3 * Copyright (C) 2012 by Larry Wall and others
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.
8 * This file contains tables and code adapted from
9 * https://bjoern.hoehrmann.de/utf-8/decoder/dfa/, which requires this
12 Copyright (c) 2008-2009 Bjoern Hoehrmann <bjoern@hoehrmann.de>
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:
21 The above copyright notice and this permission notice shall be included in all
22 copies or substantial portions of the Software.
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
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.
37 * Each section names the header file that the functions "belong" to.
40 /* ------------------------------- av.h ------------------------------- */
42 PERL_STATIC_INLINE SSize_t
43 Perl_av_top_index(pTHX_ AV *av)
45 PERL_ARGS_ASSERT_AV_TOP_INDEX;
46 assert(SvTYPE(av) == SVt_PVAV);
51 /* ------------------------------- cv.h ------------------------------- */
53 PERL_STATIC_INLINE GV *
54 Perl_CvGV(pTHX_ CV *sv)
56 PERL_ARGS_ASSERT_CVGV;
59 ? Perl_cvgv_from_hek(aTHX_ sv)
60 : ((XPVCV*)MUTABLE_PTR(SvANY(sv)))->xcv_gv_u.xcv_gv;
63 PERL_STATIC_INLINE I32 *
64 Perl_CvDEPTH(const CV * const sv)
66 PERL_ARGS_ASSERT_CVDEPTH;
67 assert(SvTYPE(sv) == SVt_PVCV || SvTYPE(sv) == SVt_PVFM);
69 return &((XPVCV*)SvANY(sv))->xcv_depth;
73 CvPROTO returns the prototype as stored, which is not necessarily what
74 the interpreter should be using. Specifically, the interpreter assumes
75 that spaces have been stripped, which has been the case if the prototype
76 was added by toke.c, but is generally not the case if it was added elsewhere.
77 Since we can't enforce the spacelessness at assignment time, this routine
78 provides a temporary copy at parse time with spaces removed.
79 I<orig> is the start of the original buffer, I<len> is the length of the
80 prototype and will be updated when this returns.
84 PERL_STATIC_INLINE char *
85 S_strip_spaces(pTHX_ const char * orig, STRLEN * const len)
89 tmpsv = newSVpvn_flags(orig, *len, SVs_TEMP);
97 *len = tmps - SvPVX(tmpsv);
102 /* ------------------------------- mg.h ------------------------------- */
104 #if defined(PERL_CORE) || defined(PERL_EXT)
105 /* assumes get-magic and stringification have already occurred */
106 PERL_STATIC_INLINE STRLEN
107 S_MgBYTEPOS(pTHX_ MAGIC *mg, SV *sv, const char *s, STRLEN len)
109 assert(mg->mg_type == PERL_MAGIC_regex_global);
110 assert(mg->mg_len != -1);
111 if (mg->mg_flags & MGf_BYTES || !DO_UTF8(sv))
112 return (STRLEN)mg->mg_len;
114 const STRLEN pos = (STRLEN)mg->mg_len;
115 /* Without this check, we may read past the end of the buffer: */
116 if (pos > sv_or_pv_len_utf8(sv, s, len)) return len+1;
117 return sv_or_pv_pos_u2b(sv, s, pos, NULL);
122 /* ------------------------------- pad.h ------------------------------ */
124 #if defined(PERL_IN_PAD_C) || defined(PERL_IN_OP_C)
125 PERL_STATIC_INLINE bool
126 S_PadnameIN_SCOPE(const PADNAME * const pn, const U32 seq)
128 PERL_ARGS_ASSERT_PADNAMEIN_SCOPE;
130 /* is seq within the range _LOW to _HIGH ?
131 * This is complicated by the fact that PL_cop_seqmax
132 * may have wrapped around at some point */
133 if (COP_SEQ_RANGE_LOW(pn) == PERL_PADSEQ_INTRO)
134 return FALSE; /* not yet introduced */
136 if (COP_SEQ_RANGE_HIGH(pn) == PERL_PADSEQ_INTRO) {
137 /* in compiling scope */
139 (seq > COP_SEQ_RANGE_LOW(pn))
140 ? (seq - COP_SEQ_RANGE_LOW(pn) < (U32_MAX >> 1))
141 : (COP_SEQ_RANGE_LOW(pn) - seq > (U32_MAX >> 1))
146 (COP_SEQ_RANGE_LOW(pn) > COP_SEQ_RANGE_HIGH(pn))
148 ( seq > COP_SEQ_RANGE_LOW(pn)
149 || seq <= COP_SEQ_RANGE_HIGH(pn))
151 : ( seq > COP_SEQ_RANGE_LOW(pn)
152 && seq <= COP_SEQ_RANGE_HIGH(pn))
159 /* ------------------------------- pp.h ------------------------------- */
161 PERL_STATIC_INLINE I32
164 DEBUG_s(DEBUG_v(PerlIO_printf(Perl_debug_log,
165 "MARK top %p %" IVdf "\n",
167 (IV)*PL_markstack_ptr)));
168 return *PL_markstack_ptr;
171 PERL_STATIC_INLINE I32
174 DEBUG_s(DEBUG_v(PerlIO_printf(Perl_debug_log,
175 "MARK pop %p %" IVdf "\n",
176 (PL_markstack_ptr-1),
177 (IV)*(PL_markstack_ptr-1))));
178 assert((PL_markstack_ptr > PL_markstack) || !"MARK underflow");
179 return *PL_markstack_ptr--;
182 /* ----------------------------- regexp.h ----------------------------- */
184 PERL_STATIC_INLINE struct regexp *
185 Perl_ReANY(const REGEXP * const re)
187 XPV* const p = (XPV*)SvANY(re);
189 PERL_ARGS_ASSERT_REANY;
190 assert(isREGEXP(re));
192 return SvTYPE(re) == SVt_PVLV ? p->xpv_len_u.xpvlenu_rx
193 : (struct regexp *)p;
196 /* ------------------------------- sv.h ------------------------------- */
198 PERL_STATIC_INLINE SV *
199 Perl_SvREFCNT_inc(SV *sv)
201 if (LIKELY(sv != NULL))
205 PERL_STATIC_INLINE SV *
206 Perl_SvREFCNT_inc_NN(SV *sv)
208 PERL_ARGS_ASSERT_SVREFCNT_INC_NN;
213 PERL_STATIC_INLINE void
214 Perl_SvREFCNT_inc_void(SV *sv)
216 if (LIKELY(sv != NULL))
219 PERL_STATIC_INLINE void
220 Perl_SvREFCNT_dec(pTHX_ SV *sv)
222 if (LIKELY(sv != NULL)) {
223 U32 rc = SvREFCNT(sv);
225 SvREFCNT(sv) = rc - 1;
227 Perl_sv_free2(aTHX_ sv, rc);
231 PERL_STATIC_INLINE void
232 Perl_SvREFCNT_dec_NN(pTHX_ SV *sv)
234 U32 rc = SvREFCNT(sv);
236 PERL_ARGS_ASSERT_SVREFCNT_DEC_NN;
239 SvREFCNT(sv) = rc - 1;
241 Perl_sv_free2(aTHX_ sv, rc);
244 PERL_STATIC_INLINE void
245 Perl_SvAMAGIC_on(SV *sv)
247 PERL_ARGS_ASSERT_SVAMAGIC_ON;
250 if (SvOBJECT(SvRV(sv))) HvAMAGIC_on(SvSTASH(SvRV(sv)));
252 PERL_STATIC_INLINE void
253 Perl_SvAMAGIC_off(SV *sv)
255 PERL_ARGS_ASSERT_SVAMAGIC_OFF;
257 if (SvROK(sv) && SvOBJECT(SvRV(sv)))
258 HvAMAGIC_off(SvSTASH(SvRV(sv)));
261 PERL_STATIC_INLINE U32
262 Perl_SvPADSTALE_on(SV *sv)
264 assert(!(SvFLAGS(sv) & SVs_PADTMP));
265 return SvFLAGS(sv) |= SVs_PADSTALE;
267 PERL_STATIC_INLINE U32
268 Perl_SvPADSTALE_off(SV *sv)
270 assert(!(SvFLAGS(sv) & SVs_PADTMP));
271 return SvFLAGS(sv) &= ~SVs_PADSTALE;
273 #if defined(PERL_CORE) || defined (PERL_EXT)
274 PERL_STATIC_INLINE STRLEN
275 S_sv_or_pv_pos_u2b(pTHX_ SV *sv, const char *pv, STRLEN pos, STRLEN *lenp)
277 PERL_ARGS_ASSERT_SV_OR_PV_POS_U2B;
279 U8 *hopped = utf8_hop((U8 *)pv, pos);
280 if (lenp) *lenp = (STRLEN)(utf8_hop(hopped, *lenp) - hopped);
281 return (STRLEN)(hopped - (U8 *)pv);
283 return sv_pos_u2b_flags(sv,pos,lenp,SV_CONST_RETURN);
287 /* ------------------------------- handy.h ------------------------------- */
289 /* saves machine code for a common noreturn idiom typically used in Newx*() */
290 GCC_DIAG_IGNORE_DECL(-Wunused-function);
292 Perl_croak_memory_wrap(void)
294 Perl_croak_nocontext("%s",PL_memory_wrap);
296 GCC_DIAG_RESTORE_DECL;
298 /* ------------------------------- utf8.h ------------------------------- */
301 =head1 Unicode Support
304 PERL_STATIC_INLINE void
305 Perl_append_utf8_from_native_byte(const U8 byte, U8** dest)
307 /* Takes an input 'byte' (Latin1 or EBCDIC) and appends it to the UTF-8
308 * encoded string at '*dest', updating '*dest' to include it */
310 PERL_ARGS_ASSERT_APPEND_UTF8_FROM_NATIVE_BYTE;
312 if (NATIVE_BYTE_IS_INVARIANT(byte))
315 *((*dest)++) = UTF8_EIGHT_BIT_HI(byte);
316 *((*dest)++) = UTF8_EIGHT_BIT_LO(byte);
321 =for apidoc valid_utf8_to_uvchr
322 Like C<L<perlapi/utf8_to_uvchr_buf>>, but should only be called when it is
323 known that the next character in the input UTF-8 string C<s> is well-formed
324 (I<e.g.>, it passes C<L<perlapi/isUTF8_CHAR>>. Surrogates, non-character code
325 points, and non-Unicode code points are allowed.
331 PERL_STATIC_INLINE UV
332 Perl_valid_utf8_to_uvchr(const U8 *s, STRLEN *retlen)
334 const UV expectlen = UTF8SKIP(s);
335 const U8* send = s + expectlen;
338 PERL_ARGS_ASSERT_VALID_UTF8_TO_UVCHR;
344 /* An invariant is trivially returned */
345 if (expectlen == 1) {
349 /* Remove the leading bits that indicate the number of bytes, leaving just
350 * the bits that are part of the value */
351 uv = NATIVE_UTF8_TO_I8(uv) & UTF_START_MASK(expectlen);
353 /* Now, loop through the remaining bytes, accumulating each into the
354 * working total as we go. (I khw tried unrolling the loop for up to 4
355 * bytes, but there was no performance improvement) */
356 for (++s; s < send; s++) {
357 uv = UTF8_ACCUMULATE(uv, *s);
360 return UNI_TO_NATIVE(uv);
365 =for apidoc is_utf8_invariant_string
367 Returns TRUE if the first C<len> bytes of the string C<s> are the same
368 regardless of the UTF-8 encoding of the string (or UTF-EBCDIC encoding on
369 EBCDIC machines); otherwise it returns FALSE. That is, it returns TRUE if they
370 are UTF-8 invariant. On ASCII-ish machines, all the ASCII characters and only
371 the ASCII characters fit this definition. On EBCDIC machines, the ASCII-range
372 characters are invariant, but so also are the C1 controls.
374 If C<len> is 0, it will be calculated using C<strlen(s)>, (which means if you
375 use this option, that C<s> can't have embedded C<NUL> characters and has to
376 have a terminating C<NUL> byte).
379 C<L</is_utf8_string>>,
380 C<L</is_utf8_string_flags>>,
381 C<L</is_utf8_string_loc>>,
382 C<L</is_utf8_string_loc_flags>>,
383 C<L</is_utf8_string_loclen>>,
384 C<L</is_utf8_string_loclen_flags>>,
385 C<L</is_utf8_fixed_width_buf_flags>>,
386 C<L</is_utf8_fixed_width_buf_loc_flags>>,
387 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
388 C<L</is_strict_utf8_string>>,
389 C<L</is_strict_utf8_string_loc>>,
390 C<L</is_strict_utf8_string_loclen>>,
391 C<L</is_c9strict_utf8_string>>,
392 C<L</is_c9strict_utf8_string_loc>>,
394 C<L</is_c9strict_utf8_string_loclen>>.
400 #define is_utf8_invariant_string(s, len) \
401 is_utf8_invariant_string_loc(s, len, NULL)
404 =for apidoc is_utf8_invariant_string_loc
406 Like C<L</is_utf8_invariant_string>> but upon failure, stores the location of
407 the first UTF-8 variant character in the C<ep> pointer; if all characters are
408 UTF-8 invariant, this function does not change the contents of C<*ep>.
414 PERL_STATIC_INLINE bool
415 Perl_is_utf8_invariant_string_loc(const U8* const s, STRLEN len, const U8 ** ep)
420 PERL_ARGS_ASSERT_IS_UTF8_INVARIANT_STRING_LOC;
423 len = strlen((const char *)s);
428 /* This looks like 0x010101... */
429 # define PERL_COUNT_MULTIPLIER (~ (UINTMAX_C(0)) / 0xFF)
431 /* This looks like 0x808080... */
432 # define PERL_VARIANTS_WORD_MASK (PERL_COUNT_MULTIPLIER * 0x80)
433 # define PERL_WORDSIZE sizeof(PERL_UINTMAX_T)
434 # define PERL_WORD_BOUNDARY_MASK (PERL_WORDSIZE - 1)
436 /* Evaluates to 0 if 'x' is at a word boundary; otherwise evaluates to 1, by
437 * or'ing together the lowest bits of 'x'. Hopefully the final term gets
438 * optimized out completely on a 32-bit system, and its mask gets optimized out
439 * on a 64-bit system */
440 # define PERL_IS_SUBWORD_ADDR(x) (1 & ( PTR2nat(x) \
441 | ( PTR2nat(x) >> 1) \
443 & PERL_WORD_BOUNDARY_MASK) >> 2))))
447 /* Do the word-at-a-time iff there is at least one usable full word. That
448 * means that after advancing to a word boundary, there still is at least a
449 * full word left. The number of bytes needed to advance is 'wordsize -
450 * offset' unless offset is 0. */
451 if ((STRLEN) (send - x) >= PERL_WORDSIZE
453 /* This term is wordsize if subword; 0 if not */
454 + PERL_WORDSIZE * PERL_IS_SUBWORD_ADDR(x)
457 - (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK))
460 /* Process per-byte until reach word boundary. XXX This loop could be
461 * eliminated if we knew that this platform had fast unaligned reads */
462 while (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK) {
463 if (! UTF8_IS_INVARIANT(*x)) {
473 /* Here, we know we have at least one full word to process. Process
474 * per-word as long as we have at least a full word left */
476 if ((* (PERL_UINTMAX_T *) x) & PERL_VARIANTS_WORD_MASK) {
478 /* Found a variant. Just return if caller doesn't want its
484 # if BYTEORDER == 0x1234 || BYTEORDER == 0x12345678 \
485 || BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
487 *ep = x + variant_byte_number(* (PERL_UINTMAX_T *) x);
488 assert(*ep >= s && *ep < send);
492 # else /* If weird byte order, drop into next loop to do byte-at-a-time
501 } while (x + PERL_WORDSIZE <= send);
504 #endif /* End of ! EBCDIC */
506 /* Process per-byte */
508 if (! UTF8_IS_INVARIANT(*x)) {
524 PERL_STATIC_INLINE unsigned int
525 Perl_variant_byte_number(PERL_UINTMAX_T word)
528 /* This returns the position in a word (0..7) of the first variant byte in
529 * it. This is a helper function. Note that there are no branches */
533 /* Get just the msb bits of each byte */
534 word &= PERL_VARIANTS_WORD_MASK;
536 # if BYTEORDER == 0x1234 || BYTEORDER == 0x12345678
538 /* Bytes are stored like
539 * Byte8 ... Byte2 Byte1
540 * 63..56...15...8 7...0
543 * https://stackoverflow.com/questions/757059/position-of-least-significant-bit-that-is-set
545 * The word will look this this, with a rightmost set bit in position 's':
546 * ('x's are don't cares)
549 * x..xx10..0 Right shift (rightmost 0 is shifted off)
550 * x..xx01..1 Subtract 1, turns all the trailing zeros into 1's and
551 * the 1 just to their left into a 0; the remainder is
553 * 0..0011..1 The xor with the original, x..xx10..0, clears that
554 * remainder, sets the bottom to all 1
555 * 0..0100..0 Add 1 to clear the word except for the bit in 's'
557 * Another method is to do 'word &= -word'; but it generates a compiler
558 * message on some platforms about taking the negative of an unsigned */
561 word = 1 + (word ^ (word - 1));
563 # elif BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
565 /* Bytes are stored like
566 * Byte1 Byte2 ... Byte8
567 * 63..56 55..47 ... 7...0
569 * Isolate the msb; http://codeforces.com/blog/entry/10330
571 * Only the most significant set bit matters. Or'ing word with its right
572 * shift of 1 makes that bit and the next one to its right both 1. Then
573 * right shifting by 2 makes for 4 1-bits in a row. ... We end with the
574 * msb and all to the right being 1. */
580 word |= word >> 32; /* This should get optimized out on 32-bit systems. */
582 /* Then subtracting the right shift by 1 clears all but the left-most of
583 * the 1 bits, which is our desired result */
587 # error Unexpected byte order
590 /* Here 'word' has a single bit set: the msb of the first byte in which it
591 * is set. Calculate that position in the word. We can use this
592 * specialized solution: https://stackoverflow.com/a/32339674/1626653,
593 * assumes an 8-bit byte. (On a 32-bit machine, the larger numbers should
594 * just get shifted off at compile time) */
595 word = (word >> 7) * ((UINTMAX_C( 7) << 56) | (UINTMAX_C(15) << 48)
596 | (UINTMAX_C(23) << 40) | (UINTMAX_C(31) << 32)
597 | (39 << 24) | (47 << 16)
598 | (55 << 8) | (63 << 0));
599 word >>= PERL_WORDSIZE * 7; /* >> by either 56 or 24 */
601 /* Here, word contains the position 7..63 of that bit. Convert to 0..7 */
602 word = ((word + 1) >> 3) - 1;
604 # if BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
606 /* And invert the result */
607 word = CHARBITS - word - 1;
611 return (unsigned int) word;
615 #if defined(PERL_CORE) || defined(PERL_EXT)
618 =for apidoc variant_under_utf8_count
620 This function looks at the sequence of bytes between C<s> and C<e>, which are
621 assumed to be encoded in ASCII/Latin1, and returns how many of them would
622 change should the string be translated into UTF-8. Due to the nature of UTF-8,
623 each of these would occupy two bytes instead of the single one in the input
624 string. Thus, this function returns the precise number of bytes the string
625 would expand by when translated to UTF-8.
627 Unlike most of the other functions that have C<utf8> in their name, the input
628 to this function is NOT a UTF-8-encoded string. The function name is slightly
629 I<odd> to emphasize this.
631 This function is internal to Perl because khw thinks that any XS code that
632 would want this is probably operating too close to the internals. Presenting a
633 valid use case could change that.
636 C<L<perlapi/is_utf8_invariant_string>>
638 C<L<perlapi/is_utf8_invariant_string_loc>>,
644 PERL_STATIC_INLINE Size_t
645 S_variant_under_utf8_count(const U8* const s, const U8* const e)
650 PERL_ARGS_ASSERT_VARIANT_UNDER_UTF8_COUNT;
654 /* Test if the string is long enough to use word-at-a-time. (Logic is the
655 * same as for is_utf8_invariant_string()) */
656 if ((STRLEN) (e - x) >= PERL_WORDSIZE
657 + PERL_WORDSIZE * PERL_IS_SUBWORD_ADDR(x)
658 - (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK))
661 /* Process per-byte until reach word boundary. XXX This loop could be
662 * eliminated if we knew that this platform had fast unaligned reads */
663 while (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK) {
664 count += ! UTF8_IS_INVARIANT(*x++);
667 /* Process per-word as long as we have at least a full word left */
668 do { /* Commit 03c1e4ab1d6ee9062fb3f94b0ba31db6698724b1 contains an
669 explanation of how this works */
670 PERL_UINTMAX_T increment
671 = ((((* (PERL_UINTMAX_T *) x) & PERL_VARIANTS_WORD_MASK) >> 7)
672 * PERL_COUNT_MULTIPLIER)
673 >> ((PERL_WORDSIZE - 1) * CHARBITS);
674 count += (Size_t) increment;
676 } while (x + PERL_WORDSIZE <= e);
681 /* Process per-byte */
683 if (! UTF8_IS_INVARIANT(*x)) {
695 #ifndef PERL_IN_REGEXEC_C /* Keep these around for that file */
696 # undef PERL_WORDSIZE
697 # undef PERL_COUNT_MULTIPLIER
698 # undef PERL_WORD_BOUNDARY_MASK
699 # undef PERL_VARIANTS_WORD_MASK
703 =for apidoc is_utf8_string
705 Returns TRUE if the first C<len> bytes of string C<s> form a valid
706 Perl-extended-UTF-8 string; returns FALSE otherwise. If C<len> is 0, it will
707 be calculated using C<strlen(s)> (which means if you use this option, that C<s>
708 can't have embedded C<NUL> characters and has to have a terminating C<NUL>
709 byte). Note that all characters being ASCII constitute 'a valid UTF-8 string'.
711 This function considers Perl's extended UTF-8 to be valid. That means that
712 code points above Unicode, surrogates, and non-character code points are
713 considered valid by this function. Use C<L</is_strict_utf8_string>>,
714 C<L</is_c9strict_utf8_string>>, or C<L</is_utf8_string_flags>> to restrict what
715 code points are considered valid.
718 C<L</is_utf8_invariant_string>>,
719 C<L</is_utf8_invariant_string_loc>>,
720 C<L</is_utf8_string_loc>>,
721 C<L</is_utf8_string_loclen>>,
722 C<L</is_utf8_fixed_width_buf_flags>>,
723 C<L</is_utf8_fixed_width_buf_loc_flags>>,
724 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
729 #define is_utf8_string(s, len) is_utf8_string_loclen(s, len, NULL, NULL)
731 #if defined(PERL_CORE) || defined (PERL_EXT)
734 =for apidoc is_utf8_non_invariant_string
736 Returns TRUE if L<perlapi/is_utf8_invariant_string> returns FALSE for the first
737 C<len> bytes of the string C<s>, but they are, nonetheless, legal Perl-extended
738 UTF-8; otherwise returns FALSE.
740 A TRUE return means that at least one code point represented by the sequence
741 either is a wide character not representable as a single byte, or the
742 representation differs depending on whether the sequence is encoded in UTF-8 or
746 C<L<perlapi/is_utf8_invariant_string>>,
747 C<L<perlapi/is_utf8_string>>
751 This is commonly used to determine if a SV's UTF-8 flag should be turned on.
752 It generally needn't be if its string is entirely UTF-8 invariant, and it
753 shouldn't be if it otherwise contains invalid UTF-8.
755 It is an internal function because khw thinks that XS code shouldn't be working
756 at this low a level. A valid use case could change that.
760 PERL_STATIC_INLINE bool
761 Perl_is_utf8_non_invariant_string(const U8* const s, STRLEN len)
763 const U8 * first_variant;
765 PERL_ARGS_ASSERT_IS_UTF8_NON_INVARIANT_STRING;
767 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
771 return is_utf8_string(first_variant, len - (first_variant - s));
777 =for apidoc is_strict_utf8_string
779 Returns TRUE if the first C<len> bytes of string C<s> form a valid
780 UTF-8-encoded string that is fully interchangeable by any application using
781 Unicode rules; otherwise it returns FALSE. If C<len> is 0, it will be
782 calculated using C<strlen(s)> (which means if you use this option, that C<s>
783 can't have embedded C<NUL> characters and has to have a terminating C<NUL>
784 byte). Note that all characters being ASCII constitute 'a valid UTF-8 string'.
786 This function returns FALSE for strings containing any
787 code points above the Unicode max of 0x10FFFF, surrogate code points, or
788 non-character code points.
791 C<L</is_utf8_invariant_string>>,
792 C<L</is_utf8_invariant_string_loc>>,
793 C<L</is_utf8_string>>,
794 C<L</is_utf8_string_flags>>,
795 C<L</is_utf8_string_loc>>,
796 C<L</is_utf8_string_loc_flags>>,
797 C<L</is_utf8_string_loclen>>,
798 C<L</is_utf8_string_loclen_flags>>,
799 C<L</is_utf8_fixed_width_buf_flags>>,
800 C<L</is_utf8_fixed_width_buf_loc_flags>>,
801 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
802 C<L</is_strict_utf8_string_loc>>,
803 C<L</is_strict_utf8_string_loclen>>,
804 C<L</is_c9strict_utf8_string>>,
805 C<L</is_c9strict_utf8_string_loc>>,
807 C<L</is_c9strict_utf8_string_loclen>>.
812 #define is_strict_utf8_string(s, len) is_strict_utf8_string_loclen(s, len, NULL, NULL)
815 =for apidoc is_c9strict_utf8_string
817 Returns TRUE if the first C<len> bytes of string C<s> form a valid
818 UTF-8-encoded string that conforms to
819 L<Unicode Corrigendum #9|http://www.unicode.org/versions/corrigendum9.html>;
820 otherwise it returns FALSE. If C<len> is 0, it will be calculated using
821 C<strlen(s)> (which means if you use this option, that C<s> can't have embedded
822 C<NUL> characters and has to have a terminating C<NUL> byte). Note that all
823 characters being ASCII constitute 'a valid UTF-8 string'.
825 This function returns FALSE for strings containing any code points above the
826 Unicode max of 0x10FFFF or surrogate code points, but accepts non-character
828 L<Corrigendum #9|http://www.unicode.org/versions/corrigendum9.html>.
831 C<L</is_utf8_invariant_string>>,
832 C<L</is_utf8_invariant_string_loc>>,
833 C<L</is_utf8_string>>,
834 C<L</is_utf8_string_flags>>,
835 C<L</is_utf8_string_loc>>,
836 C<L</is_utf8_string_loc_flags>>,
837 C<L</is_utf8_string_loclen>>,
838 C<L</is_utf8_string_loclen_flags>>,
839 C<L</is_utf8_fixed_width_buf_flags>>,
840 C<L</is_utf8_fixed_width_buf_loc_flags>>,
841 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
842 C<L</is_strict_utf8_string>>,
843 C<L</is_strict_utf8_string_loc>>,
844 C<L</is_strict_utf8_string_loclen>>,
845 C<L</is_c9strict_utf8_string_loc>>,
847 C<L</is_c9strict_utf8_string_loclen>>.
852 #define is_c9strict_utf8_string(s, len) is_c9strict_utf8_string_loclen(s, len, NULL, 0)
855 =for apidoc is_utf8_string_flags
857 Returns TRUE if the first C<len> bytes of string C<s> form a valid
858 UTF-8 string, subject to the restrictions imposed by C<flags>;
859 returns FALSE otherwise. If C<len> is 0, it will be calculated
860 using C<strlen(s)> (which means if you use this option, that C<s> can't have
861 embedded C<NUL> characters and has to have a terminating C<NUL> byte). Note
862 that all characters being ASCII constitute 'a valid UTF-8 string'.
864 If C<flags> is 0, this gives the same results as C<L</is_utf8_string>>; if
865 C<flags> is C<UTF8_DISALLOW_ILLEGAL_INTERCHANGE>, this gives the same results
866 as C<L</is_strict_utf8_string>>; and if C<flags> is
867 C<UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE>, this gives the same results as
868 C<L</is_c9strict_utf8_string>>. Otherwise C<flags> may be any
869 combination of the C<UTF8_DISALLOW_I<foo>> flags understood by
870 C<L</utf8n_to_uvchr>>, with the same meanings.
873 C<L</is_utf8_invariant_string>>,
874 C<L</is_utf8_invariant_string_loc>>,
875 C<L</is_utf8_string>>,
876 C<L</is_utf8_string_loc>>,
877 C<L</is_utf8_string_loc_flags>>,
878 C<L</is_utf8_string_loclen>>,
879 C<L</is_utf8_string_loclen_flags>>,
880 C<L</is_utf8_fixed_width_buf_flags>>,
881 C<L</is_utf8_fixed_width_buf_loc_flags>>,
882 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
883 C<L</is_strict_utf8_string>>,
884 C<L</is_strict_utf8_string_loc>>,
885 C<L</is_strict_utf8_string_loclen>>,
886 C<L</is_c9strict_utf8_string>>,
887 C<L</is_c9strict_utf8_string_loc>>,
889 C<L</is_c9strict_utf8_string_loclen>>.
894 PERL_STATIC_INLINE bool
895 Perl_is_utf8_string_flags(const U8 *s, STRLEN len, const U32 flags)
897 const U8 * first_variant;
899 PERL_ARGS_ASSERT_IS_UTF8_STRING_FLAGS;
900 assert(0 == (flags & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE
901 |UTF8_DISALLOW_PERL_EXTENDED)));
904 len = strlen((const char *)s);
908 return is_utf8_string(s, len);
911 if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
912 == UTF8_DISALLOW_ILLEGAL_INTERCHANGE)
914 return is_strict_utf8_string(s, len);
917 if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
918 == UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE)
920 return is_c9strict_utf8_string(s, len);
923 if (! is_utf8_invariant_string_loc(s, len, &first_variant)) {
924 const U8* const send = s + len;
925 const U8* x = first_variant;
928 STRLEN cur_len = isUTF8_CHAR_flags(x, send, flags);
929 if (UNLIKELY(! cur_len)) {
941 =for apidoc is_utf8_string_loc
943 Like C<L</is_utf8_string>> but stores the location of the failure (in the
944 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
945 "utf8ness success") in the C<ep> pointer.
947 See also C<L</is_utf8_string_loclen>>.
952 #define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
956 =for apidoc is_utf8_string_loclen
958 Like C<L</is_utf8_string>> but stores the location of the failure (in the
959 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
960 "utf8ness success") in the C<ep> pointer, and the number of UTF-8
961 encoded characters in the C<el> pointer.
963 See also C<L</is_utf8_string_loc>>.
968 PERL_STATIC_INLINE bool
969 Perl_is_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
971 const U8 * first_variant;
973 PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN;
976 len = strlen((const char *) s);
979 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
991 const U8* const send = s + len;
992 const U8* x = first_variant;
993 STRLEN outlen = first_variant - s;
996 const STRLEN cur_len = isUTF8_CHAR(x, send);
997 if (UNLIKELY(! cur_len)) {
1017 =for apidoc isUTF8_CHAR
1019 Evaluates to non-zero if the first few bytes of the string starting at C<s> and
1020 looking no further than S<C<e - 1>> are well-formed UTF-8, as extended by Perl,
1021 that represents some code point; otherwise it evaluates to 0. If non-zero, the
1022 value gives how many bytes starting at C<s> comprise the code point's
1023 representation. Any bytes remaining before C<e>, but beyond the ones needed to
1024 form the first code point in C<s>, are not examined.
1026 The code point can be any that will fit in an IV on this machine, using Perl's
1027 extension to official UTF-8 to represent those higher than the Unicode maximum
1028 of 0x10FFFF. That means that this macro is used to efficiently decide if the
1029 next few bytes in C<s> is legal UTF-8 for a single character.
1031 Use C<L</isSTRICT_UTF8_CHAR>> to restrict the acceptable code points to those
1032 defined by Unicode to be fully interchangeable across applications;
1033 C<L</isC9_STRICT_UTF8_CHAR>> to use the L<Unicode Corrigendum
1034 #9|http://www.unicode.org/versions/corrigendum9.html> definition of allowable
1035 code points; and C<L</isUTF8_CHAR_flags>> for a more customized definition.
1037 Use C<L</is_utf8_string>>, C<L</is_utf8_string_loc>>, and
1038 C<L</is_utf8_string_loclen>> to check entire strings.
1040 Note also that a UTF-8 "invariant" character (i.e. ASCII on non-EBCDIC
1041 machines) is a valid UTF-8 character.
1045 This uses an adaptation of the table and algorithm given in
1046 https://bjoern.hoehrmann.de/utf-8/decoder/dfa/, which provides comprehensive
1047 documentation of the original version. A copyright notice for the original
1048 version is given at the beginning of this file. The Perl adapation is
1049 documented at the definition of PL_extended_utf8_dfa_tab[].
1053 PERL_STATIC_INLINE Size_t
1054 Perl_isUTF8_CHAR(const U8 * const s0, const U8 * const e)
1059 PERL_ARGS_ASSERT_ISUTF8_CHAR;
1061 /* This dfa is fast. If it accepts the input, it was for a well-formed,
1062 * code point, which can be returned immediately. Otherwise, it is either
1063 * malformed, or for the start byte FF which the dfa doesn't handle (except
1064 * on 32-bit ASCII platforms where it trivially is an error). Call a
1065 * helper function for the other platforms. */
1067 while (s < e && LIKELY(state != 1)) {
1068 state = PL_extended_utf8_dfa_tab[256
1070 + PL_extended_utf8_dfa_tab[*s]];
1079 #if defined(UV_IS_QUAD) || defined(EBCDIC)
1081 if (NATIVE_UTF8_TO_I8(*s0) == 0xFF && e - s0 >= UTF8_MAXBYTES) {
1082 return is_utf8_char_helper(s0, e, 0);
1092 =for apidoc isSTRICT_UTF8_CHAR
1094 Evaluates to non-zero if the first few bytes of the string starting at C<s> and
1095 looking no further than S<C<e - 1>> are well-formed UTF-8 that represents some
1096 Unicode code point completely acceptable for open interchange between all
1097 applications; otherwise it evaluates to 0. If non-zero, the value gives how
1098 many bytes starting at C<s> comprise the code point's representation. Any
1099 bytes remaining before C<e>, but beyond the ones needed to form the first code
1100 point in C<s>, are not examined.
1102 The largest acceptable code point is the Unicode maximum 0x10FFFF, and must not
1103 be a surrogate nor a non-character code point. Thus this excludes any code
1104 point from Perl's extended UTF-8.
1106 This is used to efficiently decide if the next few bytes in C<s> is
1107 legal Unicode-acceptable UTF-8 for a single character.
1109 Use C<L</isC9_STRICT_UTF8_CHAR>> to use the L<Unicode Corrigendum
1110 #9|http://www.unicode.org/versions/corrigendum9.html> definition of allowable
1111 code points; C<L</isUTF8_CHAR>> to check for Perl's extended UTF-8;
1112 and C<L</isUTF8_CHAR_flags>> for a more customized definition.
1114 Use C<L</is_strict_utf8_string>>, C<L</is_strict_utf8_string_loc>>, and
1115 C<L</is_strict_utf8_string_loclen>> to check entire strings.
1119 This uses an adaptation of the tables and algorithm given in
1120 https://bjoern.hoehrmann.de/utf-8/decoder/dfa/, which provides comprehensive
1121 documentation of the original version. A copyright notice for the original
1122 version is given at the beginning of this file. The Perl adapation is
1123 documented at the definition of strict_extended_utf8_dfa_tab[].
1127 PERL_STATIC_INLINE Size_t
1128 Perl_isSTRICT_UTF8_CHAR(const U8 * const s0, const U8 * const e)
1133 PERL_ARGS_ASSERT_ISSTRICT_UTF8_CHAR;
1135 while (s < e && LIKELY(state != 1)) {
1136 state = PL_strict_utf8_dfa_tab[256 + state + PL_strict_utf8_dfa_tab[*s]];
1148 /* The dfa above drops out for certain Hanguls; handle them specially */
1149 if (is_HANGUL_ED_utf8_safe(s0, e)) {
1160 =for apidoc isC9_STRICT_UTF8_CHAR
1162 Evaluates to non-zero if the first few bytes of the string starting at C<s> and
1163 looking no further than S<C<e - 1>> are well-formed UTF-8 that represents some
1164 Unicode non-surrogate code point; otherwise it evaluates to 0. If non-zero,
1165 the value gives how many bytes starting at C<s> comprise the code point's
1166 representation. Any bytes remaining before C<e>, but beyond the ones needed to
1167 form the first code point in C<s>, are not examined.
1169 The largest acceptable code point is the Unicode maximum 0x10FFFF. This
1170 differs from C<L</isSTRICT_UTF8_CHAR>> only in that it accepts non-character
1171 code points. This corresponds to
1172 L<Unicode Corrigendum #9|http://www.unicode.org/versions/corrigendum9.html>.
1173 which said that non-character code points are merely discouraged rather than
1174 completely forbidden in open interchange. See
1175 L<perlunicode/Noncharacter code points>.
1177 Use C<L</isUTF8_CHAR>> to check for Perl's extended UTF-8; and
1178 C<L</isUTF8_CHAR_flags>> for a more customized definition.
1180 Use C<L</is_c9strict_utf8_string>>, C<L</is_c9strict_utf8_string_loc>>, and
1181 C<L</is_c9strict_utf8_string_loclen>> to check entire strings.
1185 This uses an adaptation of the tables and algorithm given in
1186 https://bjoern.hoehrmann.de/utf-8/decoder/dfa/, which provides comprehensive
1187 documentation of the original version. A copyright notice for the original
1188 version is given at the beginning of this file. The Perl adapation is
1189 documented at the definition of PL_c9_utf8_dfa_tab[].
1193 PERL_STATIC_INLINE Size_t
1194 Perl_isC9_STRICT_UTF8_CHAR(const U8 * const s0, const U8 * const e)
1199 PERL_ARGS_ASSERT_ISC9_STRICT_UTF8_CHAR;
1201 while (s < e && LIKELY(state != 1)) {
1202 state = PL_c9_utf8_dfa_tab[256 + state + PL_c9_utf8_dfa_tab[*s]];
1217 =for apidoc is_strict_utf8_string_loc
1219 Like C<L</is_strict_utf8_string>> but stores the location of the failure (in the
1220 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1221 "utf8ness success") in the C<ep> pointer.
1223 See also C<L</is_strict_utf8_string_loclen>>.
1228 #define is_strict_utf8_string_loc(s, len, ep) \
1229 is_strict_utf8_string_loclen(s, len, ep, 0)
1233 =for apidoc is_strict_utf8_string_loclen
1235 Like C<L</is_strict_utf8_string>> but stores the location of the failure (in the
1236 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1237 "utf8ness success") in the C<ep> pointer, and the number of UTF-8
1238 encoded characters in the C<el> pointer.
1240 See also C<L</is_strict_utf8_string_loc>>.
1245 PERL_STATIC_INLINE bool
1246 Perl_is_strict_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
1248 const U8 * first_variant;
1250 PERL_ARGS_ASSERT_IS_STRICT_UTF8_STRING_LOCLEN;
1253 len = strlen((const char *) s);
1256 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
1268 const U8* const send = s + len;
1269 const U8* x = first_variant;
1270 STRLEN outlen = first_variant - s;
1273 const STRLEN cur_len = isSTRICT_UTF8_CHAR(x, send);
1274 if (UNLIKELY(! cur_len)) {
1294 =for apidoc is_c9strict_utf8_string_loc
1296 Like C<L</is_c9strict_utf8_string>> but stores the location of the failure (in
1297 the case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1298 "utf8ness success") in the C<ep> pointer.
1300 See also C<L</is_c9strict_utf8_string_loclen>>.
1305 #define is_c9strict_utf8_string_loc(s, len, ep) \
1306 is_c9strict_utf8_string_loclen(s, len, ep, 0)
1310 =for apidoc is_c9strict_utf8_string_loclen
1312 Like C<L</is_c9strict_utf8_string>> but stores the location of the failure (in
1313 the case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1314 "utf8ness success") in the C<ep> pointer, and the number of UTF-8 encoded
1315 characters in the C<el> pointer.
1317 See also C<L</is_c9strict_utf8_string_loc>>.
1322 PERL_STATIC_INLINE bool
1323 Perl_is_c9strict_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
1325 const U8 * first_variant;
1327 PERL_ARGS_ASSERT_IS_C9STRICT_UTF8_STRING_LOCLEN;
1330 len = strlen((const char *) s);
1333 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
1345 const U8* const send = s + len;
1346 const U8* x = first_variant;
1347 STRLEN outlen = first_variant - s;
1350 const STRLEN cur_len = isC9_STRICT_UTF8_CHAR(x, send);
1351 if (UNLIKELY(! cur_len)) {
1371 =for apidoc is_utf8_string_loc_flags
1373 Like C<L</is_utf8_string_flags>> but stores the location of the failure (in the
1374 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1375 "utf8ness success") in the C<ep> pointer.
1377 See also C<L</is_utf8_string_loclen_flags>>.
1382 #define is_utf8_string_loc_flags(s, len, ep, flags) \
1383 is_utf8_string_loclen_flags(s, len, ep, 0, flags)
1386 /* The above 3 actual functions could have been moved into the more general one
1387 * just below, and made #defines that call it with the right 'flags'. They are
1388 * currently kept separate to increase their chances of getting inlined */
1392 =for apidoc is_utf8_string_loclen_flags
1394 Like C<L</is_utf8_string_flags>> but stores the location of the failure (in the
1395 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1396 "utf8ness success") in the C<ep> pointer, and the number of UTF-8
1397 encoded characters in the C<el> pointer.
1399 See also C<L</is_utf8_string_loc_flags>>.
1404 PERL_STATIC_INLINE bool
1405 Perl_is_utf8_string_loclen_flags(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el, const U32 flags)
1407 const U8 * first_variant;
1409 PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN_FLAGS;
1410 assert(0 == (flags & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE
1411 |UTF8_DISALLOW_PERL_EXTENDED)));
1414 len = strlen((const char *) s);
1418 return is_utf8_string_loclen(s, len, ep, el);
1421 if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
1422 == UTF8_DISALLOW_ILLEGAL_INTERCHANGE)
1424 return is_strict_utf8_string_loclen(s, len, ep, el);
1427 if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
1428 == UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE)
1430 return is_c9strict_utf8_string_loclen(s, len, ep, el);
1433 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
1445 const U8* send = s + len;
1446 const U8* x = first_variant;
1447 STRLEN outlen = first_variant - s;
1450 const STRLEN cur_len = isUTF8_CHAR_flags(x, send, flags);
1451 if (UNLIKELY(! cur_len)) {
1470 =for apidoc utf8_distance
1472 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
1475 WARNING: use only if you *know* that the pointers point inside the
1481 PERL_STATIC_INLINE IV
1482 Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
1484 PERL_ARGS_ASSERT_UTF8_DISTANCE;
1486 return (a < b) ? -1 * (IV) utf8_length(a, b) : (IV) utf8_length(b, a);
1490 =for apidoc utf8_hop
1492 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
1493 forward or backward.
1495 WARNING: do not use the following unless you *know* C<off> is within
1496 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
1497 on the first byte of character or just after the last byte of a character.
1502 PERL_STATIC_INLINE U8 *
1503 Perl_utf8_hop(const U8 *s, SSize_t off)
1505 PERL_ARGS_ASSERT_UTF8_HOP;
1507 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1508 * the bitops (especially ~) can create illegal UTF-8.
1509 * In other words: in Perl UTF-8 is not just for Unicode. */
1518 while (UTF8_IS_CONTINUATION(*s))
1522 GCC_DIAG_IGNORE(-Wcast-qual)
1528 =for apidoc utf8_hop_forward
1530 Return the UTF-8 pointer C<s> displaced by up to C<off> characters,
1533 C<off> must be non-negative.
1535 C<s> must be before or equal to C<end>.
1537 When moving forward it will not move beyond C<end>.
1539 Will not exceed this limit even if the string is not valid "UTF-8".
1544 PERL_STATIC_INLINE U8 *
1545 Perl_utf8_hop_forward(const U8 *s, SSize_t off, const U8 *end)
1547 PERL_ARGS_ASSERT_UTF8_HOP_FORWARD;
1549 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1550 * the bitops (especially ~) can create illegal UTF-8.
1551 * In other words: in Perl UTF-8 is not just for Unicode. */
1557 STRLEN skip = UTF8SKIP(s);
1558 if ((STRLEN)(end - s) <= skip) {
1559 GCC_DIAG_IGNORE(-Wcast-qual)
1566 GCC_DIAG_IGNORE(-Wcast-qual)
1572 =for apidoc utf8_hop_back
1574 Return the UTF-8 pointer C<s> displaced by up to C<off> characters,
1577 C<off> must be non-positive.
1579 C<s> must be after or equal to C<start>.
1581 When moving backward it will not move before C<start>.
1583 Will not exceed this limit even if the string is not valid "UTF-8".
1588 PERL_STATIC_INLINE U8 *
1589 Perl_utf8_hop_back(const U8 *s, SSize_t off, const U8 *start)
1591 PERL_ARGS_ASSERT_UTF8_HOP_BACK;
1593 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1594 * the bitops (especially ~) can create illegal UTF-8.
1595 * In other words: in Perl UTF-8 is not just for Unicode. */
1600 while (off++ && s > start) {
1603 } while (UTF8_IS_CONTINUATION(*s) && s > start);
1606 GCC_DIAG_IGNORE(-Wcast-qual)
1612 =for apidoc utf8_hop_safe
1614 Return the UTF-8 pointer C<s> displaced by up to C<off> characters,
1615 either forward or backward.
1617 When moving backward it will not move before C<start>.
1619 When moving forward it will not move beyond C<end>.
1621 Will not exceed those limits even if the string is not valid "UTF-8".
1626 PERL_STATIC_INLINE U8 *
1627 Perl_utf8_hop_safe(const U8 *s, SSize_t off, const U8 *start, const U8 *end)
1629 PERL_ARGS_ASSERT_UTF8_HOP_SAFE;
1631 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1632 * the bitops (especially ~) can create illegal UTF-8.
1633 * In other words: in Perl UTF-8 is not just for Unicode. */
1635 assert(start <= s && s <= end);
1638 return utf8_hop_forward(s, off, end);
1641 return utf8_hop_back(s, off, start);
1647 =for apidoc is_utf8_valid_partial_char
1649 Returns 0 if the sequence of bytes starting at C<s> and looking no further than
1650 S<C<e - 1>> is the UTF-8 encoding, as extended by Perl, for one or more code
1651 points. Otherwise, it returns 1 if there exists at least one non-empty
1652 sequence of bytes that when appended to sequence C<s>, starting at position
1653 C<e> causes the entire sequence to be the well-formed UTF-8 of some code point;
1654 otherwise returns 0.
1656 In other words this returns TRUE if C<s> points to a partial UTF-8-encoded code
1659 This is useful when a fixed-length buffer is being tested for being well-formed
1660 UTF-8, but the final few bytes in it don't comprise a full character; that is,
1661 it is split somewhere in the middle of the final code point's UTF-8
1662 representation. (Presumably when the buffer is refreshed with the next chunk
1663 of data, the new first bytes will complete the partial code point.) This
1664 function is used to verify that the final bytes in the current buffer are in
1665 fact the legal beginning of some code point, so that if they aren't, the
1666 failure can be signalled without having to wait for the next read.
1670 #define is_utf8_valid_partial_char(s, e) \
1671 is_utf8_valid_partial_char_flags(s, e, 0)
1675 =for apidoc is_utf8_valid_partial_char_flags
1677 Like C<L</is_utf8_valid_partial_char>>, it returns a boolean giving whether
1678 or not the input is a valid UTF-8 encoded partial character, but it takes an
1679 extra parameter, C<flags>, which can further restrict which code points are
1682 If C<flags> is 0, this behaves identically to
1683 C<L</is_utf8_valid_partial_char>>. Otherwise C<flags> can be any combination
1684 of the C<UTF8_DISALLOW_I<foo>> flags accepted by C<L</utf8n_to_uvchr>>. If
1685 there is any sequence of bytes that can complete the input partial character in
1686 such a way that a non-prohibited character is formed, the function returns
1687 TRUE; otherwise FALSE. Non character code points cannot be determined based on
1688 partial character input. But many of the other possible excluded types can be
1689 determined from just the first one or two bytes.
1694 PERL_STATIC_INLINE bool
1695 Perl_is_utf8_valid_partial_char_flags(const U8 * const s, const U8 * const e, const U32 flags)
1697 PERL_ARGS_ASSERT_IS_UTF8_VALID_PARTIAL_CHAR_FLAGS;
1699 assert(0 == (flags & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE
1700 |UTF8_DISALLOW_PERL_EXTENDED)));
1702 if (s >= e || s + UTF8SKIP(s) <= e) {
1706 return cBOOL(is_utf8_char_helper(s, e, flags));
1711 =for apidoc is_utf8_fixed_width_buf_flags
1713 Returns TRUE if the fixed-width buffer starting at C<s> with length C<len>
1714 is entirely valid UTF-8, subject to the restrictions given by C<flags>;
1715 otherwise it returns FALSE.
1717 If C<flags> is 0, any well-formed UTF-8, as extended by Perl, is accepted
1718 without restriction. If the final few bytes of the buffer do not form a
1719 complete code point, this will return TRUE anyway, provided that
1720 C<L</is_utf8_valid_partial_char_flags>> returns TRUE for them.
1722 If C<flags> in non-zero, it can be any combination of the
1723 C<UTF8_DISALLOW_I<foo>> flags accepted by C<L</utf8n_to_uvchr>>, and with the
1726 This function differs from C<L</is_utf8_string_flags>> only in that the latter
1727 returns FALSE if the final few bytes of the string don't form a complete code
1732 #define is_utf8_fixed_width_buf_flags(s, len, flags) \
1733 is_utf8_fixed_width_buf_loclen_flags(s, len, 0, 0, flags)
1737 =for apidoc is_utf8_fixed_width_buf_loc_flags
1739 Like C<L</is_utf8_fixed_width_buf_flags>> but stores the location of the
1740 failure in the C<ep> pointer. If the function returns TRUE, C<*ep> will point
1741 to the beginning of any partial character at the end of the buffer; if there is
1742 no partial character C<*ep> will contain C<s>+C<len>.
1744 See also C<L</is_utf8_fixed_width_buf_loclen_flags>>.
1749 #define is_utf8_fixed_width_buf_loc_flags(s, len, loc, flags) \
1750 is_utf8_fixed_width_buf_loclen_flags(s, len, loc, 0, flags)
1754 =for apidoc is_utf8_fixed_width_buf_loclen_flags
1756 Like C<L</is_utf8_fixed_width_buf_loc_flags>> but stores the number of
1757 complete, valid characters found in the C<el> pointer.
1762 PERL_STATIC_INLINE bool
1763 Perl_is_utf8_fixed_width_buf_loclen_flags(const U8 * const s,
1769 const U8 * maybe_partial;
1771 PERL_ARGS_ASSERT_IS_UTF8_FIXED_WIDTH_BUF_LOCLEN_FLAGS;
1774 ep = &maybe_partial;
1777 /* If it's entirely valid, return that; otherwise see if the only error is
1778 * that the final few bytes are for a partial character */
1779 return is_utf8_string_loclen_flags(s, len, ep, el, flags)
1780 || is_utf8_valid_partial_char_flags(*ep, s + len, flags);
1783 PERL_STATIC_INLINE UV
1784 Perl_utf8n_to_uvchr_msgs(const U8 *s,
1791 /* This is the inlined portion of utf8n_to_uvchr_msgs. It handles the
1792 * simple cases, and, if necessary calls a helper function to deal with the
1793 * more complex ones. Almost all well-formed non-problematic code points
1794 * are considered simple, so that it's unlikely that the helper function
1795 * will need to be called.
1797 * This is an adaptation of the tables and algorithm given in
1798 * https://bjoern.hoehrmann.de/utf-8/decoder/dfa/, which provides
1799 * comprehensive documentation of the original version. A copyright notice
1800 * for the original version is given at the beginning of this file. The
1801 * Perl adapation is documented at the definition of PL_strict_utf8_dfa_tab[].
1804 const U8 * const s0 = s;
1805 const U8 * send = s0 + curlen;
1806 UV uv = 0; /* The 0 silences some stupid compilers */
1809 PERL_ARGS_ASSERT_UTF8N_TO_UVCHR_MSGS;
1811 /* This dfa is fast. If it accepts the input, it was for a well-formed,
1812 * non-problematic code point, which can be returned immediately.
1813 * Otherwise we call a helper function to figure out the more complicated
1816 while (s < send && LIKELY(state != 1)) {
1817 UV type = PL_strict_utf8_dfa_tab[*s];
1820 ? ((0xff >> type) & NATIVE_UTF8_TO_I8(*s))
1821 : UTF8_ACCUMULATE(uv, *s);
1822 state = PL_strict_utf8_dfa_tab[256 + state + type];
1830 *retlen = s - s0 + 1;
1839 return UNI_TO_NATIVE(uv);
1842 /* Here is potentially problematic. Use the full mechanism */
1843 return _utf8n_to_uvchr_msgs_helper(s0, curlen, retlen, flags, errors, msgs);
1846 PERL_STATIC_INLINE UV
1847 Perl_utf8_to_uvchr_buf_helper(pTHX_ const U8 *s, const U8 *send, STRLEN *retlen)
1849 PERL_ARGS_ASSERT_UTF8_TO_UVCHR_BUF_HELPER;
1853 if (! ckWARN_d(WARN_UTF8)) {
1855 /* EMPTY is not really allowed, and asserts on debugging builds. But
1856 * on non-debugging we have to deal with it, and this causes it to
1857 * return the REPLACEMENT CHARACTER, as the documentation indicates */
1858 return utf8n_to_uvchr(s, send - s, retlen,
1859 (UTF8_ALLOW_ANY | UTF8_ALLOW_EMPTY));
1862 UV ret = utf8n_to_uvchr(s, send - s, retlen, 0);
1863 if (retlen && ret == 0 && *s != '\0') {
1864 *retlen = (STRLEN) -1;
1871 /* ------------------------------- perl.h ----------------------------- */
1874 =head1 Miscellaneous Functions
1876 =for apidoc is_safe_syscall
1878 Test that the given C<pv> (with length C<len>) doesn't contain any internal
1880 If it does, set C<errno> to C<ENOENT>, optionally warn using the C<syscalls>
1881 category, and return FALSE.
1883 Return TRUE if the name is safe.
1885 C<what> and C<op_name> are used in any warning.
1887 Used by the C<IS_SAFE_SYSCALL()> macro.
1892 PERL_STATIC_INLINE bool
1893 Perl_is_safe_syscall(pTHX_ const char *pv, STRLEN len, const char *what, const char *op_name)
1895 /* While the Windows CE API provides only UCS-16 (or UTF-16) APIs
1896 * perl itself uses xce*() functions which accept 8-bit strings.
1899 PERL_ARGS_ASSERT_IS_SAFE_SYSCALL;
1903 if (UNLIKELY((null_at = (char *)memchr(pv, 0, len-1)) != NULL)) {
1904 SETERRNO(ENOENT, LIB_INVARG);
1905 Perl_ck_warner(aTHX_ packWARN(WARN_SYSCALLS),
1906 "Invalid \\0 character in %s for %s: %s\\0%s",
1907 what, op_name, pv, null_at+1);
1917 Return true if the supplied filename has a newline character
1918 immediately before the first (hopefully only) NUL.
1920 My original look at this incorrectly used the len from SvPV(), but
1921 that's incorrect, since we allow for a NUL in pv[len-1].
1923 So instead, strlen() and work from there.
1925 This allow for the user reading a filename, forgetting to chomp it,
1928 open my $foo, "$file\0";
1934 PERL_STATIC_INLINE bool
1935 S_should_warn_nl(const char *pv)
1939 PERL_ARGS_ASSERT_SHOULD_WARN_NL;
1943 return len > 0 && pv[len-1] == '\n';
1948 #if defined(PERL_IN_PP_C) || defined(PERL_IN_PP_HOT_C)
1950 PERL_STATIC_INLINE bool
1951 S_lossless_NV_to_IV(const NV nv, IV *ivp)
1953 /* This function determines if the input NV 'nv' may be converted without
1954 * loss of data to an IV. If not, it returns FALSE taking no other action.
1955 * But if it is possible, it does the conversion, returning TRUE, and
1956 * storing the converted result in '*ivp' */
1958 PERL_ARGS_ASSERT_LOSSLESS_NV_TO_IV;
1960 # if defined(Perl_isnan)
1962 if (UNLIKELY(Perl_isnan(nv))) {
1968 if (UNLIKELY(nv < IV_MIN) || UNLIKELY(nv > IV_MAX)) {
1972 if ((IV) nv != nv) {
1982 /* ------------------ pp.c, regcomp.c, toke.c, universal.c ------------ */
1984 #if defined(PERL_IN_PP_C) || defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_TOKE_C) || defined(PERL_IN_UNIVERSAL_C)
1986 #define MAX_CHARSET_NAME_LENGTH 2
1988 PERL_STATIC_INLINE const char *
1989 S_get_regex_charset_name(const U32 flags, STRLEN* const lenp)
1991 PERL_ARGS_ASSERT_GET_REGEX_CHARSET_NAME;
1993 /* Returns a string that corresponds to the name of the regex character set
1994 * given by 'flags', and *lenp is set the length of that string, which
1995 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
1998 switch (get_regex_charset(flags)) {
1999 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
2000 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
2001 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
2002 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
2003 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
2005 return ASCII_MORE_RESTRICT_PAT_MODS;
2007 /* The NOT_REACHED; hides an assert() which has a rather complex
2008 * definition in perl.h. */
2009 NOT_REACHED; /* NOTREACHED */
2010 return "?"; /* Unknown */
2017 Return false if any get magic is on the SV other than taint magic.
2021 PERL_STATIC_INLINE bool
2022 Perl_sv_only_taint_gmagic(SV *sv)
2024 MAGIC *mg = SvMAGIC(sv);
2026 PERL_ARGS_ASSERT_SV_ONLY_TAINT_GMAGIC;
2029 if (mg->mg_type != PERL_MAGIC_taint
2030 && !(mg->mg_flags & MGf_GSKIP)
2031 && mg->mg_virtual->svt_get) {
2034 mg = mg->mg_moremagic;
2040 /* ------------------ cop.h ------------------------------------------- */
2042 /* implement GIMME_V() macro */
2044 PERL_STATIC_INLINE U8
2048 U8 gimme = (PL_op->op_flags & OPf_WANT);
2052 cxix = PL_curstackinfo->si_cxsubix;
2055 assert(cxstack[cxix].blk_gimme & G_WANT);
2056 return (cxstack[cxix].blk_gimme & G_WANT);
2060 /* Enter a block. Push a new base context and return its address. */
2062 PERL_STATIC_INLINE PERL_CONTEXT *
2063 Perl_cx_pushblock(pTHX_ U8 type, U8 gimme, SV** sp, I32 saveix)
2067 PERL_ARGS_ASSERT_CX_PUSHBLOCK;
2072 cx->blk_gimme = gimme;
2073 cx->blk_oldsaveix = saveix;
2074 cx->blk_oldsp = (I32)(sp - PL_stack_base);
2075 cx->blk_oldcop = PL_curcop;
2076 cx->blk_oldmarksp = (I32)(PL_markstack_ptr - PL_markstack);
2077 cx->blk_oldscopesp = PL_scopestack_ix;
2078 cx->blk_oldpm = PL_curpm;
2079 cx->blk_old_tmpsfloor = PL_tmps_floor;
2081 PL_tmps_floor = PL_tmps_ix;
2082 CX_DEBUG(cx, "PUSH");
2087 /* Exit a block (RETURN and LAST). */
2089 PERL_STATIC_INLINE void
2090 Perl_cx_popblock(pTHX_ PERL_CONTEXT *cx)
2092 PERL_ARGS_ASSERT_CX_POPBLOCK;
2094 CX_DEBUG(cx, "POP");
2095 /* these 3 are common to cx_popblock and cx_topblock */
2096 PL_markstack_ptr = PL_markstack + cx->blk_oldmarksp;
2097 PL_scopestack_ix = cx->blk_oldscopesp;
2098 PL_curpm = cx->blk_oldpm;
2100 /* LEAVE_SCOPE() should have made this true. /(?{})/ cheats
2101 * and leaves a CX entry lying around for repeated use, so
2102 * skip for multicall */ \
2103 assert( (CxTYPE(cx) == CXt_SUB && CxMULTICALL(cx))
2104 || PL_savestack_ix == cx->blk_oldsaveix);
2105 PL_curcop = cx->blk_oldcop;
2106 PL_tmps_floor = cx->blk_old_tmpsfloor;
2109 /* Continue a block elsewhere (e.g. NEXT, REDO, GOTO).
2110 * Whereas cx_popblock() restores the state to the point just before
2111 * cx_pushblock() was called, cx_topblock() restores it to the point just
2112 * *after* cx_pushblock() was called. */
2114 PERL_STATIC_INLINE void
2115 Perl_cx_topblock(pTHX_ PERL_CONTEXT *cx)
2117 PERL_ARGS_ASSERT_CX_TOPBLOCK;
2119 CX_DEBUG(cx, "TOP");
2120 /* these 3 are common to cx_popblock and cx_topblock */
2121 PL_markstack_ptr = PL_markstack + cx->blk_oldmarksp;
2122 PL_scopestack_ix = cx->blk_oldscopesp;
2123 PL_curpm = cx->blk_oldpm;
2125 PL_stack_sp = PL_stack_base + cx->blk_oldsp;
2129 PERL_STATIC_INLINE void
2130 Perl_cx_pushsub(pTHX_ PERL_CONTEXT *cx, CV *cv, OP *retop, bool hasargs)
2132 U8 phlags = CX_PUSHSUB_GET_LVALUE_MASK(Perl_was_lvalue_sub);
2134 PERL_ARGS_ASSERT_CX_PUSHSUB;
2136 PERL_DTRACE_PROBE_ENTRY(cv);
2137 cx->blk_sub.old_cxsubix = PL_curstackinfo->si_cxsubix;
2138 PL_curstackinfo->si_cxsubix = cx - PL_curstackinfo->si_cxstack;
2139 cx->blk_sub.cv = cv;
2140 cx->blk_sub.olddepth = CvDEPTH(cv);
2141 cx->blk_sub.prevcomppad = PL_comppad;
2142 cx->cx_type |= (hasargs) ? CXp_HASARGS : 0;
2143 cx->blk_sub.retop = retop;
2144 SvREFCNT_inc_simple_void_NN(cv);
2145 cx->blk_u16 = PL_op->op_private & (phlags|OPpDEREF);
2149 /* subsets of cx_popsub() */
2151 PERL_STATIC_INLINE void
2152 Perl_cx_popsub_common(pTHX_ PERL_CONTEXT *cx)
2156 PERL_ARGS_ASSERT_CX_POPSUB_COMMON;
2157 assert(CxTYPE(cx) == CXt_SUB);
2159 PL_comppad = cx->blk_sub.prevcomppad;
2160 PL_curpad = LIKELY(PL_comppad) ? AvARRAY(PL_comppad) : NULL;
2161 cv = cx->blk_sub.cv;
2162 CvDEPTH(cv) = cx->blk_sub.olddepth;
2163 cx->blk_sub.cv = NULL;
2165 PL_curstackinfo->si_cxsubix = cx->blk_sub.old_cxsubix;
2169 /* handle the @_ part of leaving a sub */
2171 PERL_STATIC_INLINE void
2172 Perl_cx_popsub_args(pTHX_ PERL_CONTEXT *cx)
2176 PERL_ARGS_ASSERT_CX_POPSUB_ARGS;
2177 assert(CxTYPE(cx) == CXt_SUB);
2178 assert(AvARRAY(MUTABLE_AV(
2179 PadlistARRAY(CvPADLIST(cx->blk_sub.cv))[
2180 CvDEPTH(cx->blk_sub.cv)])) == PL_curpad);
2182 CX_POP_SAVEARRAY(cx);
2183 av = MUTABLE_AV(PAD_SVl(0));
2184 if (UNLIKELY(AvREAL(av)))
2185 /* abandon @_ if it got reified */
2186 clear_defarray(av, 0);
2193 PERL_STATIC_INLINE void
2194 Perl_cx_popsub(pTHX_ PERL_CONTEXT *cx)
2196 PERL_ARGS_ASSERT_CX_POPSUB;
2197 assert(CxTYPE(cx) == CXt_SUB);
2199 PERL_DTRACE_PROBE_RETURN(cx->blk_sub.cv);
2203 cx_popsub_common(cx);
2207 PERL_STATIC_INLINE void
2208 Perl_cx_pushformat(pTHX_ PERL_CONTEXT *cx, CV *cv, OP *retop, GV *gv)
2210 PERL_ARGS_ASSERT_CX_PUSHFORMAT;
2212 cx->blk_format.old_cxsubix = PL_curstackinfo->si_cxsubix;
2213 PL_curstackinfo->si_cxsubix= cx - PL_curstackinfo->si_cxstack;
2214 cx->blk_format.cv = cv;
2215 cx->blk_format.retop = retop;
2216 cx->blk_format.gv = gv;
2217 cx->blk_format.dfoutgv = PL_defoutgv;
2218 cx->blk_format.prevcomppad = PL_comppad;
2221 SvREFCNT_inc_simple_void_NN(cv);
2223 SvREFCNT_inc_void(cx->blk_format.dfoutgv);
2227 PERL_STATIC_INLINE void
2228 Perl_cx_popformat(pTHX_ PERL_CONTEXT *cx)
2233 PERL_ARGS_ASSERT_CX_POPFORMAT;
2234 assert(CxTYPE(cx) == CXt_FORMAT);
2236 dfout = cx->blk_format.dfoutgv;
2238 cx->blk_format.dfoutgv = NULL;
2239 SvREFCNT_dec_NN(dfout);
2241 PL_comppad = cx->blk_format.prevcomppad;
2242 PL_curpad = LIKELY(PL_comppad) ? AvARRAY(PL_comppad) : NULL;
2243 cv = cx->blk_format.cv;
2244 cx->blk_format.cv = NULL;
2246 SvREFCNT_dec_NN(cv);
2247 PL_curstackinfo->si_cxsubix = cx->blk_format.old_cxsubix;
2251 PERL_STATIC_INLINE void
2252 Perl_cx_pusheval(pTHX_ PERL_CONTEXT *cx, OP *retop, SV *namesv)
2254 PERL_ARGS_ASSERT_CX_PUSHEVAL;
2256 cx->blk_eval.old_cxsubix = PL_curstackinfo->si_cxsubix;
2257 PL_curstackinfo->si_cxsubix= cx - PL_curstackinfo->si_cxstack;
2258 cx->blk_eval.retop = retop;
2259 cx->blk_eval.old_namesv = namesv;
2260 cx->blk_eval.old_eval_root = PL_eval_root;
2261 cx->blk_eval.cur_text = PL_parser ? PL_parser->linestr : NULL;
2262 cx->blk_eval.cv = NULL; /* later set by doeval_compile() */
2263 cx->blk_eval.cur_top_env = PL_top_env;
2265 assert(!(PL_in_eval & ~ 0x3F));
2266 assert(!(PL_op->op_type & ~0x1FF));
2267 cx->blk_u16 = (PL_in_eval & 0x3F) | ((U16)PL_op->op_type << 7);
2271 PERL_STATIC_INLINE void
2272 Perl_cx_popeval(pTHX_ PERL_CONTEXT *cx)
2276 PERL_ARGS_ASSERT_CX_POPEVAL;
2277 assert(CxTYPE(cx) == CXt_EVAL);
2279 PL_in_eval = CxOLD_IN_EVAL(cx);
2280 assert(!(PL_in_eval & 0xc0));
2281 PL_eval_root = cx->blk_eval.old_eval_root;
2282 sv = cx->blk_eval.cur_text;
2283 if (sv && CxEVAL_TXT_REFCNTED(cx)) {
2284 cx->blk_eval.cur_text = NULL;
2285 SvREFCNT_dec_NN(sv);
2288 sv = cx->blk_eval.old_namesv;
2290 cx->blk_eval.old_namesv = NULL;
2291 SvREFCNT_dec_NN(sv);
2293 PL_curstackinfo->si_cxsubix = cx->blk_eval.old_cxsubix;
2297 /* push a plain loop, i.e.
2299 * while (cond) { block }
2300 * for (init;cond;continue) { block }
2301 * This loop can be last/redo'ed etc.
2304 PERL_STATIC_INLINE void
2305 Perl_cx_pushloop_plain(pTHX_ PERL_CONTEXT *cx)
2307 PERL_ARGS_ASSERT_CX_PUSHLOOP_PLAIN;
2308 cx->blk_loop.my_op = cLOOP;
2312 /* push a true for loop, i.e.
2313 * for var (list) { block }
2316 PERL_STATIC_INLINE void
2317 Perl_cx_pushloop_for(pTHX_ PERL_CONTEXT *cx, void *itervarp, SV* itersave)
2319 PERL_ARGS_ASSERT_CX_PUSHLOOP_FOR;
2321 /* this one line is common with cx_pushloop_plain */
2322 cx->blk_loop.my_op = cLOOP;
2324 cx->blk_loop.itervar_u.svp = (SV**)itervarp;
2325 cx->blk_loop.itersave = itersave;
2327 cx->blk_loop.oldcomppad = PL_comppad;
2332 /* pop all loop types, including plain */
2334 PERL_STATIC_INLINE void
2335 Perl_cx_poploop(pTHX_ PERL_CONTEXT *cx)
2337 PERL_ARGS_ASSERT_CX_POPLOOP;
2339 assert(CxTYPE_is_LOOP(cx));
2340 if ( CxTYPE(cx) == CXt_LOOP_ARY
2341 || CxTYPE(cx) == CXt_LOOP_LAZYSV)
2343 /* Free ary or cur. This assumes that state_u.ary.ary
2344 * aligns with state_u.lazysv.cur. See cx_dup() */
2345 SV *sv = cx->blk_loop.state_u.lazysv.cur;
2346 cx->blk_loop.state_u.lazysv.cur = NULL;
2347 SvREFCNT_dec_NN(sv);
2348 if (CxTYPE(cx) == CXt_LOOP_LAZYSV) {
2349 sv = cx->blk_loop.state_u.lazysv.end;
2350 cx->blk_loop.state_u.lazysv.end = NULL;
2351 SvREFCNT_dec_NN(sv);
2354 if (cx->cx_type & (CXp_FOR_PAD|CXp_FOR_GV)) {
2356 SV **svp = (cx)->blk_loop.itervar_u.svp;
2357 if ((cx->cx_type & CXp_FOR_GV))
2358 svp = &GvSV((GV*)svp);
2360 *svp = cx->blk_loop.itersave;
2361 cx->blk_loop.itersave = NULL;
2362 SvREFCNT_dec(cursv);
2367 PERL_STATIC_INLINE void
2368 Perl_cx_pushwhen(pTHX_ PERL_CONTEXT *cx)
2370 PERL_ARGS_ASSERT_CX_PUSHWHEN;
2372 cx->blk_givwhen.leave_op = cLOGOP->op_other;
2376 PERL_STATIC_INLINE void
2377 Perl_cx_popwhen(pTHX_ PERL_CONTEXT *cx)
2379 PERL_ARGS_ASSERT_CX_POPWHEN;
2380 assert(CxTYPE(cx) == CXt_WHEN);
2382 PERL_UNUSED_ARG(cx);
2383 PERL_UNUSED_CONTEXT;
2384 /* currently NOOP */
2388 PERL_STATIC_INLINE void
2389 Perl_cx_pushgiven(pTHX_ PERL_CONTEXT *cx, SV *orig_defsv)
2391 PERL_ARGS_ASSERT_CX_PUSHGIVEN;
2393 cx->blk_givwhen.leave_op = cLOGOP->op_other;
2394 cx->blk_givwhen.defsv_save = orig_defsv;
2398 PERL_STATIC_INLINE void
2399 Perl_cx_popgiven(pTHX_ PERL_CONTEXT *cx)
2403 PERL_ARGS_ASSERT_CX_POPGIVEN;
2404 assert(CxTYPE(cx) == CXt_GIVEN);
2406 sv = GvSV(PL_defgv);
2407 GvSV(PL_defgv) = cx->blk_givwhen.defsv_save;
2408 cx->blk_givwhen.defsv_save = NULL;
2412 /* ------------------ util.h ------------------------------------------- */
2415 =head1 Miscellaneous Functions
2419 Returns true if the leading C<len> bytes of the strings C<s1> and C<s2> are the
2421 case-insensitively; false otherwise. Uppercase and lowercase ASCII range bytes
2422 match themselves and their opposite case counterparts. Non-cased and non-ASCII
2423 range bytes match only themselves.
2428 PERL_STATIC_INLINE I32
2429 Perl_foldEQ(const char *s1, const char *s2, I32 len)
2431 const U8 *a = (const U8 *)s1;
2432 const U8 *b = (const U8 *)s2;
2434 PERL_ARGS_ASSERT_FOLDEQ;
2439 if (*a != *b && *a != PL_fold[*b])
2446 PERL_STATIC_INLINE I32
2447 Perl_foldEQ_latin1(const char *s1, const char *s2, I32 len)
2449 /* Compare non-UTF-8 using Unicode (Latin1) semantics. Works on all folds
2450 * representable without UTF-8, except for LATIN_SMALL_LETTER_SHARP_S, and
2451 * does not check for this. Nor does it check that the strings each have
2452 * at least 'len' characters. */
2454 const U8 *a = (const U8 *)s1;
2455 const U8 *b = (const U8 *)s2;
2457 PERL_ARGS_ASSERT_FOLDEQ_LATIN1;
2462 if (*a != *b && *a != PL_fold_latin1[*b]) {
2471 =for apidoc foldEQ_locale
2473 Returns true if the leading C<len> bytes of the strings C<s1> and C<s2> are the
2474 same case-insensitively in the current locale; false otherwise.
2479 PERL_STATIC_INLINE I32
2480 Perl_foldEQ_locale(const char *s1, const char *s2, I32 len)
2483 const U8 *a = (const U8 *)s1;
2484 const U8 *b = (const U8 *)s2;
2486 PERL_ARGS_ASSERT_FOLDEQ_LOCALE;
2491 if (*a != *b && *a != PL_fold_locale[*b])
2498 #if ! defined (HAS_MEMRCHR) && (defined(PERL_CORE) || defined(PERL_EXT))
2500 PERL_STATIC_INLINE void *
2501 S_my_memrchr(const char * s, const char c, const STRLEN len)
2503 /* memrchr(), since many platforms lack it */
2505 const char * t = s + len - 1;
2507 PERL_ARGS_ASSERT_MY_MEMRCHR;
2522 * ex: set ts=8 sts=4 sw=4 et: