and "..." will be appended).
0a2ef054
JH
3338
3339The flags argument is as in pv_uni_display().
3340
d2cc3551
JH
3341The pointer to the PV of the dsv is returned.
3342
d4c19fe8
AL
3343=cut
3344*/
e6b2e755
JH
3345char *
3346Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
3347{
7918f24d
NC
3348 PERL_ARGS_ASSERT_SV_UNI_DISPLAY;
3349
cfd0369c
NC
3350 return Perl_pv_uni_display(aTHX_ dsv, (const U8*)SvPVX_const(ssv),
3351 SvCUR(ssv), pvlim, flags);
701a277b
JH
3352}
3353
d2cc3551 3354/*
e6226b18 3355=for apidoc foldEQ_utf8
d2cc3551 3356
d51c1b21 3357Returns true if the leading portions of the strings s1 and s2 (either or both
e6226b18 3358of which may be in UTF-8) are the same case-insensitively; false otherwise.
d51c1b21 3359How far into the strings to compare is determined by other input parameters.
8b35872c
KW
3360
3361If u1 is true, the string s1 is assumed to be in UTF-8-encoded Unicode;
3362otherwise it is assumed to be in native 8-bit encoding. Correspondingly for u2
3363with respect to s2.
3364
d51c1b21
KW
3365If the byte length l1 is non-zero, it says how far into s1 to check for fold
3366equality. In other words, s1+l1 will be used as a goal to reach. The
8b35872c
KW
3367scan will not be considered to be a match unless the goal is reached, and
3368scanning won't continue past that goal. Correspondingly for l2 with respect to
3369s2.
3370
3371If pe1 is non-NULL and the pointer it points to is not NULL, that pointer is
3372considered an end pointer beyond which scanning of s1 will not continue under
3373any circumstances. This means that if both l1 and pe1 are specified, and pe1
3374is less than s1+l1, the match will never be successful because it can never
d51c1b21
KW
3375get as far as its goal (and in fact is asserted against). Correspondingly for
3376pe2 with respect to s2.
8b35872c 3377
d51c1b21
KW
3378At least one of s1 and s2 must have a goal (at least one of l1 and l2 must be
3379non-zero), and if both do, both have to be
8b35872c
KW
3380reached for a successful match. Also, if the fold of a character is multiple
3381characters, all of them must be matched (see tr21 reference below for
3382'folding').
3383
e6226b18 3384Upon a successful match, if pe1 is non-NULL,
8b35872c
KW
3385it will be set to point to the beginning of the I<next> character of s1 beyond
3386what was matched. Correspondingly for pe2 and s2.
d2cc3551
JH
3387
3388For case-insensitiveness, the "casefolding" of Unicode is used
3389instead of upper/lowercasing both the characters, see
3390http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
3391
3392=cut */
a33c29bc
KW
3393
3394/* A flags parameter has been added which may change, and hence isn't
3395 * externally documented. Currently it is:
3396 * 0 for as-documented above
3397 * FOLDEQ_UTF8_NOMIX_ASCII meaning that if a non-ASCII character folds to an
3398 ASCII one, to not match
5e64d0fa
KW
3399 * FOLDEQ_UTF8_LOCALE meaning that locale rules are to be used for code
3400 * points below 256; unicode rules for above 255; and
3401 * folds that cross those boundaries are disallowed,
3402 * like the NOMIX_ASCII option
18f762c3
KW
3403 * FOLDEQ_S1_ALREADY_FOLDED s1 has already been folded before calling this
3404 * routine. This allows that step to be skipped.
3405 * FOLDEQ_S2_ALREADY_FOLDED Similarly.
a33c29bc 3406 */
701a277b 3407I32
eda9cac1 3408Perl_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 3409{
8b35872c
KW
3410 dVAR;
3411 register const U8 *p1 = (const U8*)s1; /* Point to current char */
3412 register const U8 *p2 = (const U8*)s2;
48ef279e 3413 register const U8 *g1 = NULL; /* goal for s1 */
8b35872c 3414 register const U8 *g2 = NULL;
48ef279e
KW
3415 register const U8 *e1 = NULL; /* Don't scan s1 past this */
3416 register U8 *f1 = NULL; /* Point to current folded */
8b35872c
KW
3417 register const U8 *e2 = NULL;
3418 register U8 *f2 = NULL;
48ef279e 3419 STRLEN n1 = 0, n2 = 0; /* Number of bytes in current char */
8b35872c
KW
3420 U8 foldbuf1[UTF8_MAXBYTES_CASE+1];
3421 U8 foldbuf2[UTF8_MAXBYTES_CASE+1];
48ef279e
KW
3422 U8 natbuf[2]; /* Holds native 8-bit char converted to utf8;
3423 these always fit in 2 bytes */
8b35872c 3424
eda9cac1 3425 PERL_ARGS_ASSERT_FOLDEQ_UTF8_FLAGS;
8b35872c 3426
18f762c3
KW
3427 /* The algorithm requires that input with the flags on the first line of
3428 * the assert not be pre-folded. */
3429 assert( ! ((flags & (FOLDEQ_UTF8_NOMIX_ASCII | FOLDEQ_UTF8_LOCALE))
3430 && (flags & (FOLDEQ_S1_ALREADY_FOLDED | FOLDEQ_S2_ALREADY_FOLDED))));
3431
8b35872c 3432 if (pe1) {
48ef279e 3433 e1 = *(U8**)pe1;
8b35872c
KW
3434 }
3435
3436 if (l1) {
48ef279e 3437 g1 = (const U8*)s1 + l1;
8b35872c
KW
3438 }
3439
3440 if (pe2) {
48ef279e 3441 e2 = *(U8**)pe2;
8b35872c
KW
3442 }
3443
3444 if (l2) {
48ef279e 3445 g2 = (const U8*)s2 + l2;
8b35872c
KW
3446 }
3447
3448 /* Must have at least one goal */
3449 assert(g1 || g2);
3450
3451 if (g1) {
3452
48ef279e
KW
3453 /* Will never match if goal is out-of-bounds */
3454 assert(! e1 || e1 >= g1);
8b35872c 3455
48ef279e
KW
3456 /* Here, there isn't an end pointer, or it is beyond the goal. We
3457 * only go as far as the goal */
3458 e1 = g1;
8b35872c 3459 }
313b38e5
NC
3460 else {
3461 assert(e1); /* Must have an end for looking at s1 */
3462 }
8b35872c
KW
3463
3464 /* Same for goal for s2 */
3465 if (g2) {
48ef279e
KW
3466 assert(! e2 || e2 >= g2);
3467 e2 = g2;
8b35872c 3468 }
313b38e5
NC
3469 else {
3470 assert(e2);
3471 }
8b35872c 3472
18f762c3
KW
3473 /* If both operands are already folded, we could just do a memEQ on the
3474 * whole strings at once, but it would be better if the caller realized
3475 * this and didn't even call us */
3476
8b35872c
KW
3477 /* Look through both strings, a character at a time */
3478 while (p1 < e1 && p2 < e2) {
3479
d51c1b21 3480 /* If at the beginning of a new character in s1, get its fold to use
5e64d0fa
KW
3481 * and the length of the fold. (exception: locale rules just get the
3482 * character to a single byte) */
48ef279e 3483 if (n1 == 0) {
18f762c3
KW
3484 if (flags & FOLDEQ_S1_ALREADY_FOLDED) {
3485 f1 = (U8 *) p1;
3486 n1 = UTF8SKIP(f1);
a33c29bc 3487
5e64d0fa
KW
3488 /* If in locale matching, we use two sets of rules, depending on if
3489 * the code point is above or below 255. Here, we test for and
3490 * handle locale rules */
18f762c3
KW
3491 }
3492 else {
1b4059d5
KW
3493 if ((flags & FOLDEQ_UTF8_LOCALE)
3494 && (! u1 || UTF8_IS_INVARIANT(*p1)
3495 || UTF8_IS_DOWNGRADEABLE_START(*p1)))
5e64d0fa 3496 {
1b4059d5
KW
3497 /* There is no mixing of code points above and below 255. */
3498 if (u2 && (! UTF8_IS_INVARIANT(*p2)
3499 && ! UTF8_IS_DOWNGRADEABLE_START(*p2)))
3500 {
3501 return 0;
3502 }
5e64d0fa 3503
1b4059d5
KW
3504 /* We handle locale rules by converting, if necessary, the
3505 * code point to a single byte. */
3506 if (! u1 || UTF8_IS_INVARIANT(*p1)) {
3507 *foldbuf1 = *p1;
3508 }
3509 else {
3510 *foldbuf1 = TWO_BYTE_UTF8_TO_UNI(*p1, *(p1 + 1));
3511 }
3512 n1 = 1;
5e64d0fa 3513 }
1b4059d5
KW
3514 else if (isASCII(*p1)) { /* Note, that here won't be
3515 both ASCII and using locale
3516 rules */
3517
3518 /* If trying to mix non- with ASCII, and not supposed to,
3519 * fail */
3520 if ((flags & FOLDEQ_UTF8_NOMIX_ASCII) && ! isASCII(*p2)) {
3521 return 0;
3522 }
3523 n1 = 1;
3524 *foldbuf1 = toLOWER(*p1); /* Folds in the ASCII range are
3525 just lowercased */
5e64d0fa 3526 }
1b4059d5
KW
3527 else if (u1) {
3528 to_utf8_fold(p1, foldbuf1, &n1);
a33c29bc 3529 }
1b4059d5
KW
3530 else { /* Not utf8, convert to it first and then get fold */
3531 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p1)));
3532 to_utf8_fold(natbuf, foldbuf1, &n1);
3533 }
3534 f1 = foldbuf1;
18f762c3 3535 }
48ef279e 3536 }
8b35872c 3537
48ef279e 3538 if (n2 == 0) { /* Same for s2 */
18f762c3
KW
3539 if (flags & FOLDEQ_S2_ALREADY_FOLDED) {
3540 f2 = (U8 *) p2;
3541 n2 = UTF8SKIP(f2);
3542 }
3543 else {
227968da
KW
3544 if ((flags & FOLDEQ_UTF8_LOCALE)
3545 && (! u2 || UTF8_IS_INVARIANT(*p2) || UTF8_IS_DOWNGRADEABLE_START(*p2)))
5e64d0fa 3546 {
227968da
KW
3547 /* Here, the next char in s2 is < 256. We've already
3548 * worked on s1, and if it isn't also < 256, can't match */
3549 if (u1 && (! UTF8_IS_INVARIANT(*p1)
3550 && ! UTF8_IS_DOWNGRADEABLE_START(*p1)))
3551 {
3552 return 0;
3553 }
3554 if (! u2 || UTF8_IS_INVARIANT(*p2)) {
3555 *foldbuf2 = *p2;
3556 }
3557 else {
3558 *foldbuf2 = TWO_BYTE_UTF8_TO_UNI(*p2, *(p2 + 1));
3559 }
3560
3561 /* Use another function to handle locale rules. We've made
3562 * sure that both characters to compare are single bytes */
3563 if (! foldEQ_locale((char *) f1, (char *) foldbuf2, 1)) {
3564 return 0;
3565 }
3566 n1 = n2 = 0;
5e64d0fa 3567 }
227968da
KW
3568 else if (isASCII(*p2)) {
3569 if (flags && ! isASCII(*p1)) {
3570 return 0;
3571 }
3572 n2 = 1;
3573 *foldbuf2 = toLOWER(*p2);
5e64d0fa 3574 }
227968da
KW
3575 else if (u2) {
3576 to_utf8_fold(p2, foldbuf2, &n2);
5001101e 3577 }
227968da
KW
3578 else {
3579 uvuni_to_utf8(natbuf, (UV) NATIVE_TO_UNI(((UV)*p2)));
3580 to_utf8_fold(natbuf, foldbuf2, &n2);
a33c29bc 3581 }
227968da 3582 f2 = foldbuf2;
18f762c3 3583 }
48ef279e 3584 }
8b35872c 3585
5001101e 3586 /* Here f1 and f2 point to the beginning of the strings to compare.
227968da
KW
3587 * These strings are the folds of the next character from each input
3588 * string, stored in utf8. */
5e64d0fa 3589
48ef279e
KW
3590 /* While there is more to look for in both folds, see if they
3591 * continue to match */
3592 while (n1 && n2) {
3593 U8 fold_length = UTF8SKIP(f1);
3594 if (fold_length != UTF8SKIP(f2)
3595 || (fold_length == 1 && *f1 != *f2) /* Short circuit memNE
3596 function call for single
3597 character */
3598 || memNE((char*)f1, (char*)f2, fold_length))
3599 {
e6226b18 3600 return 0; /* mismatch */
48ef279e
KW
3601 }
3602
3603 /* Here, they matched, advance past them */
3604 n1 -= fold_length;
3605 f1 += fold_length;
3606 n2 -= fold_length;
3607 f2 += fold_length;
3608 }
8b35872c 3609
48ef279e
KW
3610 /* When reach the end of any fold, advance the input past it */
3611 if (n1 == 0) {
3612 p1 += u1 ? UTF8SKIP(p1) : 1;
3613 }
3614 if (n2 == 0) {
3615 p2 += u2 ? UTF8SKIP(p2) : 1;
3616 }
8b35872c
KW
3617 } /* End of loop through both strings */
3618
3619 /* A match is defined by each scan that specified an explicit length
3620 * reaching its final goal, and the other not having matched a partial
3621 * character (which can happen when the fold of a character is more than one
3622 * character). */
3623 if (! ((g1 == 0 || p1 == g1) && (g2 == 0 || p2 == g2)) || n1 || n2) {
e6226b18 3624 return 0;
8b35872c
KW
3625 }
3626
3627 /* Successful match. Set output pointers */
3628 if (pe1) {
48ef279e 3629 *pe1 = (char*)p1;
8b35872c
KW
3630 }
3631 if (pe2) {
48ef279e 3632 *pe2 = (char*)p2;
8b35872c 3633 }
e6226b18 3634 return 1;
e6b2e755 3635}
701a277b 3636
a49f32c6
NC
3637/*
3638 * Local variables:
3639 * c-indentation-style: bsd
3640 * c-basic-offset: 4
3641 * indent-tabs-mode: t
3642 * End:
3643 *
37442d52
RGS
3644 * ex: set ts=8 sts=4 sw=4 noet:
3645 */