xref: /openbsd-src/gnu/usr.bin/perl/utf8.c (revision db3296cf5c1dd9058ceecc3a29fe4aaa0bd26000)
1 /*    utf8.c
2  *
3  *    Copyright (c) 1998-2002, Larry Wall
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"
24 #define PERL_IN_UTF8_C
25 #include "perl.h"
26 
27 static char unees[] = "Malformed UTF-8 character (unexpected end of string)";
28 
29 /*
30 =head1 Unicode Support
31 
32 =for apidoc A|U8 *|uvuni_to_utf8_flags|U8 *d|UV uv|UV flags
33 
34 Adds the UTF8 representation of the Unicode codepoint C<uv> to the end
35 of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free
36 bytes available. The return value is the pointer to the byte after the
37 end of the new character. In other words,
38 
39     d = uvuni_to_utf8_flags(d, uv, flags);
40 
41 or, in most cases,
42 
43     d = uvuni_to_utf8(d, uv);
44 
45 (which is equivalent to)
46 
47     d = uvuni_to_utf8_flags(d, uv, 0);
48 
49 is the recommended Unicode-aware way of saying
50 
51     *(d++) = uv;
52 
53 =cut
54 */
55 
56 U8 *
57 Perl_uvuni_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
58 {
59     if (ckWARN(WARN_UTF8)) {
60 	 if (UNICODE_IS_SURROGATE(uv) &&
61 	     !(flags & UNICODE_ALLOW_SURROGATE))
62 	      Perl_warner(aTHX_ packWARN(WARN_UTF8), "UTF-16 surrogate 0x%04"UVxf, uv);
63 	 else if (
64 		  ((uv >= 0xFDD0 && uv <= 0xFDEF &&
65 		    !(flags & UNICODE_ALLOW_FDD0))
66 		   ||
67 		   ((uv & 0xFFFE) == 0xFFFE && /* Either FFFE or FFFF. */
68 		    !(flags & UNICODE_ALLOW_FFFF))) &&
69 		  /* UNICODE_ALLOW_SUPER includes
70 		   * FFFEs and FFFFs beyond 0x10FFFF. */
71 		  ((uv <= PERL_UNICODE_MAX) ||
72 		   !(flags & UNICODE_ALLOW_SUPER))
73 		  )
74 	      Perl_warner(aTHX_ packWARN(WARN_UTF8),
75 			 "Unicode character 0x%04"UVxf" is illegal", uv);
76     }
77     if (UNI_IS_INVARIANT(uv)) {
78 	*d++ = (U8)UTF_TO_NATIVE(uv);
79 	return d;
80     }
81 #if defined(EBCDIC)
82     else {
83 	STRLEN len  = UNISKIP(uv);
84 	U8 *p = d+len-1;
85 	while (p > d) {
86 	    *p-- = (U8)UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK);
87 	    uv >>= UTF_ACCUMULATION_SHIFT;
88 	}
89 	*p = (U8)UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len));
90 	return d+len;
91     }
92 #else /* Non loop style */
93     if (uv < 0x800) {
94 	*d++ = (U8)(( uv >>  6)         | 0xc0);
95 	*d++ = (U8)(( uv        & 0x3f) | 0x80);
96 	return d;
97     }
98     if (uv < 0x10000) {
99 	*d++ = (U8)(( uv >> 12)         | 0xe0);
100 	*d++ = (U8)(((uv >>  6) & 0x3f) | 0x80);
101 	*d++ = (U8)(( uv        & 0x3f) | 0x80);
102 	return d;
103     }
104     if (uv < 0x200000) {
105 	*d++ = (U8)(( uv >> 18)         | 0xf0);
106 	*d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
107 	*d++ = (U8)(((uv >>  6) & 0x3f) | 0x80);
108 	*d++ = (U8)(( uv        & 0x3f) | 0x80);
109 	return d;
110     }
111     if (uv < 0x4000000) {
112 	*d++ = (U8)(( uv >> 24)         | 0xf8);
113 	*d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
114 	*d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
115 	*d++ = (U8)(((uv >>  6) & 0x3f) | 0x80);
116 	*d++ = (U8)(( uv        & 0x3f) | 0x80);
117 	return d;
118     }
119     if (uv < 0x80000000) {
120 	*d++ = (U8)(( uv >> 30)         | 0xfc);
121 	*d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
122 	*d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
123 	*d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
124 	*d++ = (U8)(((uv >>  6) & 0x3f) | 0x80);
125 	*d++ = (U8)(( uv        & 0x3f) | 0x80);
126 	return d;
127     }
128 #ifdef HAS_QUAD
129     if (uv < UTF8_QUAD_MAX)
130 #endif
131     {
132 	*d++ =                            0xfe;	/* Can't match U+FEFF! */
133 	*d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
134 	*d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
135 	*d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
136 	*d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
137 	*d++ = (U8)(((uv >>  6) & 0x3f) | 0x80);
138 	*d++ = (U8)(( uv        & 0x3f) | 0x80);
139 	return d;
140     }
141 #ifdef HAS_QUAD
142     {
143 	*d++ =                            0xff;		/* Can't match U+FFFE! */
144 	*d++ =                            0x80;		/* 6 Reserved bits */
145 	*d++ = (U8)(((uv >> 60) & 0x0f) | 0x80);	/* 2 Reserved bits */
146 	*d++ = (U8)(((uv >> 54) & 0x3f) | 0x80);
147 	*d++ = (U8)(((uv >> 48) & 0x3f) | 0x80);
148 	*d++ = (U8)(((uv >> 42) & 0x3f) | 0x80);
149 	*d++ = (U8)(((uv >> 36) & 0x3f) | 0x80);
150 	*d++ = (U8)(((uv >> 30) & 0x3f) | 0x80);
151 	*d++ = (U8)(((uv >> 24) & 0x3f) | 0x80);
152 	*d++ = (U8)(((uv >> 18) & 0x3f) | 0x80);
153 	*d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
154 	*d++ = (U8)(((uv >>  6) & 0x3f) | 0x80);
155 	*d++ = (U8)(( uv        & 0x3f) | 0x80);
156 	return d;
157     }
158 #endif
159 #endif /* Loop style */
160 }
161 
162 U8 *
163 Perl_uvuni_to_utf8(pTHX_ U8 *d, UV uv)
164 {
165     return Perl_uvuni_to_utf8_flags(aTHX_ d, uv, 0);
166 }
167 
168 
169 /*
170 =for apidoc A|STRLEN|is_utf8_char|U8 *s
171 
172 Tests if some arbitrary number of bytes begins in a valid UTF-8
173 character.  Note that an INVARIANT (i.e. ASCII) character is a valid
174 UTF-8 character.  The actual number of bytes in the UTF-8 character
175 will be returned if it is valid, otherwise 0.
176 
177 =cut */
178 STRLEN
179 Perl_is_utf8_char(pTHX_ U8 *s)
180 {
181     U8 u = *s;
182     STRLEN slen, len;
183     UV uv, ouv;
184 
185     if (UTF8_IS_INVARIANT(u))
186 	return 1;
187 
188     if (!UTF8_IS_START(u))
189 	return 0;
190 
191     len = UTF8SKIP(s);
192 
193     if (len < 2 || !UTF8_IS_CONTINUATION(s[1]))
194 	return 0;
195 
196     slen = len - 1;
197     s++;
198     u &= UTF_START_MASK(len);
199     uv  = u;
200     ouv = uv;
201     while (slen--) {
202 	if (!UTF8_IS_CONTINUATION(*s))
203 	    return 0;
204 	uv = UTF8_ACCUMULATE(uv, *s);
205 	if (uv < ouv)
206 	    return 0;
207 	ouv = uv;
208 	s++;
209     }
210 
211     if ((STRLEN)UNISKIP(uv) < len)
212 	return 0;
213 
214     return len;
215 }
216 
217 /*
218 =for apidoc A|bool|is_utf8_string|U8 *s|STRLEN len
219 
220 Returns true if first C<len> bytes of the given string form a valid UTF8
221 string, false otherwise.  Note that 'a valid UTF8 string' does not mean
222 'a string that contains UTF8' because a valid ASCII string is a valid
223 UTF8 string.
224 
225 =cut
226 */
227 
228 bool
229 Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len)
230 {
231     U8* x = s;
232     U8* send;
233     STRLEN c;
234 
235     if (!len)
236 	len = strlen((char *)s);
237     send = s + len;
238 
239     while (x < send) {
240 	 /* Inline the easy bits of is_utf8_char() here for speed... */
241 	 if (UTF8_IS_INVARIANT(*x))
242 	      c = 1;
243 	 else if (!UTF8_IS_START(*x))
244 	      return FALSE;
245 	 else {
246 	      /* ... and call is_utf8_char() only if really needed. */
247 	      c = is_utf8_char(x);
248 	      if (!c)
249 		   return FALSE;
250 	 }
251         x += c;
252     }
253     if (x != send)
254 	return FALSE;
255 
256     return TRUE;
257 }
258 
259 /*
260 =for apidoc A|UV|utf8n_to_uvuni|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
261 
262 Bottom level UTF-8 decode routine.
263 Returns the unicode code point value of the first character in the string C<s>
264 which is assumed to be in UTF8 encoding and no longer than C<curlen>;
265 C<retlen> will be set to the length, in bytes, of that character.
266 
267 If C<s> does not point to a well-formed UTF8 character, the behaviour
268 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY,
269 it is assumed that the caller will raise a warning, and this function
270 will silently just set C<retlen> to C<-1> and return zero.  If the
271 C<flags> does not contain UTF8_CHECK_ONLY, warnings about
272 malformations will be given, C<retlen> will be set to the expected
273 length of the UTF-8 character in bytes, and zero will be returned.
274 
275 The C<flags> can also contain various flags to allow deviations from
276 the strict UTF-8 encoding (see F<utf8.h>).
277 
278 Most code should use utf8_to_uvchr() rather than call this directly.
279 
280 =cut
281 */
282 
283 UV
284 Perl_utf8n_to_uvuni(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
285 {
286     U8 *s0 = s;
287     UV uv = *s, ouv = 0;
288     STRLEN len = 1;
289     bool dowarn = ckWARN_d(WARN_UTF8);
290     UV startbyte = *s;
291     STRLEN expectlen = 0;
292     U32 warning = 0;
293 
294 /* This list is a superset of the UTF8_ALLOW_XXX. */
295 
296 #define UTF8_WARN_EMPTY				 1
297 #define UTF8_WARN_CONTINUATION			 2
298 #define UTF8_WARN_NON_CONTINUATION	 	 3
299 #define UTF8_WARN_FE_FF				 4
300 #define UTF8_WARN_SHORT				 5
301 #define UTF8_WARN_OVERFLOW			 6
302 #define UTF8_WARN_SURROGATE			 7
303 #define UTF8_WARN_LONG				 8
304 #define UTF8_WARN_FFFF				 9 /* Also FFFE. */
305 
306     if (curlen == 0 &&
307 	!(flags & UTF8_ALLOW_EMPTY)) {
308 	warning = UTF8_WARN_EMPTY;
309 	goto malformed;
310     }
311 
312     if (UTF8_IS_INVARIANT(uv)) {
313 	if (retlen)
314 	    *retlen = 1;
315 	return (UV) (NATIVE_TO_UTF(*s));
316     }
317 
318     if (UTF8_IS_CONTINUATION(uv) &&
319 	!(flags & UTF8_ALLOW_CONTINUATION)) {
320 	warning = UTF8_WARN_CONTINUATION;
321 	goto malformed;
322     }
323 
324     if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) &&
325 	!(flags & UTF8_ALLOW_NON_CONTINUATION)) {
326 	warning = UTF8_WARN_NON_CONTINUATION;
327 	goto malformed;
328     }
329 
330 #ifdef EBCDIC
331     uv = NATIVE_TO_UTF(uv);
332 #else
333     if ((uv == 0xfe || uv == 0xff) &&
334 	!(flags & UTF8_ALLOW_FE_FF)) {
335 	warning = UTF8_WARN_FE_FF;
336 	goto malformed;
337     }
338 #endif
339 
340     if      (!(uv & 0x20))	{ len =  2; uv &= 0x1f; }
341     else if (!(uv & 0x10))	{ len =  3; uv &= 0x0f; }
342     else if (!(uv & 0x08))	{ len =  4; uv &= 0x07; }
343     else if (!(uv & 0x04))	{ len =  5; uv &= 0x03; }
344 #ifdef EBCDIC
345     else if (!(uv & 0x02))	{ len =  6; uv &= 0x01; }
346     else			{ len =  7; uv &= 0x01; }
347 #else
348     else if (!(uv & 0x02))	{ len =  6; uv &= 0x01; }
349     else if (!(uv & 0x01))	{ len =  7; uv = 0; }
350     else			{ len = 13; uv = 0; } /* whoa! */
351 #endif
352 
353     if (retlen)
354 	*retlen = len;
355 
356     expectlen = len;
357 
358     if ((curlen < expectlen) &&
359 	!(flags & UTF8_ALLOW_SHORT)) {
360 	warning = UTF8_WARN_SHORT;
361 	goto malformed;
362     }
363 
364     len--;
365     s++;
366     ouv = uv;
367 
368     while (len--) {
369 	if (!UTF8_IS_CONTINUATION(*s) &&
370 	    !(flags & UTF8_ALLOW_NON_CONTINUATION)) {
371 	    s--;
372 	    warning = UTF8_WARN_NON_CONTINUATION;
373 	    goto malformed;
374 	}
375 	else
376 	    uv = UTF8_ACCUMULATE(uv, *s);
377 	if (!(uv > ouv)) {
378 	    /* These cannot be allowed. */
379 	    if (uv == ouv) {
380 		if (!(flags & UTF8_ALLOW_LONG)) {
381 		    warning = UTF8_WARN_LONG;
382 		    goto malformed;
383 		}
384 	    }
385 	    else { /* uv < ouv */
386 		/* This cannot be allowed. */
387 		warning = UTF8_WARN_OVERFLOW;
388 		goto malformed;
389 	    }
390 	}
391 	s++;
392 	ouv = uv;
393     }
394 
395     if (UNICODE_IS_SURROGATE(uv) &&
396 	!(flags & UTF8_ALLOW_SURROGATE)) {
397 	warning = UTF8_WARN_SURROGATE;
398 	goto malformed;
399     } else if ((expectlen > (STRLEN)UNISKIP(uv)) &&
400 	       !(flags & UTF8_ALLOW_LONG)) {
401 	warning = UTF8_WARN_LONG;
402 	goto malformed;
403     } else if (UNICODE_IS_ILLEGAL(uv) &&
404 	       !(flags & UTF8_ALLOW_FFFF)) {
405 	warning = UTF8_WARN_FFFF;
406 	goto malformed;
407     }
408 
409     return uv;
410 
411 malformed:
412 
413     if (flags & UTF8_CHECK_ONLY) {
414 	if (retlen)
415 	    *retlen = -1;
416 	return 0;
417     }
418 
419     if (dowarn) {
420 	SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0));
421 
422 	switch (warning) {
423 	case 0: /* Intentionally empty. */ break;
424 	case UTF8_WARN_EMPTY:
425 	    Perl_sv_catpvf(aTHX_ sv, "(empty string)");
426 	    break;
427 	case UTF8_WARN_CONTINUATION:
428 	    Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv);
429 	    break;
430 	case UTF8_WARN_NON_CONTINUATION:
431 	    if (s == s0)
432 	        Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")",
433                            (UV)s[1], startbyte);
434 	    else
435 	        Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)",
436                            (UV)s[1], s - s0, s - s0 > 1 ? "s" : "", startbyte, expectlen);
437 
438 	    break;
439 	case UTF8_WARN_FE_FF:
440 	    Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv);
441 	    break;
442 	case UTF8_WARN_SHORT:
443 	    Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
444                            curlen, curlen == 1 ? "" : "s", expectlen, startbyte);
445 	    expectlen = curlen;		/* distance for caller to skip */
446 	    break;
447 	case UTF8_WARN_OVERFLOW:
448 	    Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")",
449                            ouv, *s, startbyte);
450 	    break;
451 	case UTF8_WARN_SURROGATE:
452 	    Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv);
453 	    break;
454 	case UTF8_WARN_LONG:
455 	    Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")",
456 			   expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte);
457 	    break;
458 	case UTF8_WARN_FFFF:
459 	    Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv);
460 	    break;
461 	default:
462 	    Perl_sv_catpvf(aTHX_ sv, "(unknown reason)");
463 	    break;
464 	}
465 
466 	if (warning) {
467 	    char *s = SvPVX(sv);
468 
469 	    if (PL_op)
470 		Perl_warner(aTHX_ packWARN(WARN_UTF8),
471 			    "%s in %s", s,  OP_DESC(PL_op));
472 	    else
473 		Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s);
474 	}
475     }
476 
477     if (retlen)
478 	*retlen = expectlen ? expectlen : len;
479 
480     return 0;
481 }
482 
483 /*
484 =for apidoc A|UV|utf8_to_uvchr|U8 *s|STRLEN *retlen
485 
486 Returns the native character value of the first character in the string C<s>
487 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
488 length, in bytes, of that character.
489 
490 If C<s> does not point to a well-formed UTF8 character, zero is
491 returned and retlen is set, if possible, to -1.
492 
493 =cut
494 */
495 
496 UV
497 Perl_utf8_to_uvchr(pTHX_ U8 *s, STRLEN *retlen)
498 {
499     return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXLEN, retlen,
500 			       ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
501 }
502 
503 /*
504 =for apidoc A|UV|utf8_to_uvuni|U8 *s|STRLEN *retlen
505 
506 Returns the Unicode code point of the first character in the string C<s>
507 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
508 length, in bytes, of that character.
509 
510 This function should only be used when returned UV is considered
511 an index into the Unicode semantic tables (e.g. swashes).
512 
513 If C<s> does not point to a well-formed UTF8 character, zero is
514 returned and retlen is set, if possible, to -1.
515 
516 =cut
517 */
518 
519 UV
520 Perl_utf8_to_uvuni(pTHX_ U8 *s, STRLEN *retlen)
521 {
522     /* Call the low level routine asking for checks */
523     return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXLEN, retlen,
524 			       ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
525 }
526 
527 /*
528 =for apidoc A|STRLEN|utf8_length|U8 *s|U8 *e
529 
530 Return the length of the UTF-8 char encoded string C<s> in characters.
531 Stops at C<e> (inclusive).  If C<e E<lt> s> or if the scan would end
532 up past C<e>, croaks.
533 
534 =cut
535 */
536 
537 STRLEN
538 Perl_utf8_length(pTHX_ U8 *s, U8 *e)
539 {
540     STRLEN len = 0;
541 
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 
546     if (e < s) {
547         if (ckWARN_d(WARN_UTF8)) {
548 	    if (PL_op)
549 	        Perl_warner(aTHX_ packWARN(WARN_UTF8),
550 			    "%s in %s", unees, OP_DESC(PL_op));
551 	    else
552 	        Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
553 	}
554 	return 0;
555     }
556     while (s < e) {
557 	U8 t = UTF8SKIP(s);
558 
559 	if (e - s < t) {
560 	    if (ckWARN_d(WARN_UTF8)) {
561 	        if (PL_op)
562 		    Perl_warner(aTHX_ packWARN(WARN_UTF8),
563 				unees, OP_DESC(PL_op));
564 		else
565 		    Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
566 	    }
567 	    return len;
568 	}
569 	s += t;
570 	len++;
571     }
572 
573     return len;
574 }
575 
576 /*
577 =for apidoc A|IV|utf8_distance|U8 *a|U8 *b
578 
579 Returns the number of UTF8 characters between the UTF-8 pointers C<a>
580 and C<b>.
581 
582 WARNING: use only if you *know* that the pointers point inside the
583 same UTF-8 buffer.
584 
585 =cut
586 */
587 
588 IV
589 Perl_utf8_distance(pTHX_ U8 *a, U8 *b)
590 {
591     IV off = 0;
592 
593     /* Note: cannot use UTF8_IS_...() too eagerly here since  e.g.
594      * the bitops (especially ~) can create illegal UTF-8.
595      * In other words: in Perl UTF-8 is not just for Unicode. */
596 
597     if (a < b) {
598 	while (a < b) {
599 	    U8 c = UTF8SKIP(a);
600 
601 	    if (b - a < c) {
602 	        if (ckWARN_d(WARN_UTF8)) {
603 		    if (PL_op)
604 		        Perl_warner(aTHX_ packWARN(WARN_UTF8),
605 				    "%s in %s", unees, OP_DESC(PL_op));
606 		    else
607 		        Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
608 		}
609 		return off;
610 	    }
611 	    a += c;
612 	    off--;
613 	}
614     }
615     else {
616 	while (b < a) {
617 	    U8 c = UTF8SKIP(b);
618 
619 	    if (a - b < c) {
620 	        if (ckWARN_d(WARN_UTF8)) {
621 		    if (PL_op)
622 		        Perl_warner(aTHX_ packWARN(WARN_UTF8),
623 				    "%s in %s", unees, OP_DESC(PL_op));
624 		    else
625 		        Perl_warner(aTHX_ packWARN(WARN_UTF8), unees);
626 		}
627 		return off;
628 	    }
629 	    b += c;
630 	    off++;
631 	}
632     }
633 
634     return off;
635 }
636 
637 /*
638 =for apidoc A|U8 *|utf8_hop|U8 *s|I32 off
639 
640 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
641 forward or backward.
642 
643 WARNING: do not use the following unless you *know* C<off> is within
644 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
645 on the first byte of character or just after the last byte of a character.
646 
647 =cut
648 */
649 
650 U8 *
651 Perl_utf8_hop(pTHX_ U8 *s, I32 off)
652 {
653     /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
654      * the bitops (especially ~) can create illegal UTF-8.
655      * In other words: in Perl UTF-8 is not just for Unicode. */
656 
657     if (off >= 0) {
658 	while (off--)
659 	    s += UTF8SKIP(s);
660     }
661     else {
662 	while (off++) {
663 	    s--;
664 	    while (UTF8_IS_CONTINUATION(*s))
665 		s--;
666 	}
667     }
668     return s;
669 }
670 
671 /*
672 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len
673 
674 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
675 Unlike C<bytes_to_utf8>, this over-writes the original string, and
676 updates len to contain the new length.
677 Returns zero on failure, setting C<len> to -1.
678 
679 =cut
680 */
681 
682 U8 *
683 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len)
684 {
685     U8 *send;
686     U8 *d;
687     U8 *save = s;
688 
689     /* ensure valid UTF8 and chars < 256 before updating string */
690     for (send = s + *len; s < send; ) {
691         U8 c = *s++;
692 
693         if (!UTF8_IS_INVARIANT(c) &&
694             (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send)
695 	     || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) {
696             *len = -1;
697             return 0;
698         }
699     }
700 
701     d = s = save;
702     while (s < send) {
703         STRLEN ulen;
704         *d++ = (U8)utf8_to_uvchr(s, &ulen);
705         s += ulen;
706     }
707     *d = '\0';
708     *len = d - save;
709     return save;
710 }
711 
712 /*
713 =for apidoc A|U8 *|bytes_from_utf8|U8 *s|STRLEN *len|bool *is_utf8
714 
715 Converts a string C<s> of length C<len> from UTF8 into byte encoding.
716 Unlike <utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to
717 the newly-created string, and updates C<len> to contain the new
718 length.  Returns the original string if no conversion occurs, C<len>
719 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to
720 0 if C<s> is converted or contains all 7bit characters.
721 
722 =cut
723 */
724 
725 U8 *
726 Perl_bytes_from_utf8(pTHX_ U8 *s, STRLEN *len, bool *is_utf8)
727 {
728     U8 *d;
729     U8 *start = s;
730     U8 *send;
731     I32 count = 0;
732 
733     if (!*is_utf8)
734 	return start;
735 
736     /* ensure valid UTF8 and chars < 256 before converting string */
737     for (send = s + *len; s < send;) {
738 	U8 c = *s++;
739 	if (!UTF8_IS_INVARIANT(c)) {
740 	    if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send &&
741                 (c = *s++) && UTF8_IS_CONTINUATION(c))
742 		count++;
743 	    else
744 		return start;
745 	}
746     }
747 
748     *is_utf8 = 0;
749 
750     Newz(801, d, (*len) - count + 1, U8);
751     s = start; start = d;
752     while (s < send) {
753 	U8 c = *s++;
754 	if (!UTF8_IS_INVARIANT(c)) {
755 	    /* Then it is two-byte encoded */
756 	    c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++);
757 	    c = ASCII_TO_NATIVE(c);
758 	}
759 	*d++ = c;
760     }
761     *d = '\0';
762     *len = d - start;
763     return start;
764 }
765 
766 /*
767 =for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
768 
769 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
770 Returns a pointer to the newly-created string, and sets C<len> to
771 reflect the new length.
772 
773 =cut
774 */
775 
776 U8*
777 Perl_bytes_to_utf8(pTHX_ U8 *s, STRLEN *len)
778 {
779     U8 *send;
780     U8 *d;
781     U8 *dst;
782     send = s + (*len);
783 
784     Newz(801, d, (*len) * 2 + 1, U8);
785     dst = d;
786 
787     while (s < send) {
788         UV uv = NATIVE_TO_ASCII(*s++);
789         if (UNI_IS_INVARIANT(uv))
790             *d++ = (U8)UTF_TO_NATIVE(uv);
791         else {
792             *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
793             *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
794         }
795     }
796     *d = '\0';
797     *len = d-dst;
798     return dst;
799 }
800 
801 /*
802  * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8.
803  *
804  * Destination must be pre-extended to 3/2 source.  Do not use in-place.
805  * We optimize for native, for obvious reasons. */
806 
807 U8*
808 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
809 {
810     U8* pend;
811     U8* dstart = d;
812 
813     if (bytelen & 1)
814 	Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen");
815 
816     pend = p + bytelen;
817 
818     while (p < pend) {
819 	UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */
820 	p += 2;
821 	if (uv < 0x80) {
822 	    *d++ = (U8)uv;
823 	    continue;
824 	}
825 	if (uv < 0x800) {
826 	    *d++ = (U8)(( uv >>  6)         | 0xc0);
827 	    *d++ = (U8)(( uv        & 0x3f) | 0x80);
828 	    continue;
829 	}
830 	if (uv >= 0xd800 && uv < 0xdbff) {	/* surrogates */
831 	    UV low = *p++;
832 	    if (low < 0xdc00 || low >= 0xdfff)
833 		Perl_croak(aTHX_ "Malformed UTF-16 surrogate");
834 	    uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000;
835 	}
836 	if (uv < 0x10000) {
837 	    *d++ = (U8)(( uv >> 12)         | 0xe0);
838 	    *d++ = (U8)(((uv >>  6) & 0x3f) | 0x80);
839 	    *d++ = (U8)(( uv        & 0x3f) | 0x80);
840 	    continue;
841 	}
842 	else {
843 	    *d++ = (U8)(( uv >> 18)         | 0xf0);
844 	    *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80);
845 	    *d++ = (U8)(((uv >>  6) & 0x3f) | 0x80);
846 	    *d++ = (U8)(( uv        & 0x3f) | 0x80);
847 	    continue;
848 	}
849     }
850     *newlen = d - dstart;
851     return d;
852 }
853 
854 /* Note: this one is slightly destructive of the source. */
855 
856 U8*
857 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen)
858 {
859     U8* s = (U8*)p;
860     U8* send = s + bytelen;
861     while (s < send) {
862 	U8 tmp = s[0];
863 	s[0] = s[1];
864 	s[1] = tmp;
865 	s += 2;
866     }
867     return utf16_to_utf8(p, d, bytelen, newlen);
868 }
869 
870 /* for now these are all defined (inefficiently) in terms of the utf8 versions */
871 
872 bool
873 Perl_is_uni_alnum(pTHX_ UV c)
874 {
875     U8 tmpbuf[UTF8_MAXLEN+1];
876     uvchr_to_utf8(tmpbuf, c);
877     return is_utf8_alnum(tmpbuf);
878 }
879 
880 bool
881 Perl_is_uni_alnumc(pTHX_ UV c)
882 {
883     U8 tmpbuf[UTF8_MAXLEN+1];
884     uvchr_to_utf8(tmpbuf, c);
885     return is_utf8_alnumc(tmpbuf);
886 }
887 
888 bool
889 Perl_is_uni_idfirst(pTHX_ UV c)
890 {
891     U8 tmpbuf[UTF8_MAXLEN+1];
892     uvchr_to_utf8(tmpbuf, c);
893     return is_utf8_idfirst(tmpbuf);
894 }
895 
896 bool
897 Perl_is_uni_alpha(pTHX_ UV c)
898 {
899     U8 tmpbuf[UTF8_MAXLEN+1];
900     uvchr_to_utf8(tmpbuf, c);
901     return is_utf8_alpha(tmpbuf);
902 }
903 
904 bool
905 Perl_is_uni_ascii(pTHX_ UV c)
906 {
907     U8 tmpbuf[UTF8_MAXLEN+1];
908     uvchr_to_utf8(tmpbuf, c);
909     return is_utf8_ascii(tmpbuf);
910 }
911 
912 bool
913 Perl_is_uni_space(pTHX_ UV c)
914 {
915     U8 tmpbuf[UTF8_MAXLEN+1];
916     uvchr_to_utf8(tmpbuf, c);
917     return is_utf8_space(tmpbuf);
918 }
919 
920 bool
921 Perl_is_uni_digit(pTHX_ UV c)
922 {
923     U8 tmpbuf[UTF8_MAXLEN+1];
924     uvchr_to_utf8(tmpbuf, c);
925     return is_utf8_digit(tmpbuf);
926 }
927 
928 bool
929 Perl_is_uni_upper(pTHX_ UV c)
930 {
931     U8 tmpbuf[UTF8_MAXLEN+1];
932     uvchr_to_utf8(tmpbuf, c);
933     return is_utf8_upper(tmpbuf);
934 }
935 
936 bool
937 Perl_is_uni_lower(pTHX_ UV c)
938 {
939     U8 tmpbuf[UTF8_MAXLEN+1];
940     uvchr_to_utf8(tmpbuf, c);
941     return is_utf8_lower(tmpbuf);
942 }
943 
944 bool
945 Perl_is_uni_cntrl(pTHX_ UV c)
946 {
947     U8 tmpbuf[UTF8_MAXLEN+1];
948     uvchr_to_utf8(tmpbuf, c);
949     return is_utf8_cntrl(tmpbuf);
950 }
951 
952 bool
953 Perl_is_uni_graph(pTHX_ UV c)
954 {
955     U8 tmpbuf[UTF8_MAXLEN+1];
956     uvchr_to_utf8(tmpbuf, c);
957     return is_utf8_graph(tmpbuf);
958 }
959 
960 bool
961 Perl_is_uni_print(pTHX_ UV c)
962 {
963     U8 tmpbuf[UTF8_MAXLEN+1];
964     uvchr_to_utf8(tmpbuf, c);
965     return is_utf8_print(tmpbuf);
966 }
967 
968 bool
969 Perl_is_uni_punct(pTHX_ UV c)
970 {
971     U8 tmpbuf[UTF8_MAXLEN+1];
972     uvchr_to_utf8(tmpbuf, c);
973     return is_utf8_punct(tmpbuf);
974 }
975 
976 bool
977 Perl_is_uni_xdigit(pTHX_ UV c)
978 {
979     U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
980     uvchr_to_utf8(tmpbuf, c);
981     return is_utf8_xdigit(tmpbuf);
982 }
983 
984 UV
985 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp)
986 {
987     uvchr_to_utf8(p, c);
988     return to_utf8_upper(p, p, lenp);
989 }
990 
991 UV
992 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp)
993 {
994     uvchr_to_utf8(p, c);
995     return to_utf8_title(p, p, lenp);
996 }
997 
998 UV
999 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp)
1000 {
1001     uvchr_to_utf8(p, c);
1002     return to_utf8_lower(p, p, lenp);
1003 }
1004 
1005 UV
1006 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp)
1007 {
1008     uvchr_to_utf8(p, c);
1009     return to_utf8_fold(p, p, lenp);
1010 }
1011 
1012 /* for now these all assume no locale info available for Unicode > 255 */
1013 
1014 bool
1015 Perl_is_uni_alnum_lc(pTHX_ UV c)
1016 {
1017     return is_uni_alnum(c);	/* XXX no locale support yet */
1018 }
1019 
1020 bool
1021 Perl_is_uni_alnumc_lc(pTHX_ UV c)
1022 {
1023     return is_uni_alnumc(c);	/* XXX no locale support yet */
1024 }
1025 
1026 bool
1027 Perl_is_uni_idfirst_lc(pTHX_ UV c)
1028 {
1029     return is_uni_idfirst(c);	/* XXX no locale support yet */
1030 }
1031 
1032 bool
1033 Perl_is_uni_alpha_lc(pTHX_ UV c)
1034 {
1035     return is_uni_alpha(c);	/* XXX no locale support yet */
1036 }
1037 
1038 bool
1039 Perl_is_uni_ascii_lc(pTHX_ UV c)
1040 {
1041     return is_uni_ascii(c);	/* XXX no locale support yet */
1042 }
1043 
1044 bool
1045 Perl_is_uni_space_lc(pTHX_ UV c)
1046 {
1047     return is_uni_space(c);	/* XXX no locale support yet */
1048 }
1049 
1050 bool
1051 Perl_is_uni_digit_lc(pTHX_ UV c)
1052 {
1053     return is_uni_digit(c);	/* XXX no locale support yet */
1054 }
1055 
1056 bool
1057 Perl_is_uni_upper_lc(pTHX_ UV c)
1058 {
1059     return is_uni_upper(c);	/* XXX no locale support yet */
1060 }
1061 
1062 bool
1063 Perl_is_uni_lower_lc(pTHX_ UV c)
1064 {
1065     return is_uni_lower(c);	/* XXX no locale support yet */
1066 }
1067 
1068 bool
1069 Perl_is_uni_cntrl_lc(pTHX_ UV c)
1070 {
1071     return is_uni_cntrl(c);	/* XXX no locale support yet */
1072 }
1073 
1074 bool
1075 Perl_is_uni_graph_lc(pTHX_ UV c)
1076 {
1077     return is_uni_graph(c);	/* XXX no locale support yet */
1078 }
1079 
1080 bool
1081 Perl_is_uni_print_lc(pTHX_ UV c)
1082 {
1083     return is_uni_print(c);	/* XXX no locale support yet */
1084 }
1085 
1086 bool
1087 Perl_is_uni_punct_lc(pTHX_ UV c)
1088 {
1089     return is_uni_punct(c);	/* XXX no locale support yet */
1090 }
1091 
1092 bool
1093 Perl_is_uni_xdigit_lc(pTHX_ UV c)
1094 {
1095     return is_uni_xdigit(c);	/* XXX no locale support yet */
1096 }
1097 
1098 U32
1099 Perl_to_uni_upper_lc(pTHX_ U32 c)
1100 {
1101     /* XXX returns only the first character -- do not use XXX */
1102     /* XXX no locale support yet */
1103     STRLEN len;
1104     U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
1105     return (U32)to_uni_upper(c, tmpbuf, &len);
1106 }
1107 
1108 U32
1109 Perl_to_uni_title_lc(pTHX_ U32 c)
1110 {
1111     /* XXX returns only the first character XXX -- do not use XXX */
1112     /* XXX no locale support yet */
1113     STRLEN len;
1114     U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
1115     return (U32)to_uni_title(c, tmpbuf, &len);
1116 }
1117 
1118 U32
1119 Perl_to_uni_lower_lc(pTHX_ U32 c)
1120 {
1121     /* XXX returns only the first character -- do not use XXX */
1122     /* XXX no locale support yet */
1123     STRLEN len;
1124     U8 tmpbuf[UTF8_MAXLEN_UCLC+1];
1125     return (U32)to_uni_lower(c, tmpbuf, &len);
1126 }
1127 
1128 bool
1129 Perl_is_utf8_alnum(pTHX_ U8 *p)
1130 {
1131     if (!is_utf8_char(p))
1132 	return FALSE;
1133     if (!PL_utf8_alnum)
1134 	/* NOTE: "IsWord", not "IsAlnum", since Alnum is a true
1135 	 * descendant of isalnum(3), in other words, it doesn't
1136 	 * contain the '_'. --jhi */
1137 	PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0);
1138     return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1139 /*    return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */
1140 #ifdef SURPRISINGLY_SLOWER  /* probably because alpha is usually true */
1141     if (!PL_utf8_alnum)
1142 	PL_utf8_alnum = swash_init("utf8", "",
1143 	    sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1144     return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1145 #endif
1146 }
1147 
1148 bool
1149 Perl_is_utf8_alnumc(pTHX_ U8 *p)
1150 {
1151     if (!is_utf8_char(p))
1152 	return FALSE;
1153     if (!PL_utf8_alnum)
1154 	PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
1155     return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1156 /*    return is_utf8_alpha(p) || is_utf8_digit(p); */
1157 #ifdef SURPRISINGLY_SLOWER  /* probably because alpha is usually true */
1158     if (!PL_utf8_alnum)
1159 	PL_utf8_alnum = swash_init("utf8", "",
1160 	    sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
1161     return swash_fetch(PL_utf8_alnum, p, TRUE) != 0;
1162 #endif
1163 }
1164 
1165 bool
1166 Perl_is_utf8_idfirst(pTHX_ U8 *p) /* The naming is historical. */
1167 {
1168     if (*p == '_')
1169 	return TRUE;
1170     if (!is_utf8_char(p))
1171 	return FALSE;
1172     if (!PL_utf8_idstart) /* is_utf8_idstart would be more logical. */
1173 	PL_utf8_idstart = swash_init("utf8", "IdStart", &PL_sv_undef, 0, 0);
1174     return swash_fetch(PL_utf8_idstart, p, TRUE) != 0;
1175 }
1176 
1177 bool
1178 Perl_is_utf8_idcont(pTHX_ U8 *p)
1179 {
1180     if (*p == '_')
1181 	return TRUE;
1182     if (!is_utf8_char(p))
1183 	return FALSE;
1184     if (!PL_utf8_idcont)
1185 	PL_utf8_idcont = swash_init("utf8", "IdContinue", &PL_sv_undef, 0, 0);
1186     return swash_fetch(PL_utf8_idcont, p, TRUE) != 0;
1187 }
1188 
1189 bool
1190 Perl_is_utf8_alpha(pTHX_ U8 *p)
1191 {
1192     if (!is_utf8_char(p))
1193 	return FALSE;
1194     if (!PL_utf8_alpha)
1195 	PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
1196     return swash_fetch(PL_utf8_alpha, p, TRUE) != 0;
1197 }
1198 
1199 bool
1200 Perl_is_utf8_ascii(pTHX_ U8 *p)
1201 {
1202     if (!is_utf8_char(p))
1203 	return FALSE;
1204     if (!PL_utf8_ascii)
1205 	PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
1206     return swash_fetch(PL_utf8_ascii, p, TRUE) != 0;
1207 }
1208 
1209 bool
1210 Perl_is_utf8_space(pTHX_ U8 *p)
1211 {
1212     if (!is_utf8_char(p))
1213 	return FALSE;
1214     if (!PL_utf8_space)
1215 	PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0);
1216     return swash_fetch(PL_utf8_space, p, TRUE) != 0;
1217 }
1218 
1219 bool
1220 Perl_is_utf8_digit(pTHX_ U8 *p)
1221 {
1222     if (!is_utf8_char(p))
1223 	return FALSE;
1224     if (!PL_utf8_digit)
1225 	PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
1226     return swash_fetch(PL_utf8_digit, p, TRUE) != 0;
1227 }
1228 
1229 bool
1230 Perl_is_utf8_upper(pTHX_ U8 *p)
1231 {
1232     if (!is_utf8_char(p))
1233 	return FALSE;
1234     if (!PL_utf8_upper)
1235 	PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
1236     return swash_fetch(PL_utf8_upper, p, TRUE) != 0;
1237 }
1238 
1239 bool
1240 Perl_is_utf8_lower(pTHX_ U8 *p)
1241 {
1242     if (!is_utf8_char(p))
1243 	return FALSE;
1244     if (!PL_utf8_lower)
1245 	PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
1246     return swash_fetch(PL_utf8_lower, p, TRUE) != 0;
1247 }
1248 
1249 bool
1250 Perl_is_utf8_cntrl(pTHX_ U8 *p)
1251 {
1252     if (!is_utf8_char(p))
1253 	return FALSE;
1254     if (!PL_utf8_cntrl)
1255 	PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
1256     return swash_fetch(PL_utf8_cntrl, p, TRUE) != 0;
1257 }
1258 
1259 bool
1260 Perl_is_utf8_graph(pTHX_ U8 *p)
1261 {
1262     if (!is_utf8_char(p))
1263 	return FALSE;
1264     if (!PL_utf8_graph)
1265 	PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
1266     return swash_fetch(PL_utf8_graph, p, TRUE) != 0;
1267 }
1268 
1269 bool
1270 Perl_is_utf8_print(pTHX_ U8 *p)
1271 {
1272     if (!is_utf8_char(p))
1273 	return FALSE;
1274     if (!PL_utf8_print)
1275 	PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
1276     return swash_fetch(PL_utf8_print, p, TRUE) != 0;
1277 }
1278 
1279 bool
1280 Perl_is_utf8_punct(pTHX_ U8 *p)
1281 {
1282     if (!is_utf8_char(p))
1283 	return FALSE;
1284     if (!PL_utf8_punct)
1285 	PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
1286     return swash_fetch(PL_utf8_punct, p, TRUE) != 0;
1287 }
1288 
1289 bool
1290 Perl_is_utf8_xdigit(pTHX_ U8 *p)
1291 {
1292     if (!is_utf8_char(p))
1293 	return FALSE;
1294     if (!PL_utf8_xdigit)
1295 	PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
1296     return swash_fetch(PL_utf8_xdigit, p, TRUE) != 0;
1297 }
1298 
1299 bool
1300 Perl_is_utf8_mark(pTHX_ U8 *p)
1301 {
1302     if (!is_utf8_char(p))
1303 	return FALSE;
1304     if (!PL_utf8_mark)
1305 	PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
1306     return swash_fetch(PL_utf8_mark, p, TRUE) != 0;
1307 }
1308 
1309 /*
1310 =for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special
1311 
1312 The "p" contains the pointer to the UTF-8 string encoding
1313 the character that is being converted.
1314 
1315 The "ustrp" is a pointer to the character buffer to put the
1316 conversion result to.  The "lenp" is a pointer to the length
1317 of the result.
1318 
1319 The "swashp" is a pointer to the swash to use.
1320 
1321 Both the special and normal mappings are stored lib/unicore/To/Foo.pl,
1322 and loaded by SWASHGET, using lib/utf8_heavy.pl.  The special (usually,
1323 but not always, a multicharacter mapping), is tried first.
1324 
1325 The "special" is a string like "utf8::ToSpecLower", which means the
1326 hash %utf8::ToSpecLower.  The access to the hash is through
1327 Perl_to_utf8_case().
1328 
1329 The "normal" is a string like "ToLower" which means the swash
1330 %utf8::ToLower.
1331 
1332 =cut */
1333 
1334 UV
1335 Perl_to_utf8_case(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp, char *normal, char *special)
1336 {
1337     UV uv0, uv1;
1338     U8 tmpbuf[UTF8_MAXLEN_FOLD+1];
1339     STRLEN len = 0;
1340 
1341     uv0 = utf8_to_uvchr(p, 0);
1342     /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings
1343      * are necessary in EBCDIC, they are redundant no-ops
1344      * in ASCII-ish platforms, and hopefully optimized away. */
1345     uv1 = NATIVE_TO_UNI(uv0);
1346     uvuni_to_utf8(tmpbuf, uv1);
1347 
1348     if (!*swashp) /* load on-demand */
1349          *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0);
1350 
1351     if (special) {
1352          /* It might be "special" (sometimes, but not always,
1353 	  * a multicharacter mapping) */
1354 	 HV *hv;
1355 	 SV *keysv;
1356 	 HE *he;
1357 	 SV *val;
1358 
1359 	 if ((hv    = get_hv(special, FALSE)) &&
1360 	     (keysv = sv_2mortal(Perl_newSVpvf(aTHX_ "%04"UVXf, uv1))) &&
1361 	     (he    = hv_fetch_ent(hv, keysv, FALSE, 0)) &&
1362 	     (val   = HeVAL(he))) {
1363 	     char *s;
1364 
1365 	      s = SvPV(val, len);
1366 	      if (len == 1)
1367 		   len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp;
1368 	      else {
1369 #ifdef EBCDIC
1370 		   /* If we have EBCDIC we need to remap the characters
1371 		    * since any characters in the low 256 are Unicode
1372 		    * code points, not EBCDIC. */
1373 		   U8 *t = (U8*)s, *tend = t + len, *d;
1374 
1375 		   d = tmpbuf;
1376 		   if (SvUTF8(val)) {
1377 			STRLEN tlen = 0;
1378 
1379 			while (t < tend) {
1380 			     UV c = utf8_to_uvchr(t, &tlen);
1381 			     if (tlen > 0) {
1382 				  d = uvchr_to_utf8(d, UNI_TO_NATIVE(c));
1383 				  t += tlen;
1384 			     }
1385 			     else
1386 				  break;
1387 			}
1388 		   }
1389 		   else {
1390 			while (t < tend) {
1391 			     d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t));
1392 			     t++;
1393 			}
1394 		   }
1395 		   len = d - tmpbuf;
1396 		   Copy(tmpbuf, ustrp, len, U8);
1397 #else
1398 		   Copy(s, ustrp, len, U8);
1399 #endif
1400 	      }
1401 	 }
1402     }
1403 
1404     if (!len && *swashp) {
1405 	 UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE);
1406 
1407 	 if (uv2) {
1408 	      /* It was "normal" (a single character mapping). */
1409 	      UV uv3 = UNI_TO_NATIVE(uv2);
1410 
1411 	      len = uvchr_to_utf8(ustrp, uv3) - ustrp;
1412 	 }
1413     }
1414 
1415     if (!len) /* Neither: just copy. */
1416 	 len = uvchr_to_utf8(ustrp, uv0) - ustrp;
1417 
1418     if (lenp)
1419 	 *lenp = len;
1420 
1421     return len ? utf8_to_uvchr(ustrp, 0) : 0;
1422 }
1423 
1424 /*
1425 =for apidoc A|UV|to_utf8_upper|U8 *p|U8 *ustrp|STRLEN *lenp
1426 
1427 Convert the UTF-8 encoded character at p to its uppercase version and
1428 store that in UTF-8 in ustrp and its length in bytes in lenp.  Note
1429 that the ustrp needs to be at least UTF8_MAXLEN_UCLC+1 bytes since the
1430 uppercase version may be longer than the original character (up to two
1431 characters).
1432 
1433 The first character of the uppercased version is returned
1434 (but note, as explained above, that there may be more.)
1435 
1436 =cut */
1437 
1438 UV
1439 Perl_to_utf8_upper(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1440 {
1441     return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1442                              &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper");
1443 }
1444 
1445 /*
1446 =for apidoc A|UV|to_utf8_title|U8 *p|U8 *ustrp|STRLEN *lenp
1447 
1448 Convert the UTF-8 encoded character at p to its titlecase version and
1449 store that in UTF-8 in ustrp and its length in bytes in lenp.  Note
1450 that the ustrp needs to be at least UTF8_MAXLEN_UCLC+1 bytes since the
1451 titlecase version may be longer than the original character (up to two
1452 characters).
1453 
1454 The first character of the titlecased version is returned
1455 (but note, as explained above, that there may be more.)
1456 
1457 =cut */
1458 
1459 UV
1460 Perl_to_utf8_title(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1461 {
1462     return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1463                              &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle");
1464 }
1465 
1466 /*
1467 =for apidoc A|UV|to_utf8_lower|U8 *p|U8 *ustrp|STRLEN *lenp
1468 
1469 Convert the UTF-8 encoded character at p to its lowercase version and
1470 store that in UTF-8 in ustrp and its length in bytes in lenp.  Note
1471 that the ustrp needs to be at least UTF8_MAXLEN_UCLC+1 bytes since the
1472 lowercase version may be longer than the original character (up to two
1473 characters).
1474 
1475 The first character of the lowercased version is returned
1476 (but note, as explained above, that there may be more.)
1477 
1478 =cut */
1479 
1480 UV
1481 Perl_to_utf8_lower(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1482 {
1483     return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1484                              &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower");
1485 }
1486 
1487 /*
1488 =for apidoc A|UV|to_utf8_fold|U8 *p|U8 *ustrp|STRLEN *lenp
1489 
1490 Convert the UTF-8 encoded character at p to its foldcase version and
1491 store that in UTF-8 in ustrp and its length in bytes in lenp.  Note
1492 that the ustrp needs to be at least UTF8_MAXLEN_FOLD+1 bytes since the
1493 foldcase version may be longer than the original character (up to
1494 three characters).
1495 
1496 The first character of the foldcased version is returned
1497 (but note, as explained above, that there may be more.)
1498 
1499 =cut */
1500 
1501 UV
1502 Perl_to_utf8_fold(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp)
1503 {
1504     return Perl_to_utf8_case(aTHX_ p, ustrp, lenp,
1505                              &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold");
1506 }
1507 
1508 /* a "swash" is a swatch hash */
1509 
1510 SV*
1511 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none)
1512 {
1513     SV* retval;
1514     SV* tokenbufsv = sv_2mortal(NEWSV(0,0));
1515     dSP;
1516     HV *stash = gv_stashpvn(pkg, strlen(pkg), FALSE);
1517     SV* errsv_save;
1518 
1519     if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) {	/* demand load utf8 */
1520 	ENTER;
1521 	errsv_save = newSVsv(ERRSV);
1522 	Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv);
1523 	if (!SvTRUE(ERRSV))
1524 	    sv_setsv(ERRSV, errsv_save);
1525 	SvREFCNT_dec(errsv_save);
1526 	LEAVE;
1527     }
1528     SPAGAIN;
1529     PUSHSTACKi(PERLSI_MAGIC);
1530     PUSHMARK(SP);
1531     EXTEND(SP,5);
1532     PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg))));
1533     PUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
1534     PUSHs(listsv);
1535     PUSHs(sv_2mortal(newSViv(minbits)));
1536     PUSHs(sv_2mortal(newSViv(none)));
1537     PUTBACK;
1538     ENTER;
1539     SAVEI32(PL_hints);
1540     PL_hints = 0;
1541     save_re_context();
1542     if (PL_curcop == &PL_compiling) {
1543 	/* XXX ought to be handled by lex_start */
1544 	SAVEI32(PL_in_my);
1545 	sv_setpv(tokenbufsv, PL_tokenbuf);
1546     }
1547     errsv_save = newSVsv(ERRSV);
1548     if (call_method("SWASHNEW", G_SCALAR))
1549 	retval = newSVsv(*PL_stack_sp--);
1550     else
1551 	retval = &PL_sv_undef;
1552     if (!SvTRUE(ERRSV))
1553 	sv_setsv(ERRSV, errsv_save);
1554     SvREFCNT_dec(errsv_save);
1555     LEAVE;
1556     POPSTACK;
1557     if (PL_curcop == &PL_compiling) {
1558 	STRLEN len;
1559 	char* pv = SvPV(tokenbufsv, len);
1560 
1561 	Copy(pv, PL_tokenbuf, len+1, char);
1562 	PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1563     }
1564     if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) {
1565         if (SvPOK(retval))
1566 	    Perl_croak(aTHX_ "Can't find Unicode property definition \"%s\"",
1567 		       SvPV_nolen(retval));
1568 	Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref");
1569     }
1570     return retval;
1571 }
1572 
1573 
1574 /* This API is wrong for special case conversions since we may need to
1575  * return several Unicode characters for a single Unicode character
1576  * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is
1577  * the lower-level routine, and it is similarly broken for returning
1578  * multiple values.  --jhi */
1579 UV
1580 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr, bool do_utf8)
1581 {
1582     HV* hv = (HV*)SvRV(sv);
1583     U32 klen;
1584     U32 off;
1585     STRLEN slen;
1586     STRLEN needents;
1587     U8 *tmps = NULL;
1588     U32 bit;
1589     SV *retval;
1590     U8 tmputf8[2];
1591     UV c = NATIVE_TO_ASCII(*ptr);
1592 
1593     if (!do_utf8 && !UNI_IS_INVARIANT(c)) {
1594         tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c);
1595         tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c);
1596         ptr = tmputf8;
1597     }
1598     /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ
1599      * then the "swatch" is a vec() for al the chars which start
1600      * with 0xAA..0xYY
1601      * So the key in the hash (klen) is length of encoded char -1
1602      */
1603     klen = UTF8SKIP(ptr) - 1;
1604     off  = ptr[klen];
1605 
1606     if (klen == 0)
1607      {
1608       /* If char in invariant then swatch is for all the invariant chars
1609        * In both UTF-8 and UTF8-MOD that happens to be UTF_CONTINUATION_MARK
1610        */
1611       needents = UTF_CONTINUATION_MARK;
1612       off      = NATIVE_TO_UTF(ptr[klen]);
1613      }
1614     else
1615      {
1616       /* If char is encoded then swatch is for the prefix */
1617       needents = (1 << UTF_ACCUMULATION_SHIFT);
1618       off      = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK;
1619      }
1620 
1621     /*
1622      * This single-entry cache saves about 1/3 of the utf8 overhead in test
1623      * suite.  (That is, only 7-8% overall over just a hash cache.  Still,
1624      * it's nothing to sniff at.)  Pity we usually come through at least
1625      * two function calls to get here...
1626      *
1627      * NB: this code assumes that swatches are never modified, once generated!
1628      */
1629 
1630     if (hv   == PL_last_swash_hv &&
1631 	klen == PL_last_swash_klen &&
1632 	(!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) )
1633     {
1634 	tmps = PL_last_swash_tmps;
1635 	slen = PL_last_swash_slen;
1636     }
1637     else {
1638 	/* Try our second-level swatch cache, kept in a hash. */
1639 	SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE);
1640 
1641 	/* If not cached, generate it via utf8::SWASHGET */
1642 	if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) {
1643 	    dSP;
1644 	    /* We use utf8n_to_uvuni() as we want an index into
1645 	       Unicode tables, not a native character number.
1646 	     */
1647 	    UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXLEN, 0,
1648 					   ckWARN(WARN_UTF8) ?
1649 					   0 : UTF8_ALLOW_ANY);
1650 	    SV *errsv_save;
1651 	    ENTER;
1652 	    SAVETMPS;
1653 	    save_re_context();
1654 	    PUSHSTACKi(PERLSI_MAGIC);
1655 	    PUSHMARK(SP);
1656 	    EXTEND(SP,3);
1657 	    PUSHs((SV*)sv);
1658 	    /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */
1659 	    PUSHs(sv_2mortal(newSViv((klen) ?
1660 				     (code_point & ~(needents - 1)) : 0)));
1661 	    PUSHs(sv_2mortal(newSViv(needents)));
1662 	    PUTBACK;
1663 	    errsv_save = newSVsv(ERRSV);
1664 	    if (call_method("SWASHGET", G_SCALAR))
1665 		retval = newSVsv(*PL_stack_sp--);
1666 	    else
1667 		retval = &PL_sv_undef;
1668 	    if (!SvTRUE(ERRSV))
1669 		sv_setsv(ERRSV, errsv_save);
1670 	    SvREFCNT_dec(errsv_save);
1671 	    POPSTACK;
1672 	    FREETMPS;
1673 	    LEAVE;
1674 	    if (PL_curcop == &PL_compiling)
1675 		PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK);
1676 
1677 	    svp = hv_store(hv, (char*)ptr, klen, retval, 0);
1678 
1679 	    if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || (slen << 3) < needents)
1680 		Perl_croak(aTHX_ "SWASHGET didn't return result of proper length");
1681 	}
1682 
1683 	PL_last_swash_hv = hv;
1684 	PL_last_swash_klen = klen;
1685 	PL_last_swash_tmps = tmps;
1686 	PL_last_swash_slen = slen;
1687 	if (klen)
1688 	    Copy(ptr, PL_last_swash_key, klen, U8);
1689     }
1690 
1691     switch ((int)((slen << 3) / needents)) {
1692     case 1:
1693 	bit = 1 << (off & 7);
1694 	off >>= 3;
1695 	return (tmps[off] & bit) != 0;
1696     case 8:
1697 	return tmps[off];
1698     case 16:
1699 	off <<= 1;
1700 	return (tmps[off] << 8) + tmps[off + 1] ;
1701     case 32:
1702 	off <<= 2;
1703 	return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ;
1704     }
1705     Perl_croak(aTHX_ "panic: swash_fetch");
1706     return 0;
1707 }
1708 
1709 
1710 /*
1711 =for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv
1712 
1713 Adds the UTF8 representation of the Native codepoint C<uv> to the end
1714 of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free
1715 bytes available. The return value is the pointer to the byte after the
1716 end of the new character. In other words,
1717 
1718     d = uvchr_to_utf8(d, uv);
1719 
1720 is the recommended wide native character-aware way of saying
1721 
1722     *(d++) = uv;
1723 
1724 =cut
1725 */
1726 
1727 /* On ASCII machines this is normally a macro but we want a
1728    real function in case XS code wants it
1729 */
1730 #undef Perl_uvchr_to_utf8
1731 U8 *
1732 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv)
1733 {
1734     return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0);
1735 }
1736 
1737 U8 *
1738 Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags)
1739 {
1740     return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags);
1741 }
1742 
1743 /*
1744 =for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags
1745 
1746 Returns the native character value of the first character in the string C<s>
1747 which is assumed to be in UTF8 encoding; C<retlen> will be set to the
1748 length, in bytes, of that character.
1749 
1750 Allows length and flags to be passed to low level routine.
1751 
1752 =cut
1753 */
1754 /* On ASCII machines this is normally a macro but we want
1755    a real function in case XS code wants it
1756 */
1757 #undef Perl_utf8n_to_uvchr
1758 UV
1759 Perl_utf8n_to_uvchr(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags)
1760 {
1761     UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags);
1762     return UNI_TO_NATIVE(uv);
1763 }
1764 
1765 /*
1766 =for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags
1767 
1768 Build to the scalar dsv a displayable version of the string spv,
1769 length len, the displayable version being at most pvlim bytes long
1770 (if longer, the rest is truncated and "..." will be appended).
1771 
1772 The flags argument can have UNI_DISPLAY_ISPRINT set to display
1773 isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH
1774 to display the \\[nrfta\\] as the backslashed versions (like '\n')
1775 (UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\).
1776 UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both
1777 UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on.
1778 
1779 The pointer to the PV of the dsv is returned.
1780 
1781 =cut */
1782 char *
1783 Perl_pv_uni_display(pTHX_ SV *dsv, U8 *spv, STRLEN len, STRLEN pvlim, UV flags)
1784 {
1785     int truncated = 0;
1786     char *s, *e;
1787 
1788     sv_setpvn(dsv, "", 0);
1789     for (s = (char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) {
1790 	 UV u;
1791 	 bool ok = FALSE;
1792 
1793 	 if (pvlim && SvCUR(dsv) >= pvlim) {
1794 	      truncated++;
1795 	      break;
1796 	 }
1797 	 u = utf8_to_uvchr((U8*)s, 0);
1798 	 if (u < 256) {
1799 	     if (!ok && (flags & UNI_DISPLAY_BACKSLASH)) {
1800 	         switch (u & 0xFF) {
1801 		 case '\n':
1802 		     Perl_sv_catpvf(aTHX_ dsv, "\\n"); ok = TRUE; break;
1803 		 case '\r':
1804 		     Perl_sv_catpvf(aTHX_ dsv, "\\r"); ok = TRUE; break;
1805 		 case '\t':
1806 		     Perl_sv_catpvf(aTHX_ dsv, "\\t"); ok = TRUE; break;
1807 		 case '\f':
1808 		     Perl_sv_catpvf(aTHX_ dsv, "\\f"); ok = TRUE; break;
1809 		 case '\a':
1810 		     Perl_sv_catpvf(aTHX_ dsv, "\\a"); ok = TRUE; break;
1811 		 case '\\':
1812 		     Perl_sv_catpvf(aTHX_ dsv, "\\\\" ); ok = TRUE; break;
1813 		 default: break;
1814 		 }
1815 	     }
1816 	     /* isPRINT() is the locale-blind version. */
1817 	     if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(u & 0xFF)) {
1818 	         Perl_sv_catpvf(aTHX_ dsv, "%c", (char)(u & 0xFF));
1819 		 ok = TRUE;
1820 	     }
1821 	 }
1822 	 if (!ok)
1823 	     Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u);
1824     }
1825     if (truncated)
1826 	 sv_catpvn(dsv, "...", 3);
1827 
1828     return SvPVX(dsv);
1829 }
1830 
1831 /*
1832 =for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags
1833 
1834 Build to the scalar dsv a displayable version of the scalar sv,
1835 the displayable version being at most pvlim bytes long
1836 (if longer, the rest is truncated and "..." will be appended).
1837 
1838 The flags argument is as in pv_uni_display().
1839 
1840 The pointer to the PV of the dsv is returned.
1841 
1842 =cut */
1843 char *
1844 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
1845 {
1846      return Perl_pv_uni_display(aTHX_ dsv, (U8*)SvPVX(ssv), SvCUR(ssv),
1847 				pvlim, flags);
1848 }
1849 
1850 /*
1851 =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
1852 
1853 Return true if the strings s1 and s2 differ case-insensitively, false
1854 if not (if they are equal case-insensitively).  If u1 is true, the
1855 string s1 is assumed to be in UTF-8-encoded Unicode.  If u2 is true,
1856 the string s2 is assumed to be in UTF-8-encoded Unicode.  If u1 or u2
1857 are false, the respective string is assumed to be in native 8-bit
1858 encoding.
1859 
1860 If the pe1 and pe2 are non-NULL, the scanning pointers will be copied
1861 in there (they will point at the beginning of the I<next> character).
1862 If the pointers behind pe1 or pe2 are non-NULL, they are the end
1863 pointers beyond which scanning will not continue under any
1864 circustances.  If the byte lengths l1 and l2 are non-zero, s1+l1 and
1865 s2+l2 will be used as goal end pointers that will also stop the scan,
1866 and which qualify towards defining a successful match: all the scans
1867 that define an explicit length must reach their goal pointers for
1868 a match to succeed).
1869 
1870 For case-insensitiveness, the "casefolding" of Unicode is used
1871 instead of upper/lowercasing both the characters, see
1872 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
1873 
1874 =cut */
1875 I32
1876 Perl_ibcmp_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2)
1877 {
1878      register U8 *p1  = (U8*)s1;
1879      register U8 *p2  = (U8*)s2;
1880      register U8 *e1 = 0, *f1 = 0, *q1 = 0;
1881      register U8 *e2 = 0, *f2 = 0, *q2 = 0;
1882      STRLEN n1 = 0, n2 = 0;
1883      U8 foldbuf1[UTF8_MAXLEN_FOLD+1];
1884      U8 foldbuf2[UTF8_MAXLEN_FOLD+1];
1885      U8 natbuf[1+1];
1886      STRLEN foldlen1, foldlen2;
1887      bool match;
1888 
1889      if (pe1)
1890 	  e1 = *(U8**)pe1;
1891      if (e1 == 0 || (l1 && l1 < (UV)(e1 - (U8*)s1)))
1892 	  f1 = (U8*)s1 + l1;
1893      if (pe2)
1894 	  e2 = *(U8**)pe2;
1895      if (e2 == 0 || (l2 && l2 < (UV)(e2 - (U8*)s2)))
1896 	  f2 = (U8*)s2 + l2;
1897 
1898      if ((e1 == 0 && f1 == 0) || (e2 == 0 && f2 == 0) || (f1 == 0 && f2 == 0))
1899 	  return 1; /* mismatch; possible infinite loop or false positive */
1900 
1901      if (!u1 || !u2)
1902 	  natbuf[1] = 0; /* Need to terminate the buffer. */
1903 
1904      while ((e1 == 0 || p1 < e1) &&
1905 	    (f1 == 0 || p1 < f1) &&
1906 	    (e2 == 0 || p2 < e2) &&
1907 	    (f2 == 0 || p2 < f2)) {
1908 	  if (n1 == 0) {
1909 	       if (u1)
1910 		    to_utf8_fold(p1, foldbuf1, &foldlen1);
1911 	       else {
1912 		    natbuf[0] = *p1;
1913 		    to_utf8_fold(natbuf, foldbuf1, &foldlen1);
1914 	       }
1915 	       q1 = foldbuf1;
1916 	       n1 = foldlen1;
1917 	  }
1918 	  if (n2 == 0) {
1919 	       if (u2)
1920 		    to_utf8_fold(p2, foldbuf2, &foldlen2);
1921 	       else {
1922 		    natbuf[0] = *p2;
1923 		    to_utf8_fold(natbuf, foldbuf2, &foldlen2);
1924 	       }
1925 	       q2 = foldbuf2;
1926 	       n2 = foldlen2;
1927 	  }
1928 	  while (n1 && n2) {
1929 	       if ( UTF8SKIP(q1) != UTF8SKIP(q2) ||
1930 		   (UTF8SKIP(q1) == 1 && *q1 != *q2) ||
1931 		    memNE((char*)q1, (char*)q2, UTF8SKIP(q1)) )
1932 		   return 1; /* mismatch */
1933 	       n1 -= UTF8SKIP(q1);
1934 	       q1 += UTF8SKIP(q1);
1935 	       n2 -= UTF8SKIP(q2);
1936 	       q2 += UTF8SKIP(q2);
1937 	  }
1938 	  if (n1 == 0)
1939 	       p1 += u1 ? UTF8SKIP(p1) : 1;
1940 	  if (n2 == 0)
1941 	       p2 += u2 ? UTF8SKIP(p2) : 1;
1942 
1943      }
1944 
1945      /* A match is defined by all the scans that specified
1946       * an explicit length reaching their final goals. */
1947      match = (f1 == 0 || p1 == f1) && (f2 == 0 || p2 == f2);
1948 
1949      if (match) {
1950 	  if (pe1)
1951 	       *pe1 = (char*)p1;
1952 	  if (pe2)
1953 	       *pe2 = (char*)p2;
1954      }
1955 
1956      return match ? 0 : 1; /* 0 match, 1 mismatch */
1957 }
1958 
1959