1 /*    inline.h
2  *
3  *    Copyright (C) 2012 by Larry Wall and others
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  *    This file contains tables and code adapted from
9  *    http://bjoern.hoehrmann.de/utf-8/decoder/dfa/, which requires this
10  *    copyright notice:
11 
12 Copyright (c) 2008-2009 Bjoern Hoehrmann <bjoern@hoehrmann.de>
13 
14 Permission is hereby granted, free of charge, to any person obtaining a copy of
15 this software and associated documentation files (the "Software"), to deal in
16 the Software without restriction, including without limitation the rights to
17 use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
18 of the Software, and to permit persons to whom the Software is furnished to do
19 so, subject to the following conditions:
20 
21 The above copyright notice and this permission notice shall be included in all
22 copies or substantial portions of the Software.
23 
24 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
25 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
26 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
27 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
28 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
29 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
30 SOFTWARE.
31 
32  *
33  * This file is a home for static inline functions that cannot go in other
34  * header files, because they depend on proto.h (included after most other
35  * headers) or struct definitions.
36  *
37  * Each section names the header file that the functions "belong" to.
38  */
39 
40 /* ------------------------------- av.h ------------------------------- */
41 
42 PERL_STATIC_INLINE SSize_t
S_av_top_index(pTHX_ AV * av)43 S_av_top_index(pTHX_ AV *av)
44 {
45     PERL_ARGS_ASSERT_AV_TOP_INDEX;
46     assert(SvTYPE(av) == SVt_PVAV);
47 
48     return AvFILL(av);
49 }
50 
51 /* ------------------------------- cv.h ------------------------------- */
52 
53 PERL_STATIC_INLINE GV *
S_CvGV(pTHX_ CV * sv)54 S_CvGV(pTHX_ CV *sv)
55 {
56     return CvNAMED(sv)
57 	? Perl_cvgv_from_hek(aTHX_ sv)
58 	: ((XPVCV*)MUTABLE_PTR(SvANY(sv)))->xcv_gv_u.xcv_gv;
59 }
60 
61 PERL_STATIC_INLINE I32 *
S_CvDEPTHp(const CV * const sv)62 S_CvDEPTHp(const CV * const sv)
63 {
64     assert(SvTYPE(sv) == SVt_PVCV || SvTYPE(sv) == SVt_PVFM);
65     return &((XPVCV*)SvANY(sv))->xcv_depth;
66 }
67 
68 /*
69  CvPROTO returns the prototype as stored, which is not necessarily what
70  the interpreter should be using. Specifically, the interpreter assumes
71  that spaces have been stripped, which has been the case if the prototype
72  was added by toke.c, but is generally not the case if it was added elsewhere.
73  Since we can't enforce the spacelessness at assignment time, this routine
74  provides a temporary copy at parse time with spaces removed.
75  I<orig> is the start of the original buffer, I<len> is the length of the
76  prototype and will be updated when this returns.
77  */
78 
79 #ifdef PERL_CORE
80 PERL_STATIC_INLINE char *
S_strip_spaces(pTHX_ const char * orig,STRLEN * const len)81 S_strip_spaces(pTHX_ const char * orig, STRLEN * const len)
82 {
83     SV * tmpsv;
84     char * tmps;
85     tmpsv = newSVpvn_flags(orig, *len, SVs_TEMP);
86     tmps = SvPVX(tmpsv);
87     while ((*len)--) {
88 	if (!isSPACE(*orig))
89 	    *tmps++ = *orig;
90 	orig++;
91     }
92     *tmps = '\0';
93     *len = tmps - SvPVX(tmpsv);
94 		return SvPVX(tmpsv);
95 }
96 #endif
97 
98 /* ------------------------------- mg.h ------------------------------- */
99 
100 #if defined(PERL_CORE) || defined(PERL_EXT)
101 /* assumes get-magic and stringification have already occurred */
102 PERL_STATIC_INLINE STRLEN
S_MgBYTEPOS(pTHX_ MAGIC * mg,SV * sv,const char * s,STRLEN len)103 S_MgBYTEPOS(pTHX_ MAGIC *mg, SV *sv, const char *s, STRLEN len)
104 {
105     assert(mg->mg_type == PERL_MAGIC_regex_global);
106     assert(mg->mg_len != -1);
107     if (mg->mg_flags & MGf_BYTES || !DO_UTF8(sv))
108 	return (STRLEN)mg->mg_len;
109     else {
110 	const STRLEN pos = (STRLEN)mg->mg_len;
111 	/* Without this check, we may read past the end of the buffer: */
112 	if (pos > sv_or_pv_len_utf8(sv, s, len)) return len+1;
113 	return sv_or_pv_pos_u2b(sv, s, pos, NULL);
114     }
115 }
116 #endif
117 
118 /* ------------------------------- pad.h ------------------------------ */
119 
120 #if defined(PERL_IN_PAD_C) || defined(PERL_IN_OP_C)
121 PERL_STATIC_INLINE bool
PadnameIN_SCOPE(const PADNAME * const pn,const U32 seq)122 PadnameIN_SCOPE(const PADNAME * const pn, const U32 seq)
123 {
124     /* is seq within the range _LOW to _HIGH ?
125      * This is complicated by the fact that PL_cop_seqmax
126      * may have wrapped around at some point */
127     if (COP_SEQ_RANGE_LOW(pn) == PERL_PADSEQ_INTRO)
128 	return FALSE; /* not yet introduced */
129 
130     if (COP_SEQ_RANGE_HIGH(pn) == PERL_PADSEQ_INTRO) {
131     /* in compiling scope */
132 	if (
133 	    (seq >  COP_SEQ_RANGE_LOW(pn))
134 	    ? (seq - COP_SEQ_RANGE_LOW(pn) < (U32_MAX >> 1))
135 	    : (COP_SEQ_RANGE_LOW(pn) - seq > (U32_MAX >> 1))
136 	)
137 	    return TRUE;
138     }
139     else if (
140 	(COP_SEQ_RANGE_LOW(pn) > COP_SEQ_RANGE_HIGH(pn))
141 	?
142 	    (  seq >  COP_SEQ_RANGE_LOW(pn)
143 	    || seq <= COP_SEQ_RANGE_HIGH(pn))
144 
145 	:    (  seq >  COP_SEQ_RANGE_LOW(pn)
146 	     && seq <= COP_SEQ_RANGE_HIGH(pn))
147     )
148 	return TRUE;
149     return FALSE;
150 }
151 #endif
152 
153 /* ------------------------------- pp.h ------------------------------- */
154 
155 PERL_STATIC_INLINE I32
S_TOPMARK(pTHX)156 S_TOPMARK(pTHX)
157 {
158     DEBUG_s(DEBUG_v(PerlIO_printf(Perl_debug_log,
159 				 "MARK top  %p %" IVdf "\n",
160 				  PL_markstack_ptr,
161 				  (IV)*PL_markstack_ptr)));
162     return *PL_markstack_ptr;
163 }
164 
165 PERL_STATIC_INLINE I32
S_POPMARK(pTHX)166 S_POPMARK(pTHX)
167 {
168     DEBUG_s(DEBUG_v(PerlIO_printf(Perl_debug_log,
169 				 "MARK pop  %p %" IVdf "\n",
170 				  (PL_markstack_ptr-1),
171 				  (IV)*(PL_markstack_ptr-1))));
172     assert((PL_markstack_ptr > PL_markstack) || !"MARK underflow");
173     return *PL_markstack_ptr--;
174 }
175 
176 /* ----------------------------- regexp.h ----------------------------- */
177 
178 PERL_STATIC_INLINE struct regexp *
S_ReANY(const REGEXP * const re)179 S_ReANY(const REGEXP * const re)
180 {
181     XPV* const p = (XPV*)SvANY(re);
182     assert(isREGEXP(re));
183     return SvTYPE(re) == SVt_PVLV ? p->xpv_len_u.xpvlenu_rx
184                                    : (struct regexp *)p;
185 }
186 
187 /* ------------------------------- sv.h ------------------------------- */
188 
189 PERL_STATIC_INLINE SV *
S_SvREFCNT_inc(SV * sv)190 S_SvREFCNT_inc(SV *sv)
191 {
192     if (LIKELY(sv != NULL))
193 	SvREFCNT(sv)++;
194     return sv;
195 }
196 PERL_STATIC_INLINE SV *
S_SvREFCNT_inc_NN(SV * sv)197 S_SvREFCNT_inc_NN(SV *sv)
198 {
199     SvREFCNT(sv)++;
200     return sv;
201 }
202 PERL_STATIC_INLINE void
S_SvREFCNT_inc_void(SV * sv)203 S_SvREFCNT_inc_void(SV *sv)
204 {
205     if (LIKELY(sv != NULL))
206 	SvREFCNT(sv)++;
207 }
208 PERL_STATIC_INLINE void
S_SvREFCNT_dec(pTHX_ SV * sv)209 S_SvREFCNT_dec(pTHX_ SV *sv)
210 {
211     if (LIKELY(sv != NULL)) {
212 	U32 rc = SvREFCNT(sv);
213 	if (LIKELY(rc > 1))
214 	    SvREFCNT(sv) = rc - 1;
215 	else
216 	    Perl_sv_free2(aTHX_ sv, rc);
217     }
218 }
219 
220 PERL_STATIC_INLINE void
S_SvREFCNT_dec_NN(pTHX_ SV * sv)221 S_SvREFCNT_dec_NN(pTHX_ SV *sv)
222 {
223     U32 rc = SvREFCNT(sv);
224     if (LIKELY(rc > 1))
225 	SvREFCNT(sv) = rc - 1;
226     else
227 	Perl_sv_free2(aTHX_ sv, rc);
228 }
229 
230 PERL_STATIC_INLINE void
SvAMAGIC_on(SV * sv)231 SvAMAGIC_on(SV *sv)
232 {
233     assert(SvROK(sv));
234     if (SvOBJECT(SvRV(sv))) HvAMAGIC_on(SvSTASH(SvRV(sv)));
235 }
236 PERL_STATIC_INLINE void
SvAMAGIC_off(SV * sv)237 SvAMAGIC_off(SV *sv)
238 {
239     if (SvROK(sv) && SvOBJECT(SvRV(sv)))
240 	HvAMAGIC_off(SvSTASH(SvRV(sv)));
241 }
242 
243 PERL_STATIC_INLINE U32
S_SvPADSTALE_on(SV * sv)244 S_SvPADSTALE_on(SV *sv)
245 {
246     assert(!(SvFLAGS(sv) & SVs_PADTMP));
247     return SvFLAGS(sv) |= SVs_PADSTALE;
248 }
249 PERL_STATIC_INLINE U32
S_SvPADSTALE_off(SV * sv)250 S_SvPADSTALE_off(SV *sv)
251 {
252     assert(!(SvFLAGS(sv) & SVs_PADTMP));
253     return SvFLAGS(sv) &= ~SVs_PADSTALE;
254 }
255 #if defined(PERL_CORE) || defined (PERL_EXT)
256 PERL_STATIC_INLINE STRLEN
S_sv_or_pv_pos_u2b(pTHX_ SV * sv,const char * pv,STRLEN pos,STRLEN * lenp)257 S_sv_or_pv_pos_u2b(pTHX_ SV *sv, const char *pv, STRLEN pos, STRLEN *lenp)
258 {
259     PERL_ARGS_ASSERT_SV_OR_PV_POS_U2B;
260     if (SvGAMAGIC(sv)) {
261 	U8 *hopped = utf8_hop((U8 *)pv, pos);
262 	if (lenp) *lenp = (STRLEN)(utf8_hop(hopped, *lenp) - hopped);
263 	return (STRLEN)(hopped - (U8 *)pv);
264     }
265     return sv_pos_u2b_flags(sv,pos,lenp,SV_CONST_RETURN);
266 }
267 #endif
268 
269 /* ------------------------------- handy.h ------------------------------- */
270 
271 /* saves machine code for a common noreturn idiom typically used in Newx*() */
272 GCC_DIAG_IGNORE_DECL(-Wunused-function);
273 static void
S_croak_memory_wrap(void)274 S_croak_memory_wrap(void)
275 {
276     Perl_croak_nocontext("%s",PL_memory_wrap);
277 }
278 GCC_DIAG_RESTORE_DECL;
279 
280 /* ------------------------------- utf8.h ------------------------------- */
281 
282 /*
283 =head1 Unicode Support
284 */
285 
286 PERL_STATIC_INLINE void
S_append_utf8_from_native_byte(const U8 byte,U8 ** dest)287 S_append_utf8_from_native_byte(const U8 byte, U8** dest)
288 {
289     /* Takes an input 'byte' (Latin1 or EBCDIC) and appends it to the UTF-8
290      * encoded string at '*dest', updating '*dest' to include it */
291 
292     PERL_ARGS_ASSERT_APPEND_UTF8_FROM_NATIVE_BYTE;
293 
294     if (NATIVE_BYTE_IS_INVARIANT(byte))
295         *((*dest)++) = byte;
296     else {
297         *((*dest)++) = UTF8_EIGHT_BIT_HI(byte);
298         *((*dest)++) = UTF8_EIGHT_BIT_LO(byte);
299     }
300 }
301 
302 /*
303 =for apidoc valid_utf8_to_uvchr
304 Like C<L</utf8_to_uvchr_buf>>, but should only be called when it is known that
305 the next character in the input UTF-8 string C<s> is well-formed (I<e.g.>,
306 it passes C<L</isUTF8_CHAR>>.  Surrogates, non-character code points, and
307 non-Unicode code points are allowed.
308 
309 =cut
310 
311  */
312 
313 PERL_STATIC_INLINE UV
Perl_valid_utf8_to_uvchr(const U8 * s,STRLEN * retlen)314 Perl_valid_utf8_to_uvchr(const U8 *s, STRLEN *retlen)
315 {
316     const UV expectlen = UTF8SKIP(s);
317     const U8* send = s + expectlen;
318     UV uv = *s;
319 
320     PERL_ARGS_ASSERT_VALID_UTF8_TO_UVCHR;
321 
322     if (retlen) {
323         *retlen = expectlen;
324     }
325 
326     /* An invariant is trivially returned */
327     if (expectlen == 1) {
328 	return uv;
329     }
330 
331     /* Remove the leading bits that indicate the number of bytes, leaving just
332      * the bits that are part of the value */
333     uv = NATIVE_UTF8_TO_I8(uv) & UTF_START_MASK(expectlen);
334 
335     /* Now, loop through the remaining bytes, accumulating each into the
336      * working total as we go.  (I khw tried unrolling the loop for up to 4
337      * bytes, but there was no performance improvement) */
338     for (++s; s < send; s++) {
339         uv = UTF8_ACCUMULATE(uv, *s);
340     }
341 
342     return UNI_TO_NATIVE(uv);
343 
344 }
345 
346 /*
347 =for apidoc is_utf8_invariant_string
348 
349 Returns TRUE if the first C<len> bytes of the string C<s> are the same
350 regardless of the UTF-8 encoding of the string (or UTF-EBCDIC encoding on
351 EBCDIC machines); otherwise it returns FALSE.  That is, it returns TRUE if they
352 are UTF-8 invariant.  On ASCII-ish machines, all the ASCII characters and only
353 the ASCII characters fit this definition.  On EBCDIC machines, the ASCII-range
354 characters are invariant, but so also are the C1 controls.
355 
356 If C<len> is 0, it will be calculated using C<strlen(s)>, (which means if you
357 use this option, that C<s> can't have embedded C<NUL> characters and has to
358 have a terminating C<NUL> byte).
359 
360 See also
361 C<L</is_utf8_string>>,
362 C<L</is_utf8_string_flags>>,
363 C<L</is_utf8_string_loc>>,
364 C<L</is_utf8_string_loc_flags>>,
365 C<L</is_utf8_string_loclen>>,
366 C<L</is_utf8_string_loclen_flags>>,
367 C<L</is_utf8_fixed_width_buf_flags>>,
368 C<L</is_utf8_fixed_width_buf_loc_flags>>,
369 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
370 C<L</is_strict_utf8_string>>,
371 C<L</is_strict_utf8_string_loc>>,
372 C<L</is_strict_utf8_string_loclen>>,
373 C<L</is_c9strict_utf8_string>>,
374 C<L</is_c9strict_utf8_string_loc>>,
375 and
376 C<L</is_c9strict_utf8_string_loclen>>.
377 
378 =cut
379 
380 */
381 
382 #define is_utf8_invariant_string(s, len)                                    \
383                                 is_utf8_invariant_string_loc(s, len, NULL)
384 
385 /*
386 =for apidoc is_utf8_invariant_string_loc
387 
388 Like C<L</is_utf8_invariant_string>> but upon failure, stores the location of
389 the first UTF-8 variant character in the C<ep> pointer; if all characters are
390 UTF-8 invariant, this function does not change the contents of C<*ep>.
391 
392 =cut
393 
394 */
395 
396 PERL_STATIC_INLINE bool
S_is_utf8_invariant_string_loc(const U8 * const s,STRLEN len,const U8 ** ep)397 S_is_utf8_invariant_string_loc(const U8* const s, STRLEN len, const U8 ** ep)
398 {
399     const U8* send;
400     const U8* x = s;
401 
402     PERL_ARGS_ASSERT_IS_UTF8_INVARIANT_STRING_LOC;
403 
404     if (len == 0) {
405         len = strlen((const char *)s);
406     }
407 
408     send = s + len;
409 
410 /* This looks like 0x010101... */
411 #  define PERL_COUNT_MULTIPLIER   (~ (UINTMAX_C(0)) / 0xFF)
412 
413 /* This looks like 0x808080... */
414 #  define PERL_VARIANTS_WORD_MASK (PERL_COUNT_MULTIPLIER * 0x80)
415 #  define PERL_WORDSIZE            sizeof(PERL_UINTMAX_T)
416 #  define PERL_WORD_BOUNDARY_MASK (PERL_WORDSIZE - 1)
417 
418 /* Evaluates to 0 if 'x' is at a word boundary; otherwise evaluates to 1, by
419  * or'ing together the lowest bits of 'x'.  Hopefully the final term gets
420  * optimized out completely on a 32-bit system, and its mask gets optimized out
421  * on a 64-bit system */
422 #  define PERL_IS_SUBWORD_ADDR(x) (1 & (       PTR2nat(x)                     \
423                                       |   (  PTR2nat(x) >> 1)                 \
424                                       | ( ( (PTR2nat(x)                       \
425                                            & PERL_WORD_BOUNDARY_MASK) >> 2))))
426 
427 #ifndef EBCDIC
428 
429     /* Do the word-at-a-time iff there is at least one usable full word.  That
430      * means that after advancing to a word boundary, there still is at least a
431      * full word left.  The number of bytes needed to advance is 'wordsize -
432      * offset' unless offset is 0. */
433     if ((STRLEN) (send - x) >= PERL_WORDSIZE
434 
435                             /* This term is wordsize if subword; 0 if not */
436                           + PERL_WORDSIZE * PERL_IS_SUBWORD_ADDR(x)
437 
438                             /* 'offset' */
439                           - (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK))
440     {
441 
442         /* Process per-byte until reach word boundary.  XXX This loop could be
443          * eliminated if we knew that this platform had fast unaligned reads */
444         while (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK) {
445             if (! UTF8_IS_INVARIANT(*x)) {
446                 if (ep) {
447                     *ep = x;
448                 }
449 
450                 return FALSE;
451             }
452             x++;
453         }
454 
455         /* Here, we know we have at least one full word to process.  Process
456          * per-word as long as we have at least a full word left */
457         do {
458             if ((* (PERL_UINTMAX_T *) x) & PERL_VARIANTS_WORD_MASK)  {
459 
460                 /* Found a variant.  Just return if caller doesn't want its
461                  * exact position */
462                 if (! ep) {
463                     return FALSE;
464                 }
465 
466 #  if   BYTEORDER == 0x1234 || BYTEORDER == 0x12345678    \
467      || BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
468 
469                 *ep = x + _variant_byte_number(* (PERL_UINTMAX_T *) x);
470                 assert(*ep >= s && *ep < send);
471 
472                 return FALSE;
473 
474 #  else   /* If weird byte order, drop into next loop to do byte-at-a-time
475            checks. */
476 
477                 break;
478 #  endif
479             }
480 
481             x += PERL_WORDSIZE;
482 
483         } while (x + PERL_WORDSIZE <= send);
484     }
485 
486 #endif      /* End of ! EBCDIC */
487 
488     /* Process per-byte */
489     while (x < send) {
490 	if (! UTF8_IS_INVARIANT(*x)) {
491             if (ep) {
492                 *ep = x;
493             }
494 
495             return FALSE;
496         }
497 
498         x++;
499     }
500 
501     return TRUE;
502 }
503 
504 #ifndef EBCDIC
505 
506 PERL_STATIC_INLINE unsigned int
S__variant_byte_number(PERL_UINTMAX_T word)507 S__variant_byte_number(PERL_UINTMAX_T word)
508 {
509 
510     /* This returns the position in a word (0..7) of the first variant byte in
511      * it.  This is a helper function.  Note that there are no branches */
512 
513     assert(word);
514 
515     /* Get just the msb bits of each byte */
516     word &= PERL_VARIANTS_WORD_MASK;
517 
518 #  ifdef USING_MSVC6    /* VC6 has some issues with the normal code, and the
519                            easiest thing is to hide that from the callers */
520     {
521         unsigned int i;
522         const U8 * s = (U8 *) &word;
523         dTHX;
524 
525         for (i = 0; i < sizeof(word); i++ ) {
526             if (s[i]) {
527                 return i;
528             }
529         }
530 
531         Perl_croak(aTHX_ "panic: %s: %d: unexpected zero word\n",
532                                  __FILE__, __LINE__);
533     }
534 
535 #  elif BYTEORDER == 0x1234 || BYTEORDER == 0x12345678
536 
537     /* Bytes are stored like
538      *  Byte8 ... Byte2 Byte1
539      *  63..56...15...8 7...0
540      *
541      *  Isolate the lsb;
542      * https://stackoverflow.com/questions/757059/position-of-least-significant-bit-that-is-set
543      *
544      * The word will look this this, with a rightmost set bit in position 's':
545      * ('x's are don't cares)
546      *      s
547      *  x..x100..0
548      *  x..xx10..0      Right shift (rightmost 0 is shifted off)
549      *  x..xx01..1      Subtract 1, turns all the trailing zeros into 1's and
550      *                  the 1 just to their left into a 0; the remainder is
551      *                  untouched
552      *  0..0011..1      The xor with the original, x..xx10..0, clears that
553      *                  remainder, sets the bottom to all 1
554      *  0..0100..0      Add 1 to clear the word except for the bit in 's'
555      *
556      * Another method is to do 'word &= -word'; but it generates a compiler
557      * message on some platforms about taking the negative of an unsigned */
558 
559     word >>= 1;
560     word = 1 + (word ^ (word - 1));
561 
562 #  elif BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
563 
564     /* Bytes are stored like
565      *  Byte1 Byte2  ... Byte8
566      * 63..56 55..47 ... 7...0
567      *
568      * Isolate the msb; http://codeforces.com/blog/entry/10330
569      *
570      * Only the most significant set bit matters.  Or'ing word with its right
571      * shift of 1 makes that bit and the next one to its right both 1.  Then
572      * right shifting by 2 makes for 4 1-bits in a row. ...  We end with the
573      * msb and all to the right being 1. */
574     word |= word >>  1;
575     word |= word >>  2;
576     word |= word >>  4;
577     word |= word >>  8;
578     word |= word >> 16;
579     word |= word >> 32;  /* This should get optimized out on 32-bit systems. */
580 
581     /* Then subtracting the right shift by 1 clears all but the left-most of
582      * the 1 bits, which is our desired result */
583     word -= (word >> 1);
584 
585 #  else
586 #    error Unexpected byte order
587 #  endif
588 
589     /* Here 'word' has a single bit set: the  msb of the first byte in which it
590      * is set.  Calculate that position in the word.  We can use this
591      * specialized solution: https://stackoverflow.com/a/32339674/1626653,
592      * assumes an 8-bit byte.  (On a 32-bit machine, the larger numbers should
593      * just get shifted off at compile time) */
594     word = (word >> 7) * ((UINTMAX_C( 7) << 56) | (UINTMAX_C(15) << 48)
595                         | (UINTMAX_C(23) << 40) | (UINTMAX_C(31) << 32)
596                         |           (39 <<  24) |           (47 <<  16)
597                         |           (55 <<   8) |           (63 <<   0));
598     word >>= PERL_WORDSIZE * 7; /* >> by either 56 or 24 */
599 
600     /* Here, word contains the position 7..63 of that bit.  Convert to 0..7 */
601     word = ((word + 1) >> 3) - 1;
602 
603 #  if BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
604 
605     /* And invert the result */
606     word = CHARBITS - word - 1;
607 
608 #  endif
609 
610     return (unsigned int) word;
611 }
612 
613 #endif
614 #if defined(PERL_CORE) || defined(PERL_EXT)
615 
616 /*
617 =for apidoc variant_under_utf8_count
618 
619 This function looks at the sequence of bytes between C<s> and C<e>, which are
620 assumed to be encoded in ASCII/Latin1, and returns how many of them would
621 change should the string be translated into UTF-8.  Due to the nature of UTF-8,
622 each of these would occupy two bytes instead of the single one in the input
623 string.  Thus, this function returns the precise number of bytes the string
624 would expand by when translated to UTF-8.
625 
626 Unlike most of the other functions that have C<utf8> in their name, the input
627 to this function is NOT a UTF-8-encoded string.  The function name is slightly
628 I<odd> to emphasize this.
629 
630 This function is internal to Perl because khw thinks that any XS code that
631 would want this is probably operating too close to the internals.  Presenting a
632 valid use case could change that.
633 
634 See also
635 C<L<perlapi/is_utf8_invariant_string>>
636 and
637 C<L<perlapi/is_utf8_invariant_string_loc>>,
638 
639 =cut
640 
641 */
642 
643 PERL_STATIC_INLINE Size_t
S_variant_under_utf8_count(const U8 * const s,const U8 * const e)644 S_variant_under_utf8_count(const U8* const s, const U8* const e)
645 {
646     const U8* x = s;
647     Size_t count = 0;
648 
649     PERL_ARGS_ASSERT_VARIANT_UNDER_UTF8_COUNT;
650 
651 #  ifndef EBCDIC
652 
653     /* Test if the string is long enough to use word-at-a-time.  (Logic is the
654      * same as for is_utf8_invariant_string()) */
655     if ((STRLEN) (e - x) >= PERL_WORDSIZE
656                           + PERL_WORDSIZE * PERL_IS_SUBWORD_ADDR(x)
657                           - (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK))
658     {
659 
660         /* Process per-byte until reach word boundary.  XXX This loop could be
661          * eliminated if we knew that this platform had fast unaligned reads */
662         while (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK) {
663             count += ! UTF8_IS_INVARIANT(*x++);
664         }
665 
666         /* Process per-word as long as we have at least a full word left */
667         do {    /* Commit 03c1e4ab1d6ee9062fb3f94b0ba31db6698724b1 contains an
668                    explanation of how this works */
669             PERL_UINTMAX_T increment
670                 = ((((* (PERL_UINTMAX_T *) x) & PERL_VARIANTS_WORD_MASK) >> 7)
671                       * PERL_COUNT_MULTIPLIER)
672                     >> ((PERL_WORDSIZE - 1) * CHARBITS);
673             count += (Size_t) increment;
674             x += PERL_WORDSIZE;
675         } while (x + PERL_WORDSIZE <= e);
676     }
677 
678 #  endif
679 
680     /* Process per-byte */
681     while (x < e) {
682 	if (! UTF8_IS_INVARIANT(*x)) {
683             count++;
684         }
685 
686         x++;
687     }
688 
689     return count;
690 }
691 
692 #endif
693 
694 #ifndef PERL_IN_REGEXEC_C   /* Keep  these around for that file */
695 #  undef PERL_WORDSIZE
696 #  undef PERL_COUNT_MULTIPLIER
697 #  undef PERL_WORD_BOUNDARY_MASK
698 #  undef PERL_VARIANTS_WORD_MASK
699 #endif
700 
701 /*
702 =for apidoc is_utf8_string
703 
704 Returns TRUE if the first C<len> bytes of string C<s> form a valid
705 Perl-extended-UTF-8 string; returns FALSE otherwise.  If C<len> is 0, it will
706 be calculated using C<strlen(s)> (which means if you use this option, that C<s>
707 can't have embedded C<NUL> characters and has to have a terminating C<NUL>
708 byte).  Note that all characters being ASCII constitute 'a valid UTF-8 string'.
709 
710 This function considers Perl's extended UTF-8 to be valid.  That means that
711 code points above Unicode, surrogates, and non-character code points are
712 considered valid by this function.  Use C<L</is_strict_utf8_string>>,
713 C<L</is_c9strict_utf8_string>>, or C<L</is_utf8_string_flags>> to restrict what
714 code points are considered valid.
715 
716 See also
717 C<L</is_utf8_invariant_string>>,
718 C<L</is_utf8_invariant_string_loc>>,
719 C<L</is_utf8_string_loc>>,
720 C<L</is_utf8_string_loclen>>,
721 C<L</is_utf8_fixed_width_buf_flags>>,
722 C<L</is_utf8_fixed_width_buf_loc_flags>>,
723 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
724 
725 =cut
726 */
727 
728 #define is_utf8_string(s, len)  is_utf8_string_loclen(s, len, NULL, NULL)
729 
730 #if defined(PERL_CORE) || defined (PERL_EXT)
731 
732 /*
733 =for apidoc is_utf8_non_invariant_string
734 
735 Returns TRUE if L<perlapi/is_utf8_invariant_string> returns FALSE for the first
736 C<len> bytes of the string C<s>, but they are, nonetheless, legal Perl-extended
737 UTF-8; otherwise returns FALSE.
738 
739 A TRUE return means that at least one code point represented by the sequence
740 either is a wide character not representable as a single byte, or the
741 representation differs depending on whether the sequence is encoded in UTF-8 or
742 not.
743 
744 See also
745 C<L<perlapi/is_utf8_invariant_string>>,
746 C<L<perlapi/is_utf8_string>>
747 
748 =cut
749 
750 This is commonly used to determine if a SV's UTF-8 flag should be turned on.
751 It generally needn't be if its string is entirely UTF-8 invariant, and it
752 shouldn't be if it otherwise contains invalid UTF-8.
753 
754 It is an internal function because khw thinks that XS code shouldn't be working
755 at this low a level.  A valid use case could change that.
756 
757 */
758 
759 PERL_STATIC_INLINE bool
S_is_utf8_non_invariant_string(const U8 * const s,STRLEN len)760 S_is_utf8_non_invariant_string(const U8* const s, STRLEN len)
761 {
762     const U8 * first_variant;
763 
764     PERL_ARGS_ASSERT_IS_UTF8_NON_INVARIANT_STRING;
765 
766     if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
767         return FALSE;
768     }
769 
770     return is_utf8_string(first_variant, len - (first_variant - s));
771 }
772 
773 #endif
774 
775 /*
776 =for apidoc is_strict_utf8_string
777 
778 Returns TRUE if the first C<len> bytes of string C<s> form a valid
779 UTF-8-encoded string that is fully interchangeable by any application using
780 Unicode rules; otherwise it returns FALSE.  If C<len> is 0, it will be
781 calculated using C<strlen(s)> (which means if you use this option, that C<s>
782 can't have embedded C<NUL> characters and has to have a terminating C<NUL>
783 byte).  Note that all characters being ASCII constitute 'a valid UTF-8 string'.
784 
785 This function returns FALSE for strings containing any
786 code points above the Unicode max of 0x10FFFF, surrogate code points, or
787 non-character code points.
788 
789 See also
790 C<L</is_utf8_invariant_string>>,
791 C<L</is_utf8_invariant_string_loc>>,
792 C<L</is_utf8_string>>,
793 C<L</is_utf8_string_flags>>,
794 C<L</is_utf8_string_loc>>,
795 C<L</is_utf8_string_loc_flags>>,
796 C<L</is_utf8_string_loclen>>,
797 C<L</is_utf8_string_loclen_flags>>,
798 C<L</is_utf8_fixed_width_buf_flags>>,
799 C<L</is_utf8_fixed_width_buf_loc_flags>>,
800 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
801 C<L</is_strict_utf8_string_loc>>,
802 C<L</is_strict_utf8_string_loclen>>,
803 C<L</is_c9strict_utf8_string>>,
804 C<L</is_c9strict_utf8_string_loc>>,
805 and
806 C<L</is_c9strict_utf8_string_loclen>>.
807 
808 =cut
809 */
810 
811 #define is_strict_utf8_string(s, len)  is_strict_utf8_string_loclen(s, len, NULL, NULL)
812 
813 /*
814 =for apidoc is_c9strict_utf8_string
815 
816 Returns TRUE if the first C<len> bytes of string C<s> form a valid
817 UTF-8-encoded string that conforms to
818 L<Unicode Corrigendum #9|http://www.unicode.org/versions/corrigendum9.html>;
819 otherwise it returns FALSE.  If C<len> is 0, it will be calculated using
820 C<strlen(s)> (which means if you use this option, that C<s> can't have embedded
821 C<NUL> characters and has to have a terminating C<NUL> byte).  Note that all
822 characters being ASCII constitute 'a valid UTF-8 string'.
823 
824 This function returns FALSE for strings containing any code points above the
825 Unicode max of 0x10FFFF or surrogate code points, but accepts non-character
826 code points per
827 L<Corrigendum #9|http://www.unicode.org/versions/corrigendum9.html>.
828 
829 See also
830 C<L</is_utf8_invariant_string>>,
831 C<L</is_utf8_invariant_string_loc>>,
832 C<L</is_utf8_string>>,
833 C<L</is_utf8_string_flags>>,
834 C<L</is_utf8_string_loc>>,
835 C<L</is_utf8_string_loc_flags>>,
836 C<L</is_utf8_string_loclen>>,
837 C<L</is_utf8_string_loclen_flags>>,
838 C<L</is_utf8_fixed_width_buf_flags>>,
839 C<L</is_utf8_fixed_width_buf_loc_flags>>,
840 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
841 C<L</is_strict_utf8_string>>,
842 C<L</is_strict_utf8_string_loc>>,
843 C<L</is_strict_utf8_string_loclen>>,
844 C<L</is_c9strict_utf8_string_loc>>,
845 and
846 C<L</is_c9strict_utf8_string_loclen>>.
847 
848 =cut
849 */
850 
851 #define is_c9strict_utf8_string(s, len)  is_c9strict_utf8_string_loclen(s, len, NULL, 0)
852 
853 /*
854 =for apidoc is_utf8_string_flags
855 
856 Returns TRUE if the first C<len> bytes of string C<s> form a valid
857 UTF-8 string, subject to the restrictions imposed by C<flags>;
858 returns FALSE otherwise.  If C<len> is 0, it will be calculated
859 using C<strlen(s)> (which means if you use this option, that C<s> can't have
860 embedded C<NUL> characters and has to have a terminating C<NUL> byte).  Note
861 that all characters being ASCII constitute 'a valid UTF-8 string'.
862 
863 If C<flags> is 0, this gives the same results as C<L</is_utf8_string>>; if
864 C<flags> is C<UTF8_DISALLOW_ILLEGAL_INTERCHANGE>, this gives the same results
865 as C<L</is_strict_utf8_string>>; and if C<flags> is
866 C<UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE>, this gives the same results as
867 C<L</is_c9strict_utf8_string>>.  Otherwise C<flags> may be any
868 combination of the C<UTF8_DISALLOW_I<foo>> flags understood by
869 C<L</utf8n_to_uvchr>>, with the same meanings.
870 
871 See also
872 C<L</is_utf8_invariant_string>>,
873 C<L</is_utf8_invariant_string_loc>>,
874 C<L</is_utf8_string>>,
875 C<L</is_utf8_string_loc>>,
876 C<L</is_utf8_string_loc_flags>>,
877 C<L</is_utf8_string_loclen>>,
878 C<L</is_utf8_string_loclen_flags>>,
879 C<L</is_utf8_fixed_width_buf_flags>>,
880 C<L</is_utf8_fixed_width_buf_loc_flags>>,
881 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
882 C<L</is_strict_utf8_string>>,
883 C<L</is_strict_utf8_string_loc>>,
884 C<L</is_strict_utf8_string_loclen>>,
885 C<L</is_c9strict_utf8_string>>,
886 C<L</is_c9strict_utf8_string_loc>>,
887 and
888 C<L</is_c9strict_utf8_string_loclen>>.
889 
890 =cut
891 */
892 
893 PERL_STATIC_INLINE bool
S_is_utf8_string_flags(const U8 * s,STRLEN len,const U32 flags)894 S_is_utf8_string_flags(const U8 *s, STRLEN len, const U32 flags)
895 {
896     const U8 * first_variant;
897 
898     PERL_ARGS_ASSERT_IS_UTF8_STRING_FLAGS;
899     assert(0 == (flags & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE
900                           |UTF8_DISALLOW_PERL_EXTENDED)));
901 
902     if (len == 0) {
903         len = strlen((const char *)s);
904     }
905 
906     if (flags == 0) {
907         return is_utf8_string(s, len);
908     }
909 
910     if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
911                                         == UTF8_DISALLOW_ILLEGAL_INTERCHANGE)
912     {
913         return is_strict_utf8_string(s, len);
914     }
915 
916     if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
917                                        == UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE)
918     {
919         return is_c9strict_utf8_string(s, len);
920     }
921 
922     if (! is_utf8_invariant_string_loc(s, len, &first_variant)) {
923         const U8* const send = s + len;
924         const U8* x = first_variant;
925 
926         while (x < send) {
927             STRLEN cur_len = isUTF8_CHAR_flags(x, send, flags);
928             if (UNLIKELY(! cur_len)) {
929                 return FALSE;
930             }
931             x += cur_len;
932         }
933     }
934 
935     return TRUE;
936 }
937 
938 /*
939 
940 =for apidoc is_utf8_string_loc
941 
942 Like C<L</is_utf8_string>> but stores the location of the failure (in the
943 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
944 "utf8ness success") in the C<ep> pointer.
945 
946 See also C<L</is_utf8_string_loclen>>.
947 
948 =cut
949 */
950 
951 #define is_utf8_string_loc(s, len, ep)  is_utf8_string_loclen(s, len, ep, 0)
952 
953 /*
954 
955 =for apidoc is_utf8_string_loclen
956 
957 Like C<L</is_utf8_string>> but stores the location of the failure (in the
958 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
959 "utf8ness success") in the C<ep> pointer, and the number of UTF-8
960 encoded characters in the C<el> pointer.
961 
962 See also C<L</is_utf8_string_loc>>.
963 
964 =cut
965 */
966 
967 PERL_STATIC_INLINE bool
Perl_is_utf8_string_loclen(const U8 * s,STRLEN len,const U8 ** ep,STRLEN * el)968 Perl_is_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
969 {
970     const U8 * first_variant;
971 
972     PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN;
973 
974     if (len == 0) {
975         len = strlen((const char *) s);
976     }
977 
978     if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
979         if (el)
980             *el = len;
981 
982         if (ep) {
983             *ep = s + len;
984         }
985 
986         return TRUE;
987     }
988 
989     {
990         const U8* const send = s + len;
991         const U8* x = first_variant;
992         STRLEN outlen = first_variant - s;
993 
994         while (x < send) {
995             const STRLEN cur_len = isUTF8_CHAR(x, send);
996             if (UNLIKELY(! cur_len)) {
997                 break;
998             }
999             x += cur_len;
1000             outlen++;
1001         }
1002 
1003         if (el)
1004             *el = outlen;
1005 
1006         if (ep) {
1007             *ep = x;
1008         }
1009 
1010         return (x == send);
1011     }
1012 }
1013 
1014 /*
1015 
1016 =for apidoc Am|STRLEN|isUTF8_CHAR|const U8 *s|const U8 *e
1017 
1018 Evaluates to non-zero if the first few bytes of the string starting at C<s> and
1019 looking no further than S<C<e - 1>> are well-formed UTF-8, as extended by Perl,
1020 that represents some code point; otherwise it evaluates to 0.  If non-zero, the
1021 value gives how many bytes starting at C<s> comprise the code point's
1022 representation.  Any bytes remaining before C<e>, but beyond the ones needed to
1023 form the first code point in C<s>, are not examined.
1024 
1025 The code point can be any that will fit in an IV on this machine, using Perl's
1026 extension to official UTF-8 to represent those higher than the Unicode maximum
1027 of 0x10FFFF.  That means that this macro is used to efficiently decide if the
1028 next few bytes in C<s> is legal UTF-8 for a single character.
1029 
1030 Use C<L</isSTRICT_UTF8_CHAR>> to restrict the acceptable code points to those
1031 defined by Unicode to be fully interchangeable across applications;
1032 C<L</isC9_STRICT_UTF8_CHAR>> to use the L<Unicode Corrigendum
1033 #9|http://www.unicode.org/versions/corrigendum9.html> definition of allowable
1034 code points; and C<L</isUTF8_CHAR_flags>> for a more customized definition.
1035 
1036 Use C<L</is_utf8_string>>, C<L</is_utf8_string_loc>>, and
1037 C<L</is_utf8_string_loclen>> to check entire strings.
1038 
1039 Note also that a UTF-8 "invariant" character (i.e. ASCII on non-EBCDIC
1040 machines) is a valid UTF-8 character.
1041 
1042 =cut
1043 
1044 This uses an adaptation of the table and algorithm given in
1045 http://bjoern.hoehrmann.de/utf-8/decoder/dfa/, which provides comprehensive
1046 documentation of the original version.  A copyright notice for the original
1047 version is given at the beginning of this file.  The Perl adapation is
1048 documented at the definition of PL_extended_utf8_dfa_tab[].
1049 
1050 */
1051 
1052 PERL_STATIC_INLINE Size_t
S_isUTF8_CHAR(const U8 * const s0,const U8 * const e)1053 S_isUTF8_CHAR(const U8 * const s0, const U8 * const e)
1054 {
1055     const U8 * s = s0;
1056     UV state = 0;
1057 
1058     PERL_ARGS_ASSERT_ISUTF8_CHAR;
1059 
1060     /* This dfa is fast.  If it accepts the input, it was for a well-formed,
1061      * code point, which can be returned immediately.  Otherwise, it is either
1062      * malformed, or for the start byte FF which the dfa doesn't handle (except
1063      * on 32-bit ASCII platforms where it trivially is an error).  Call a
1064      * helper function for the other platforms. */
1065 
1066     while (s < e && LIKELY(state != 1)) {
1067         state = PL_extended_utf8_dfa_tab[256
1068                                          + state
1069                                          + PL_extended_utf8_dfa_tab[*s]];
1070         if (state != 0) {
1071             s++;
1072             continue;
1073         }
1074 
1075         return s - s0 + 1;
1076     }
1077 
1078 #if defined(UV_IS_QUAD) || defined(EBCDIC)
1079 
1080     if (NATIVE_UTF8_TO_I8(*s0) == 0xFF && e - s0 >= UTF8_MAXBYTES) {
1081        return _is_utf8_char_helper(s0, e, 0);
1082     }
1083 
1084 #endif
1085 
1086     return 0;
1087 }
1088 
1089 /*
1090 
1091 =for apidoc isSTRICT_UTF8_CHAR
1092 
1093 Evaluates to non-zero if the first few bytes of the string starting at C<s> and
1094 looking no further than S<C<e - 1>> are well-formed UTF-8 that represents some
1095 Unicode code point completely acceptable for open interchange between all
1096 applications; otherwise it evaluates to 0.  If non-zero, the value gives how
1097 many bytes starting at C<s> comprise the code point's representation.  Any
1098 bytes remaining before C<e>, but beyond the ones needed to form the first code
1099 point in C<s>, are not examined.
1100 
1101 The largest acceptable code point is the Unicode maximum 0x10FFFF, and must not
1102 be a surrogate nor a non-character code point.  Thus this excludes any code
1103 point from Perl's extended UTF-8.
1104 
1105 This is used to efficiently decide if the next few bytes in C<s> is
1106 legal Unicode-acceptable UTF-8 for a single character.
1107 
1108 Use C<L</isC9_STRICT_UTF8_CHAR>> to use the L<Unicode Corrigendum
1109 #9|http://www.unicode.org/versions/corrigendum9.html> definition of allowable
1110 code points; C<L</isUTF8_CHAR>> to check for Perl's extended UTF-8;
1111 and C<L</isUTF8_CHAR_flags>> for a more customized definition.
1112 
1113 Use C<L</is_strict_utf8_string>>, C<L</is_strict_utf8_string_loc>>, and
1114 C<L</is_strict_utf8_string_loclen>> to check entire strings.
1115 
1116 =cut
1117 
1118 This uses an adaptation of the tables and algorithm given in
1119 http://bjoern.hoehrmann.de/utf-8/decoder/dfa/, which provides comprehensive
1120 documentation of the original version.  A copyright notice for the original
1121 version is given at the beginning of this file.  The Perl adapation is
1122 documented at the definition of strict_extended_utf8_dfa_tab[].
1123 
1124 */
1125 
1126 PERL_STATIC_INLINE Size_t
S_isSTRICT_UTF8_CHAR(const U8 * const s0,const U8 * const e)1127 S_isSTRICT_UTF8_CHAR(const U8 * const s0, const U8 * const e)
1128 {
1129     const U8 * s = s0;
1130     UV state = 0;
1131 
1132     PERL_ARGS_ASSERT_ISSTRICT_UTF8_CHAR;
1133 
1134     while (s < e && LIKELY(state != 1)) {
1135         state = PL_strict_utf8_dfa_tab[256 + state + PL_strict_utf8_dfa_tab[*s]];
1136 
1137         if (state != 0) {
1138             s++;
1139             continue;
1140         }
1141 
1142         return s - s0 + 1;
1143     }
1144 
1145 #ifndef EBCDIC
1146 
1147     /* The dfa above drops out for certain Hanguls; handle them specially */
1148     if (is_HANGUL_ED_utf8_safe(s0, e)) {
1149         return 3;
1150     }
1151 
1152 #endif
1153 
1154     return 0;
1155 }
1156 
1157 /*
1158 
1159 =for apidoc Am|STRLEN|isC9_STRICT_UTF8_CHAR|const U8 *s|const U8 *e
1160 
1161 Evaluates to non-zero if the first few bytes of the string starting at C<s> and
1162 looking no further than S<C<e - 1>> are well-formed UTF-8 that represents some
1163 Unicode non-surrogate code point; otherwise it evaluates to 0.  If non-zero,
1164 the value gives how many bytes starting at C<s> comprise the code point's
1165 representation.  Any bytes remaining before C<e>, but beyond the ones needed to
1166 form the first code point in C<s>, are not examined.
1167 
1168 The largest acceptable code point is the Unicode maximum 0x10FFFF.  This
1169 differs from C<L</isSTRICT_UTF8_CHAR>> only in that it accepts non-character
1170 code points.  This corresponds to
1171 L<Unicode Corrigendum #9|http://www.unicode.org/versions/corrigendum9.html>.
1172 which said that non-character code points are merely discouraged rather than
1173 completely forbidden in open interchange.  See
1174 L<perlunicode/Noncharacter code points>.
1175 
1176 Use C<L</isUTF8_CHAR>> to check for Perl's extended UTF-8; and
1177 C<L</isUTF8_CHAR_flags>> for a more customized definition.
1178 
1179 Use C<L</is_c9strict_utf8_string>>, C<L</is_c9strict_utf8_string_loc>>, and
1180 C<L</is_c9strict_utf8_string_loclen>> to check entire strings.
1181 
1182 =cut
1183 
1184 This uses an adaptation of the tables and algorithm given in
1185 http://bjoern.hoehrmann.de/utf-8/decoder/dfa/, which provides comprehensive
1186 documentation of the original version.  A copyright notice for the original
1187 version is given at the beginning of this file.  The Perl adapation is
1188 documented at the definition of PL_c9_utf8_dfa_tab[].
1189 
1190 */
1191 
1192 PERL_STATIC_INLINE Size_t
S_isC9_STRICT_UTF8_CHAR(const U8 * const s0,const U8 * const e)1193 S_isC9_STRICT_UTF8_CHAR(const U8 * const s0, const U8 * const e)
1194 {
1195     const U8 * s = s0;
1196     UV state = 0;
1197 
1198     PERL_ARGS_ASSERT_ISC9_STRICT_UTF8_CHAR;
1199 
1200     while (s < e && LIKELY(state != 1)) {
1201         state = PL_c9_utf8_dfa_tab[256 + state + PL_c9_utf8_dfa_tab[*s]];
1202 
1203         if (state != 0) {
1204             s++;
1205             continue;
1206         }
1207 
1208         return s - s0 + 1;
1209     }
1210 
1211     return 0;
1212 }
1213 
1214 /*
1215 
1216 =for apidoc is_strict_utf8_string_loc
1217 
1218 Like C<L</is_strict_utf8_string>> but stores the location of the failure (in the
1219 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1220 "utf8ness success") in the C<ep> pointer.
1221 
1222 See also C<L</is_strict_utf8_string_loclen>>.
1223 
1224 =cut
1225 */
1226 
1227 #define is_strict_utf8_string_loc(s, len, ep)                               \
1228                                 is_strict_utf8_string_loclen(s, len, ep, 0)
1229 
1230 /*
1231 
1232 =for apidoc is_strict_utf8_string_loclen
1233 
1234 Like C<L</is_strict_utf8_string>> but stores the location of the failure (in the
1235 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1236 "utf8ness success") in the C<ep> pointer, and the number of UTF-8
1237 encoded characters in the C<el> pointer.
1238 
1239 See also C<L</is_strict_utf8_string_loc>>.
1240 
1241 =cut
1242 */
1243 
1244 PERL_STATIC_INLINE bool
S_is_strict_utf8_string_loclen(const U8 * s,STRLEN len,const U8 ** ep,STRLEN * el)1245 S_is_strict_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
1246 {
1247     const U8 * first_variant;
1248 
1249     PERL_ARGS_ASSERT_IS_STRICT_UTF8_STRING_LOCLEN;
1250 
1251     if (len == 0) {
1252         len = strlen((const char *) s);
1253     }
1254 
1255     if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
1256         if (el)
1257             *el = len;
1258 
1259         if (ep) {
1260             *ep = s + len;
1261         }
1262 
1263         return TRUE;
1264     }
1265 
1266     {
1267         const U8* const send = s + len;
1268         const U8* x = first_variant;
1269         STRLEN outlen = first_variant - s;
1270 
1271         while (x < send) {
1272             const STRLEN cur_len = isSTRICT_UTF8_CHAR(x, send);
1273             if (UNLIKELY(! cur_len)) {
1274                 break;
1275             }
1276             x += cur_len;
1277             outlen++;
1278         }
1279 
1280         if (el)
1281             *el = outlen;
1282 
1283         if (ep) {
1284             *ep = x;
1285         }
1286 
1287         return (x == send);
1288     }
1289 }
1290 
1291 /*
1292 
1293 =for apidoc is_c9strict_utf8_string_loc
1294 
1295 Like C<L</is_c9strict_utf8_string>> but stores the location of the failure (in
1296 the case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1297 "utf8ness success") in the C<ep> pointer.
1298 
1299 See also C<L</is_c9strict_utf8_string_loclen>>.
1300 
1301 =cut
1302 */
1303 
1304 #define is_c9strict_utf8_string_loc(s, len, ep)	                            \
1305                             is_c9strict_utf8_string_loclen(s, len, ep, 0)
1306 
1307 /*
1308 
1309 =for apidoc is_c9strict_utf8_string_loclen
1310 
1311 Like C<L</is_c9strict_utf8_string>> but stores the location of the failure (in
1312 the case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1313 "utf8ness success") in the C<ep> pointer, and the number of UTF-8 encoded
1314 characters in the C<el> pointer.
1315 
1316 See also C<L</is_c9strict_utf8_string_loc>>.
1317 
1318 =cut
1319 */
1320 
1321 PERL_STATIC_INLINE bool
S_is_c9strict_utf8_string_loclen(const U8 * s,STRLEN len,const U8 ** ep,STRLEN * el)1322 S_is_c9strict_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
1323 {
1324     const U8 * first_variant;
1325 
1326     PERL_ARGS_ASSERT_IS_C9STRICT_UTF8_STRING_LOCLEN;
1327 
1328     if (len == 0) {
1329         len = strlen((const char *) s);
1330     }
1331 
1332     if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
1333         if (el)
1334             *el = len;
1335 
1336         if (ep) {
1337             *ep = s + len;
1338         }
1339 
1340         return TRUE;
1341     }
1342 
1343     {
1344         const U8* const send = s + len;
1345         const U8* x = first_variant;
1346         STRLEN outlen = first_variant - s;
1347 
1348         while (x < send) {
1349             const STRLEN cur_len = isC9_STRICT_UTF8_CHAR(x, send);
1350             if (UNLIKELY(! cur_len)) {
1351                 break;
1352             }
1353             x += cur_len;
1354             outlen++;
1355         }
1356 
1357         if (el)
1358             *el = outlen;
1359 
1360         if (ep) {
1361             *ep = x;
1362         }
1363 
1364         return (x == send);
1365     }
1366 }
1367 
1368 /*
1369 
1370 =for apidoc is_utf8_string_loc_flags
1371 
1372 Like C<L</is_utf8_string_flags>> but stores the location of the failure (in the
1373 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1374 "utf8ness success") in the C<ep> pointer.
1375 
1376 See also C<L</is_utf8_string_loclen_flags>>.
1377 
1378 =cut
1379 */
1380 
1381 #define is_utf8_string_loc_flags(s, len, ep, flags)                         \
1382                         is_utf8_string_loclen_flags(s, len, ep, 0, flags)
1383 
1384 
1385 /* The above 3 actual functions could have been moved into the more general one
1386  * just below, and made #defines that call it with the right 'flags'.  They are
1387  * currently kept separate to increase their chances of getting inlined */
1388 
1389 /*
1390 
1391 =for apidoc is_utf8_string_loclen_flags
1392 
1393 Like C<L</is_utf8_string_flags>> but stores the location of the failure (in the
1394 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1395 "utf8ness success") in the C<ep> pointer, and the number of UTF-8
1396 encoded characters in the C<el> pointer.
1397 
1398 See also C<L</is_utf8_string_loc_flags>>.
1399 
1400 =cut
1401 */
1402 
1403 PERL_STATIC_INLINE bool
S_is_utf8_string_loclen_flags(const U8 * s,STRLEN len,const U8 ** ep,STRLEN * el,const U32 flags)1404 S_is_utf8_string_loclen_flags(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el, const U32 flags)
1405 {
1406     const U8 * first_variant;
1407 
1408     PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN_FLAGS;
1409     assert(0 == (flags & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE
1410                           |UTF8_DISALLOW_PERL_EXTENDED)));
1411 
1412     if (len == 0) {
1413         len = strlen((const char *) s);
1414     }
1415 
1416     if (flags == 0) {
1417         return is_utf8_string_loclen(s, len, ep, el);
1418     }
1419 
1420     if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
1421                                         == UTF8_DISALLOW_ILLEGAL_INTERCHANGE)
1422     {
1423         return is_strict_utf8_string_loclen(s, len, ep, el);
1424     }
1425 
1426     if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
1427                                     == UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE)
1428     {
1429         return is_c9strict_utf8_string_loclen(s, len, ep, el);
1430     }
1431 
1432     if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
1433         if (el)
1434             *el = len;
1435 
1436         if (ep) {
1437             *ep = s + len;
1438         }
1439 
1440         return TRUE;
1441     }
1442 
1443     {
1444         const U8* send = s + len;
1445         const U8* x = first_variant;
1446         STRLEN outlen = first_variant - s;
1447 
1448         while (x < send) {
1449             const STRLEN cur_len = isUTF8_CHAR_flags(x, send, flags);
1450             if (UNLIKELY(! cur_len)) {
1451                 break;
1452             }
1453             x += cur_len;
1454             outlen++;
1455         }
1456 
1457         if (el)
1458             *el = outlen;
1459 
1460         if (ep) {
1461             *ep = x;
1462         }
1463 
1464         return (x == send);
1465     }
1466 }
1467 
1468 /*
1469 =for apidoc utf8_distance
1470 
1471 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
1472 and C<b>.
1473 
1474 WARNING: use only if you *know* that the pointers point inside the
1475 same UTF-8 buffer.
1476 
1477 =cut
1478 */
1479 
1480 PERL_STATIC_INLINE IV
Perl_utf8_distance(pTHX_ const U8 * a,const U8 * b)1481 Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
1482 {
1483     PERL_ARGS_ASSERT_UTF8_DISTANCE;
1484 
1485     return (a < b) ? -1 * (IV) utf8_length(a, b) : (IV) utf8_length(b, a);
1486 }
1487 
1488 /*
1489 =for apidoc utf8_hop
1490 
1491 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
1492 forward or backward.
1493 
1494 WARNING: do not use the following unless you *know* C<off> is within
1495 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
1496 on the first byte of character or just after the last byte of a character.
1497 
1498 =cut
1499 */
1500 
1501 PERL_STATIC_INLINE U8 *
Perl_utf8_hop(const U8 * s,SSize_t off)1502 Perl_utf8_hop(const U8 *s, SSize_t off)
1503 {
1504     PERL_ARGS_ASSERT_UTF8_HOP;
1505 
1506     /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1507      * the bitops (especially ~) can create illegal UTF-8.
1508      * In other words: in Perl UTF-8 is not just for Unicode. */
1509 
1510     if (off >= 0) {
1511 	while (off--)
1512 	    s += UTF8SKIP(s);
1513     }
1514     else {
1515 	while (off++) {
1516 	    s--;
1517 	    while (UTF8_IS_CONTINUATION(*s))
1518 		s--;
1519 	}
1520     }
1521     GCC_DIAG_IGNORE(-Wcast-qual)
1522     return (U8 *)s;
1523     GCC_DIAG_RESTORE
1524 }
1525 
1526 /*
1527 =for apidoc utf8_hop_forward
1528 
1529 Return the UTF-8 pointer C<s> displaced by up to C<off> characters,
1530 forward.
1531 
1532 C<off> must be non-negative.
1533 
1534 C<s> must be before or equal to C<end>.
1535 
1536 When moving forward it will not move beyond C<end>.
1537 
1538 Will not exceed this limit even if the string is not valid "UTF-8".
1539 
1540 =cut
1541 */
1542 
1543 PERL_STATIC_INLINE U8 *
Perl_utf8_hop_forward(const U8 * s,SSize_t off,const U8 * end)1544 Perl_utf8_hop_forward(const U8 *s, SSize_t off, const U8 *end)
1545 {
1546     PERL_ARGS_ASSERT_UTF8_HOP_FORWARD;
1547 
1548     /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1549      * the bitops (especially ~) can create illegal UTF-8.
1550      * In other words: in Perl UTF-8 is not just for Unicode. */
1551 
1552     assert(s <= end);
1553     assert(off >= 0);
1554 
1555     while (off--) {
1556         STRLEN skip = UTF8SKIP(s);
1557         if ((STRLEN)(end - s) <= skip) {
1558             GCC_DIAG_IGNORE(-Wcast-qual)
1559             return (U8 *)end;
1560             GCC_DIAG_RESTORE
1561         }
1562         s += skip;
1563     }
1564 
1565     GCC_DIAG_IGNORE(-Wcast-qual)
1566     return (U8 *)s;
1567     GCC_DIAG_RESTORE
1568 }
1569 
1570 /*
1571 =for apidoc utf8_hop_back
1572 
1573 Return the UTF-8 pointer C<s> displaced by up to C<off> characters,
1574 backward.
1575 
1576 C<off> must be non-positive.
1577 
1578 C<s> must be after or equal to C<start>.
1579 
1580 When moving backward it will not move before C<start>.
1581 
1582 Will not exceed this limit even if the string is not valid "UTF-8".
1583 
1584 =cut
1585 */
1586 
1587 PERL_STATIC_INLINE U8 *
Perl_utf8_hop_back(const U8 * s,SSize_t off,const U8 * start)1588 Perl_utf8_hop_back(const U8 *s, SSize_t off, const U8 *start)
1589 {
1590     PERL_ARGS_ASSERT_UTF8_HOP_BACK;
1591 
1592     /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1593      * the bitops (especially ~) can create illegal UTF-8.
1594      * In other words: in Perl UTF-8 is not just for Unicode. */
1595 
1596     assert(start <= s);
1597     assert(off <= 0);
1598 
1599     while (off++ && s > start) {
1600         do {
1601             s--;
1602         } while (UTF8_IS_CONTINUATION(*s) && s > start);
1603     }
1604 
1605     GCC_DIAG_IGNORE(-Wcast-qual)
1606     return (U8 *)s;
1607     GCC_DIAG_RESTORE
1608 }
1609 
1610 /*
1611 =for apidoc utf8_hop_safe
1612 
1613 Return the UTF-8 pointer C<s> displaced by up to C<off> characters,
1614 either forward or backward.
1615 
1616 When moving backward it will not move before C<start>.
1617 
1618 When moving forward it will not move beyond C<end>.
1619 
1620 Will not exceed those limits even if the string is not valid "UTF-8".
1621 
1622 =cut
1623 */
1624 
1625 PERL_STATIC_INLINE U8 *
Perl_utf8_hop_safe(const U8 * s,SSize_t off,const U8 * start,const U8 * end)1626 Perl_utf8_hop_safe(const U8 *s, SSize_t off, const U8 *start, const U8 *end)
1627 {
1628     PERL_ARGS_ASSERT_UTF8_HOP_SAFE;
1629 
1630     /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1631      * the bitops (especially ~) can create illegal UTF-8.
1632      * In other words: in Perl UTF-8 is not just for Unicode. */
1633 
1634     assert(start <= s && s <= end);
1635 
1636     if (off >= 0) {
1637         return utf8_hop_forward(s, off, end);
1638     }
1639     else {
1640         return utf8_hop_back(s, off, start);
1641     }
1642 }
1643 
1644 /*
1645 
1646 =for apidoc is_utf8_valid_partial_char
1647 
1648 Returns 0 if the sequence of bytes starting at C<s> and looking no further than
1649 S<C<e - 1>> is the UTF-8 encoding, as extended by Perl, for one or more code
1650 points.  Otherwise, it returns 1 if there exists at least one non-empty
1651 sequence of bytes that when appended to sequence C<s>, starting at position
1652 C<e> causes the entire sequence to be the well-formed UTF-8 of some code point;
1653 otherwise returns 0.
1654 
1655 In other words this returns TRUE if C<s> points to a partial UTF-8-encoded code
1656 point.
1657 
1658 This is useful when a fixed-length buffer is being tested for being well-formed
1659 UTF-8, but the final few bytes in it don't comprise a full character; that is,
1660 it is split somewhere in the middle of the final code point's UTF-8
1661 representation.  (Presumably when the buffer is refreshed with the next chunk
1662 of data, the new first bytes will complete the partial code point.)   This
1663 function is used to verify that the final bytes in the current buffer are in
1664 fact the legal beginning of some code point, so that if they aren't, the
1665 failure can be signalled without having to wait for the next read.
1666 
1667 =cut
1668 */
1669 #define is_utf8_valid_partial_char(s, e)                                    \
1670                                 is_utf8_valid_partial_char_flags(s, e, 0)
1671 
1672 /*
1673 
1674 =for apidoc is_utf8_valid_partial_char_flags
1675 
1676 Like C<L</is_utf8_valid_partial_char>>, it returns a boolean giving whether
1677 or not the input is a valid UTF-8 encoded partial character, but it takes an
1678 extra parameter, C<flags>, which can further restrict which code points are
1679 considered valid.
1680 
1681 If C<flags> is 0, this behaves identically to
1682 C<L</is_utf8_valid_partial_char>>.  Otherwise C<flags> can be any combination
1683 of the C<UTF8_DISALLOW_I<foo>> flags accepted by C<L</utf8n_to_uvchr>>.  If
1684 there is any sequence of bytes that can complete the input partial character in
1685 such a way that a non-prohibited character is formed, the function returns
1686 TRUE; otherwise FALSE.  Non character code points cannot be determined based on
1687 partial character input.  But many  of the other possible excluded types can be
1688 determined from just the first one or two bytes.
1689 
1690 =cut
1691  */
1692 
1693 PERL_STATIC_INLINE bool
S_is_utf8_valid_partial_char_flags(const U8 * const s,const U8 * const e,const U32 flags)1694 S_is_utf8_valid_partial_char_flags(const U8 * const s, const U8 * const e, const U32 flags)
1695 {
1696     PERL_ARGS_ASSERT_IS_UTF8_VALID_PARTIAL_CHAR_FLAGS;
1697 
1698     assert(0 == (flags & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE
1699                           |UTF8_DISALLOW_PERL_EXTENDED)));
1700 
1701     if (s >= e || s + UTF8SKIP(s) <= e) {
1702         return FALSE;
1703     }
1704 
1705     return cBOOL(_is_utf8_char_helper(s, e, flags));
1706 }
1707 
1708 /*
1709 
1710 =for apidoc is_utf8_fixed_width_buf_flags
1711 
1712 Returns TRUE if the fixed-width buffer starting at C<s> with length C<len>
1713 is entirely valid UTF-8, subject to the restrictions given by C<flags>;
1714 otherwise it returns FALSE.
1715 
1716 If C<flags> is 0, any well-formed UTF-8, as extended by Perl, is accepted
1717 without restriction.  If the final few bytes of the buffer do not form a
1718 complete code point, this will return TRUE anyway, provided that
1719 C<L</is_utf8_valid_partial_char_flags>> returns TRUE for them.
1720 
1721 If C<flags> in non-zero, it can be any combination of the
1722 C<UTF8_DISALLOW_I<foo>> flags accepted by C<L</utf8n_to_uvchr>>, and with the
1723 same meanings.
1724 
1725 This function differs from C<L</is_utf8_string_flags>> only in that the latter
1726 returns FALSE if the final few bytes of the string don't form a complete code
1727 point.
1728 
1729 =cut
1730  */
1731 #define is_utf8_fixed_width_buf_flags(s, len, flags)                        \
1732                 is_utf8_fixed_width_buf_loclen_flags(s, len, 0, 0, flags)
1733 
1734 /*
1735 
1736 =for apidoc is_utf8_fixed_width_buf_loc_flags
1737 
1738 Like C<L</is_utf8_fixed_width_buf_flags>> but stores the location of the
1739 failure in the C<ep> pointer.  If the function returns TRUE, C<*ep> will point
1740 to the beginning of any partial character at the end of the buffer; if there is
1741 no partial character C<*ep> will contain C<s>+C<len>.
1742 
1743 See also C<L</is_utf8_fixed_width_buf_loclen_flags>>.
1744 
1745 =cut
1746 */
1747 
1748 #define is_utf8_fixed_width_buf_loc_flags(s, len, loc, flags)               \
1749                 is_utf8_fixed_width_buf_loclen_flags(s, len, loc, 0, flags)
1750 
1751 /*
1752 
1753 =for apidoc is_utf8_fixed_width_buf_loclen_flags
1754 
1755 Like C<L</is_utf8_fixed_width_buf_loc_flags>> but stores the number of
1756 complete, valid characters found in the C<el> pointer.
1757 
1758 =cut
1759 */
1760 
1761 PERL_STATIC_INLINE bool
S_is_utf8_fixed_width_buf_loclen_flags(const U8 * const s,STRLEN len,const U8 ** ep,STRLEN * el,const U32 flags)1762 S_is_utf8_fixed_width_buf_loclen_flags(const U8 * const s,
1763                                        STRLEN len,
1764                                        const U8 **ep,
1765                                        STRLEN *el,
1766                                        const U32 flags)
1767 {
1768     const U8 * maybe_partial;
1769 
1770     PERL_ARGS_ASSERT_IS_UTF8_FIXED_WIDTH_BUF_LOCLEN_FLAGS;
1771 
1772     if (! ep) {
1773         ep  = &maybe_partial;
1774     }
1775 
1776     /* If it's entirely valid, return that; otherwise see if the only error is
1777      * that the final few bytes are for a partial character */
1778     return    is_utf8_string_loclen_flags(s, len, ep, el, flags)
1779            || is_utf8_valid_partial_char_flags(*ep, s + len, flags);
1780 }
1781 
1782 PERL_STATIC_INLINE UV
S_utf8n_to_uvchr_msgs(const U8 * s,STRLEN curlen,STRLEN * retlen,const U32 flags,U32 * errors,AV ** msgs)1783 S_utf8n_to_uvchr_msgs(const U8 *s,
1784                       STRLEN curlen,
1785                       STRLEN *retlen,
1786                       const U32 flags,
1787                       U32 * errors,
1788                       AV ** msgs)
1789 {
1790     /* This is the inlined portion of utf8n_to_uvchr_msgs.  It handles the
1791      * simple cases, and, if necessary calls a helper function to deal with the
1792      * more complex ones.  Almost all well-formed non-problematic code points
1793      * are considered simple, so that it's unlikely that the helper function
1794      * will need to be called.
1795      *
1796      * This is an adaptation of the tables and algorithm given in
1797      * http://bjoern.hoehrmann.de/utf-8/decoder/dfa/, which provides
1798      * comprehensive documentation of the original version.  A copyright notice
1799      * for the original version is given at the beginning of this file.  The
1800      * Perl adapation is documented at the definition of PL_strict_utf8_dfa_tab[].
1801      */
1802 
1803     const U8 * const s0 = s;
1804     const U8 * send = s0 + curlen;
1805     UV uv = 0;      /* The 0 silences some stupid compilers */
1806     UV state = 0;
1807 
1808     PERL_ARGS_ASSERT_UTF8N_TO_UVCHR_MSGS;
1809 
1810     /* This dfa is fast.  If it accepts the input, it was for a well-formed,
1811      * non-problematic code point, which can be returned immediately.
1812      * Otherwise we call a helper function to figure out the more complicated
1813      * cases. */
1814 
1815     while (s < send && LIKELY(state != 1)) {
1816         UV type = PL_strict_utf8_dfa_tab[*s];
1817 
1818         uv = (state == 0)
1819              ?  ((0xff >> type) & NATIVE_UTF8_TO_I8(*s))
1820              : UTF8_ACCUMULATE(uv, *s);
1821         state = PL_strict_utf8_dfa_tab[256 + state + type];
1822 
1823         if (state != 0) {
1824             s++;
1825             continue;
1826         }
1827 
1828         if (retlen) {
1829             *retlen = s - s0 + 1;
1830         }
1831         if (errors) {
1832             *errors = 0;
1833         }
1834         if (msgs) {
1835             *msgs = NULL;
1836         }
1837 
1838         return uv;
1839     }
1840 
1841     /* Here is potentially problematic.  Use the full mechanism */
1842     return _utf8n_to_uvchr_msgs_helper(s0, curlen, retlen, flags, errors, msgs);
1843 }
1844 
1845 /* ------------------------------- perl.h ----------------------------- */
1846 
1847 /*
1848 =head1 Miscellaneous Functions
1849 
1850 =for apidoc AiR|bool|is_safe_syscall|const char *pv|STRLEN len|const char *what|const char *op_name
1851 
1852 Test that the given C<pv> doesn't contain any internal C<NUL> characters.
1853 If it does, set C<errno> to C<ENOENT>, optionally warn, and return FALSE.
1854 
1855 Return TRUE if the name is safe.
1856 
1857 Used by the C<IS_SAFE_SYSCALL()> macro.
1858 
1859 =cut
1860 */
1861 
1862 PERL_STATIC_INLINE bool
S_is_safe_syscall(pTHX_ const char * pv,STRLEN len,const char * what,const char * op_name)1863 S_is_safe_syscall(pTHX_ const char *pv, STRLEN len, const char *what, const char *op_name) {
1864     /* While the Windows CE API provides only UCS-16 (or UTF-16) APIs
1865      * perl itself uses xce*() functions which accept 8-bit strings.
1866      */
1867 
1868     PERL_ARGS_ASSERT_IS_SAFE_SYSCALL;
1869 
1870     if (len > 1) {
1871         char *null_at;
1872         if (UNLIKELY((null_at = (char *)memchr(pv, 0, len-1)) != NULL)) {
1873                 SETERRNO(ENOENT, LIB_INVARG);
1874                 Perl_ck_warner(aTHX_ packWARN(WARN_SYSCALLS),
1875                                    "Invalid \\0 character in %s for %s: %s\\0%s",
1876                                    what, op_name, pv, null_at+1);
1877                 return FALSE;
1878         }
1879     }
1880 
1881     return TRUE;
1882 }
1883 
1884 /*
1885 
1886 Return true if the supplied filename has a newline character
1887 immediately before the first (hopefully only) NUL.
1888 
1889 My original look at this incorrectly used the len from SvPV(), but
1890 that's incorrect, since we allow for a NUL in pv[len-1].
1891 
1892 So instead, strlen() and work from there.
1893 
1894 This allow for the user reading a filename, forgetting to chomp it,
1895 then calling:
1896 
1897   open my $foo, "$file\0";
1898 
1899 */
1900 
1901 #ifdef PERL_CORE
1902 
1903 PERL_STATIC_INLINE bool
S_should_warn_nl(const char * pv)1904 S_should_warn_nl(const char *pv) {
1905     STRLEN len;
1906 
1907     PERL_ARGS_ASSERT_SHOULD_WARN_NL;
1908 
1909     len = strlen(pv);
1910 
1911     return len > 0 && pv[len-1] == '\n';
1912 }
1913 
1914 #endif
1915 
1916 /* ------------------ pp.c, regcomp.c, toke.c, universal.c ------------ */
1917 
1918 #define MAX_CHARSET_NAME_LENGTH 2
1919 
1920 PERL_STATIC_INLINE const char *
get_regex_charset_name(const U32 flags,STRLEN * const lenp)1921 get_regex_charset_name(const U32 flags, STRLEN* const lenp)
1922 {
1923     /* Returns a string that corresponds to the name of the regex character set
1924      * given by 'flags', and *lenp is set the length of that string, which
1925      * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
1926 
1927     *lenp = 1;
1928     switch (get_regex_charset(flags)) {
1929         case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
1930         case REGEX_LOCALE_CHARSET:  return LOCALE_PAT_MODS;
1931         case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
1932 	case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
1933 	case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
1934 	    *lenp = 2;
1935 	    return ASCII_MORE_RESTRICT_PAT_MODS;
1936     }
1937     /* The NOT_REACHED; hides an assert() which has a rather complex
1938      * definition in perl.h. */
1939     NOT_REACHED; /* NOTREACHED */
1940     return "?";	    /* Unknown */
1941 }
1942 
1943 /*
1944 
1945 Return false if any get magic is on the SV other than taint magic.
1946 
1947 */
1948 
1949 PERL_STATIC_INLINE bool
S_sv_only_taint_gmagic(SV * sv)1950 S_sv_only_taint_gmagic(SV *sv) {
1951     MAGIC *mg = SvMAGIC(sv);
1952 
1953     PERL_ARGS_ASSERT_SV_ONLY_TAINT_GMAGIC;
1954 
1955     while (mg) {
1956         if (mg->mg_type != PERL_MAGIC_taint
1957             && !(mg->mg_flags & MGf_GSKIP)
1958             && mg->mg_virtual->svt_get) {
1959             return FALSE;
1960         }
1961         mg = mg->mg_moremagic;
1962     }
1963 
1964     return TRUE;
1965 }
1966 
1967 /* ------------------ cop.h ------------------------------------------- */
1968 
1969 
1970 /* Enter a block. Push a new base context and return its address. */
1971 
1972 PERL_STATIC_INLINE PERL_CONTEXT *
S_cx_pushblock(pTHX_ U8 type,U8 gimme,SV ** sp,I32 saveix)1973 S_cx_pushblock(pTHX_ U8 type, U8 gimme, SV** sp, I32 saveix)
1974 {
1975     PERL_CONTEXT * cx;
1976 
1977     PERL_ARGS_ASSERT_CX_PUSHBLOCK;
1978 
1979     CXINC;
1980     cx = CX_CUR();
1981     cx->cx_type        = type;
1982     cx->blk_gimme      = gimme;
1983     cx->blk_oldsaveix  = saveix;
1984     cx->blk_oldsp      = (I32)(sp - PL_stack_base);
1985     cx->blk_oldcop     = PL_curcop;
1986     cx->blk_oldmarksp  = (I32)(PL_markstack_ptr - PL_markstack);
1987     cx->blk_oldscopesp = PL_scopestack_ix;
1988     cx->blk_oldpm      = PL_curpm;
1989     cx->blk_old_tmpsfloor = PL_tmps_floor;
1990 
1991     PL_tmps_floor        = PL_tmps_ix;
1992     CX_DEBUG(cx, "PUSH");
1993     return cx;
1994 }
1995 
1996 
1997 /* Exit a block (RETURN and LAST). */
1998 
1999 PERL_STATIC_INLINE void
S_cx_popblock(pTHX_ PERL_CONTEXT * cx)2000 S_cx_popblock(pTHX_ PERL_CONTEXT *cx)
2001 {
2002     PERL_ARGS_ASSERT_CX_POPBLOCK;
2003 
2004     CX_DEBUG(cx, "POP");
2005     /* these 3 are common to cx_popblock and cx_topblock */
2006     PL_markstack_ptr = PL_markstack + cx->blk_oldmarksp;
2007     PL_scopestack_ix = cx->blk_oldscopesp;
2008     PL_curpm         = cx->blk_oldpm;
2009 
2010     /* LEAVE_SCOPE() should have made this true. /(?{})/ cheats
2011      * and leaves a CX entry lying around for repeated use, so
2012      * skip for multicall */                  \
2013     assert(   (CxTYPE(cx) == CXt_SUB && CxMULTICALL(cx))
2014             || PL_savestack_ix == cx->blk_oldsaveix);
2015     PL_curcop     = cx->blk_oldcop;
2016     PL_tmps_floor = cx->blk_old_tmpsfloor;
2017 }
2018 
2019 /* Continue a block elsewhere (e.g. NEXT, REDO, GOTO).
2020  * Whereas cx_popblock() restores the state to the point just before
2021  * cx_pushblock() was called,  cx_topblock() restores it to the point just
2022  * *after* cx_pushblock() was called. */
2023 
2024 PERL_STATIC_INLINE void
S_cx_topblock(pTHX_ PERL_CONTEXT * cx)2025 S_cx_topblock(pTHX_ PERL_CONTEXT *cx)
2026 {
2027     PERL_ARGS_ASSERT_CX_TOPBLOCK;
2028 
2029     CX_DEBUG(cx, "TOP");
2030     /* these 3 are common to cx_popblock and cx_topblock */
2031     PL_markstack_ptr = PL_markstack + cx->blk_oldmarksp;
2032     PL_scopestack_ix = cx->blk_oldscopesp;
2033     PL_curpm         = cx->blk_oldpm;
2034 
2035     PL_stack_sp      = PL_stack_base + cx->blk_oldsp;
2036 }
2037 
2038 
2039 PERL_STATIC_INLINE void
S_cx_pushsub(pTHX_ PERL_CONTEXT * cx,CV * cv,OP * retop,bool hasargs)2040 S_cx_pushsub(pTHX_ PERL_CONTEXT *cx, CV *cv, OP *retop, bool hasargs)
2041 {
2042     U8 phlags = CX_PUSHSUB_GET_LVALUE_MASK(Perl_was_lvalue_sub);
2043 
2044     PERL_ARGS_ASSERT_CX_PUSHSUB;
2045 
2046     PERL_DTRACE_PROBE_ENTRY(cv);
2047     cx->blk_sub.cv = cv;
2048     cx->blk_sub.olddepth = CvDEPTH(cv);
2049     cx->blk_sub.prevcomppad = PL_comppad;
2050     cx->cx_type |= (hasargs) ? CXp_HASARGS : 0;
2051     cx->blk_sub.retop = retop;
2052     SvREFCNT_inc_simple_void_NN(cv);
2053     cx->blk_u16 = PL_op->op_private & (phlags|OPpDEREF);
2054 }
2055 
2056 
2057 /* subsets of cx_popsub() */
2058 
2059 PERL_STATIC_INLINE void
S_cx_popsub_common(pTHX_ PERL_CONTEXT * cx)2060 S_cx_popsub_common(pTHX_ PERL_CONTEXT *cx)
2061 {
2062     CV *cv;
2063 
2064     PERL_ARGS_ASSERT_CX_POPSUB_COMMON;
2065     assert(CxTYPE(cx) == CXt_SUB);
2066 
2067     PL_comppad = cx->blk_sub.prevcomppad;
2068     PL_curpad = LIKELY(PL_comppad) ? AvARRAY(PL_comppad) : NULL;
2069     cv = cx->blk_sub.cv;
2070     CvDEPTH(cv) = cx->blk_sub.olddepth;
2071     cx->blk_sub.cv = NULL;
2072     SvREFCNT_dec(cv);
2073 }
2074 
2075 
2076 /* handle the @_ part of leaving a sub */
2077 
2078 PERL_STATIC_INLINE void
S_cx_popsub_args(pTHX_ PERL_CONTEXT * cx)2079 S_cx_popsub_args(pTHX_ PERL_CONTEXT *cx)
2080 {
2081     AV *av;
2082 
2083     PERL_ARGS_ASSERT_CX_POPSUB_ARGS;
2084     assert(CxTYPE(cx) == CXt_SUB);
2085     assert(AvARRAY(MUTABLE_AV(
2086         PadlistARRAY(CvPADLIST(cx->blk_sub.cv))[
2087                 CvDEPTH(cx->blk_sub.cv)])) == PL_curpad);
2088 
2089     CX_POP_SAVEARRAY(cx);
2090     av = MUTABLE_AV(PAD_SVl(0));
2091     if (UNLIKELY(AvREAL(av)))
2092         /* abandon @_ if it got reified */
2093         clear_defarray(av, 0);
2094     else {
2095         CLEAR_ARGARRAY(av);
2096     }
2097 }
2098 
2099 
2100 PERL_STATIC_INLINE void
S_cx_popsub(pTHX_ PERL_CONTEXT * cx)2101 S_cx_popsub(pTHX_ PERL_CONTEXT *cx)
2102 {
2103     PERL_ARGS_ASSERT_CX_POPSUB;
2104     assert(CxTYPE(cx) == CXt_SUB);
2105 
2106     PERL_DTRACE_PROBE_RETURN(cx->blk_sub.cv);
2107 
2108     if (CxHASARGS(cx))
2109         cx_popsub_args(cx);
2110     cx_popsub_common(cx);
2111 }
2112 
2113 
2114 PERL_STATIC_INLINE void
S_cx_pushformat(pTHX_ PERL_CONTEXT * cx,CV * cv,OP * retop,GV * gv)2115 S_cx_pushformat(pTHX_ PERL_CONTEXT *cx, CV *cv, OP *retop, GV *gv)
2116 {
2117     PERL_ARGS_ASSERT_CX_PUSHFORMAT;
2118 
2119     cx->blk_format.cv          = cv;
2120     cx->blk_format.retop       = retop;
2121     cx->blk_format.gv          = gv;
2122     cx->blk_format.dfoutgv     = PL_defoutgv;
2123     cx->blk_format.prevcomppad = PL_comppad;
2124     cx->blk_u16                = 0;
2125 
2126     SvREFCNT_inc_simple_void_NN(cv);
2127     CvDEPTH(cv)++;
2128     SvREFCNT_inc_void(cx->blk_format.dfoutgv);
2129 }
2130 
2131 
2132 PERL_STATIC_INLINE void
S_cx_popformat(pTHX_ PERL_CONTEXT * cx)2133 S_cx_popformat(pTHX_ PERL_CONTEXT *cx)
2134 {
2135     CV *cv;
2136     GV *dfout;
2137 
2138     PERL_ARGS_ASSERT_CX_POPFORMAT;
2139     assert(CxTYPE(cx) == CXt_FORMAT);
2140 
2141     dfout = cx->blk_format.dfoutgv;
2142     setdefout(dfout);
2143     cx->blk_format.dfoutgv = NULL;
2144     SvREFCNT_dec_NN(dfout);
2145 
2146     PL_comppad = cx->blk_format.prevcomppad;
2147     PL_curpad = LIKELY(PL_comppad) ? AvARRAY(PL_comppad) : NULL;
2148     cv = cx->blk_format.cv;
2149     cx->blk_format.cv = NULL;
2150     --CvDEPTH(cv);
2151     SvREFCNT_dec_NN(cv);
2152 }
2153 
2154 
2155 PERL_STATIC_INLINE void
S_cx_pusheval(pTHX_ PERL_CONTEXT * cx,OP * retop,SV * namesv)2156 S_cx_pusheval(pTHX_ PERL_CONTEXT *cx, OP *retop, SV *namesv)
2157 {
2158     PERL_ARGS_ASSERT_CX_PUSHEVAL;
2159 
2160     cx->blk_eval.retop         = retop;
2161     cx->blk_eval.old_namesv    = namesv;
2162     cx->blk_eval.old_eval_root = PL_eval_root;
2163     cx->blk_eval.cur_text      = PL_parser ? PL_parser->linestr : NULL;
2164     cx->blk_eval.cv            = NULL; /* later set by doeval_compile() */
2165     cx->blk_eval.cur_top_env   = PL_top_env;
2166 
2167     assert(!(PL_in_eval     & ~ 0x3F));
2168     assert(!(PL_op->op_type & ~0x1FF));
2169     cx->blk_u16 = (PL_in_eval & 0x3F) | ((U16)PL_op->op_type << 7);
2170 }
2171 
2172 
2173 PERL_STATIC_INLINE void
S_cx_popeval(pTHX_ PERL_CONTEXT * cx)2174 S_cx_popeval(pTHX_ PERL_CONTEXT *cx)
2175 {
2176     SV *sv;
2177 
2178     PERL_ARGS_ASSERT_CX_POPEVAL;
2179     assert(CxTYPE(cx) == CXt_EVAL);
2180 
2181     PL_in_eval = CxOLD_IN_EVAL(cx);
2182     assert(!(PL_in_eval & 0xc0));
2183     PL_eval_root = cx->blk_eval.old_eval_root;
2184     sv = cx->blk_eval.cur_text;
2185     if (sv && CxEVAL_TXT_REFCNTED(cx)) {
2186         cx->blk_eval.cur_text = NULL;
2187         SvREFCNT_dec_NN(sv);
2188     }
2189 
2190     sv = cx->blk_eval.old_namesv;
2191     if (sv) {
2192         cx->blk_eval.old_namesv = NULL;
2193         SvREFCNT_dec_NN(sv);
2194     }
2195 }
2196 
2197 
2198 /* push a plain loop, i.e.
2199  *     { block }
2200  *     while (cond) { block }
2201  *     for (init;cond;continue) { block }
2202  * This loop can be last/redo'ed etc.
2203  */
2204 
2205 PERL_STATIC_INLINE void
S_cx_pushloop_plain(pTHX_ PERL_CONTEXT * cx)2206 S_cx_pushloop_plain(pTHX_ PERL_CONTEXT *cx)
2207 {
2208     PERL_ARGS_ASSERT_CX_PUSHLOOP_PLAIN;
2209     cx->blk_loop.my_op = cLOOP;
2210 }
2211 
2212 
2213 /* push a true for loop, i.e.
2214  *     for var (list) { block }
2215  */
2216 
2217 PERL_STATIC_INLINE void
S_cx_pushloop_for(pTHX_ PERL_CONTEXT * cx,void * itervarp,SV * itersave)2218 S_cx_pushloop_for(pTHX_ PERL_CONTEXT *cx, void *itervarp, SV* itersave)
2219 {
2220     PERL_ARGS_ASSERT_CX_PUSHLOOP_FOR;
2221 
2222     /* this one line is common with cx_pushloop_plain */
2223     cx->blk_loop.my_op = cLOOP;
2224 
2225     cx->blk_loop.itervar_u.svp = (SV**)itervarp;
2226     cx->blk_loop.itersave      = itersave;
2227 #ifdef USE_ITHREADS
2228     cx->blk_loop.oldcomppad = PL_comppad;
2229 #endif
2230 }
2231 
2232 
2233 /* pop all loop types, including plain */
2234 
2235 PERL_STATIC_INLINE void
S_cx_poploop(pTHX_ PERL_CONTEXT * cx)2236 S_cx_poploop(pTHX_ PERL_CONTEXT *cx)
2237 {
2238     PERL_ARGS_ASSERT_CX_POPLOOP;
2239 
2240     assert(CxTYPE_is_LOOP(cx));
2241     if (  CxTYPE(cx) == CXt_LOOP_ARY
2242        || CxTYPE(cx) == CXt_LOOP_LAZYSV)
2243     {
2244         /* Free ary or cur. This assumes that state_u.ary.ary
2245          * aligns with state_u.lazysv.cur. See cx_dup() */
2246         SV *sv = cx->blk_loop.state_u.lazysv.cur;
2247         cx->blk_loop.state_u.lazysv.cur = NULL;
2248         SvREFCNT_dec_NN(sv);
2249         if (CxTYPE(cx) == CXt_LOOP_LAZYSV) {
2250             sv = cx->blk_loop.state_u.lazysv.end;
2251             cx->blk_loop.state_u.lazysv.end = NULL;
2252             SvREFCNT_dec_NN(sv);
2253         }
2254     }
2255     if (cx->cx_type & (CXp_FOR_PAD|CXp_FOR_GV)) {
2256         SV *cursv;
2257         SV **svp = (cx)->blk_loop.itervar_u.svp;
2258         if ((cx->cx_type & CXp_FOR_GV))
2259             svp = &GvSV((GV*)svp);
2260         cursv = *svp;
2261         *svp = cx->blk_loop.itersave;
2262         cx->blk_loop.itersave = NULL;
2263         SvREFCNT_dec(cursv);
2264     }
2265 }
2266 
2267 
2268 PERL_STATIC_INLINE void
S_cx_pushwhen(pTHX_ PERL_CONTEXT * cx)2269 S_cx_pushwhen(pTHX_ PERL_CONTEXT *cx)
2270 {
2271     PERL_ARGS_ASSERT_CX_PUSHWHEN;
2272 
2273     cx->blk_givwhen.leave_op = cLOGOP->op_other;
2274 }
2275 
2276 
2277 PERL_STATIC_INLINE void
S_cx_popwhen(pTHX_ PERL_CONTEXT * cx)2278 S_cx_popwhen(pTHX_ PERL_CONTEXT *cx)
2279 {
2280     PERL_ARGS_ASSERT_CX_POPWHEN;
2281     assert(CxTYPE(cx) == CXt_WHEN);
2282 
2283     PERL_UNUSED_ARG(cx);
2284     PERL_UNUSED_CONTEXT;
2285     /* currently NOOP */
2286 }
2287 
2288 
2289 PERL_STATIC_INLINE void
S_cx_pushgiven(pTHX_ PERL_CONTEXT * cx,SV * orig_defsv)2290 S_cx_pushgiven(pTHX_ PERL_CONTEXT *cx, SV *orig_defsv)
2291 {
2292     PERL_ARGS_ASSERT_CX_PUSHGIVEN;
2293 
2294     cx->blk_givwhen.leave_op = cLOGOP->op_other;
2295     cx->blk_givwhen.defsv_save = orig_defsv;
2296 }
2297 
2298 
2299 PERL_STATIC_INLINE void
S_cx_popgiven(pTHX_ PERL_CONTEXT * cx)2300 S_cx_popgiven(pTHX_ PERL_CONTEXT *cx)
2301 {
2302     SV *sv;
2303 
2304     PERL_ARGS_ASSERT_CX_POPGIVEN;
2305     assert(CxTYPE(cx) == CXt_GIVEN);
2306 
2307     sv = GvSV(PL_defgv);
2308     GvSV(PL_defgv) = cx->blk_givwhen.defsv_save;
2309     cx->blk_givwhen.defsv_save = NULL;
2310     SvREFCNT_dec(sv);
2311 }
2312 
2313 /* ------------------ util.h ------------------------------------------- */
2314 
2315 /*
2316 =head1 Miscellaneous Functions
2317 
2318 =for apidoc foldEQ
2319 
2320 Returns true if the leading C<len> bytes of the strings C<s1> and C<s2> are the
2321 same
2322 case-insensitively; false otherwise.  Uppercase and lowercase ASCII range bytes
2323 match themselves and their opposite case counterparts.  Non-cased and non-ASCII
2324 range bytes match only themselves.
2325 
2326 =cut
2327 */
2328 
2329 PERL_STATIC_INLINE I32
Perl_foldEQ(const char * s1,const char * s2,I32 len)2330 Perl_foldEQ(const char *s1, const char *s2, I32 len)
2331 {
2332     const U8 *a = (const U8 *)s1;
2333     const U8 *b = (const U8 *)s2;
2334 
2335     PERL_ARGS_ASSERT_FOLDEQ;
2336 
2337     assert(len >= 0);
2338 
2339     while (len--) {
2340 	if (*a != *b && *a != PL_fold[*b])
2341 	    return 0;
2342 	a++,b++;
2343     }
2344     return 1;
2345 }
2346 
2347 PERL_STATIC_INLINE I32
Perl_foldEQ_latin1(const char * s1,const char * s2,I32 len)2348 Perl_foldEQ_latin1(const char *s1, const char *s2, I32 len)
2349 {
2350     /* Compare non-UTF-8 using Unicode (Latin1) semantics.  Works on all folds
2351      * representable without UTF-8, except for LATIN_SMALL_LETTER_SHARP_S, and
2352      * does not check for this.  Nor does it check that the strings each have
2353      * at least 'len' characters. */
2354 
2355     const U8 *a = (const U8 *)s1;
2356     const U8 *b = (const U8 *)s2;
2357 
2358     PERL_ARGS_ASSERT_FOLDEQ_LATIN1;
2359 
2360     assert(len >= 0);
2361 
2362     while (len--) {
2363 	if (*a != *b && *a != PL_fold_latin1[*b]) {
2364 	    return 0;
2365 	}
2366 	a++, b++;
2367     }
2368     return 1;
2369 }
2370 
2371 /*
2372 =for apidoc foldEQ_locale
2373 
2374 Returns true if the leading C<len> bytes of the strings C<s1> and C<s2> are the
2375 same case-insensitively in the current locale; false otherwise.
2376 
2377 =cut
2378 */
2379 
2380 PERL_STATIC_INLINE I32
Perl_foldEQ_locale(const char * s1,const char * s2,I32 len)2381 Perl_foldEQ_locale(const char *s1, const char *s2, I32 len)
2382 {
2383     dVAR;
2384     const U8 *a = (const U8 *)s1;
2385     const U8 *b = (const U8 *)s2;
2386 
2387     PERL_ARGS_ASSERT_FOLDEQ_LOCALE;
2388 
2389     assert(len >= 0);
2390 
2391     while (len--) {
2392 	if (*a != *b && *a != PL_fold_locale[*b])
2393 	    return 0;
2394 	a++,b++;
2395     }
2396     return 1;
2397 }
2398 
2399 #if ! defined (HAS_MEMRCHR) && (defined(PERL_CORE) || defined(PERL_EXT))
2400 
2401 PERL_STATIC_INLINE void *
S_my_memrchr(const char * s,const char c,const STRLEN len)2402 S_my_memrchr(const char * s, const char c, const STRLEN len)
2403 {
2404     /* memrchr(), since many platforms lack it */
2405 
2406     const char * t = s + len - 1;
2407 
2408     PERL_ARGS_ASSERT_MY_MEMRCHR;
2409 
2410     while (t >= s) {
2411         if (*t == c) {
2412             return (void *) t;
2413         }
2414         t--;
2415     }
2416 
2417     return NULL;
2418 }
2419 
2420 #endif
2421 
2422 /*
2423  * ex: set ts=8 sts=4 sw=4 et:
2424  */
2425