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 * http://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 S_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 *
57 ? Perl_cvgv_from_hek(aTHX_ sv)
58 : ((XPVCV*)MUTABLE_PTR(SvANY(sv)))->xcv_gv_u.xcv_gv;
61 PERL_STATIC_INLINE I32 *
62 S_CvDEPTHp(const CV * const sv)
64 assert(SvTYPE(sv) == SVt_PVCV || SvTYPE(sv) == SVt_PVFM);
65 return &((XPVCV*)SvANY(sv))->xcv_depth;
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.
80 PERL_STATIC_INLINE char *
81 S_strip_spaces(pTHX_ const char * orig, STRLEN * const len)
85 tmpsv = newSVpvn_flags(orig, *len, SVs_TEMP);
93 *len = tmps - SvPVX(tmpsv);
98 /* ------------------------------- mg.h ------------------------------- */
100 #if defined(PERL_CORE) || defined(PERL_EXT)
101 /* assumes get-magic and stringification have already occurred */
102 PERL_STATIC_INLINE STRLEN
103 S_MgBYTEPOS(pTHX_ MAGIC *mg, SV *sv, const char *s, STRLEN len)
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;
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);
118 /* ------------------------------- pad.h ------------------------------ */
120 #if defined(PERL_IN_PAD_C) || defined(PERL_IN_OP_C)
121 PERL_STATIC_INLINE bool
122 PadnameIN_SCOPE(const PADNAME * const pn, const U32 seq)
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 */
130 if (COP_SEQ_RANGE_HIGH(pn) == PERL_PADSEQ_INTRO) {
131 /* in compiling scope */
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))
140 (COP_SEQ_RANGE_LOW(pn) > COP_SEQ_RANGE_HIGH(pn))
142 ( seq > COP_SEQ_RANGE_LOW(pn)
143 || seq <= COP_SEQ_RANGE_HIGH(pn))
145 : ( seq > COP_SEQ_RANGE_LOW(pn)
146 && seq <= COP_SEQ_RANGE_HIGH(pn))
153 /* ------------------------------- pp.h ------------------------------- */
155 PERL_STATIC_INLINE I32
158 DEBUG_s(DEBUG_v(PerlIO_printf(Perl_debug_log,
159 "MARK top %p %" IVdf "\n",
161 (IV)*PL_markstack_ptr)));
162 return *PL_markstack_ptr;
165 PERL_STATIC_INLINE I32
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--;
176 /* ----------------------------- regexp.h ----------------------------- */
178 PERL_STATIC_INLINE struct regexp *
179 S_ReANY(const REGEXP * const re)
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;
187 /* ------------------------------- sv.h ------------------------------- */
189 PERL_STATIC_INLINE SV *
190 S_SvREFCNT_inc(SV *sv)
192 if (LIKELY(sv != NULL))
196 PERL_STATIC_INLINE SV *
197 S_SvREFCNT_inc_NN(SV *sv)
202 PERL_STATIC_INLINE void
203 S_SvREFCNT_inc_void(SV *sv)
205 if (LIKELY(sv != NULL))
208 PERL_STATIC_INLINE void
209 S_SvREFCNT_dec(pTHX_ SV *sv)
211 if (LIKELY(sv != NULL)) {
212 U32 rc = SvREFCNT(sv);
214 SvREFCNT(sv) = rc - 1;
216 Perl_sv_free2(aTHX_ sv, rc);
220 PERL_STATIC_INLINE void
221 S_SvREFCNT_dec_NN(pTHX_ SV *sv)
223 U32 rc = SvREFCNT(sv);
225 SvREFCNT(sv) = rc - 1;
227 Perl_sv_free2(aTHX_ sv, rc);
230 PERL_STATIC_INLINE void
234 if (SvOBJECT(SvRV(sv))) HvAMAGIC_on(SvSTASH(SvRV(sv)));
236 PERL_STATIC_INLINE void
239 if (SvROK(sv) && SvOBJECT(SvRV(sv)))
240 HvAMAGIC_off(SvSTASH(SvRV(sv)));
243 PERL_STATIC_INLINE U32
244 S_SvPADSTALE_on(SV *sv)
246 assert(!(SvFLAGS(sv) & SVs_PADTMP));
247 return SvFLAGS(sv) |= SVs_PADSTALE;
249 PERL_STATIC_INLINE U32
250 S_SvPADSTALE_off(SV *sv)
252 assert(!(SvFLAGS(sv) & SVs_PADTMP));
253 return SvFLAGS(sv) &= ~SVs_PADSTALE;
255 #if defined(PERL_CORE) || defined (PERL_EXT)
256 PERL_STATIC_INLINE STRLEN
257 S_sv_or_pv_pos_u2b(pTHX_ SV *sv, const char *pv, STRLEN pos, STRLEN *lenp)
259 PERL_ARGS_ASSERT_SV_OR_PV_POS_U2B;
261 U8 *hopped = utf8_hop((U8 *)pv, pos);
262 if (lenp) *lenp = (STRLEN)(utf8_hop(hopped, *lenp) - hopped);
263 return (STRLEN)(hopped - (U8 *)pv);
265 return sv_pos_u2b_flags(sv,pos,lenp,SV_CONST_RETURN);
269 /* ------------------------------- handy.h ------------------------------- */
271 /* saves machine code for a common noreturn idiom typically used in Newx*() */
272 GCC_DIAG_IGNORE_DECL(-Wunused-function);
274 S_croak_memory_wrap(void)
276 Perl_croak_nocontext("%s",PL_memory_wrap);
278 GCC_DIAG_RESTORE_DECL;
280 /* ------------------------------- utf8.h ------------------------------- */
283 =head1 Unicode Support
286 PERL_STATIC_INLINE void
287 S_append_utf8_from_native_byte(const U8 byte, U8** dest)
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 */
292 PERL_ARGS_ASSERT_APPEND_UTF8_FROM_NATIVE_BYTE;
294 if (NATIVE_BYTE_IS_INVARIANT(byte))
297 *((*dest)++) = UTF8_EIGHT_BIT_HI(byte);
298 *((*dest)++) = UTF8_EIGHT_BIT_LO(byte);
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.
313 PERL_STATIC_INLINE UV
314 Perl_valid_utf8_to_uvchr(const U8 *s, STRLEN *retlen)
316 const UV expectlen = UTF8SKIP(s);
317 const U8* send = s + expectlen;
320 PERL_ARGS_ASSERT_VALID_UTF8_TO_UVCHR;
326 /* An invariant is trivially returned */
327 if (expectlen == 1) {
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);
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);
342 return UNI_TO_NATIVE(uv);
347 =for apidoc is_utf8_invariant_string
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.
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).
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>>,
376 C<L</is_c9strict_utf8_string_loclen>>.
382 #define is_utf8_invariant_string(s, len) \
383 is_utf8_invariant_string_loc(s, len, NULL)
386 =for apidoc is_utf8_invariant_string_loc
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>.
396 PERL_STATIC_INLINE bool
397 S_is_utf8_invariant_string_loc(const U8* const s, STRLEN len, const U8 ** ep)
402 PERL_ARGS_ASSERT_IS_UTF8_INVARIANT_STRING_LOC;
405 len = strlen((const char *)s);
410 /* This looks like 0x010101... */
411 # define PERL_COUNT_MULTIPLIER (~ (UINTMAX_C(0)) / 0xFF)
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)
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) \
425 & PERL_WORD_BOUNDARY_MASK) >> 2))))
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
435 /* This term is wordsize if subword; 0 if not */
436 + PERL_WORDSIZE * PERL_IS_SUBWORD_ADDR(x)
439 - (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK))
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)) {
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 */
458 if ((* (PERL_UINTMAX_T *) x) & PERL_VARIANTS_WORD_MASK) {
460 /* Found a variant. Just return if caller doesn't want its
466 # if BYTEORDER == 0x1234 || BYTEORDER == 0x12345678 \
467 || BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
469 *ep = x + _variant_byte_number(* (PERL_UINTMAX_T *) x);
470 assert(*ep >= s && *ep < send);
474 # else /* If weird byte order, drop into next loop to do byte-at-a-time
483 } while (x + PERL_WORDSIZE <= send);
486 #endif /* End of ! EBCDIC */
488 /* Process per-byte */
490 if (! UTF8_IS_INVARIANT(*x)) {
506 PERL_STATIC_INLINE unsigned int
507 S__variant_byte_number(PERL_UINTMAX_T word)
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 */
515 /* Get just the msb bits of each byte */
516 word &= PERL_VARIANTS_WORD_MASK;
518 # ifdef USING_MSVC6 /* VC6 has some issues with the normal code, and the
519 easiest thing is to hide that from the callers */
522 const U8 * s = (U8 *) &word;
525 for (i = 0; i < sizeof(word); i++ ) {
531 Perl_croak(aTHX_ "panic: %s: %d: unexpected zero word\n",
535 # elif BYTEORDER == 0x1234 || BYTEORDER == 0x12345678
537 /* Bytes are stored like
538 * Byte8 ... Byte2 Byte1
539 * 63..56...15...8 7...0
542 * https://stackoverflow.com/questions/757059/position-of-least-significant-bit-that-is-set
544 * The word will look this this, with a rightmost set bit in position 's':
545 * ('x's are don't cares)
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
552 * 0..0011..1 The xor with x..xx10..0 clears that remainder, sets
554 * 0..0100..0 Add 1 to clear the word except for the bit in 's'
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 */
560 word = 1 + (word ^ (word - 1));
562 # elif BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
564 /* Bytes are stored like
565 * Byte1 Byte2 ... Byte8
566 * 63..56 55..47 ... 7...0
568 * Isolate the msb; http://codeforces.com/blog/entry/10330
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. */
579 word |= word >> 32; /* This should get optimized out on 32-bit systems. */
581 /* Then subtracting the right shift by 1 clears all but the left-most of
582 * the 1 bits, which is our desired result */
586 # error Unexpected byte order
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 */
600 /* Here, word contains the position 7..63 of that bit. Convert to 0..7 */
601 word = ((word + 1) >> 3) - 1;
603 # if BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
605 /* And invert the result */
606 word = CHARBITS - word - 1;
610 return (unsigned int) word;
614 #if defined(PERL_CORE) || defined(PERL_EXT)
617 =for apidoc variant_under_utf8_count
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.
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.
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.
635 C<L<perlapi/is_utf8_invariant_string>>
637 C<L<perlapi/is_utf8_invariant_string_loc>>,
643 PERL_STATIC_INLINE Size_t
644 S_variant_under_utf8_count(const U8* const s, const U8* const e)
649 PERL_ARGS_ASSERT_VARIANT_UNDER_UTF8_COUNT;
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))
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++);
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;
675 } while (x + PERL_WORDSIZE <= e);
680 /* Process per-byte */
682 if (! UTF8_IS_INVARIANT(*x)) {
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
702 =for apidoc is_utf8_string
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'.
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.
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>>,
728 #define is_utf8_string(s, len) is_utf8_string_loclen(s, len, NULL, NULL)
730 #if defined(PERL_CORE) || defined (PERL_EXT)
733 =for apidoc is_utf8_non_invariant_string
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.
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
745 C<L<perlapi/is_utf8_invariant_string>>,
746 C<L<perlapi/is_utf8_string>>
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.
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.
759 PERL_STATIC_INLINE bool
760 S_is_utf8_non_invariant_string(const U8* const s, STRLEN len)
762 const U8 * first_variant;
764 PERL_ARGS_ASSERT_IS_UTF8_NON_INVARIANT_STRING;
766 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
770 return is_utf8_string(first_variant, len - (first_variant - s));
776 =for apidoc is_strict_utf8_string
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'.
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.
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>>,
806 C<L</is_c9strict_utf8_string_loclen>>.
811 #define is_strict_utf8_string(s, len) is_strict_utf8_string_loclen(s, len, NULL, NULL)
814 =for apidoc is_c9strict_utf8_string
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'.
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
827 L<Corrigendum #9|http://www.unicode.org/versions/corrigendum9.html>.
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>>,
846 C<L</is_c9strict_utf8_string_loclen>>.
851 #define is_c9strict_utf8_string(s, len) is_c9strict_utf8_string_loclen(s, len, NULL, 0)
854 =for apidoc is_utf8_string_flags
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'.
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.
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>>,
888 C<L</is_c9strict_utf8_string_loclen>>.
893 PERL_STATIC_INLINE bool
894 S_is_utf8_string_flags(const U8 *s, STRLEN len, const U32 flags)
896 const U8 * first_variant;
898 PERL_ARGS_ASSERT_IS_UTF8_STRING_FLAGS;
899 assert(0 == (flags & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE
900 |UTF8_DISALLOW_PERL_EXTENDED)));
903 len = strlen((const char *)s);
907 return is_utf8_string(s, len);
910 if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
911 == UTF8_DISALLOW_ILLEGAL_INTERCHANGE)
913 return is_strict_utf8_string(s, len);
916 if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
917 == UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE)
919 return is_c9strict_utf8_string(s, len);
922 if (! is_utf8_invariant_string_loc(s, len, &first_variant)) {
923 const U8* const send = s + len;
924 const U8* x = first_variant;
927 STRLEN cur_len = isUTF8_CHAR_flags(x, send, flags);
928 if (UNLIKELY(! cur_len)) {
940 =for apidoc is_utf8_string_loc
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.
946 See also C<L</is_utf8_string_loclen>>.
951 #define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
955 =for apidoc is_utf8_string_loclen
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.
962 See also C<L</is_utf8_string_loc>>.
967 PERL_STATIC_INLINE bool
968 Perl_is_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
970 const U8 * first_variant;
972 PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN;
975 len = strlen((const char *) s);
978 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
990 const U8* const send = s + len;
991 const U8* x = first_variant;
992 STRLEN outlen = first_variant - s;
995 const STRLEN cur_len = isUTF8_CHAR(x, send);
996 if (UNLIKELY(! cur_len)) {
1016 =for apidoc Am|STRLEN|isUTF8_CHAR|const U8 *s|const U8 *e
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.
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.
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.
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.
1039 Note also that a UTF-8 "invariant" character (i.e. ASCII on non-EBCDIC
1040 machines) is a valid UTF-8 character.
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 perl_extended_utf8_dfa_tab[].
1052 PERL_STATIC_INLINE Size_t
1053 S_isUTF8_CHAR(const U8 * const s0, const U8 * const e)
1058 PERL_ARGS_ASSERT_ISUTF8_CHAR;
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. */
1066 while (s < e && LIKELY(state != 1)) {
1067 state = perl_extended_utf8_dfa_tab[256
1069 + perl_extended_utf8_dfa_tab[*s]];
1078 #if defined(UV_IS_QUAD) || defined(EBCDIC)
1080 if (NATIVE_UTF8_TO_I8(*s0) == 0xFF && e - s0 >= UTF8_MAXBYTES) {
1081 return _is_utf8_char_helper(s0, e, 0);
1091 =for apidoc is_strict_utf8_string_loc
1093 Like C<L</is_strict_utf8_string>> but stores the location of the failure (in the
1094 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1095 "utf8ness success") in the C<ep> pointer.
1097 See also C<L</is_strict_utf8_string_loclen>>.
1102 #define is_strict_utf8_string_loc(s, len, ep) \
1103 is_strict_utf8_string_loclen(s, len, ep, 0)
1107 =for apidoc is_strict_utf8_string_loclen
1109 Like C<L</is_strict_utf8_string>> but stores the location of the failure (in the
1110 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1111 "utf8ness success") in the C<ep> pointer, and the number of UTF-8
1112 encoded characters in the C<el> pointer.
1114 See also C<L</is_strict_utf8_string_loc>>.
1119 PERL_STATIC_INLINE bool
1120 S_is_strict_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
1122 const U8 * first_variant;
1124 PERL_ARGS_ASSERT_IS_STRICT_UTF8_STRING_LOCLEN;
1127 len = strlen((const char *) s);
1130 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
1142 const U8* const send = s + len;
1143 const U8* x = first_variant;
1144 STRLEN outlen = first_variant - s;
1147 const STRLEN cur_len = isSTRICT_UTF8_CHAR(x, send);
1148 if (UNLIKELY(! cur_len)) {
1168 =for apidoc is_c9strict_utf8_string_loc
1170 Like C<L</is_c9strict_utf8_string>> but stores the location of the failure (in
1171 the case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1172 "utf8ness success") in the C<ep> pointer.
1174 See also C<L</is_c9strict_utf8_string_loclen>>.
1179 #define is_c9strict_utf8_string_loc(s, len, ep) \
1180 is_c9strict_utf8_string_loclen(s, len, ep, 0)
1184 =for apidoc is_c9strict_utf8_string_loclen
1186 Like C<L</is_c9strict_utf8_string>> but stores the location of the failure (in
1187 the case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1188 "utf8ness success") in the C<ep> pointer, and the number of UTF-8 encoded
1189 characters in the C<el> pointer.
1191 See also C<L</is_c9strict_utf8_string_loc>>.
1196 PERL_STATIC_INLINE bool
1197 S_is_c9strict_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
1199 const U8 * first_variant;
1201 PERL_ARGS_ASSERT_IS_C9STRICT_UTF8_STRING_LOCLEN;
1204 len = strlen((const char *) s);
1207 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
1219 const U8* const send = s + len;
1220 const U8* x = first_variant;
1221 STRLEN outlen = first_variant - s;
1224 const STRLEN cur_len = isC9_STRICT_UTF8_CHAR(x, send);
1225 if (UNLIKELY(! cur_len)) {
1245 =for apidoc is_utf8_string_loc_flags
1247 Like C<L</is_utf8_string_flags>> but stores the location of the failure (in the
1248 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1249 "utf8ness success") in the C<ep> pointer.
1251 See also C<L</is_utf8_string_loclen_flags>>.
1256 #define is_utf8_string_loc_flags(s, len, ep, flags) \
1257 is_utf8_string_loclen_flags(s, len, ep, 0, flags)
1260 /* The above 3 actual functions could have been moved into the more general one
1261 * just below, and made #defines that call it with the right 'flags'. They are
1262 * currently kept separate to increase their chances of getting inlined */
1266 =for apidoc is_utf8_string_loclen_flags
1268 Like C<L</is_utf8_string_flags>> but stores the location of the failure (in the
1269 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1270 "utf8ness success") in the C<ep> pointer, and the number of UTF-8
1271 encoded characters in the C<el> pointer.
1273 See also C<L</is_utf8_string_loc_flags>>.
1278 PERL_STATIC_INLINE bool
1279 S_is_utf8_string_loclen_flags(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el, const U32 flags)
1281 const U8 * first_variant;
1283 PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN_FLAGS;
1284 assert(0 == (flags & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE
1285 |UTF8_DISALLOW_PERL_EXTENDED)));
1288 len = strlen((const char *) s);
1292 return is_utf8_string_loclen(s, len, ep, el);
1295 if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
1296 == UTF8_DISALLOW_ILLEGAL_INTERCHANGE)
1298 return is_strict_utf8_string_loclen(s, len, ep, el);
1301 if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
1302 == UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE)
1304 return is_c9strict_utf8_string_loclen(s, len, ep, el);
1307 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
1319 const U8* send = s + len;
1320 const U8* x = first_variant;
1321 STRLEN outlen = first_variant - s;
1324 const STRLEN cur_len = isUTF8_CHAR_flags(x, send, flags);
1325 if (UNLIKELY(! cur_len)) {
1344 =for apidoc utf8_distance
1346 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
1349 WARNING: use only if you *know* that the pointers point inside the
1355 PERL_STATIC_INLINE IV
1356 Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
1358 PERL_ARGS_ASSERT_UTF8_DISTANCE;
1360 return (a < b) ? -1 * (IV) utf8_length(a, b) : (IV) utf8_length(b, a);
1364 =for apidoc utf8_hop
1366 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
1367 forward or backward.
1369 WARNING: do not use the following unless you *know* C<off> is within
1370 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
1371 on the first byte of character or just after the last byte of a character.
1376 PERL_STATIC_INLINE U8 *
1377 Perl_utf8_hop(const U8 *s, SSize_t off)
1379 PERL_ARGS_ASSERT_UTF8_HOP;
1381 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1382 * the bitops (especially ~) can create illegal UTF-8.
1383 * In other words: in Perl UTF-8 is not just for Unicode. */
1392 while (UTF8_IS_CONTINUATION(*s))
1396 GCC_DIAG_IGNORE(-Wcast-qual)
1402 =for apidoc utf8_hop_forward
1404 Return the UTF-8 pointer C<s> displaced by up to C<off> characters,
1407 C<off> must be non-negative.
1409 C<s> must be before or equal to C<end>.
1411 When moving forward it will not move beyond C<end>.
1413 Will not exceed this limit even if the string is not valid "UTF-8".
1418 PERL_STATIC_INLINE U8 *
1419 Perl_utf8_hop_forward(const U8 *s, SSize_t off, const U8 *end)
1421 PERL_ARGS_ASSERT_UTF8_HOP_FORWARD;
1423 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1424 * the bitops (especially ~) can create illegal UTF-8.
1425 * In other words: in Perl UTF-8 is not just for Unicode. */
1431 STRLEN skip = UTF8SKIP(s);
1432 if ((STRLEN)(end - s) <= skip) {
1433 GCC_DIAG_IGNORE(-Wcast-qual)
1440 GCC_DIAG_IGNORE(-Wcast-qual)
1446 =for apidoc utf8_hop_back
1448 Return the UTF-8 pointer C<s> displaced by up to C<off> characters,
1451 C<off> must be non-positive.
1453 C<s> must be after or equal to C<start>.
1455 When moving backward it will not move before C<start>.
1457 Will not exceed this limit even if the string is not valid "UTF-8".
1462 PERL_STATIC_INLINE U8 *
1463 Perl_utf8_hop_back(const U8 *s, SSize_t off, const U8 *start)
1465 PERL_ARGS_ASSERT_UTF8_HOP_BACK;
1467 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1468 * the bitops (especially ~) can create illegal UTF-8.
1469 * In other words: in Perl UTF-8 is not just for Unicode. */
1474 while (off++ && s > start) {
1476 while (UTF8_IS_CONTINUATION(*s) && s > start)
1480 GCC_DIAG_IGNORE(-Wcast-qual)
1486 =for apidoc utf8_hop_safe
1488 Return the UTF-8 pointer C<s> displaced by up to C<off> characters,
1489 either forward or backward.
1491 When moving backward it will not move before C<start>.
1493 When moving forward it will not move beyond C<end>.
1495 Will not exceed those limits even if the string is not valid "UTF-8".
1500 PERL_STATIC_INLINE U8 *
1501 Perl_utf8_hop_safe(const U8 *s, SSize_t off, const U8 *start, const U8 *end)
1503 PERL_ARGS_ASSERT_UTF8_HOP_SAFE;
1505 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1506 * the bitops (especially ~) can create illegal UTF-8.
1507 * In other words: in Perl UTF-8 is not just for Unicode. */
1509 assert(start <= s && s <= end);
1512 return utf8_hop_forward(s, off, end);
1515 return utf8_hop_back(s, off, start);
1521 =for apidoc is_utf8_valid_partial_char
1523 Returns 0 if the sequence of bytes starting at C<s> and looking no further than
1524 S<C<e - 1>> is the UTF-8 encoding, as extended by Perl, for one or more code
1525 points. Otherwise, it returns 1 if there exists at least one non-empty
1526 sequence of bytes that when appended to sequence C<s>, starting at position
1527 C<e> causes the entire sequence to be the well-formed UTF-8 of some code point;
1528 otherwise returns 0.
1530 In other words this returns TRUE if C<s> points to a partial UTF-8-encoded code
1533 This is useful when a fixed-length buffer is being tested for being well-formed
1534 UTF-8, but the final few bytes in it don't comprise a full character; that is,
1535 it is split somewhere in the middle of the final code point's UTF-8
1536 representation. (Presumably when the buffer is refreshed with the next chunk
1537 of data, the new first bytes will complete the partial code point.) This
1538 function is used to verify that the final bytes in the current buffer are in
1539 fact the legal beginning of some code point, so that if they aren't, the
1540 failure can be signalled without having to wait for the next read.
1544 #define is_utf8_valid_partial_char(s, e) \
1545 is_utf8_valid_partial_char_flags(s, e, 0)
1549 =for apidoc is_utf8_valid_partial_char_flags
1551 Like C<L</is_utf8_valid_partial_char>>, it returns a boolean giving whether
1552 or not the input is a valid UTF-8 encoded partial character, but it takes an
1553 extra parameter, C<flags>, which can further restrict which code points are
1556 If C<flags> is 0, this behaves identically to
1557 C<L</is_utf8_valid_partial_char>>. Otherwise C<flags> can be any combination
1558 of the C<UTF8_DISALLOW_I<foo>> flags accepted by C<L</utf8n_to_uvchr>>. If
1559 there is any sequence of bytes that can complete the input partial character in
1560 such a way that a non-prohibited character is formed, the function returns
1561 TRUE; otherwise FALSE. Non character code points cannot be determined based on
1562 partial character input. But many of the other possible excluded types can be
1563 determined from just the first one or two bytes.
1568 PERL_STATIC_INLINE bool
1569 S_is_utf8_valid_partial_char_flags(const U8 * const s, const U8 * const e, const U32 flags)
1571 PERL_ARGS_ASSERT_IS_UTF8_VALID_PARTIAL_CHAR_FLAGS;
1573 assert(0 == (flags & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE
1574 |UTF8_DISALLOW_PERL_EXTENDED)));
1576 if (s >= e || s + UTF8SKIP(s) <= e) {
1580 return cBOOL(_is_utf8_char_helper(s, e, flags));
1585 =for apidoc is_utf8_fixed_width_buf_flags
1587 Returns TRUE if the fixed-width buffer starting at C<s> with length C<len>
1588 is entirely valid UTF-8, subject to the restrictions given by C<flags>;
1589 otherwise it returns FALSE.
1591 If C<flags> is 0, any well-formed UTF-8, as extended by Perl, is accepted
1592 without restriction. If the final few bytes of the buffer do not form a
1593 complete code point, this will return TRUE anyway, provided that
1594 C<L</is_utf8_valid_partial_char_flags>> returns TRUE for them.
1596 If C<flags> in non-zero, it can be any combination of the
1597 C<UTF8_DISALLOW_I<foo>> flags accepted by C<L</utf8n_to_uvchr>>, and with the
1600 This function differs from C<L</is_utf8_string_flags>> only in that the latter
1601 returns FALSE if the final few bytes of the string don't form a complete code
1606 #define is_utf8_fixed_width_buf_flags(s, len, flags) \
1607 is_utf8_fixed_width_buf_loclen_flags(s, len, 0, 0, flags)
1611 =for apidoc is_utf8_fixed_width_buf_loc_flags
1613 Like C<L</is_utf8_fixed_width_buf_flags>> but stores the location of the
1614 failure in the C<ep> pointer. If the function returns TRUE, C<*ep> will point
1615 to the beginning of any partial character at the end of the buffer; if there is
1616 no partial character C<*ep> will contain C<s>+C<len>.
1618 See also C<L</is_utf8_fixed_width_buf_loclen_flags>>.
1623 #define is_utf8_fixed_width_buf_loc_flags(s, len, loc, flags) \
1624 is_utf8_fixed_width_buf_loclen_flags(s, len, loc, 0, flags)
1628 =for apidoc is_utf8_fixed_width_buf_loclen_flags
1630 Like C<L</is_utf8_fixed_width_buf_loc_flags>> but stores the number of
1631 complete, valid characters found in the C<el> pointer.
1636 PERL_STATIC_INLINE bool
1637 S_is_utf8_fixed_width_buf_loclen_flags(const U8 * const s,
1643 const U8 * maybe_partial;
1645 PERL_ARGS_ASSERT_IS_UTF8_FIXED_WIDTH_BUF_LOCLEN_FLAGS;
1648 ep = &maybe_partial;
1651 /* If it's entirely valid, return that; otherwise see if the only error is
1652 * that the final few bytes are for a partial character */
1653 return is_utf8_string_loclen_flags(s, len, ep, el, flags)
1654 || is_utf8_valid_partial_char_flags(*ep, s + len, flags);
1657 /* ------------------------------- perl.h ----------------------------- */
1660 =head1 Miscellaneous Functions
1662 =for apidoc AiR|bool|is_safe_syscall|const char *pv|STRLEN len|const char *what|const char *op_name
1664 Test that the given C<pv> doesn't contain any internal C<NUL> characters.
1665 If it does, set C<errno> to C<ENOENT>, optionally warn, and return FALSE.
1667 Return TRUE if the name is safe.
1669 Used by the C<IS_SAFE_SYSCALL()> macro.
1674 PERL_STATIC_INLINE bool
1675 S_is_safe_syscall(pTHX_ const char *pv, STRLEN len, const char *what, const char *op_name) {
1676 /* While the Windows CE API provides only UCS-16 (or UTF-16) APIs
1677 * perl itself uses xce*() functions which accept 8-bit strings.
1680 PERL_ARGS_ASSERT_IS_SAFE_SYSCALL;
1684 if (UNLIKELY((null_at = (char *)memchr(pv, 0, len-1)) != NULL)) {
1685 SETERRNO(ENOENT, LIB_INVARG);
1686 Perl_ck_warner(aTHX_ packWARN(WARN_SYSCALLS),
1687 "Invalid \\0 character in %s for %s: %s\\0%s",
1688 what, op_name, pv, null_at+1);
1698 Return true if the supplied filename has a newline character
1699 immediately before the first (hopefully only) NUL.
1701 My original look at this incorrectly used the len from SvPV(), but
1702 that's incorrect, since we allow for a NUL in pv[len-1].
1704 So instead, strlen() and work from there.
1706 This allow for the user reading a filename, forgetting to chomp it,
1709 open my $foo, "$file\0";
1715 PERL_STATIC_INLINE bool
1716 S_should_warn_nl(const char *pv) {
1719 PERL_ARGS_ASSERT_SHOULD_WARN_NL;
1723 return len > 0 && pv[len-1] == '\n';
1728 /* ------------------ pp.c, regcomp.c, toke.c, universal.c ------------ */
1730 #define MAX_CHARSET_NAME_LENGTH 2
1732 PERL_STATIC_INLINE const char *
1733 get_regex_charset_name(const U32 flags, STRLEN* const lenp)
1735 /* Returns a string that corresponds to the name of the regex character set
1736 * given by 'flags', and *lenp is set the length of that string, which
1737 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
1740 switch (get_regex_charset(flags)) {
1741 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
1742 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
1743 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
1744 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
1745 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
1747 return ASCII_MORE_RESTRICT_PAT_MODS;
1749 /* The NOT_REACHED; hides an assert() which has a rather complex
1750 * definition in perl.h. */
1751 NOT_REACHED; /* NOTREACHED */
1752 return "?"; /* Unknown */
1757 Return false if any get magic is on the SV other than taint magic.
1761 PERL_STATIC_INLINE bool
1762 S_sv_only_taint_gmagic(SV *sv) {
1763 MAGIC *mg = SvMAGIC(sv);
1765 PERL_ARGS_ASSERT_SV_ONLY_TAINT_GMAGIC;
1768 if (mg->mg_type != PERL_MAGIC_taint
1769 && !(mg->mg_flags & MGf_GSKIP)
1770 && mg->mg_virtual->svt_get) {
1773 mg = mg->mg_moremagic;
1779 /* ------------------ cop.h ------------------------------------------- */
1782 /* Enter a block. Push a new base context and return its address. */
1784 PERL_STATIC_INLINE PERL_CONTEXT *
1785 S_cx_pushblock(pTHX_ U8 type, U8 gimme, SV** sp, I32 saveix)
1789 PERL_ARGS_ASSERT_CX_PUSHBLOCK;
1794 cx->blk_gimme = gimme;
1795 cx->blk_oldsaveix = saveix;
1796 cx->blk_oldsp = (I32)(sp - PL_stack_base);
1797 cx->blk_oldcop = PL_curcop;
1798 cx->blk_oldmarksp = (I32)(PL_markstack_ptr - PL_markstack);
1799 cx->blk_oldscopesp = PL_scopestack_ix;
1800 cx->blk_oldpm = PL_curpm;
1801 cx->blk_old_tmpsfloor = PL_tmps_floor;
1803 PL_tmps_floor = PL_tmps_ix;
1804 CX_DEBUG(cx, "PUSH");
1809 /* Exit a block (RETURN and LAST). */
1811 PERL_STATIC_INLINE void
1812 S_cx_popblock(pTHX_ PERL_CONTEXT *cx)
1814 PERL_ARGS_ASSERT_CX_POPBLOCK;
1816 CX_DEBUG(cx, "POP");
1817 /* these 3 are common to cx_popblock and cx_topblock */
1818 PL_markstack_ptr = PL_markstack + cx->blk_oldmarksp;
1819 PL_scopestack_ix = cx->blk_oldscopesp;
1820 PL_curpm = cx->blk_oldpm;
1822 /* LEAVE_SCOPE() should have made this true. /(?{})/ cheats
1823 * and leaves a CX entry lying around for repeated use, so
1824 * skip for multicall */ \
1825 assert( (CxTYPE(cx) == CXt_SUB && CxMULTICALL(cx))
1826 || PL_savestack_ix == cx->blk_oldsaveix);
1827 PL_curcop = cx->blk_oldcop;
1828 PL_tmps_floor = cx->blk_old_tmpsfloor;
1831 /* Continue a block elsewhere (e.g. NEXT, REDO, GOTO).
1832 * Whereas cx_popblock() restores the state to the point just before
1833 * cx_pushblock() was called, cx_topblock() restores it to the point just
1834 * *after* cx_pushblock() was called. */
1836 PERL_STATIC_INLINE void
1837 S_cx_topblock(pTHX_ PERL_CONTEXT *cx)
1839 PERL_ARGS_ASSERT_CX_TOPBLOCK;
1841 CX_DEBUG(cx, "TOP");
1842 /* these 3 are common to cx_popblock and cx_topblock */
1843 PL_markstack_ptr = PL_markstack + cx->blk_oldmarksp;
1844 PL_scopestack_ix = cx->blk_oldscopesp;
1845 PL_curpm = cx->blk_oldpm;
1847 PL_stack_sp = PL_stack_base + cx->blk_oldsp;
1851 PERL_STATIC_INLINE void
1852 S_cx_pushsub(pTHX_ PERL_CONTEXT *cx, CV *cv, OP *retop, bool hasargs)
1854 U8 phlags = CX_PUSHSUB_GET_LVALUE_MASK(Perl_was_lvalue_sub);
1856 PERL_ARGS_ASSERT_CX_PUSHSUB;
1858 PERL_DTRACE_PROBE_ENTRY(cv);
1859 cx->blk_sub.cv = cv;
1860 cx->blk_sub.olddepth = CvDEPTH(cv);
1861 cx->blk_sub.prevcomppad = PL_comppad;
1862 cx->cx_type |= (hasargs) ? CXp_HASARGS : 0;
1863 cx->blk_sub.retop = retop;
1864 SvREFCNT_inc_simple_void_NN(cv);
1865 cx->blk_u16 = PL_op->op_private & (phlags|OPpDEREF);
1869 /* subsets of cx_popsub() */
1871 PERL_STATIC_INLINE void
1872 S_cx_popsub_common(pTHX_ PERL_CONTEXT *cx)
1876 PERL_ARGS_ASSERT_CX_POPSUB_COMMON;
1877 assert(CxTYPE(cx) == CXt_SUB);
1879 PL_comppad = cx->blk_sub.prevcomppad;
1880 PL_curpad = LIKELY(PL_comppad) ? AvARRAY(PL_comppad) : NULL;
1881 cv = cx->blk_sub.cv;
1882 CvDEPTH(cv) = cx->blk_sub.olddepth;
1883 cx->blk_sub.cv = NULL;
1888 /* handle the @_ part of leaving a sub */
1890 PERL_STATIC_INLINE void
1891 S_cx_popsub_args(pTHX_ PERL_CONTEXT *cx)
1895 PERL_ARGS_ASSERT_CX_POPSUB_ARGS;
1896 assert(CxTYPE(cx) == CXt_SUB);
1897 assert(AvARRAY(MUTABLE_AV(
1898 PadlistARRAY(CvPADLIST(cx->blk_sub.cv))[
1899 CvDEPTH(cx->blk_sub.cv)])) == PL_curpad);
1901 CX_POP_SAVEARRAY(cx);
1902 av = MUTABLE_AV(PAD_SVl(0));
1903 if (UNLIKELY(AvREAL(av)))
1904 /* abandon @_ if it got reified */
1905 clear_defarray(av, 0);
1912 PERL_STATIC_INLINE void
1913 S_cx_popsub(pTHX_ PERL_CONTEXT *cx)
1915 PERL_ARGS_ASSERT_CX_POPSUB;
1916 assert(CxTYPE(cx) == CXt_SUB);
1918 PERL_DTRACE_PROBE_RETURN(cx->blk_sub.cv);
1922 cx_popsub_common(cx);
1926 PERL_STATIC_INLINE void
1927 S_cx_pushformat(pTHX_ PERL_CONTEXT *cx, CV *cv, OP *retop, GV *gv)
1929 PERL_ARGS_ASSERT_CX_PUSHFORMAT;
1931 cx->blk_format.cv = cv;
1932 cx->blk_format.retop = retop;
1933 cx->blk_format.gv = gv;
1934 cx->blk_format.dfoutgv = PL_defoutgv;
1935 cx->blk_format.prevcomppad = PL_comppad;
1938 SvREFCNT_inc_simple_void_NN(cv);
1940 SvREFCNT_inc_void(cx->blk_format.dfoutgv);
1944 PERL_STATIC_INLINE void
1945 S_cx_popformat(pTHX_ PERL_CONTEXT *cx)
1950 PERL_ARGS_ASSERT_CX_POPFORMAT;
1951 assert(CxTYPE(cx) == CXt_FORMAT);
1953 dfout = cx->blk_format.dfoutgv;
1955 cx->blk_format.dfoutgv = NULL;
1956 SvREFCNT_dec_NN(dfout);
1958 PL_comppad = cx->blk_format.prevcomppad;
1959 PL_curpad = LIKELY(PL_comppad) ? AvARRAY(PL_comppad) : NULL;
1960 cv = cx->blk_format.cv;
1961 cx->blk_format.cv = NULL;
1963 SvREFCNT_dec_NN(cv);
1967 PERL_STATIC_INLINE void
1968 S_cx_pusheval(pTHX_ PERL_CONTEXT *cx, OP *retop, SV *namesv)
1970 PERL_ARGS_ASSERT_CX_PUSHEVAL;
1972 cx->blk_eval.retop = retop;
1973 cx->blk_eval.old_namesv = namesv;
1974 cx->blk_eval.old_eval_root = PL_eval_root;
1975 cx->blk_eval.cur_text = PL_parser ? PL_parser->linestr : NULL;
1976 cx->blk_eval.cv = NULL; /* later set by doeval_compile() */
1977 cx->blk_eval.cur_top_env = PL_top_env;
1979 assert(!(PL_in_eval & ~ 0x3F));
1980 assert(!(PL_op->op_type & ~0x1FF));
1981 cx->blk_u16 = (PL_in_eval & 0x3F) | ((U16)PL_op->op_type << 7);
1985 PERL_STATIC_INLINE void
1986 S_cx_popeval(pTHX_ PERL_CONTEXT *cx)
1990 PERL_ARGS_ASSERT_CX_POPEVAL;
1991 assert(CxTYPE(cx) == CXt_EVAL);
1993 PL_in_eval = CxOLD_IN_EVAL(cx);
1994 assert(!(PL_in_eval & 0xc0));
1995 PL_eval_root = cx->blk_eval.old_eval_root;
1996 sv = cx->blk_eval.cur_text;
1997 if (sv && CxEVAL_TXT_REFCNTED(cx)) {
1998 cx->blk_eval.cur_text = NULL;
1999 SvREFCNT_dec_NN(sv);
2002 sv = cx->blk_eval.old_namesv;
2004 cx->blk_eval.old_namesv = NULL;
2005 SvREFCNT_dec_NN(sv);
2010 /* push a plain loop, i.e.
2012 * while (cond) { block }
2013 * for (init;cond;continue) { block }
2014 * This loop can be last/redo'ed etc.
2017 PERL_STATIC_INLINE void
2018 S_cx_pushloop_plain(pTHX_ PERL_CONTEXT *cx)
2020 PERL_ARGS_ASSERT_CX_PUSHLOOP_PLAIN;
2021 cx->blk_loop.my_op = cLOOP;
2025 /* push a true for loop, i.e.
2026 * for var (list) { block }
2029 PERL_STATIC_INLINE void
2030 S_cx_pushloop_for(pTHX_ PERL_CONTEXT *cx, void *itervarp, SV* itersave)
2032 PERL_ARGS_ASSERT_CX_PUSHLOOP_FOR;
2034 /* this one line is common with cx_pushloop_plain */
2035 cx->blk_loop.my_op = cLOOP;
2037 cx->blk_loop.itervar_u.svp = (SV**)itervarp;
2038 cx->blk_loop.itersave = itersave;
2040 cx->blk_loop.oldcomppad = PL_comppad;
2045 /* pop all loop types, including plain */
2047 PERL_STATIC_INLINE void
2048 S_cx_poploop(pTHX_ PERL_CONTEXT *cx)
2050 PERL_ARGS_ASSERT_CX_POPLOOP;
2052 assert(CxTYPE_is_LOOP(cx));
2053 if ( CxTYPE(cx) == CXt_LOOP_ARY
2054 || CxTYPE(cx) == CXt_LOOP_LAZYSV)
2056 /* Free ary or cur. This assumes that state_u.ary.ary
2057 * aligns with state_u.lazysv.cur. See cx_dup() */
2058 SV *sv = cx->blk_loop.state_u.lazysv.cur;
2059 cx->blk_loop.state_u.lazysv.cur = NULL;
2060 SvREFCNT_dec_NN(sv);
2061 if (CxTYPE(cx) == CXt_LOOP_LAZYSV) {
2062 sv = cx->blk_loop.state_u.lazysv.end;
2063 cx->blk_loop.state_u.lazysv.end = NULL;
2064 SvREFCNT_dec_NN(sv);
2067 if (cx->cx_type & (CXp_FOR_PAD|CXp_FOR_GV)) {
2069 SV **svp = (cx)->blk_loop.itervar_u.svp;
2070 if ((cx->cx_type & CXp_FOR_GV))
2071 svp = &GvSV((GV*)svp);
2073 *svp = cx->blk_loop.itersave;
2074 cx->blk_loop.itersave = NULL;
2075 SvREFCNT_dec(cursv);
2080 PERL_STATIC_INLINE void
2081 S_cx_pushwhen(pTHX_ PERL_CONTEXT *cx)
2083 PERL_ARGS_ASSERT_CX_PUSHWHEN;
2085 cx->blk_givwhen.leave_op = cLOGOP->op_other;
2089 PERL_STATIC_INLINE void
2090 S_cx_popwhen(pTHX_ PERL_CONTEXT *cx)
2092 PERL_ARGS_ASSERT_CX_POPWHEN;
2093 assert(CxTYPE(cx) == CXt_WHEN);
2095 PERL_UNUSED_ARG(cx);
2096 PERL_UNUSED_CONTEXT;
2097 /* currently NOOP */
2101 PERL_STATIC_INLINE void
2102 S_cx_pushgiven(pTHX_ PERL_CONTEXT *cx, SV *orig_defsv)
2104 PERL_ARGS_ASSERT_CX_PUSHGIVEN;
2106 cx->blk_givwhen.leave_op = cLOGOP->op_other;
2107 cx->blk_givwhen.defsv_save = orig_defsv;
2111 PERL_STATIC_INLINE void
2112 S_cx_popgiven(pTHX_ PERL_CONTEXT *cx)
2116 PERL_ARGS_ASSERT_CX_POPGIVEN;
2117 assert(CxTYPE(cx) == CXt_GIVEN);
2119 sv = GvSV(PL_defgv);
2120 GvSV(PL_defgv) = cx->blk_givwhen.defsv_save;
2121 cx->blk_givwhen.defsv_save = NULL;
2125 /* ------------------ util.h ------------------------------------------- */
2128 =head1 Miscellaneous Functions
2132 Returns true if the leading C<len> bytes of the strings C<s1> and C<s2> are the
2134 case-insensitively; false otherwise. Uppercase and lowercase ASCII range bytes
2135 match themselves and their opposite case counterparts. Non-cased and non-ASCII
2136 range bytes match only themselves.
2141 PERL_STATIC_INLINE I32
2142 Perl_foldEQ(const char *s1, const char *s2, I32 len)
2144 const U8 *a = (const U8 *)s1;
2145 const U8 *b = (const U8 *)s2;
2147 PERL_ARGS_ASSERT_FOLDEQ;
2152 if (*a != *b && *a != PL_fold[*b])
2159 PERL_STATIC_INLINE I32
2160 Perl_foldEQ_latin1(const char *s1, const char *s2, I32 len)
2162 /* Compare non-utf8 using Unicode (Latin1) semantics. Does not work on
2163 * MICRO_SIGN, LATIN_SMALL_LETTER_SHARP_S, nor
2164 * LATIN_SMALL_LETTER_Y_WITH_DIAERESIS, and does not check for these. Nor
2165 * does it check that the strings each have at least 'len' characters */
2167 const U8 *a = (const U8 *)s1;
2168 const U8 *b = (const U8 *)s2;
2170 PERL_ARGS_ASSERT_FOLDEQ_LATIN1;
2175 if (*a != *b && *a != PL_fold_latin1[*b]) {
2184 =for apidoc foldEQ_locale
2186 Returns true if the leading C<len> bytes of the strings C<s1> and C<s2> are the
2187 same case-insensitively in the current locale; false otherwise.
2192 PERL_STATIC_INLINE I32
2193 Perl_foldEQ_locale(const char *s1, const char *s2, I32 len)
2196 const U8 *a = (const U8 *)s1;
2197 const U8 *b = (const U8 *)s2;
2199 PERL_ARGS_ASSERT_FOLDEQ_LOCALE;
2204 if (*a != *b && *a != PL_fold_locale[*b])
2211 #if ! defined (HAS_MEMRCHR) && (defined(PERL_CORE) || defined(PERL_EXT))
2213 PERL_STATIC_INLINE void *
2214 S_my_memrchr(const char * s, const char c, const STRLEN len)
2216 /* memrchr(), since many platforms lack it */
2218 const char * t = s + len - 1;
2220 PERL_ARGS_ASSERT_MY_MEMRCHR;
2235 * ex: set ts=8 sts=4 sw=4 et: