This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Small tweaks to get the test running (fix @INC and
[perl5.git] / utf8.c
CommitLineData
a0ed51b3
LW
1/* utf8.c
2 *
be3c0a43 3 * Copyright (c) 1998-2002, Larry Wall
a0ed51b3
LW
4 *
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
7 *
8 */
9
10/*
11 * 'What a fix!' said Sam. 'That's the one place in all the lands we've ever
12 * heard of that we don't want to see any closer; and that's the one place
13 * we're trying to get to! And that's just where we can't get, nohow.'
14 *
15 * 'Well do I understand your speech,' he answered in the same language;
16 * 'yet few strangers do so. Why then do you not speak in the Common Tongue,
17 * as is the custom in the West, if you wish to be answered?'
18 *
19 * ...the travellers perceived that the floor was paved with stones of many
20 * hues; branching runes and strange devices intertwined beneath their feet.
21 */
22
23#include "EXTERN.h"
864dbfa3 24#define PERL_IN_UTF8_C
a0ed51b3
LW
25#include "perl.h"
26
ccfc67b7
JH
27/*
28=head1 Unicode Support
a0ed51b3 29
b851fbc1 30=for apidoc A|U8 *|uvuni_to_utf8_flags|U8 *d|UV uv|UV flags
eebe1485
SC
31
32Adds the UTF8 representation of the Unicode codepoint C<uv> to the end
33of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free
34bytes available. The return value is the pointer to the byte after the
9041c2e3 35end of the new character. In other words,
eebe1485 36
b851fbc1
JH
37 d = uvuni_to_utf8_flags(d, uv, flags);
38
39or, in most cases,
40
9041c2e3 41 d = uvuni_to_utf8(d, uv);
eebe1485 42
b851fbc1
JH
43(which is equivalent to)
44
45 d = uvuni_to_utf8_flags(d, uv, 0);
46
eebe1485
SC
47is the recommended Unicode-aware way of saying
48
49 *(d++) = uv;
50
51=cut
52*/
53
dfe13c55 54U8 *
b851fbc1 55Perl_uvuni_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
a0ed51b3 56{
62961d2e 57 if (ckWARN(WARN_UTF8)) {
b851fbc1
JH
58 if (UNICODE_IS_SURROGATE(uv) &&
59 !(flags & UNICODE_ALLOW_SURROGATE))
507b9800 60 Perl_warner(aTHX_ WARN_UTF8, "UTF-16 surrogate 0x%04"UVxf, uv);
b851fbc1
JH
61 else if (
62 ((uv >= 0xFDD0 && uv <= 0xFDEF &&
63 !(flags & UNICODE_ALLOW_FDD0))
64 ||
65 ((uv & 0xFFFF) == 0xFFFE &&
66 !(flags & UNICODE_ALLOW_FFFE))
67 ||
68 ((uv & 0xFFFF) == 0xFFFF &&
69 !(flags & UNICODE_ALLOW_FFFF))) &&
70 /* UNICODE_ALLOW_SUPER includes
71 * FFFEs and FFFFs beyond 0x10FFFF. */
72 ((uv <= PERL_UNICODE_MAX) ||
73 !(flags & UNICODE_ALLOW_SUPER))
74 )
507b9800
JH
75 Perl_warner(aTHX_ WARN_UTF8,
76 "Unicode character 0x%04"UVxf" is illegal", uv);
77 }
c4d5f83a
NIS
78 if (UNI_IS_INVARIANT(uv)) {
79 *d++ = UTF_TO_NATIVE(uv);
a0ed51b3
LW
80 return d;
81 }
2d331972 82#if defined(EBCDIC)
1d72bdf6
NIS
83 else {
84 STRLEN len = UNISKIP(uv);
85 U8 *p = d+len-1;
86 while (p > d) {
87 *p-- = UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
88 uv >>= UTF_ACCUMULATION_SHIFT;
89 }
90 *p = UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
91 return d+len;
92 }
93#else /* Non loop style */
a0ed51b3
LW
94 if (uv < 0x800) {
95 *d++ = (( uv >> 6) | 0xc0);
96 *d++ = (( uv & 0x3f) | 0x80);
97 return d;
98 }
99 if (uv < 0x10000) {
100 *d++ = (( uv >> 12) | 0xe0);
101 *d++ = (((uv >> 6) & 0x3f) | 0x80);
102 *d++ = (( uv & 0x3f) | 0x80);
103 return d;
104 }
105 if (uv < 0x200000) {
106 *d++ = (( uv >> 18) | 0xf0);
107 *d++ = (((uv >> 12) & 0x3f) | 0x80);
108 *d++ = (((uv >> 6) & 0x3f) | 0x80);
109 *d++ = (( uv & 0x3f) | 0x80);
110 return d;
111 }
112 if (uv < 0x4000000) {
113 *d++ = (( uv >> 24) | 0xf8);
114 *d++ = (((uv >> 18) & 0x3f) | 0x80);
115 *d++ = (((uv >> 12) & 0x3f) | 0x80);
116 *d++ = (((uv >> 6) & 0x3f) | 0x80);
117 *d++ = (( uv & 0x3f) | 0x80);
118 return d;
119 }
120 if (uv < 0x80000000) {
121 *d++ = (( uv >> 30) | 0xfc);
122 *d++ = (((uv >> 24) & 0x3f) | 0x80);
123 *d++ = (((uv >> 18) & 0x3f) | 0x80);
124 *d++ = (((uv >> 12) & 0x3f) | 0x80);
125 *d++ = (((uv >> 6) & 0x3f) | 0x80);
126 *d++ = (( uv & 0x3f) | 0x80);
127 return d;
128 }
6b8eaf93 129#ifdef HAS_QUAD
d7578b48 130 if (uv < UTF8_QUAD_MAX)
a0ed51b3
LW
131#endif
132 {
133 *d++ = 0xfe; /* Can't match U+FEFF! */
134 *d++ = (((uv >> 30) & 0x3f) | 0x80);
135 *d++ = (((uv >> 24) & 0x3f) | 0x80);
136 *d++ = (((uv >> 18) & 0x3f) | 0x80);
137 *d++ = (((uv >> 12) & 0x3f) | 0x80);
138 *d++ = (((uv >> 6) & 0x3f) | 0x80);
139 *d++ = (( uv & 0x3f) | 0x80);
140 return d;
141 }
6b8eaf93 142#ifdef HAS_QUAD
a0ed51b3
LW
143 {
144 *d++ = 0xff; /* Can't match U+FFFE! */
3c77ea2b
GS
145 *d++ = 0x80; /* 6 Reserved bits */
146 *d++ = (((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */
147 *d++ = (((uv >> 54) & 0x3f) | 0x80);
148 *d++ = (((uv >> 48) & 0x3f) | 0x80);
149 *d++ = (((uv >> 42) & 0x3f) | 0x80);
a0ed51b3
LW
150 *d++ = (((uv >> 36) & 0x3f) | 0x80);
151 *d++ = (((uv >> 30) & 0x3f) | 0x80);
152 *d++ = (((uv >> 24) & 0x3f) | 0x80);
153 *d++ = (((uv >> 18) & 0x3f) | 0x80);
154 *d++ = (((uv >> 12) & 0x3f) | 0x80);
155 *d++ = (((uv >> 6) & 0x3f) | 0x80);
156 *d++ = (( uv & 0x3f) | 0x80);
157 return d;
158 }
159#endif
1d72bdf6 160#endif /* Loop style */
a0ed51b3 161}
b851fbc1
JH
162
163U8 *
164Perl_uvuni_to_utf8(pTHX_ U8 *d, UV uv)
165{
166 return Perl_uvuni_to_utf8_flags(aTHX_ d, uv, 0);
167}
9041c2e3
NIS
168
169
170/*
eebe1485
SC
171=for apidoc A|STRLEN|is_utf8_char|U8 *s
172
5da9da9e 173Tests if some arbitrary number of bytes begins in a valid UTF-8
1d72bdf6 174character. Note that an INVARIANT (i.e. ASCII) character is a valid UTF-8 character.
5da9da9e
JH
175The actual number of bytes in the UTF-8 character will be returned if
176it is valid, otherwise 0.
9041c2e3 177
37607a96
PK
178=cut
179*/
067a85ef 180STRLEN
386d01d6
GS
181Perl_is_utf8_char(pTHX_ U8 *s)
182{
183 U8 u = *s;
067a85ef
A
184 STRLEN slen, len;
185 UV uv, ouv;
386d01d6 186
1d72bdf6 187 if (UTF8_IS_INVARIANT(u))
386d01d6
GS
188 return 1;
189
60006e79 190 if (!UTF8_IS_START(u))
386d01d6
GS
191 return 0;
192
9f07fdcd 193 len = UTF8SKIP(s);
386d01d6 194
60006e79 195 if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
067a85ef
A
196 return 0;
197
386d01d6
GS
198 slen = len - 1;
199 s++;
6eb6869e 200 u &= UTF_START_MASK(len);
1d72bdf6 201 uv = u;
067a85ef 202 ouv = uv;
386d01d6 203 while (slen--) {
60006e79 204 if (!UTF8_IS_CONTINUATION(*s))
386d01d6 205 return 0;
8850bf83 206 uv = UTF8_ACCUMULATE(uv, *s);
209a9bc1 207 if (uv < ouv)
067a85ef
A
208 return 0;
209 ouv = uv;
386d01d6
GS
210 s++;
211 }
067a85ef 212
5bbb0b5a 213 if (UNISKIP(uv) < len)
067a85ef
A
214 return 0;
215
386d01d6
GS
216 return len;
217}
218
6662521e 219/*
eebe1485 220=for apidoc A|bool|is_utf8_string|U8 *s|STRLEN len
6662521e 221
5da9da9e
JH
222Returns true if first C<len> bytes of the given string form a valid UTF8
223string, false otherwise. Note that 'a valid UTF8 string' does not mean
224'a string that contains UTF8' because a valid ASCII string is a valid
225UTF8 string.
6662521e
GS
226
227=cut
228*/
229
8e84507e 230bool
6662521e
GS
231Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len)
232{
067a85ef 233 U8* x = s;
1aa99e6b 234 U8* send;
067a85ef
A
235 STRLEN c;
236
1aa99e6b 237 if (!len)
6cd5fe39 238 len = strlen((char *)s);
1aa99e6b
IH
239 send = s + len;
240
6662521e
GS
241 while (x < send) {
242 c = is_utf8_char(x);
067a85ef
A
243 if (!c)
244 return FALSE;
6662521e 245 x += c;
6662521e 246 }
60006e79
JH
247 if (x != send)
248 return FALSE;
067a85ef
A
249
250 return TRUE;
6662521e
GS
251}
252
67e989fb 253/*
9041c2e3 254=for apidoc A|UV|utf8n_to_uvuni|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
67e989fb 255
9041c2e3
NIS
256Bottom level UTF-8 decode routine.
257Returns the unicode code point value of the first character in the string C<s>
ba210ebe 258which is assumed to be in UTF8 encoding and no longer than C<curlen>;
7df053ec 259C<retlen> will be set to the length, in bytes, of that character.
67e989fb
JH
260
261If C<s> does not point to a well-formed UTF8 character, the behaviour
dcad2880
JH
262is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
263it is assumed that the caller will raise a warning, and this function
28d3d195
JH
264will silently just set C<retlen> to C<-1> and return zero. If the
265C<flags> does not contain UTF8_CHECK_ONLY, warnings about
266malformations will be given, C<retlen> will be set to the expected
267length of the UTF-8 character in bytes, and zero will be returned.
268
269The C<flags> can also contain various flags to allow deviations from
270the strict UTF-8 encoding (see F<utf8.h>).
67e989fb 271
9041c2e3
NIS
272Most code should use utf8_to_uvchr() rather than call this directly.
273
37607a96
PK
274=cut
275*/
67e989fb 276
a0ed51b3 277UV
37607a96 278Perl_utf8n_to_uvuni(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
a0ed51b3 279{
097fb8e2 280 U8 *s0 = s;
9c5ffd7c 281 UV uv = *s, ouv = 0;
ba210ebe
JH
282 STRLEN len = 1;
283 bool dowarn = ckWARN_d(WARN_UTF8);
4f849cb6 284 UV startbyte = *s;
ba210ebe 285 STRLEN expectlen = 0;
a0dbb045
JH
286 U32 warning = 0;
287
288/* This list is a superset of the UTF8_ALLOW_XXX. */
289
290#define UTF8_WARN_EMPTY 1
291#define UTF8_WARN_CONTINUATION 2
292#define UTF8_WARN_NON_CONTINUATION 3
293#define UTF8_WARN_FE_FF 4
294#define UTF8_WARN_SHORT 5
295#define UTF8_WARN_OVERFLOW 6
296#define UTF8_WARN_SURROGATE 7
297#define UTF8_WARN_BOM 8
298#define UTF8_WARN_LONG 9
299#define UTF8_WARN_FFFF 10
300
301 if (curlen == 0 &&
302 !(flags & UTF8_ALLOW_EMPTY)) {
303 warning = UTF8_WARN_EMPTY;
0c443dc2
JH
304 goto malformed;
305 }
306
1d72bdf6 307 if (UTF8_IS_INVARIANT(uv)) {
a0ed51b3
LW
308 if (retlen)
309 *retlen = 1;
c4d5f83a 310 return (UV) (NATIVE_TO_UTF(*s));
a0ed51b3 311 }
67e989fb 312
421a8bf2 313 if (UTF8_IS_CONTINUATION(uv) &&
fcc8fcf6 314 !(flags & UTF8_ALLOW_CONTINUATION)) {
a0dbb045 315 warning = UTF8_WARN_CONTINUATION;
ba210ebe
JH
316 goto malformed;
317 }
318
421a8bf2 319 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
fcc8fcf6 320 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
a0dbb045 321 warning = UTF8_WARN_NON_CONTINUATION;
ba210ebe
JH
322 goto malformed;
323 }
9041c2e3 324
1d72bdf6 325#ifdef EBCDIC
75383841 326 uv = NATIVE_TO_UTF(uv);
1d72bdf6 327#else
fcc8fcf6
JH
328 if ((uv == 0xfe || uv == 0xff) &&
329 !(flags & UTF8_ALLOW_FE_FF)) {
a0dbb045 330 warning = UTF8_WARN_FE_FF;
ba210ebe 331 goto malformed;
a0ed51b3 332 }
1d72bdf6
NIS
333#endif
334
ba210ebe
JH
335 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; }
336 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; }
337 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; }
338 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; }
1d72bdf6
NIS
339#ifdef EBCDIC
340 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
341 else { len = 7; uv &= 0x01; }
342#else
ba210ebe
JH
343 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; }
344 else if (!(uv & 0x01)) { len = 7; uv = 0; }
1d72bdf6
NIS
345 else { len = 13; uv = 0; } /* whoa! */
346#endif
347
a0ed51b3
LW
348 if (retlen)
349 *retlen = len;
9041c2e3 350
ba210ebe
JH
351 expectlen = len;
352
fcc8fcf6
JH
353 if ((curlen < expectlen) &&
354 !(flags & UTF8_ALLOW_SHORT)) {
a0dbb045 355 warning = UTF8_WARN_SHORT;
ba210ebe
JH
356 goto malformed;
357 }
358
359 len--;
a0ed51b3 360 s++;
ba210ebe
JH
361 ouv = uv;
362
a0ed51b3 363 while (len--) {
421a8bf2
JH
364 if (!UTF8_IS_CONTINUATION(*s) &&
365 !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
a0dbb045
JH
366 s--;
367 warning = UTF8_WARN_NON_CONTINUATION;
ba210ebe 368 goto malformed;
a0ed51b3
LW
369 }
370 else
8850bf83 371 uv = UTF8_ACCUMULATE(uv, *s);
a0dbb045
JH
372 if (!(uv > ouv)) {
373 /* These cannot be allowed. */
374 if (uv == ouv) {
375 if (!(flags & UTF8_ALLOW_LONG)) {
376 warning = UTF8_WARN_LONG;
377 goto malformed;
378 }
379 }
380 else { /* uv < ouv */
381 /* This cannot be allowed. */
382 warning = UTF8_WARN_OVERFLOW;
383 goto malformed;
384 }
ba210ebe
JH
385 }
386 s++;
387 ouv = uv;
388 }
389
421a8bf2 390 if (UNICODE_IS_SURROGATE(uv) &&
fcc8fcf6 391 !(flags & UTF8_ALLOW_SURROGATE)) {
a0dbb045 392 warning = UTF8_WARN_SURROGATE;
ba210ebe 393 goto malformed;
421a8bf2 394 } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) &&
fcc8fcf6 395 !(flags & UTF8_ALLOW_BOM)) {
a0dbb045 396 warning = UTF8_WARN_BOM;
ba210ebe 397 goto malformed;
fcc8fcf6
JH
398 } else if ((expectlen > UNISKIP(uv)) &&
399 !(flags & UTF8_ALLOW_LONG)) {
a0dbb045 400 warning = UTF8_WARN_LONG;
ba210ebe 401 goto malformed;
421a8bf2 402 } else if (UNICODE_IS_ILLEGAL(uv) &&
a9917092 403 !(flags & UTF8_ALLOW_FFFF)) {
a0dbb045 404 warning = UTF8_WARN_FFFF;
a9917092 405 goto malformed;
a0ed51b3 406 }
ba210ebe 407
a0ed51b3 408 return uv;
ba210ebe
JH
409
410malformed:
411
fcc8fcf6 412 if (flags & UTF8_CHECK_ONLY) {
ba210ebe 413 if (retlen)
cc366d4b 414 *retlen = -1;
ba210ebe
JH
415 return 0;
416 }
417
a0dbb045
JH
418 if (dowarn) {
419 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
420
421 switch (warning) {
422 case 0: /* Intentionally empty. */ break;
423 case UTF8_WARN_EMPTY:
424 Perl_sv_catpvf(aTHX_ sv, "(empty string)");
425 break;
426 case UTF8_WARN_CONTINUATION:
097fb8e2 427 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
a0dbb045
JH
428 break;
429 case UTF8_WARN_NON_CONTINUATION:
097fb8e2
JH
430 if (s == s0)
431 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
432 (UV)s[1], startbyte);
433 else
434 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
435 (UV)s[1], s - s0, s - s0 > 1 ? "s" : "", startbyte, expectlen);
436
a0dbb045
JH
437 break;
438 case UTF8_WARN_FE_FF:
439 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
440 break;
441 case UTF8_WARN_SHORT:
097fb8e2
JH
442 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
443 curlen, curlen == 1 ? "" : "s", expectlen, startbyte);
b31f83c2 444 expectlen = curlen; /* distance for caller to skip */
a0dbb045
JH
445 break;
446 case UTF8_WARN_OVERFLOW:
097fb8e2
JH
447 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
448 ouv, *s, startbyte);
a0dbb045
JH
449 break;
450 case UTF8_WARN_SURROGATE:
451 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
452 break;
453 case UTF8_WARN_BOM:
454 Perl_sv_catpvf(aTHX_ sv, "(byte order mark 0x%04"UVxf")", uv);
455 break;
456 case UTF8_WARN_LONG:
097fb8e2
JH
457 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
458 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
a0dbb045
JH
459 break;
460 case UTF8_WARN_FFFF:
461 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
462 break;
463 default:
464 Perl_sv_catpvf(aTHX_ sv, "(unknown reason)");
465 break;
466 }
467
468 if (warning) {
469 char *s = SvPVX(sv);
470
471 if (PL_op)
472 Perl_warner(aTHX_ WARN_UTF8,
53e06cf0 473 "%s in %s", s, OP_DESC(PL_op));
a0dbb045
JH
474 else
475 Perl_warner(aTHX_ WARN_UTF8, "%s", s);
476 }
477 }
478
ba210ebe 479 if (retlen)
28d3d195 480 *retlen = expectlen ? expectlen : len;
ba210ebe 481
28d3d195 482 return 0;
a0ed51b3
LW
483}
484
8e84507e 485/*
37607a96 486=for apidoc A|UV|utf8_to_uvchr|U8 *s|STRLEN *retlen
9041c2e3
NIS
487
488Returns the native character value of the first character in the string C<s>
489which is assumed to be in UTF8 encoding; C<retlen> will be set to the
490length, in bytes, of that character.
491
492If C<s> does not point to a well-formed UTF8 character, zero is
493returned and retlen is set, if possible, to -1.
494
495=cut
496*/
497
498UV
37607a96 499Perl_utf8_to_uvchr(pTHX_ U8 *s, STRLEN *retlen)
9041c2e3
NIS
500{
501 return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXLEN, retlen, 0);
502}
503
504/*
37607a96 505=for apidoc A|UV|utf8_to_uvuni|U8 *s|STRLEN *retlen
9041c2e3
NIS
506
507Returns the Unicode code point of the first character in the string C<s>
508which is assumed to be in UTF8 encoding; C<retlen> will be set to the
509length, in bytes, of that character.
510
511This function should only be used when returned UV is considered
512an index into the Unicode semantic tables (e.g. swashes).
513
ba210ebe
JH
514If C<s> does not point to a well-formed UTF8 character, zero is
515returned and retlen is set, if possible, to -1.
8e84507e
NIS
516
517=cut
518*/
519
520UV
37607a96 521Perl_utf8_to_uvuni(pTHX_ U8 *s, STRLEN *retlen)
8e84507e 522{
9041c2e3
NIS
523 /* Call the low level routine asking for checks */
524 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXLEN, retlen, 0);
8e84507e
NIS
525}
526
b76347f2 527/*
37607a96 528=for apidoc A|STRLEN|utf8_length|U8 *s|U8 *e
b76347f2
JH
529
530Return the length of the UTF-8 char encoded string C<s> in characters.
02eb7b47
JH
531Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end
532up past C<e>, croaks.
b76347f2
JH
533
534=cut
535*/
536
537STRLEN
37607a96 538Perl_utf8_length(pTHX_ U8 *s, U8 *e)
b76347f2
JH
539{
540 STRLEN len = 0;
541
8850bf83
JH
542 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
543 * the bitops (especially ~) can create illegal UTF-8.
544 * In other words: in Perl UTF-8 is not just for Unicode. */
545
b76347f2 546 if (e < s)
02eb7b47 547 Perl_croak(aTHX_ "panic: utf8_length: unexpected end");
b76347f2 548 while (s < e) {
02eb7b47 549 U8 t = UTF8SKIP(s);
b76347f2
JH
550
551 if (e - s < t)
0064a8a9 552 Perl_croak(aTHX_ "panic: utf8_length: unaligned end");
b76347f2
JH
553 s += t;
554 len++;
555 }
556
557 return len;
558}
559
b06226ff 560/*
eebe1485 561=for apidoc A|IV|utf8_distance|U8 *a|U8 *b
b06226ff
JH
562
563Returns the number of UTF8 characters between the UTF-8 pointers C<a>
564and C<b>.
565
566WARNING: use only if you *know* that the pointers point inside the
567same UTF-8 buffer.
568
37607a96
PK
569=cut
570*/
a0ed51b3 571
02eb7b47 572IV
864dbfa3 573Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
a0ed51b3 574{
02eb7b47
JH
575 IV off = 0;
576
8850bf83
JH
577 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g.
578 * the bitops (especially ~) can create illegal UTF-8.
579 * In other words: in Perl UTF-8 is not just for Unicode. */
580
a0ed51b3
LW
581 if (a < b) {
582 while (a < b) {
02eb7b47
JH
583 U8 c = UTF8SKIP(a);
584
585 if (b - a < c)
586 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
587 a += c;
a0ed51b3
LW
588 off--;
589 }
590 }
591 else {
592 while (b < a) {
02eb7b47
JH
593 U8 c = UTF8SKIP(b);
594
595 if (a - b < c)
596 Perl_croak(aTHX_ "panic: utf8_distance: unaligned end");
597 b += c;
a0ed51b3
LW
598 off++;
599 }
600 }
02eb7b47 601
a0ed51b3
LW
602 return off;
603}
604
b06226ff 605/*
37607a96 606=for apidoc A|U8 *|utf8_hop|U8 *s|I32 off
b06226ff 607
8850bf83
JH
608Return the UTF-8 pointer C<s> displaced by C<off> characters, either
609forward or backward.
b06226ff
JH
610
611WARNING: do not use the following unless you *know* C<off> is within
8850bf83
JH
612the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
613on the first byte of character or just after the last byte of a character.
b06226ff 614
37607a96
PK
615=cut
616*/
a0ed51b3
LW
617
618U8 *
864dbfa3 619Perl_utf8_hop(pTHX_ U8 *s, I32 off)
a0ed51b3 620{
8850bf83
JH
621 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
622 * the bitops (especially ~) can create illegal UTF-8.
623 * In other words: in Perl UTF-8 is not just for Unicode. */
624
a0ed51b3
LW
625 if (off >= 0) {
626 while (off--)
627 s += UTF8SKIP(s);
628 }
629 else {
630 while (off++) {
631 s--;
8850bf83
JH
632 while (UTF8_IS_CONTINUATION(*s))
633 s--;
a0ed51b3
LW
634 }
635 }
636 return s;
637}
638
6940069f 639/*
eebe1485 640=for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
6940069f 641
246fae53
MG
642Converts a string C<s> of length C<len> from UTF8 into byte encoding.
643Unlike C<bytes_to_utf8>, this over-writes the original string, and
644updates len to contain the new length.
67e989fb 645Returns zero on failure, setting C<len> to -1.
6940069f
GS
646
647=cut
648*/
649
650U8 *
37607a96 651Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
6940069f 652{
6940069f
GS
653 U8 *send;
654 U8 *d;
dcad2880 655 U8 *save = s;
246fae53
MG
656
657 /* ensure valid UTF8 and chars < 256 before updating string */
dcad2880
JH
658 for (send = s + *len; s < send; ) {
659 U8 c = *s++;
660
1d72bdf6
NIS
661 if (!UTF8_IS_INVARIANT(c) &&
662 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
663 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
dcad2880
JH
664 *len = -1;
665 return 0;
666 }
246fae53 667 }
dcad2880
JH
668
669 d = s = save;
6940069f 670 while (s < send) {
ed646e6e 671 STRLEN ulen;
9041c2e3 672 *d++ = (U8)utf8_to_uvchr(s, &ulen);
ed646e6e 673 s += ulen;
6940069f
GS
674 }
675 *d = '\0';
246fae53 676 *len = d - save;
6940069f
GS
677 return save;
678}
679
680/*
f9a63242
JH
681=for apidoc A|U8 *|bytes_from_utf8|U8 *s|STRLEN *len|bool *is_utf8
682
683Converts a string C<s> of length C<len> from UTF8 into byte encoding.
684Unlike <utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
ef9edfd0
JH
685the newly-created string, and updates C<len> to contain the new
686length. Returns the original string if no conversion occurs, C<len>
687is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
6880 if C<s> is converted or contains all 7bit characters.
f9a63242 689
37607a96
PK
690=cut
691*/
f9a63242
JH
692
693U8 *
37607a96 694Perl_bytes_from_utf8(pTHX_ U8 *s, STRLEN *len, bool *is_utf8)
f9a63242 695{
f9a63242
JH
696 U8 *d;
697 U8 *start = s;
db42d148 698 U8 *send;
f9a63242
JH
699 I32 count = 0;
700
701 if (!*is_utf8)
702 return start;
703
ef9edfd0 704 /* ensure valid UTF8 and chars < 256 before converting string */
f9a63242
JH
705 for (send = s + *len; s < send;) {
706 U8 c = *s++;
1d72bdf6 707 if (!UTF8_IS_INVARIANT(c)) {
db42d148
NIS
708 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
709 (c = *s++) && UTF8_IS_CONTINUATION(c))
710 count++;
711 else
f9a63242 712 return start;
db42d148 713 }
f9a63242
JH
714 }
715
716 *is_utf8 = 0;
717
f9a63242 718 Newz(801, d, (*len) - count + 1, U8);
ef9edfd0 719 s = start; start = d;
f9a63242
JH
720 while (s < send) {
721 U8 c = *s++;
c4d5f83a
NIS
722 if (!UTF8_IS_INVARIANT(c)) {
723 /* Then it is two-byte encoded */
724 c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++);
725 c = ASCII_TO_NATIVE(c);
726 }
727 *d++ = c;
f9a63242
JH
728 }
729 *d = '\0';
730 *len = d - start;
731 return start;
732}
733
734/*
eebe1485 735=for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
6940069f
GS
736
737Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
6662521e
GS
738Returns a pointer to the newly-created string, and sets C<len> to
739reflect the new length.
6940069f 740
497711e7 741=cut
6940069f
GS
742*/
743
744U8*
37607a96 745Perl_bytes_to_utf8(pTHX_ U8 *s, STRLEN *len)
6940069f 746{
6940069f
GS
747 U8 *send;
748 U8 *d;
749 U8 *dst;
6662521e 750 send = s + (*len);
6940069f 751
6662521e 752 Newz(801, d, (*len) * 2 + 1, U8);
6940069f
GS
753 dst = d;
754
755 while (s < send) {
db42d148 756 UV uv = NATIVE_TO_ASCII(*s++);
c4d5f83a
NIS
757 if (UNI_IS_INVARIANT(uv))
758 *d++ = UTF_TO_NATIVE(uv);
6940069f 759 else {
90f44359
JH
760 *d++ = UTF8_EIGHT_BIT_HI(uv);
761 *d++ = UTF8_EIGHT_BIT_LO(uv);
6940069f
GS
762 }
763 }
764 *d = '\0';
6662521e 765 *len = d-dst;
6940069f
GS
766 return dst;
767}
768
a0ed51b3 769/*
dea0fc0b 770 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
a0ed51b3
LW
771 *
772 * Destination must be pre-extended to 3/2 source. Do not use in-place.
773 * We optimize for native, for obvious reasons. */
774
775U8*
dea0fc0b 776Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
a0ed51b3 777{
dea0fc0b
JH
778 U8* pend;
779 U8* dstart = d;
780
781 if (bytelen & 1)
a7867d0a 782 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
dea0fc0b
JH
783
784 pend = p + bytelen;
785
a0ed51b3 786 while (p < pend) {
dea0fc0b
JH
787 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
788 p += 2;
a0ed51b3
LW
789 if (uv < 0x80) {
790 *d++ = uv;
791 continue;
792 }
793 if (uv < 0x800) {
794 *d++ = (( uv >> 6) | 0xc0);
795 *d++ = (( uv & 0x3f) | 0x80);
796 continue;
797 }
798 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */
dea0fc0b
JH
799 UV low = *p++;
800 if (low < 0xdc00 || low >= 0xdfff)
801 Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
a0ed51b3
LW
802 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
803 }
804 if (uv < 0x10000) {
805 *d++ = (( uv >> 12) | 0xe0);
806 *d++ = (((uv >> 6) & 0x3f) | 0x80);
807 *d++ = (( uv & 0x3f) | 0x80);
808 continue;
809 }
810 else {
811 *d++ = (( uv >> 18) | 0xf0);
812 *d++ = (((uv >> 12) & 0x3f) | 0x80);
813 *d++ = (((uv >> 6) & 0x3f) | 0x80);
814 *d++ = (( uv & 0x3f) | 0x80);
815 continue;
816 }
817 }
dea0fc0b 818 *newlen = d - dstart;
a0ed51b3
LW
819 return d;
820}
821
822/* Note: this one is slightly destructive of the source. */
823
824U8*
dea0fc0b 825Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
a0ed51b3
LW
826{
827 U8* s = (U8*)p;
828 U8* send = s + bytelen;
829 while (s < send) {
830 U8 tmp = s[0];
831 s[0] = s[1];
832 s[1] = tmp;
833 s += 2;
834 }
dea0fc0b 835 return utf16_to_utf8(p, d, bytelen, newlen);
a0ed51b3
LW
836}
837
838/* for now these are all defined (inefficiently) in terms of the utf8 versions */
839
840bool
84afefe6 841Perl_is_uni_alnum(pTHX_ UV c)
a0ed51b3 842{
ad391ad9 843 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 844 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
845 return is_utf8_alnum(tmpbuf);
846}
847
848bool
84afefe6 849Perl_is_uni_alnumc(pTHX_ UV c)
b8c5462f 850{
ad391ad9 851 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 852 uvchr_to_utf8(tmpbuf, c);
b8c5462f
JH
853 return is_utf8_alnumc(tmpbuf);
854}
855
856bool
84afefe6 857Perl_is_uni_idfirst(pTHX_ UV c)
a0ed51b3 858{
ad391ad9 859 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 860 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
861 return is_utf8_idfirst(tmpbuf);
862}
863
864bool
84afefe6 865Perl_is_uni_alpha(pTHX_ UV c)
a0ed51b3 866{
ad391ad9 867 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 868 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
869 return is_utf8_alpha(tmpbuf);
870}
871
872bool
84afefe6 873Perl_is_uni_ascii(pTHX_ UV c)
4d61ec05 874{
ad391ad9 875 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 876 uvchr_to_utf8(tmpbuf, c);
4d61ec05
GS
877 return is_utf8_ascii(tmpbuf);
878}
879
880bool
84afefe6 881Perl_is_uni_space(pTHX_ UV c)
a0ed51b3 882{
ad391ad9 883 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 884 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
885 return is_utf8_space(tmpbuf);
886}
887
888bool
84afefe6 889Perl_is_uni_digit(pTHX_ UV c)
a0ed51b3 890{
ad391ad9 891 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 892 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
893 return is_utf8_digit(tmpbuf);
894}
895
896bool
84afefe6 897Perl_is_uni_upper(pTHX_ UV c)
a0ed51b3 898{
ad391ad9 899 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 900 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
901 return is_utf8_upper(tmpbuf);
902}
903
904bool
84afefe6 905Perl_is_uni_lower(pTHX_ UV c)
a0ed51b3 906{
ad391ad9 907 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 908 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
909 return is_utf8_lower(tmpbuf);
910}
911
912bool
84afefe6 913Perl_is_uni_cntrl(pTHX_ UV c)
b8c5462f 914{
ad391ad9 915 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 916 uvchr_to_utf8(tmpbuf, c);
b8c5462f
JH
917 return is_utf8_cntrl(tmpbuf);
918}
919
920bool
84afefe6 921Perl_is_uni_graph(pTHX_ UV c)
b8c5462f 922{
ad391ad9 923 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 924 uvchr_to_utf8(tmpbuf, c);
b8c5462f
JH
925 return is_utf8_graph(tmpbuf);
926}
927
928bool
84afefe6 929Perl_is_uni_print(pTHX_ UV c)
a0ed51b3 930{
ad391ad9 931 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 932 uvchr_to_utf8(tmpbuf, c);
a0ed51b3
LW
933 return is_utf8_print(tmpbuf);
934}
935
b8c5462f 936bool
84afefe6 937Perl_is_uni_punct(pTHX_ UV c)
b8c5462f 938{
ad391ad9 939 U8 tmpbuf[UTF8_MAXLEN+1];
230880c1 940 uvchr_to_utf8(tmpbuf, c);
b8c5462f
JH
941 return is_utf8_punct(tmpbuf);
942}
943
4d61ec05 944bool
84afefe6 945Perl_is_uni_xdigit(pTHX_ UV c)
4d61ec05 946{
e7ae6809 947 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
230880c1 948 uvchr_to_utf8(tmpbuf, c);
4d61ec05
GS
949 return is_utf8_xdigit(tmpbuf);
950}
951
84afefe6
JH
952UV
953Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 954{
e7ae6809 955 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
230880c1 956 uvchr_to_utf8(tmpbuf, c);
a2a2844f 957 return to_utf8_upper(tmpbuf, p, lenp);
a0ed51b3
LW
958}
959
84afefe6
JH
960UV
961Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 962{
e7ae6809 963 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
230880c1 964 uvchr_to_utf8(tmpbuf, c);
a2a2844f 965 return to_utf8_title(tmpbuf, p, lenp);
a0ed51b3
LW
966}
967
84afefe6
JH
968UV
969Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
a0ed51b3 970{
d90824e8 971 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
230880c1 972 uvchr_to_utf8(tmpbuf, c);
a2a2844f 973 return to_utf8_lower(tmpbuf, p, lenp);
a0ed51b3
LW
974}
975
84afefe6
JH
976UV
977Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
978{
d90824e8 979 U8 tmpbuf[UTF8_MAXLEN_FOLD+1];
230880c1 980 uvchr_to_utf8(tmpbuf, c);
84afefe6
JH
981 return to_utf8_fold(tmpbuf, p, lenp);
982}
983
a0ed51b3
LW
984/* for now these all assume no locale info available for Unicode > 255 */
985
986bool
84afefe6 987Perl_is_uni_alnum_lc(pTHX_ UV c)
a0ed51b3
LW
988{
989 return is_uni_alnum(c); /* XXX no locale support yet */
990}
991
992bool
84afefe6 993Perl_is_uni_alnumc_lc(pTHX_ UV c)
b8c5462f
JH
994{
995 return is_uni_alnumc(c); /* XXX no locale support yet */
996}
997
998bool
84afefe6 999Perl_is_uni_idfirst_lc(pTHX_ UV c)
a0ed51b3
LW
1000{
1001 return is_uni_idfirst(c); /* XXX no locale support yet */
1002}
1003
1004bool
84afefe6 1005Perl_is_uni_alpha_lc(pTHX_ UV c)
a0ed51b3
LW
1006{
1007 return is_uni_alpha(c); /* XXX no locale support yet */
1008}
1009
1010bool
84afefe6 1011Perl_is_uni_ascii_lc(pTHX_ UV c)
4d61ec05
GS
1012{
1013 return is_uni_ascii(c); /* XXX no locale support yet */
1014}
1015
1016bool
84afefe6 1017Perl_is_uni_space_lc(pTHX_ UV c)
a0ed51b3
LW
1018{
1019 return is_uni_space(c); /* XXX no locale support yet */
1020}
1021
1022bool
84afefe6 1023Perl_is_uni_digit_lc(pTHX_ UV c)
a0ed51b3
LW
1024{
1025 return is_uni_digit(c); /* XXX no locale support yet */
1026}
1027
1028bool
84afefe6 1029Perl_is_uni_upper_lc(pTHX_ UV c)
a0ed51b3
LW
1030{
1031 return is_uni_upper(c); /* XXX no locale support yet */
1032}
1033
1034bool
84afefe6 1035Perl_is_uni_lower_lc(pTHX_ UV c)
a0ed51b3
LW
1036{
1037 return is_uni_lower(c); /* XXX no locale support yet */
1038}
1039
1040bool
84afefe6 1041Perl_is_uni_cntrl_lc(pTHX_ UV c)
b8c5462f
JH
1042{
1043 return is_uni_cntrl(c); /* XXX no locale support yet */
1044}
1045
1046bool
84afefe6 1047Perl_is_uni_graph_lc(pTHX_ UV c)
b8c5462f
JH
1048{
1049 return is_uni_graph(c); /* XXX no locale support yet */
1050}
1051
1052bool
84afefe6 1053Perl_is_uni_print_lc(pTHX_ UV c)
a0ed51b3
LW
1054{
1055 return is_uni_print(c); /* XXX no locale support yet */
1056}
1057
b8c5462f 1058bool
84afefe6 1059Perl_is_uni_punct_lc(pTHX_ UV c)
b8c5462f
JH
1060{
1061 return is_uni_punct(c); /* XXX no locale support yet */
1062}
1063
4d61ec05 1064bool
84afefe6 1065Perl_is_uni_xdigit_lc(pTHX_ UV c)
4d61ec05
GS
1066{
1067 return is_uni_xdigit(c); /* XXX no locale support yet */
1068}
1069
b7ac61fa
JH
1070U32
1071Perl_to_uni_upper_lc(pTHX_ U32 c)
1072{
ee099d14
JH
1073 /* XXX returns only the first character -- do not use XXX */
1074 /* XXX no locale support yet */
1075 STRLEN len;
1076 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
1077 return (U32)to_uni_upper(c, tmpbuf, &len);
b7ac61fa
JH
1078}
1079
1080U32
1081Perl_to_uni_title_lc(pTHX_ U32 c)
1082{
ee099d14
JH
1083 /* XXX returns only the first character XXX -- do not use XXX */
1084 /* XXX no locale support yet */
1085 STRLEN len;
1086 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
1087 return (U32)to_uni_title(c, tmpbuf, &len);
b7ac61fa
JH
1088}
1089
1090U32
1091Perl_to_uni_lower_lc(pTHX_ U32 c)
1092{
ee099d14
JH
1093 /* XXX returns only the first character -- do not use XXX */
1094 /* XXX no locale support yet */
1095 STRLEN len;
1096 U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
1097 return (U32)to_uni_lower(c, tmpbuf, &len);
b7ac61fa
JH
1098}
1099
a0ed51b3 1100bool
864dbfa3 1101Perl_is_utf8_alnum(pTHX_ U8 *p)
a0ed51b3 1102{
386d01d6
GS
1103 if (!is_utf8_char(p))
1104 return FALSE;
a0ed51b3 1105 if (!PL_utf8_alnum)
289d4f09
ML
1106 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1107 * descendant of isalnum(3), in other words, it doesn't
1108 * contain the '_'. --jhi */
1109 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
3568d838 1110 return swash_fetch(PL_utf8_alnum, p, TRUE);
a0ed51b3
LW
1111/* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1112#ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1113 if (!PL_utf8_alnum)
1114 PL_utf8_alnum = swash_init("utf8", "",
1115 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
3568d838 1116 return swash_fetch(PL_utf8_alnum, p, TRUE);
a0ed51b3
LW
1117#endif
1118}
1119
1120bool
b8c5462f
JH
1121Perl_is_utf8_alnumc(pTHX_ U8 *p)
1122{
386d01d6
GS
1123 if (!is_utf8_char(p))
1124 return FALSE;
b8c5462f
JH
1125 if (!PL_utf8_alnum)
1126 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
3568d838 1127 return swash_fetch(PL_utf8_alnum, p, TRUE);
b8c5462f
JH
1128/* return is_utf8_alpha(p) || is_utf8_digit(p); */
1129#ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */
1130 if (!PL_utf8_alnum)
1131 PL_utf8_alnum = swash_init("utf8", "",
1132 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
3568d838 1133 return swash_fetch(PL_utf8_alnum, p, TRUE);
b8c5462f
JH
1134#endif
1135}
1136
1137bool
864dbfa3 1138Perl_is_utf8_idfirst(pTHX_ U8 *p)
a0ed51b3
LW
1139{
1140 return *p == '_' || is_utf8_alpha(p);
1141}
1142
1143bool
864dbfa3 1144Perl_is_utf8_alpha(pTHX_ U8 *p)
a0ed51b3 1145{
386d01d6
GS
1146 if (!is_utf8_char(p))
1147 return FALSE;
a0ed51b3 1148 if (!PL_utf8_alpha)
e24b16f9 1149 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
3568d838 1150 return swash_fetch(PL_utf8_alpha, p, TRUE);
a0ed51b3
LW
1151}
1152
1153bool
b8c5462f
JH
1154Perl_is_utf8_ascii(pTHX_ U8 *p)
1155{
386d01d6
GS
1156 if (!is_utf8_char(p))
1157 return FALSE;
b8c5462f
JH
1158 if (!PL_utf8_ascii)
1159 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
3568d838 1160 return swash_fetch(PL_utf8_ascii, p, TRUE);
b8c5462f
JH
1161}
1162
1163bool
864dbfa3 1164Perl_is_utf8_space(pTHX_ U8 *p)
a0ed51b3 1165{
386d01d6
GS
1166 if (!is_utf8_char(p))
1167 return FALSE;
a0ed51b3 1168 if (!PL_utf8_space)
3bec3564 1169 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
3568d838 1170 return swash_fetch(PL_utf8_space, p, TRUE);
a0ed51b3
LW
1171}
1172
1173bool
864dbfa3 1174Perl_is_utf8_digit(pTHX_ U8 *p)
a0ed51b3 1175{
386d01d6
GS
1176 if (!is_utf8_char(p))
1177 return FALSE;
a0ed51b3 1178 if (!PL_utf8_digit)
e24b16f9 1179 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
3568d838 1180 return swash_fetch(PL_utf8_digit, p, TRUE);
a0ed51b3
LW
1181}
1182
1183bool
864dbfa3 1184Perl_is_utf8_upper(pTHX_ U8 *p)
a0ed51b3 1185{
386d01d6
GS
1186 if (!is_utf8_char(p))
1187 return FALSE;
a0ed51b3 1188 if (!PL_utf8_upper)
e24b16f9 1189 PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
3568d838 1190 return swash_fetch(PL_utf8_upper, p, TRUE);
a0ed51b3
LW
1191}
1192
1193bool
864dbfa3 1194Perl_is_utf8_lower(pTHX_ U8 *p)
a0ed51b3 1195{
386d01d6
GS
1196 if (!is_utf8_char(p))
1197 return FALSE;
a0ed51b3 1198 if (!PL_utf8_lower)
e24b16f9 1199 PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
3568d838 1200 return swash_fetch(PL_utf8_lower, p, TRUE);
a0ed51b3
LW
1201}
1202
1203bool
b8c5462f
JH
1204Perl_is_utf8_cntrl(pTHX_ U8 *p)
1205{
386d01d6
GS
1206 if (!is_utf8_char(p))
1207 return FALSE;
b8c5462f
JH
1208 if (!PL_utf8_cntrl)
1209 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
3568d838 1210 return swash_fetch(PL_utf8_cntrl, p, TRUE);
b8c5462f
JH
1211}
1212
1213bool
1214Perl_is_utf8_graph(pTHX_ U8 *p)
1215{
386d01d6
GS
1216 if (!is_utf8_char(p))
1217 return FALSE;
b8c5462f
JH
1218 if (!PL_utf8_graph)
1219 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
3568d838 1220 return swash_fetch(PL_utf8_graph, p, TRUE);
b8c5462f
JH
1221}
1222
1223bool
864dbfa3 1224Perl_is_utf8_print(pTHX_ U8 *p)
a0ed51b3 1225{
386d01d6
GS
1226 if (!is_utf8_char(p))
1227 return FALSE;
a0ed51b3 1228 if (!PL_utf8_print)
e24b16f9 1229 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
3568d838 1230 return swash_fetch(PL_utf8_print, p, TRUE);
a0ed51b3
LW
1231}
1232
1233bool
b8c5462f
JH
1234Perl_is_utf8_punct(pTHX_ U8 *p)
1235{
386d01d6
GS
1236 if (!is_utf8_char(p))
1237 return FALSE;
b8c5462f
JH
1238 if (!PL_utf8_punct)
1239 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
3568d838 1240 return swash_fetch(PL_utf8_punct, p, TRUE);
b8c5462f
JH
1241}
1242
1243bool
1244Perl_is_utf8_xdigit(pTHX_ U8 *p)
1245{
386d01d6
GS
1246 if (!is_utf8_char(p))
1247 return FALSE;
b8c5462f
JH
1248 if (!PL_utf8_xdigit)
1249 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
3568d838 1250 return swash_fetch(PL_utf8_xdigit, p, TRUE);
b8c5462f
JH
1251}
1252
1253bool
864dbfa3 1254Perl_is_utf8_mark(pTHX_ U8 *p)
a0ed51b3 1255{
386d01d6
GS
1256 if (!is_utf8_char(p))
1257 return FALSE;
a0ed51b3 1258 if (!PL_utf8_mark)
e24b16f9 1259 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
3568d838 1260 return swash_fetch(PL_utf8_mark, p, TRUE);
a0ed51b3
LW
1261}
1262
6b5c0936
JH
1263/*
1264=for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special
1265
1266The "p" contains the pointer to the UTF-8 string encoding
1267the character that is being converted.
1268
1269The "ustrp" is a pointer to the character buffer to put the
1270conversion result to. The "lenp" is a pointer to the length
1271of the result.
1272
1273The "swash" is a pointer to the swash to use.
1274
1275The "normal" is a string like "ToLower" which means the swash
1276$utf8::ToLower, which is stored in lib/unicore/To/Lower.pl,
1277and loaded by SWASHGET, using lib/utf8_heavy.pl.
1278
1279The "special" is a string like "utf8::ToSpecLower", which means
d1be9408 1280the hash %utf8::ToSpecLower, which is stored in the same file,
6b5c0936
JH
1281lib/unicore/To/Lower.pl, and also loaded by SWASHGET. The access
1282to the hash is by Perl_to_utf8_case().
1283
1284=cut
1285 */
1286
2104c8d9 1287UV
a6872d42 1288Perl_to_utf8_case(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp, char *normal, char *special)
a0ed51b3 1289{
1feea2c7
JH
1290 UV uv0, uv1, uv2;
1291 U8 tmpbuf[UTF8_MAXLEN_FOLD+1];
1292 STRLEN len;
a0ed51b3 1293
983ffd37
JH
1294 if (!*swashp)
1295 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1feea2c7
JH
1296 uv0 = utf8_to_uvchr(p, 0);
1297 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1298 * are necessary in EBCDIC, they are redundant no-ops
1299 * in ASCII-ish platforms, and hopefully optimized away. */
1300 uv1 = NATIVE_TO_UNI(uv0);
1301 uvuni_to_utf8(tmpbuf, uv1);
1302 uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1303 if (uv2) {
1304 /* It was "normal" (single character mapping). */
1305 UV uv3 = UNI_TO_NATIVE(uv2);
1306
1307 len = uvchr_to_utf8(ustrp, uv3) - ustrp;
1308 if (lenp)
1309 *lenp = len;
1310
1311 return uv3;
1312 }
1313 else {
983ffd37
JH
1314 HV *hv;
1315 SV *keysv;
1316 HE *he;
1317
983ffd37 1318 if ((hv = get_hv(special, FALSE)) &&
1feea2c7 1319 (keysv = sv_2mortal(Perl_newSVpvf(aTHX_ "%04"UVXf, uv1))) &&
983ffd37
JH
1320 (he = hv_fetch_ent(hv, keysv, FALSE, 0))) {
1321 SV *val = HeVAL(he);
d2dcd0fb 1322 char *s = SvPV(val, len);
a6872d42 1323
d2dcd0fb
JH
1324 if (len > 1) {
1325 Copy(s, ustrp, len, U8);
a241e37b 1326#ifdef EBCDIC
d2dcd0fb
JH
1327 {
1328 /* If we have EBCDIC we need to remap the
1329 * characters coming in from the "special"
1330 * (usually, but not always multicharacter)
1331 * mapping, since any characters in the low 256
1332 * are in Unicode code points, not EBCDIC.
d2dcd0fb 1333 * --jhi */
d2dcd0fb
JH
1334 U8 *d = tmpbuf;
1335 U8 *t, *tend;
d2dcd0fb 1336
8c19e1b5
JH
1337 if (SvUTF8(val)) {
1338 STRLEN tlen = 0;
1339
1340 for (t = ustrp, tend = t + len;
1341 t < tend; t += tlen) {
1342 UV c = utf8_to_uvchr(t, &tlen);
1343
1344 if (tlen > 0)
1345 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1346 else
1347 break;
1348 }
1349 } else {
1350 for (t = ustrp, tend = t + len;
1351 t < tend; t++)
1352 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
d2dcd0fb
JH
1353 }
1354 len = d - tmpbuf;
1355 Copy(tmpbuf, ustrp, len, U8);
a241e37b 1356 }
a241e37b 1357#endif
d2dcd0fb 1358 }
ddc50ff6
JH
1359 else
1360 len = uvchr_to_utf8(ustrp, UNI_TO_NATIVE(*(U8*)s)) - ustrp;
d2dcd0fb
JH
1361 if (lenp)
1362 *lenp = len;
1feea2c7 1363
d3e79532 1364 return utf8_to_uvchr(ustrp, 0);
983ffd37 1365 }
1feea2c7
JH
1366
1367 /* So it was not "special": just copy it. */
1368 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1369 if (lenp)
1370 *lenp = len;
1371
1372 return uv0;
983ffd37 1373 }
a0ed51b3
LW
1374}
1375
d3e79532
JH
1376/*
1377=for apidoc A|UV|to_utf8_upper|U8 *p|U8 *ustrp|STRLEN *lenp
1378
1379Convert the UTF-8 encoded character at p to its uppercase version and
1380store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1381that the ustrp needs to be at least UTF8_MAXLEN_UCLC+1 bytes since the
1382uppercase version may be longer than the original character (up to two
1383characters).
1384
1385The first character of the uppercased version is returned
1386(but note, as explained above, that there may be more.)
1387
1388=cut */
1389
2104c8d9 1390UV
983ffd37 1391Perl_to_utf8_upper(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
a0ed51b3 1392{
983ffd37 1393 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
b4e400f9 1394 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
983ffd37 1395}
a0ed51b3 1396
d3e79532
JH
1397/*
1398=for apidoc A|UV|to_utf8_title|U8 *p|U8 *ustrp|STRLEN *lenp
1399
1400Convert the UTF-8 encoded character at p to its titlecase version and
1401store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1402that the ustrp needs to be at least UTF8_MAXLEN_UCLC+1 bytes since the
1403titlecase version may be longer than the original character (up to two
1404characters).
1405
1406The first character of the titlecased version is returned
1407(but note, as explained above, that there may be more.)
1408
1409=cut */
1410
983ffd37
JH
1411UV
1412Perl_to_utf8_title(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1413{
1414 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
b4e400f9 1415 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
a0ed51b3
LW
1416}
1417
d3e79532
JH
1418/*
1419=for apidoc A|UV|to_utf8_lower|U8 *p|U8 *ustrp|STRLEN *lenp
1420
1421Convert the UTF-8 encoded character at p to its lowercase version and
1422store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1423that the ustrp needs to be at least UTF8_MAXLEN_UCLC+1 bytes since the
1424lowercase version may be longer than the original character (up to two
1425characters).
1426
1427The first character of the lowercased version is returned
1428(but note, as explained above, that there may be more.)
1429
1430=cut */
1431
2104c8d9 1432UV
a2a2844f 1433Perl_to_utf8_lower(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
a0ed51b3 1434{
983ffd37 1435 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
b4e400f9
JH
1436 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1437}
1438
d3e79532
JH
1439/*
1440=for apidoc A|UV|to_utf8_fold|U8 *p|U8 *ustrp|STRLEN *lenp
1441
1442Convert the UTF-8 encoded character at p to its foldcase version and
1443store that in UTF-8 in ustrp and its length in bytes in lenp. Note
1444that the ustrp needs to be at least UTF8_MAXLEN_FOLD+1 bytes since the
1445foldcase version may be longer than the original character (up to
1446three characters).
1447
1448The first character of the foldcased version is returned
1449(but note, as explained above, that there may be more.)
1450
1451=cut */
1452
b4e400f9
JH
1453UV
1454Perl_to_utf8_fold(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1455{
1456 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1457 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
a0ed51b3
LW
1458}
1459
1460/* a "swash" is a swatch hash */
1461
1462SV*
864dbfa3 1463Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
a0ed51b3
LW
1464{
1465 SV* retval;
bf1fed83 1466 SV* tokenbufsv = sv_2mortal(NEWSV(0,0));
8e84507e 1467 dSP;
1b026014 1468 HV *stash = gv_stashpvn(pkg, strlen(pkg), FALSE);
f8be5cf0 1469 SV* errsv_save;
ce3b816e 1470
1b026014 1471 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */
ce3b816e 1472 ENTER;
f8be5cf0 1473 errsv_save = newSVsv(ERRSV);
ce3b816e 1474 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
f8be5cf0
JH
1475 if (!SvTRUE(ERRSV))
1476 sv_setsv(ERRSV, errsv_save);
1477 SvREFCNT_dec(errsv_save);
ce3b816e
GS
1478 LEAVE;
1479 }
1480 SPAGAIN;
a0ed51b3
LW
1481 PUSHSTACKi(PERLSI_MAGIC);
1482 PUSHMARK(SP);
1483 EXTEND(SP,5);
1484 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1485 PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1486 PUSHs(listsv);
1487 PUSHs(sv_2mortal(newSViv(minbits)));
1488 PUSHs(sv_2mortal(newSViv(none)));
1489 PUTBACK;
1490 ENTER;
1491 SAVEI32(PL_hints);
1492 PL_hints = 0;
1493 save_re_context();
bf1fed83
JH
1494 if (PL_curcop == &PL_compiling)
1495 /* XXX ought to be handled by lex_start */
1496 sv_setpv(tokenbufsv, PL_tokenbuf);
f8be5cf0 1497 errsv_save = newSVsv(ERRSV);
864dbfa3 1498 if (call_method("SWASHNEW", G_SCALAR))
8e84507e 1499 retval = newSVsv(*PL_stack_sp--);
a0ed51b3 1500 else
e24b16f9 1501 retval = &PL_sv_undef;
f8be5cf0
JH
1502 if (!SvTRUE(ERRSV))
1503 sv_setsv(ERRSV, errsv_save);
1504 SvREFCNT_dec(errsv_save);
a0ed51b3
LW
1505 LEAVE;
1506 POPSTACK;
e24b16f9 1507 if (PL_curcop == &PL_compiling) {
bf1fed83
JH
1508 STRLEN len;
1509 char* pv = SvPV(tokenbufsv, len);
1510
1511 Copy(pv, PL_tokenbuf, len+1, char);
e24b16f9 1512 PL_curcop->op_private = PL_hints;
a0ed51b3
LW
1513 }
1514 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV)
cea2e8a9 1515 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
a0ed51b3
LW
1516 return retval;
1517}
1518
035d37be
JH
1519
1520/* This API is wrong for special case conversions since we may need to
1521 * return several Unicode characters for a single Unicode character
1522 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1523 * the lower-level routine, and it is similarly broken for returning
1524 * multiple values. --jhi */
a0ed51b3 1525UV
3568d838 1526Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr, bool do_utf8)
a0ed51b3
LW
1527{
1528 HV* hv = (HV*)SvRV(sv);
3568d838
JH
1529 U32 klen;
1530 U32 off;
a0ed51b3 1531 STRLEN slen;
7d85a32c 1532 STRLEN needents;
4ea42e7f 1533 U8 *tmps = NULL;
a0ed51b3
LW
1534 U32 bit;
1535 SV *retval;
3568d838
JH
1536 U8 tmputf8[2];
1537 UV c = NATIVE_TO_ASCII(*ptr);
1538
1539 if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1540 tmputf8[0] = UTF8_EIGHT_BIT_HI(c);
1541 tmputf8[1] = UTF8_EIGHT_BIT_LO(c);
1542 ptr = tmputf8;
1543 }
1544 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
1545 * then the "swatch" is a vec() for al the chars which start
1546 * with 0xAA..0xYY
1547 * So the key in the hash (klen) is length of encoded char -1
1548 */
1549 klen = UTF8SKIP(ptr) - 1;
1550 off = ptr[klen];
a0ed51b3 1551
7d85a32c
JH
1552 if (klen == 0)
1553 {
1554 /* If char in invariant then swatch is for all the invariant chars
1555 * In both UTF-8 and UTF8-MOD that happens to be UTF_CONTINUATION_MARK
1556 */
1557 needents = UTF_CONTINUATION_MARK;
1558 off = NATIVE_TO_UTF(ptr[klen]);
1559 }
1560 else
1561 {
1562 /* If char is encoded then swatch is for the prefix */
1563 needents = (1 << UTF_ACCUMULATION_SHIFT);
1564 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
1565 }
1566
a0ed51b3
LW
1567 /*
1568 * This single-entry cache saves about 1/3 of the utf8 overhead in test
1569 * suite. (That is, only 7-8% overall over just a hash cache. Still,
1570 * it's nothing to sniff at.) Pity we usually come through at least
1571 * two function calls to get here...
1572 *
1573 * NB: this code assumes that swatches are never modified, once generated!
1574 */
1575
3568d838 1576 if (hv == PL_last_swash_hv &&
a0ed51b3 1577 klen == PL_last_swash_klen &&
3568d838 1578 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
a0ed51b3
LW
1579 {
1580 tmps = PL_last_swash_tmps;
1581 slen = PL_last_swash_slen;
1582 }
1583 else {
1584 /* Try our second-level swatch cache, kept in a hash. */
dfe13c55 1585 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
a0ed51b3
LW
1586
1587 /* If not cached, generate it via utf8::SWASHGET */
dfe13c55 1588 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
a0ed51b3 1589 dSP;
2b9d42f0
NIS
1590 /* We use utf8n_to_uvuni() as we want an index into
1591 Unicode tables, not a native character number.
1592 */
1593 UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXLEN, NULL, 0);
f8be5cf0 1594 SV *errsv_save;
a0ed51b3
LW
1595 ENTER;
1596 SAVETMPS;
1597 save_re_context();
1598 PUSHSTACKi(PERLSI_MAGIC);
1599 PUSHMARK(SP);
1600 EXTEND(SP,3);
1601 PUSHs((SV*)sv);
ffbc6a93 1602 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
3568d838
JH
1603 PUSHs(sv_2mortal(newSViv((klen) ?
1604 (code_point & ~(needents - 1)) : 0)));
a0ed51b3
LW
1605 PUSHs(sv_2mortal(newSViv(needents)));
1606 PUTBACK;
f8be5cf0 1607 errsv_save = newSVsv(ERRSV);
864dbfa3 1608 if (call_method("SWASHGET", G_SCALAR))
8e84507e 1609 retval = newSVsv(*PL_stack_sp--);
a0ed51b3 1610 else
e24b16f9 1611 retval = &PL_sv_undef;
f8be5cf0
JH
1612 if (!SvTRUE(ERRSV))
1613 sv_setsv(ERRSV, errsv_save);
1614 SvREFCNT_dec(errsv_save);
a0ed51b3
LW
1615 POPSTACK;
1616 FREETMPS;
1617 LEAVE;
e24b16f9
GS
1618 if (PL_curcop == &PL_compiling)
1619 PL_curcop->op_private = PL_hints;
a0ed51b3 1620
dfe13c55 1621 svp = hv_store(hv, (char*)ptr, klen, retval, 0);
a0ed51b3 1622
7d85a32c 1623 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || (slen << 3) < needents)
cea2e8a9 1624 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
a0ed51b3
LW
1625 }
1626
1627 PL_last_swash_hv = hv;
1628 PL_last_swash_klen = klen;
1629 PL_last_swash_tmps = tmps;
1630 PL_last_swash_slen = slen;
1631 if (klen)
1632 Copy(ptr, PL_last_swash_key, klen, U8);
1633 }
1634
9faf8d75 1635 switch ((int)((slen << 3) / needents)) {
a0ed51b3
LW
1636 case 1:
1637 bit = 1 << (off & 7);
1638 off >>= 3;
1639 return (tmps[off] & bit) != 0;
1640 case 8:
1641 return tmps[off];
1642 case 16:
1643 off <<= 1;
1644 return (tmps[off] << 8) + tmps[off + 1] ;
1645 case 32:
1646 off <<= 2;
1647 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1648 }
cea2e8a9 1649 Perl_croak(aTHX_ "panic: swash_fetch");
a0ed51b3
LW
1650 return 0;
1651}
2b9d42f0
NIS
1652
1653
1654/*
37607a96 1655=for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv
2b9d42f0
NIS
1656
1657Adds the UTF8 representation of the Native codepoint C<uv> to the end
1658of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free
1659bytes available. The return value is the pointer to the byte after the
1660end of the new character. In other words,
1661
1662 d = uvchr_to_utf8(d, uv);
1663
1664is the recommended wide native character-aware way of saying
1665
1666 *(d++) = uv;
1667
1668=cut
1669*/
1670
1671/* On ASCII machines this is normally a macro but we want a
1672 real function in case XS code wants it
1673*/
1674#undef Perl_uvchr_to_utf8
1675U8 *
1676Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
1677{
b851fbc1 1678 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
2b9d42f0
NIS
1679}
1680
b851fbc1
JH
1681U8 *
1682Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
1683{
1684 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
1685}
2b9d42f0
NIS
1686
1687/*
37607a96 1688=for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
2b9d42f0
NIS
1689
1690Returns the native character value of the first character in the string C<s>
1691which is assumed to be in UTF8 encoding; C<retlen> will be set to the
1692length, in bytes, of that character.
1693
1694Allows length and flags to be passed to low level routine.
1695
1696=cut
1697*/
0a2ef054
JH
1698/* On ASCII machines this is normally a macro but we want
1699 a real function in case XS code wants it
2b9d42f0
NIS
1700*/
1701#undef Perl_utf8n_to_uvchr
1702UV
37607a96 1703Perl_utf8n_to_uvchr(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
2b9d42f0
NIS
1704{
1705 UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
1706 return UNI_TO_NATIVE(uv);
1707}
1708
d2cc3551
JH
1709/*
1710=for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags
1711
1712Build to the scalar dsv a displayable version of the string spv,
1713length len, the displayable version being at most pvlim bytes long
1714(if longer, the rest is truncated and "..." will be appended).
0a2ef054 1715
9e55ce06 1716The flags argument can have UNI_DISPLAY_ISPRINT set to display
00e86452 1717isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
0a2ef054
JH
1718to display the \\[nrfta\\] as the backslashed versions (like '\n')
1719(UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
1720UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
1721UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
1722
d2cc3551
JH
1723The pointer to the PV of the dsv is returned.
1724
1725=cut */
e6b2e755
JH
1726char *
1727Perl_pv_uni_display(pTHX_ SV *dsv, U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
1728{
1729 int truncated = 0;
1730 char *s, *e;
1731
1732 sv_setpvn(dsv, "", 0);
1733 for (s = (char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
1734 UV u;
c728cb41
JH
1735 bool ok = FALSE;
1736
e6b2e755
JH
1737 if (pvlim && SvCUR(dsv) >= pvlim) {
1738 truncated++;
1739 break;
1740 }
1741 u = utf8_to_uvchr((U8*)s, 0);
c728cb41 1742 if (u < 256) {
c728cb41
JH
1743 if (!ok && (flags & UNI_DISPLAY_BACKSLASH)) {
1744 switch (u & 0xFF) {
1745 case '\n':
1746 Perl_sv_catpvf(aTHX_ dsv, "\\n"); ok = TRUE; break;
1747 case '\r':
1748 Perl_sv_catpvf(aTHX_ dsv, "\\r"); ok = TRUE; break;
1749 case '\t':
1750 Perl_sv_catpvf(aTHX_ dsv, "\\t"); ok = TRUE; break;
1751 case '\f':
1752 Perl_sv_catpvf(aTHX_ dsv, "\\f"); ok = TRUE; break;
1753 case '\a':
1754 Perl_sv_catpvf(aTHX_ dsv, "\\a"); ok = TRUE; break;
1755 case '\\':
1756 Perl_sv_catpvf(aTHX_ dsv, "\\" ); ok = TRUE; break;
1757 default: break;
1758 }
1759 }
00e86452
JH
1760 /* isPRINT() is the locale-blind version. */
1761 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(u & 0xFF)) {
2c4547fe 1762 Perl_sv_catpvf(aTHX_ dsv, "%c", (char)(u & 0xFF));
0a2ef054
JH
1763 ok = TRUE;
1764 }
c728cb41
JH
1765 }
1766 if (!ok)
9e55ce06 1767 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
e6b2e755
JH
1768 }
1769 if (truncated)
1770 sv_catpvn(dsv, "...", 3);
1771
1772 return SvPVX(dsv);
1773}
2b9d42f0 1774
d2cc3551
JH
1775/*
1776=for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags
1777
1778Build to the scalar dsv a displayable version of the scalar sv,
0a2ef054 1779the displayable version being at most pvlim bytes long
d2cc3551 1780(if longer, the rest is truncated and "..." will be appended).
0a2ef054
JH
1781
1782The flags argument is as in pv_uni_display().
1783
d2cc3551
JH
1784The pointer to the PV of the dsv is returned.
1785
1786=cut */
e6b2e755
JH
1787char *
1788Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
1789{
701a277b
JH
1790 return Perl_pv_uni_display(aTHX_ dsv, (U8*)SvPVX(ssv), SvCUR(ssv),
1791 pvlim, flags);
1792}
1793
d2cc3551 1794/*
d07ddd77 1795=for apidoc A|I32|ibcmp_utf8|const char *s1|char **pe1|register UV l1|bool u1|const char *s2|char **pe2|register UV l2|bool u2
d2cc3551
JH
1796
1797Return true if the strings s1 and s2 differ case-insensitively, false
1798if not (if they are equal case-insensitively). If u1 is true, the
1799string s1 is assumed to be in UTF-8-encoded Unicode. If u2 is true,
d07ddd77
JH
1800the string s2 is assumed to be in UTF-8-encoded Unicode. If u1 or u2
1801are false, the respective string is assumed to be in native 8-bit
1802encoding.
1803
1804If the pe1 and pe2 are non-NULL, the scanning pointers will be copied
1805in there (they will point at the beginning of the I<next> character).
1806If the pointers behind pe1 or pe2 are non-NULL, they are the end
1807pointers beyond which scanning will not continue under any
1808circustances. If the byte lengths l1 and l2 are non-zero, s1+l1 and
1809s2+l2 will be used as goal end pointers that will also stop the scan,
1810and which qualify towards defining a successful match: all the scans
1811that define an explicit length must reach their goal pointers for
1812a match to succeed).
d2cc3551
JH
1813
1814For case-insensitiveness, the "casefolding" of Unicode is used
1815instead of upper/lowercasing both the characters, see
1816http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
1817
1818=cut */
701a277b 1819I32
d07ddd77 1820Perl_ibcmp_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
332ddc25 1821{
5469e704
JH
1822 register U8 *p1 = (U8*)s1;
1823 register U8 *p2 = (U8*)s2;
d07ddd77
JH
1824 register U8 *e1 = 0, *f1 = 0, *q1 = 0;
1825 register U8 *e2 = 0, *f2 = 0, *q2 = 0;
1826 STRLEN n1 = 0, n2 = 0;
ffce6cc2
JH
1827 U8 foldbuf1[UTF8_MAXLEN_FOLD+1];
1828 U8 foldbuf2[UTF8_MAXLEN_FOLD+1];
d7f013c8
JH
1829 U8 natbuf[1+1];
1830 STRLEN foldlen1, foldlen2;
d07ddd77 1831 bool match;
332ddc25 1832
d07ddd77
JH
1833 if (pe1)
1834 e1 = *(U8**)pe1;
1835 if (e1 == 0 || (l1 && l1 < e1 - (U8*)s1))
1836 f1 = (U8*)s1 + l1;
1837 if (pe2)
1838 e2 = *(U8**)pe2;
1839 if (e2 == 0 || (l2 && l2 < e2 - (U8*)s2))
1840 f2 = (U8*)s2 + l2;
1841
1842 if ((e1 == 0 && f1 == 0) || (e2 == 0 && f2 == 0) || (f1 == 0 && f2 == 0))
1843 return 1; /* mismatch; possible infinite loop or false positive */
1844
a6872d42
JH
1845 if (!u1 || !u2)
1846 natbuf[1] = 0; /* Need to terminate the buffer. */
1847
d07ddd77
JH
1848 while ((e1 == 0 || p1 < e1) &&
1849 (f1 == 0 || p1 < f1) &&
1850 (e2 == 0 || p2 < e2) &&
1851 (f2 == 0 || p2 < f2)) {
1852 if (n1 == 0) {
d7f013c8
JH
1853 if (u1)
1854 to_utf8_fold(p1, foldbuf1, &foldlen1);
1855 else {
f5cee151 1856 natbuf[0] = *p1;
d7f013c8
JH
1857 to_utf8_fold(natbuf, foldbuf1, &foldlen1);
1858 }
1859 q1 = foldbuf1;
d07ddd77 1860 n1 = foldlen1;
332ddc25 1861 }
d07ddd77 1862 if (n2 == 0) {
d7f013c8
JH
1863 if (u2)
1864 to_utf8_fold(p2, foldbuf2, &foldlen2);
1865 else {
f5cee151 1866 natbuf[0] = *p2;
d7f013c8
JH
1867 to_utf8_fold(natbuf, foldbuf2, &foldlen2);
1868 }
1869 q2 = foldbuf2;
d07ddd77 1870 n2 = foldlen2;
332ddc25 1871 }
d07ddd77
JH
1872 while (n1 && n2) {
1873 if ( UTF8SKIP(q1) != UTF8SKIP(q2) ||
1874 (UTF8SKIP(q1) == 1 && *q1 != *q2) ||
1875 memNE((char*)q1, (char*)q2, UTF8SKIP(q1)) )
d7f013c8 1876 return 1; /* mismatch */
d07ddd77 1877 n1 -= UTF8SKIP(q1);
d7f013c8 1878 q1 += UTF8SKIP(q1);
d07ddd77 1879 n2 -= UTF8SKIP(q2);
d7f013c8 1880 q2 += UTF8SKIP(q2);
701a277b 1881 }
d07ddd77 1882 if (n1 == 0)
d7f013c8 1883 p1 += u1 ? UTF8SKIP(p1) : 1;
d07ddd77 1884 if (n2 == 0)
d7f013c8
JH
1885 p2 += u2 ? UTF8SKIP(p2) : 1;
1886
d2cc3551 1887 }
5469e704 1888
d07ddd77
JH
1889 /* A match is defined by all the scans that specified
1890 * an explicit length reaching their final goals. */
1891 match = (f1 == 0 || p1 == f1) && (f2 == 0 || p2 == f2);
5469e704
JH
1892
1893 if (match) {
d07ddd77
JH
1894 if (pe1)
1895 *pe1 = (char*)p1;
1896 if (pe2)
1897 *pe2 = (char*)p2;
5469e704
JH
1898 }
1899
1900 return match ? 0 : 1; /* 0 match, 1 mismatch */
e6b2e755 1901}
701a277b 1902