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 is a home for static inline functions that cannot go in other
9 * headers files, because they depend on proto.h (included after most other
10 * headers) or struct definitions.
12 * Each section names the header file that the functions "belong" to.
15 /* ------------------------------- av.h ------------------------------- */
17 PERL_STATIC_INLINE SSize_t
18 S_av_top_index(pTHX_ AV *av)
20 PERL_ARGS_ASSERT_AV_TOP_INDEX;
21 assert(SvTYPE(av) == SVt_PVAV);
26 /* ------------------------------- cv.h ------------------------------- */
28 PERL_STATIC_INLINE GV *
32 ? Perl_cvgv_from_hek(aTHX_ sv)
33 : ((XPVCV*)MUTABLE_PTR(SvANY(sv)))->xcv_gv_u.xcv_gv;
36 PERL_STATIC_INLINE I32 *
37 S_CvDEPTHp(const CV * const sv)
39 assert(SvTYPE(sv) == SVt_PVCV || SvTYPE(sv) == SVt_PVFM);
40 return &((XPVCV*)SvANY(sv))->xcv_depth;
44 CvPROTO returns the prototype as stored, which is not necessarily what
45 the interpreter should be using. Specifically, the interpreter assumes
46 that spaces have been stripped, which has been the case if the prototype
47 was added by toke.c, but is generally not the case if it was added elsewhere.
48 Since we can't enforce the spacelessness at assignment time, this routine
49 provides a temporary copy at parse time with spaces removed.
50 I<orig> is the start of the original buffer, I<len> is the length of the
51 prototype and will be updated when this returns.
55 PERL_STATIC_INLINE char *
56 S_strip_spaces(pTHX_ const char * orig, STRLEN * const len)
60 tmpsv = newSVpvn_flags(orig, *len, SVs_TEMP);
68 *len = tmps - SvPVX(tmpsv);
73 /* ------------------------------- mg.h ------------------------------- */
75 #if defined(PERL_CORE) || defined(PERL_EXT)
76 /* assumes get-magic and stringification have already occurred */
77 PERL_STATIC_INLINE STRLEN
78 S_MgBYTEPOS(pTHX_ MAGIC *mg, SV *sv, const char *s, STRLEN len)
80 assert(mg->mg_type == PERL_MAGIC_regex_global);
81 assert(mg->mg_len != -1);
82 if (mg->mg_flags & MGf_BYTES || !DO_UTF8(sv))
83 return (STRLEN)mg->mg_len;
85 const STRLEN pos = (STRLEN)mg->mg_len;
86 /* Without this check, we may read past the end of the buffer: */
87 if (pos > sv_or_pv_len_utf8(sv, s, len)) return len+1;
88 return sv_or_pv_pos_u2b(sv, s, pos, NULL);
93 /* ------------------------------- pad.h ------------------------------ */
95 #if defined(PERL_IN_PAD_C) || defined(PERL_IN_OP_C)
96 PERL_STATIC_INLINE bool
97 PadnameIN_SCOPE(const PADNAME * const pn, const U32 seq)
99 /* is seq within the range _LOW to _HIGH ?
100 * This is complicated by the fact that PL_cop_seqmax
101 * may have wrapped around at some point */
102 if (COP_SEQ_RANGE_LOW(pn) == PERL_PADSEQ_INTRO)
103 return FALSE; /* not yet introduced */
105 if (COP_SEQ_RANGE_HIGH(pn) == PERL_PADSEQ_INTRO) {
106 /* in compiling scope */
108 (seq > COP_SEQ_RANGE_LOW(pn))
109 ? (seq - COP_SEQ_RANGE_LOW(pn) < (U32_MAX >> 1))
110 : (COP_SEQ_RANGE_LOW(pn) - seq > (U32_MAX >> 1))
115 (COP_SEQ_RANGE_LOW(pn) > COP_SEQ_RANGE_HIGH(pn))
117 ( seq > COP_SEQ_RANGE_LOW(pn)
118 || seq <= COP_SEQ_RANGE_HIGH(pn))
120 : ( seq > COP_SEQ_RANGE_LOW(pn)
121 && seq <= COP_SEQ_RANGE_HIGH(pn))
128 /* ------------------------------- pp.h ------------------------------- */
130 PERL_STATIC_INLINE I32
133 DEBUG_s(DEBUG_v(PerlIO_printf(Perl_debug_log,
134 "MARK top %p %" IVdf "\n",
136 (IV)*PL_markstack_ptr)));
137 return *PL_markstack_ptr;
140 PERL_STATIC_INLINE I32
143 DEBUG_s(DEBUG_v(PerlIO_printf(Perl_debug_log,
144 "MARK pop %p %" IVdf "\n",
145 (PL_markstack_ptr-1),
146 (IV)*(PL_markstack_ptr-1))));
147 assert((PL_markstack_ptr > PL_markstack) || !"MARK underflow");
148 return *PL_markstack_ptr--;
151 /* ----------------------------- regexp.h ----------------------------- */
153 PERL_STATIC_INLINE struct regexp *
154 S_ReANY(const REGEXP * const re)
156 XPV* const p = (XPV*)SvANY(re);
157 assert(isREGEXP(re));
158 return SvTYPE(re) == SVt_PVLV ? p->xpv_len_u.xpvlenu_rx
159 : (struct regexp *)p;
162 /* ------------------------------- sv.h ------------------------------- */
164 PERL_STATIC_INLINE SV *
165 S_SvREFCNT_inc(SV *sv)
167 if (LIKELY(sv != NULL))
171 PERL_STATIC_INLINE SV *
172 S_SvREFCNT_inc_NN(SV *sv)
177 PERL_STATIC_INLINE void
178 S_SvREFCNT_inc_void(SV *sv)
180 if (LIKELY(sv != NULL))
183 PERL_STATIC_INLINE void
184 S_SvREFCNT_dec(pTHX_ SV *sv)
186 if (LIKELY(sv != NULL)) {
187 U32 rc = SvREFCNT(sv);
189 SvREFCNT(sv) = rc - 1;
191 Perl_sv_free2(aTHX_ sv, rc);
195 PERL_STATIC_INLINE void
196 S_SvREFCNT_dec_NN(pTHX_ SV *sv)
198 U32 rc = SvREFCNT(sv);
200 SvREFCNT(sv) = rc - 1;
202 Perl_sv_free2(aTHX_ sv, rc);
205 PERL_STATIC_INLINE void
209 if (SvOBJECT(SvRV(sv))) HvAMAGIC_on(SvSTASH(SvRV(sv)));
211 PERL_STATIC_INLINE void
214 if (SvROK(sv) && SvOBJECT(SvRV(sv)))
215 HvAMAGIC_off(SvSTASH(SvRV(sv)));
218 PERL_STATIC_INLINE U32
219 S_SvPADSTALE_on(SV *sv)
221 assert(!(SvFLAGS(sv) & SVs_PADTMP));
222 return SvFLAGS(sv) |= SVs_PADSTALE;
224 PERL_STATIC_INLINE U32
225 S_SvPADSTALE_off(SV *sv)
227 assert(!(SvFLAGS(sv) & SVs_PADTMP));
228 return SvFLAGS(sv) &= ~SVs_PADSTALE;
230 #if defined(PERL_CORE) || defined (PERL_EXT)
231 PERL_STATIC_INLINE STRLEN
232 S_sv_or_pv_pos_u2b(pTHX_ SV *sv, const char *pv, STRLEN pos, STRLEN *lenp)
234 PERL_ARGS_ASSERT_SV_OR_PV_POS_U2B;
236 U8 *hopped = utf8_hop((U8 *)pv, pos);
237 if (lenp) *lenp = (STRLEN)(utf8_hop(hopped, *lenp) - hopped);
238 return (STRLEN)(hopped - (U8 *)pv);
240 return sv_pos_u2b_flags(sv,pos,lenp,SV_CONST_RETURN);
244 /* ------------------------------- handy.h ------------------------------- */
246 /* saves machine code for a common noreturn idiom typically used in Newx*() */
247 GCC_DIAG_IGNORE_DECL(-Wunused-function);
249 S_croak_memory_wrap(void)
251 Perl_croak_nocontext("%s",PL_memory_wrap);
253 GCC_DIAG_RESTORE_DECL;
255 /* ------------------------------- utf8.h ------------------------------- */
258 =head1 Unicode Support
261 PERL_STATIC_INLINE void
262 S_append_utf8_from_native_byte(const U8 byte, U8** dest)
264 /* Takes an input 'byte' (Latin1 or EBCDIC) and appends it to the UTF-8
265 * encoded string at '*dest', updating '*dest' to include it */
267 PERL_ARGS_ASSERT_APPEND_UTF8_FROM_NATIVE_BYTE;
269 if (NATIVE_BYTE_IS_INVARIANT(byte))
272 *((*dest)++) = UTF8_EIGHT_BIT_HI(byte);
273 *((*dest)++) = UTF8_EIGHT_BIT_LO(byte);
278 =for apidoc valid_utf8_to_uvchr
279 Like C<L</utf8_to_uvchr_buf>>, but should only be called when it is known that
280 the next character in the input UTF-8 string C<s> is well-formed (I<e.g.>,
281 it passes C<L</isUTF8_CHAR>>. Surrogates, non-character code points, and
282 non-Unicode code points are allowed.
288 PERL_STATIC_INLINE UV
289 Perl_valid_utf8_to_uvchr(const U8 *s, STRLEN *retlen)
291 const UV expectlen = UTF8SKIP(s);
292 const U8* send = s + expectlen;
295 PERL_ARGS_ASSERT_VALID_UTF8_TO_UVCHR;
301 /* An invariant is trivially returned */
302 if (expectlen == 1) {
306 /* Remove the leading bits that indicate the number of bytes, leaving just
307 * the bits that are part of the value */
308 uv = NATIVE_UTF8_TO_I8(uv) & UTF_START_MASK(expectlen);
310 /* Now, loop through the remaining bytes, accumulating each into the
311 * working total as we go. (I khw tried unrolling the loop for up to 4
312 * bytes, but there was no performance improvement) */
313 for (++s; s < send; s++) {
314 uv = UTF8_ACCUMULATE(uv, *s);
317 return UNI_TO_NATIVE(uv);
322 =for apidoc is_utf8_invariant_string
324 Returns TRUE if the first C<len> bytes of the string C<s> are the same
325 regardless of the UTF-8 encoding of the string (or UTF-EBCDIC encoding on
326 EBCDIC machines); otherwise it returns FALSE. That is, it returns TRUE if they
327 are UTF-8 invariant. On ASCII-ish machines, all the ASCII characters and only
328 the ASCII characters fit this definition. On EBCDIC machines, the ASCII-range
329 characters are invariant, but so also are the C1 controls.
331 If C<len> is 0, it will be calculated using C<strlen(s)>, (which means if you
332 use this option, that C<s> can't have embedded C<NUL> characters and has to
333 have a terminating C<NUL> byte).
336 C<L</is_utf8_string>>,
337 C<L</is_utf8_string_flags>>,
338 C<L</is_utf8_string_loc>>,
339 C<L</is_utf8_string_loc_flags>>,
340 C<L</is_utf8_string_loclen>>,
341 C<L</is_utf8_string_loclen_flags>>,
342 C<L</is_utf8_fixed_width_buf_flags>>,
343 C<L</is_utf8_fixed_width_buf_loc_flags>>,
344 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
345 C<L</is_strict_utf8_string>>,
346 C<L</is_strict_utf8_string_loc>>,
347 C<L</is_strict_utf8_string_loclen>>,
348 C<L</is_c9strict_utf8_string>>,
349 C<L</is_c9strict_utf8_string_loc>>,
351 C<L</is_c9strict_utf8_string_loclen>>.
357 #define is_utf8_invariant_string(s, len) \
358 is_utf8_invariant_string_loc(s, len, NULL)
361 =for apidoc is_utf8_invariant_string_loc
363 Like C<L</is_utf8_invariant_string>> but upon failure, stores the location of
364 the first UTF-8 variant character in the C<ep> pointer; if all characters are
365 UTF-8 invariant, this function does not change the contents of C<*ep>.
371 PERL_STATIC_INLINE bool
372 S_is_utf8_invariant_string_loc(const U8* const s, STRLEN len, const U8 ** ep)
377 PERL_ARGS_ASSERT_IS_UTF8_INVARIANT_STRING_LOC;
380 len = strlen((const char *)s);
385 /* This looks like 0x010101... */
386 # define PERL_COUNT_MULTIPLIER (~ (UINTMAX_C(0)) / 0xFF)
388 /* This looks like 0x808080... */
389 # define PERL_VARIANTS_WORD_MASK (PERL_COUNT_MULTIPLIER * 0x80)
390 # define PERL_WORDSIZE sizeof(PERL_COUNT_MULTIPLIER)
391 # define PERL_WORD_BOUNDARY_MASK (PERL_WORDSIZE - 1)
393 /* Evaluates to 0 if 'x' is at a word boundary; otherwise evaluates to 1, by
394 * or'ing together the lowest bits of 'x'. Hopefully the final term gets
395 * optimized out completely on a 32-bit system, and its mask gets optimized out
396 * on a 64-bit system */
397 # define PERL_IS_SUBWORD_ADDR(x) (1 & ( PTR2nat(x) \
398 | ( PTR2nat(x) >> 1) \
400 & PERL_WORD_BOUNDARY_MASK) >> 2))))
404 /* Do the word-at-a-time iff there is at least one usable full word. That
405 * means that after advancing to a word boundary, there still is at least a
406 * full word left. The number of bytes needed to advance is 'wordsize -
407 * offset' unless offset is 0. */
408 if ((STRLEN) (send - x) >= PERL_WORDSIZE
410 /* This term is wordsize if subword; 0 if not */
411 + PERL_WORDSIZE * PERL_IS_SUBWORD_ADDR(x)
414 - (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK))
417 /* Process per-byte until reach word boundary. XXX This loop could be
418 * eliminated if we knew that this platform had fast unaligned reads */
419 while (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK) {
420 if (! UTF8_IS_INVARIANT(*x)) {
430 /* Here, we know we have at least one full word to process. Process
431 * per-word as long as we have at least a full word left */
433 if ((* (PERL_UINTMAX_T *) x) & PERL_VARIANTS_WORD_MASK) {
435 /* Found a variant. Just return if caller doesn't want its
441 # if BYTEORDER == 0x1234 || BYTEORDER == 0x12345678 \
442 || BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
444 *ep = x + _variant_byte_number(* (PERL_UINTMAX_T *) x);
445 assert(*ep >= s && *ep < send);
449 # else /* If weird byte order, drop into next loop to do byte-at-a-time
458 } while (x + PERL_WORDSIZE <= send);
461 #endif /* End of ! EBCDIC */
463 /* Process per-byte */
465 if (! UTF8_IS_INVARIANT(*x)) {
481 PERL_STATIC_INLINE unsigned int
482 S__variant_byte_number(PERL_UINTMAX_T word)
485 /* This returns the position in a word (0..7) of the first variant byte in
486 * it. This is a helper function. Note that there are no branches */
490 /* Get just the msb bits of each byte */
491 word &= PERL_VARIANTS_WORD_MASK;
493 # ifdef USING_MSVC6 /* VC6 has some issues with the normal code, and the
494 easiest thing is to hide that from the callers */
497 const U8 * s = (U8 *) &word;
500 for (i = 0; i < sizeof(word); i++ ) {
506 Perl_croak(aTHX_ "panic: %s: %d: unexpected zero word\n",
510 # elif BYTEORDER == 0x1234 || BYTEORDER == 0x12345678
512 /* Bytes are stored like
513 * Byte8 ... Byte2 Byte1
514 * 63..56...15...8 7...0
517 * https://stackoverflow.com/questions/757059/position-of-least-significant-bit-that-is-set
519 * The word will look this this, with a rightmost set bit in position 's':
520 * ('x's are don't cares)
523 * x..xx10..0 Right shift (rightmost 0 is shifted off)
524 * x..xx01..1 Subtract 1, turns all the trailing zeros into 1's and
525 * the 1 just to their left into a 0; the remainder is
527 * 0..0011..1 The xor with x..xx10..0 clears that remainder, sets
529 * 0..0100..0 Add 1 to clear the word except for the bit in 's'
531 * Another method is to do 'word &= -word'; but it generates a compiler
532 * message on some platforms about taking the negative of an unsigned */
535 word = 1 + (word ^ (word - 1));
537 # elif BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
539 /* Bytes are stored like
540 * Byte1 Byte2 ... Byte8
541 * 63..56 55..47 ... 7...0
543 * Isolate the msb; http://codeforces.com/blog/entry/10330
545 * Only the most significant set bit matters. Or'ing word with its right
546 * shift of 1 makes that bit and the next one to its right both 1. Then
547 * right shifting by 2 makes for 4 1-bits in a row. ... We end with the
548 * msb and all to the right being 1. */
554 word |= word >> 32; /* This should get optimized out on 32-bit systems. */
556 /* Then subtracting the right shift by 1 clears all but the left-most of
557 * the 1 bits, which is our desired result */
561 # error Unexpected byte order
564 /* Here 'word' has a single bit set: the msb of the first byte in which it
565 * is set. Calculate that position in the word. We can use this
566 * specialized solution: https://stackoverflow.com/a/32339674/1626653,
567 * assumes an 8-bit byte. (On a 32-bit machine, the larger numbers should
568 * just get shifted off at compile time) */
569 word = (word >> 7) * ((UINTMAX_C( 7) << 56) | (UINTMAX_C(15) << 48)
570 | (UINTMAX_C(23) << 40) | (UINTMAX_C(31) << 32)
571 | (39 << 24) | (47 << 16)
572 | (55 << 8) | (63 << 0));
573 word >>= PERL_WORDSIZE * 7; /* >> by either 56 or 24 */
575 /* Here, word contains the position 7..63 of that bit. Convert to 0..7 */
576 word = ((word + 1) >> 3) - 1;
578 # if BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
580 /* And invert the result */
581 word = CHARBITS - word - 1;
585 return (unsigned int) word;
589 #if defined(PERL_CORE) || defined(PERL_EXT)
592 =for apidoc variant_under_utf8_count
594 This function looks at the sequence of bytes between C<s> and C<e>, which are
595 assumed to be encoded in ASCII/Latin1, and returns how many of them would
596 change should the string be translated into UTF-8. Due to the nature of UTF-8,
597 each of these would occupy two bytes instead of the single one in the input
598 string. Thus, this function returns the precise number of bytes the string
599 would expand by when translated to UTF-8.
601 Unlike most of the other functions that have C<utf8> in their name, the input
602 to this function is NOT a UTF-8-encoded string. The function name is slightly
603 I<odd> to emphasize this.
605 This function is internal to Perl because khw thinks that any XS code that
606 would want this is probably operating too close to the internals. Presenting a
607 valid use case could change that.
610 C<L<perlapi/is_utf8_invariant_string>>
612 C<L<perlapi/is_utf8_invariant_string_loc>>,
618 PERL_STATIC_INLINE Size_t
619 S_variant_under_utf8_count(const U8* const s, const U8* const e)
624 PERL_ARGS_ASSERT_VARIANT_UNDER_UTF8_COUNT;
628 /* Test if the string is long enough to use word-at-a-time. (Logic is the
629 * same as for is_utf8_invariant_string()) */
630 if ((STRLEN) (e - x) >= PERL_WORDSIZE
631 + PERL_WORDSIZE * PERL_IS_SUBWORD_ADDR(x)
632 - (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK))
635 /* Process per-byte until reach word boundary. XXX This loop could be
636 * eliminated if we knew that this platform had fast unaligned reads */
637 while (PTR2nat(x) & PERL_WORD_BOUNDARY_MASK) {
638 count += ! UTF8_IS_INVARIANT(*x++);
641 /* Process per-word as long as we have at least a full word left */
642 do { /* Commit 03c1e4ab1d6ee9062fb3f94b0ba31db6698724b1 contains an
643 explanation of how this works */
644 PERL_UINTMAX_T increment
645 = ((((* (PERL_UINTMAX_T *) x) & PERL_VARIANTS_WORD_MASK) >> 7)
646 * PERL_COUNT_MULTIPLIER)
647 >> ((PERL_WORDSIZE - 1) * CHARBITS);
648 count += (Size_t) increment;
650 } while (x + PERL_WORDSIZE <= e);
655 /* Process per-byte */
657 if (! UTF8_IS_INVARIANT(*x)) {
669 #ifndef PERL_IN_REGEXEC_C /* Keep these around for that file */
670 # undef PERL_WORDSIZE
671 # undef PERL_COUNT_MULTIPLIER
672 # undef PERL_WORD_BOUNDARY_MASK
673 # undef PERL_VARIANTS_WORD_MASK
677 =for apidoc is_utf8_string
679 Returns TRUE if the first C<len> bytes of string C<s> form a valid
680 Perl-extended-UTF-8 string; returns FALSE otherwise. If C<len> is 0, it will
681 be calculated using C<strlen(s)> (which means if you use this option, that C<s>
682 can't have embedded C<NUL> characters and has to have a terminating C<NUL>
683 byte). Note that all characters being ASCII constitute 'a valid UTF-8 string'.
685 This function considers Perl's extended UTF-8 to be valid. That means that
686 code points above Unicode, surrogates, and non-character code points are
687 considered valid by this function. Use C<L</is_strict_utf8_string>>,
688 C<L</is_c9strict_utf8_string>>, or C<L</is_utf8_string_flags>> to restrict what
689 code points are considered valid.
692 C<L</is_utf8_invariant_string>>,
693 C<L</is_utf8_invariant_string_loc>>,
694 C<L</is_utf8_string_loc>>,
695 C<L</is_utf8_string_loclen>>,
696 C<L</is_utf8_fixed_width_buf_flags>>,
697 C<L</is_utf8_fixed_width_buf_loc_flags>>,
698 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
703 #define is_utf8_string(s, len) is_utf8_string_loclen(s, len, NULL, NULL)
705 #if defined(PERL_CORE) || defined (PERL_EXT)
708 =for apidoc is_utf8_non_invariant_string
710 Returns TRUE if L<perlapi/is_utf8_invariant_string> returns FALSE for the first
711 C<len> bytes of the string C<s>, but they are, nonetheless, legal Perl-extended
712 UTF-8; otherwise returns FALSE.
714 A TRUE return means that at least one code point represented by the sequence
715 either is a wide character not representable as a single byte, or the
716 representation differs depending on whether the sequence is encoded in UTF-8 or
720 C<L<perlapi/is_utf8_invariant_string>>,
721 C<L<perlapi/is_utf8_string>>
725 This is commonly used to determine if a SV's UTF-8 flag should be turned on.
726 It generally needn't be if its string is entirely UTF-8 invariant, and it
727 shouldn't be if it otherwise contains invalid UTF-8.
729 It is an internal function because khw thinks that XS code shouldn't be working
730 at this low a level. A valid use case could change that.
734 PERL_STATIC_INLINE bool
735 S_is_utf8_non_invariant_string(const U8* const s, STRLEN len)
737 const U8 * first_variant;
739 PERL_ARGS_ASSERT_IS_UTF8_NON_INVARIANT_STRING;
741 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
745 return is_utf8_string(first_variant, len - (first_variant - s));
751 =for apidoc is_strict_utf8_string
753 Returns TRUE if the first C<len> bytes of string C<s> form a valid
754 UTF-8-encoded string that is fully interchangeable by any application using
755 Unicode rules; otherwise it returns FALSE. If C<len> is 0, it will be
756 calculated using C<strlen(s)> (which means if you use this option, that C<s>
757 can't have embedded C<NUL> characters and has to have a terminating C<NUL>
758 byte). Note that all characters being ASCII constitute 'a valid UTF-8 string'.
760 This function returns FALSE for strings containing any
761 code points above the Unicode max of 0x10FFFF, surrogate code points, or
762 non-character code points.
765 C<L</is_utf8_invariant_string>>,
766 C<L</is_utf8_invariant_string_loc>>,
767 C<L</is_utf8_string>>,
768 C<L</is_utf8_string_flags>>,
769 C<L</is_utf8_string_loc>>,
770 C<L</is_utf8_string_loc_flags>>,
771 C<L</is_utf8_string_loclen>>,
772 C<L</is_utf8_string_loclen_flags>>,
773 C<L</is_utf8_fixed_width_buf_flags>>,
774 C<L</is_utf8_fixed_width_buf_loc_flags>>,
775 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
776 C<L</is_strict_utf8_string_loc>>,
777 C<L</is_strict_utf8_string_loclen>>,
778 C<L</is_c9strict_utf8_string>>,
779 C<L</is_c9strict_utf8_string_loc>>,
781 C<L</is_c9strict_utf8_string_loclen>>.
786 #define is_strict_utf8_string(s, len) is_strict_utf8_string_loclen(s, len, NULL, NULL)
789 =for apidoc is_c9strict_utf8_string
791 Returns TRUE if the first C<len> bytes of string C<s> form a valid
792 UTF-8-encoded string that conforms to
793 L<Unicode Corrigendum #9|http://www.unicode.org/versions/corrigendum9.html>;
794 otherwise it returns FALSE. If C<len> is 0, it will be calculated using
795 C<strlen(s)> (which means if you use this option, that C<s> can't have embedded
796 C<NUL> characters and has to have a terminating C<NUL> byte). Note that all
797 characters being ASCII constitute 'a valid UTF-8 string'.
799 This function returns FALSE for strings containing any code points above the
800 Unicode max of 0x10FFFF or surrogate code points, but accepts non-character
802 L<Corrigendum #9|http://www.unicode.org/versions/corrigendum9.html>.
805 C<L</is_utf8_invariant_string>>,
806 C<L</is_utf8_invariant_string_loc>>,
807 C<L</is_utf8_string>>,
808 C<L</is_utf8_string_flags>>,
809 C<L</is_utf8_string_loc>>,
810 C<L</is_utf8_string_loc_flags>>,
811 C<L</is_utf8_string_loclen>>,
812 C<L</is_utf8_string_loclen_flags>>,
813 C<L</is_utf8_fixed_width_buf_flags>>,
814 C<L</is_utf8_fixed_width_buf_loc_flags>>,
815 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
816 C<L</is_strict_utf8_string>>,
817 C<L</is_strict_utf8_string_loc>>,
818 C<L</is_strict_utf8_string_loclen>>,
819 C<L</is_c9strict_utf8_string_loc>>,
821 C<L</is_c9strict_utf8_string_loclen>>.
826 #define is_c9strict_utf8_string(s, len) is_c9strict_utf8_string_loclen(s, len, NULL, 0)
829 =for apidoc is_utf8_string_flags
831 Returns TRUE if the first C<len> bytes of string C<s> form a valid
832 UTF-8 string, subject to the restrictions imposed by C<flags>;
833 returns FALSE otherwise. If C<len> is 0, it will be calculated
834 using C<strlen(s)> (which means if you use this option, that C<s> can't have
835 embedded C<NUL> characters and has to have a terminating C<NUL> byte). Note
836 that all characters being ASCII constitute 'a valid UTF-8 string'.
838 If C<flags> is 0, this gives the same results as C<L</is_utf8_string>>; if
839 C<flags> is C<UTF8_DISALLOW_ILLEGAL_INTERCHANGE>, this gives the same results
840 as C<L</is_strict_utf8_string>>; and if C<flags> is
841 C<UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE>, this gives the same results as
842 C<L</is_c9strict_utf8_string>>. Otherwise C<flags> may be any
843 combination of the C<UTF8_DISALLOW_I<foo>> flags understood by
844 C<L</utf8n_to_uvchr>>, with the same meanings.
847 C<L</is_utf8_invariant_string>>,
848 C<L</is_utf8_invariant_string_loc>>,
849 C<L</is_utf8_string>>,
850 C<L</is_utf8_string_loc>>,
851 C<L</is_utf8_string_loc_flags>>,
852 C<L</is_utf8_string_loclen>>,
853 C<L</is_utf8_string_loclen_flags>>,
854 C<L</is_utf8_fixed_width_buf_flags>>,
855 C<L</is_utf8_fixed_width_buf_loc_flags>>,
856 C<L</is_utf8_fixed_width_buf_loclen_flags>>,
857 C<L</is_strict_utf8_string>>,
858 C<L</is_strict_utf8_string_loc>>,
859 C<L</is_strict_utf8_string_loclen>>,
860 C<L</is_c9strict_utf8_string>>,
861 C<L</is_c9strict_utf8_string_loc>>,
863 C<L</is_c9strict_utf8_string_loclen>>.
868 PERL_STATIC_INLINE bool
869 S_is_utf8_string_flags(const U8 *s, STRLEN len, const U32 flags)
871 const U8 * first_variant;
873 PERL_ARGS_ASSERT_IS_UTF8_STRING_FLAGS;
874 assert(0 == (flags & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE
875 |UTF8_DISALLOW_PERL_EXTENDED)));
878 len = strlen((const char *)s);
882 return is_utf8_string(s, len);
885 if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
886 == UTF8_DISALLOW_ILLEGAL_INTERCHANGE)
888 return is_strict_utf8_string(s, len);
891 if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
892 == UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE)
894 return is_c9strict_utf8_string(s, len);
897 if (! is_utf8_invariant_string_loc(s, len, &first_variant)) {
898 const U8* const send = s + len;
899 const U8* x = first_variant;
902 STRLEN cur_len = isUTF8_CHAR_flags(x, send, flags);
903 if (UNLIKELY(! cur_len)) {
915 =for apidoc is_utf8_string_loc
917 Like C<L</is_utf8_string>> but stores the location of the failure (in the
918 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
919 "utf8ness success") in the C<ep> pointer.
921 See also C<L</is_utf8_string_loclen>>.
926 #define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
930 =for apidoc is_utf8_string_loclen
932 Like C<L</is_utf8_string>> but stores the location of the failure (in the
933 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
934 "utf8ness success") in the C<ep> pointer, and the number of UTF-8
935 encoded characters in the C<el> pointer.
937 See also C<L</is_utf8_string_loc>>.
942 PERL_STATIC_INLINE bool
943 Perl_is_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
945 const U8 * first_variant;
947 PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN;
950 len = strlen((const char *) s);
953 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
965 const U8* const send = s + len;
966 const U8* x = first_variant;
967 STRLEN outlen = first_variant - s;
970 const STRLEN cur_len = isUTF8_CHAR(x, send);
971 if (UNLIKELY(! cur_len)) {
991 =for apidoc is_strict_utf8_string_loc
993 Like C<L</is_strict_utf8_string>> but stores the location of the failure (in the
994 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
995 "utf8ness success") in the C<ep> pointer.
997 See also C<L</is_strict_utf8_string_loclen>>.
1002 #define is_strict_utf8_string_loc(s, len, ep) \
1003 is_strict_utf8_string_loclen(s, len, ep, 0)
1007 =for apidoc is_strict_utf8_string_loclen
1009 Like C<L</is_strict_utf8_string>> but stores the location of the failure (in the
1010 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1011 "utf8ness success") in the C<ep> pointer, and the number of UTF-8
1012 encoded characters in the C<el> pointer.
1014 See also C<L</is_strict_utf8_string_loc>>.
1019 PERL_STATIC_INLINE bool
1020 S_is_strict_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
1022 const U8 * first_variant;
1024 PERL_ARGS_ASSERT_IS_STRICT_UTF8_STRING_LOCLEN;
1027 len = strlen((const char *) s);
1030 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
1042 const U8* const send = s + len;
1043 const U8* x = first_variant;
1044 STRLEN outlen = first_variant - s;
1047 const STRLEN cur_len = isSTRICT_UTF8_CHAR(x, send);
1048 if (UNLIKELY(! cur_len)) {
1068 =for apidoc is_c9strict_utf8_string_loc
1070 Like C<L</is_c9strict_utf8_string>> but stores the location of the failure (in
1071 the case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1072 "utf8ness success") in the C<ep> pointer.
1074 See also C<L</is_c9strict_utf8_string_loclen>>.
1079 #define is_c9strict_utf8_string_loc(s, len, ep) \
1080 is_c9strict_utf8_string_loclen(s, len, ep, 0)
1084 =for apidoc is_c9strict_utf8_string_loclen
1086 Like C<L</is_c9strict_utf8_string>> but stores the location of the failure (in
1087 the case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1088 "utf8ness success") in the C<ep> pointer, and the number of UTF-8 encoded
1089 characters in the C<el> pointer.
1091 See also C<L</is_c9strict_utf8_string_loc>>.
1096 PERL_STATIC_INLINE bool
1097 S_is_c9strict_utf8_string_loclen(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el)
1099 const U8 * first_variant;
1101 PERL_ARGS_ASSERT_IS_C9STRICT_UTF8_STRING_LOCLEN;
1104 len = strlen((const char *) s);
1107 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
1119 const U8* const send = s + len;
1120 const U8* x = first_variant;
1121 STRLEN outlen = first_variant - s;
1124 const STRLEN cur_len = isC9_STRICT_UTF8_CHAR(x, send);
1125 if (UNLIKELY(! cur_len)) {
1145 =for apidoc is_utf8_string_loc_flags
1147 Like C<L</is_utf8_string_flags>> but stores the location of the failure (in the
1148 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1149 "utf8ness success") in the C<ep> pointer.
1151 See also C<L</is_utf8_string_loclen_flags>>.
1156 #define is_utf8_string_loc_flags(s, len, ep, flags) \
1157 is_utf8_string_loclen_flags(s, len, ep, 0, flags)
1160 /* The above 3 actual functions could have been moved into the more general one
1161 * just below, and made #defines that call it with the right 'flags'. They are
1162 * currently kept separate to increase their chances of getting inlined */
1166 =for apidoc is_utf8_string_loclen_flags
1168 Like C<L</is_utf8_string_flags>> but stores the location of the failure (in the
1169 case of "utf8ness failure") or the location C<s>+C<len> (in the case of
1170 "utf8ness success") in the C<ep> pointer, and the number of UTF-8
1171 encoded characters in the C<el> pointer.
1173 See also C<L</is_utf8_string_loc_flags>>.
1178 PERL_STATIC_INLINE bool
1179 S_is_utf8_string_loclen_flags(const U8 *s, STRLEN len, const U8 **ep, STRLEN *el, const U32 flags)
1181 const U8 * first_variant;
1183 PERL_ARGS_ASSERT_IS_UTF8_STRING_LOCLEN_FLAGS;
1184 assert(0 == (flags & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE
1185 |UTF8_DISALLOW_PERL_EXTENDED)));
1188 len = strlen((const char *) s);
1192 return is_utf8_string_loclen(s, len, ep, el);
1195 if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
1196 == UTF8_DISALLOW_ILLEGAL_INTERCHANGE)
1198 return is_strict_utf8_string_loclen(s, len, ep, el);
1201 if ((flags & ~UTF8_DISALLOW_PERL_EXTENDED)
1202 == UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE)
1204 return is_c9strict_utf8_string_loclen(s, len, ep, el);
1207 if (is_utf8_invariant_string_loc(s, len, &first_variant)) {
1219 const U8* send = s + len;
1220 const U8* x = first_variant;
1221 STRLEN outlen = first_variant - s;
1224 const STRLEN cur_len = isUTF8_CHAR_flags(x, send, flags);
1225 if (UNLIKELY(! cur_len)) {
1244 =for apidoc utf8_distance
1246 Returns the number of UTF-8 characters between the UTF-8 pointers C<a>
1249 WARNING: use only if you *know* that the pointers point inside the
1255 PERL_STATIC_INLINE IV
1256 Perl_utf8_distance(pTHX_ const U8 *a, const U8 *b)
1258 PERL_ARGS_ASSERT_UTF8_DISTANCE;
1260 return (a < b) ? -1 * (IV) utf8_length(a, b) : (IV) utf8_length(b, a);
1264 =for apidoc utf8_hop
1266 Return the UTF-8 pointer C<s> displaced by C<off> characters, either
1267 forward or backward.
1269 WARNING: do not use the following unless you *know* C<off> is within
1270 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned
1271 on the first byte of character or just after the last byte of a character.
1276 PERL_STATIC_INLINE U8 *
1277 Perl_utf8_hop(const U8 *s, SSize_t off)
1279 PERL_ARGS_ASSERT_UTF8_HOP;
1281 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1282 * the bitops (especially ~) can create illegal UTF-8.
1283 * In other words: in Perl UTF-8 is not just for Unicode. */
1292 while (UTF8_IS_CONTINUATION(*s))
1296 GCC_DIAG_IGNORE_STMT(-Wcast-qual);
1298 GCC_DIAG_RESTORE_STMT;
1302 =for apidoc utf8_hop_forward
1304 Return the UTF-8 pointer C<s> displaced by up to C<off> characters,
1307 C<off> must be non-negative.
1309 C<s> must be before or equal to C<end>.
1311 When moving forward it will not move beyond C<end>.
1313 Will not exceed this limit even if the string is not valid "UTF-8".
1318 PERL_STATIC_INLINE U8 *
1319 Perl_utf8_hop_forward(const U8 *s, SSize_t off, const U8 *end)
1321 PERL_ARGS_ASSERT_UTF8_HOP_FORWARD;
1323 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1324 * the bitops (especially ~) can create illegal UTF-8.
1325 * In other words: in Perl UTF-8 is not just for Unicode. */
1331 STRLEN skip = UTF8SKIP(s);
1332 if ((STRLEN)(end - s) <= skip) {
1333 GCC_DIAG_IGNORE_STMT(-Wcast-qual);
1335 GCC_DIAG_RESTORE_STMT;
1340 GCC_DIAG_IGNORE_STMT(-Wcast-qual);
1342 GCC_DIAG_RESTORE_STMT;
1346 =for apidoc utf8_hop_back
1348 Return the UTF-8 pointer C<s> displaced by up to C<off> characters,
1351 C<off> must be non-positive.
1353 C<s> must be after or equal to C<start>.
1355 When moving backward it will not move before C<start>.
1357 Will not exceed this limit even if the string is not valid "UTF-8".
1362 PERL_STATIC_INLINE U8 *
1363 Perl_utf8_hop_back(const U8 *s, SSize_t off, const U8 *start)
1365 PERL_ARGS_ASSERT_UTF8_HOP_BACK;
1367 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1368 * the bitops (especially ~) can create illegal UTF-8.
1369 * In other words: in Perl UTF-8 is not just for Unicode. */
1374 while (off++ && s > start) {
1376 while (UTF8_IS_CONTINUATION(*s) && s > start)
1380 GCC_DIAG_IGNORE_STMT(-Wcast-qual);
1382 GCC_DIAG_RESTORE_STMT;
1386 =for apidoc utf8_hop_safe
1388 Return the UTF-8 pointer C<s> displaced by up to C<off> characters,
1389 either forward or backward.
1391 When moving backward it will not move before C<start>.
1393 When moving forward it will not move beyond C<end>.
1395 Will not exceed those limits even if the string is not valid "UTF-8".
1400 PERL_STATIC_INLINE U8 *
1401 Perl_utf8_hop_safe(const U8 *s, SSize_t off, const U8 *start, const U8 *end)
1403 PERL_ARGS_ASSERT_UTF8_HOP_SAFE;
1405 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g
1406 * the bitops (especially ~) can create illegal UTF-8.
1407 * In other words: in Perl UTF-8 is not just for Unicode. */
1409 assert(start <= s && s <= end);
1412 return utf8_hop_forward(s, off, end);
1415 return utf8_hop_back(s, off, start);
1421 =for apidoc is_utf8_valid_partial_char
1423 Returns 0 if the sequence of bytes starting at C<s> and looking no further than
1424 S<C<e - 1>> is the UTF-8 encoding, as extended by Perl, for one or more code
1425 points. Otherwise, it returns 1 if there exists at least one non-empty
1426 sequence of bytes that when appended to sequence C<s>, starting at position
1427 C<e> causes the entire sequence to be the well-formed UTF-8 of some code point;
1428 otherwise returns 0.
1430 In other words this returns TRUE if C<s> points to a partial UTF-8-encoded code
1433 This is useful when a fixed-length buffer is being tested for being well-formed
1434 UTF-8, but the final few bytes in it don't comprise a full character; that is,
1435 it is split somewhere in the middle of the final code point's UTF-8
1436 representation. (Presumably when the buffer is refreshed with the next chunk
1437 of data, the new first bytes will complete the partial code point.) This
1438 function is used to verify that the final bytes in the current buffer are in
1439 fact the legal beginning of some code point, so that if they aren't, the
1440 failure can be signalled without having to wait for the next read.
1444 #define is_utf8_valid_partial_char(s, e) \
1445 is_utf8_valid_partial_char_flags(s, e, 0)
1449 =for apidoc is_utf8_valid_partial_char_flags
1451 Like C<L</is_utf8_valid_partial_char>>, it returns a boolean giving whether
1452 or not the input is a valid UTF-8 encoded partial character, but it takes an
1453 extra parameter, C<flags>, which can further restrict which code points are
1456 If C<flags> is 0, this behaves identically to
1457 C<L</is_utf8_valid_partial_char>>. Otherwise C<flags> can be any combination
1458 of the C<UTF8_DISALLOW_I<foo>> flags accepted by C<L</utf8n_to_uvchr>>. If
1459 there is any sequence of bytes that can complete the input partial character in
1460 such a way that a non-prohibited character is formed, the function returns
1461 TRUE; otherwise FALSE. Non character code points cannot be determined based on
1462 partial character input. But many of the other possible excluded types can be
1463 determined from just the first one or two bytes.
1468 PERL_STATIC_INLINE bool
1469 S_is_utf8_valid_partial_char_flags(const U8 * const s, const U8 * const e, const U32 flags)
1471 PERL_ARGS_ASSERT_IS_UTF8_VALID_PARTIAL_CHAR_FLAGS;
1473 assert(0 == (flags & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE
1474 |UTF8_DISALLOW_PERL_EXTENDED)));
1476 if (s >= e || s + UTF8SKIP(s) <= e) {
1480 return cBOOL(_is_utf8_char_helper(s, e, flags));
1485 =for apidoc is_utf8_fixed_width_buf_flags
1487 Returns TRUE if the fixed-width buffer starting at C<s> with length C<len>
1488 is entirely valid UTF-8, subject to the restrictions given by C<flags>;
1489 otherwise it returns FALSE.
1491 If C<flags> is 0, any well-formed UTF-8, as extended by Perl, is accepted
1492 without restriction. If the final few bytes of the buffer do not form a
1493 complete code point, this will return TRUE anyway, provided that
1494 C<L</is_utf8_valid_partial_char_flags>> returns TRUE for them.
1496 If C<flags> in non-zero, it can be any combination of the
1497 C<UTF8_DISALLOW_I<foo>> flags accepted by C<L</utf8n_to_uvchr>>, and with the
1500 This function differs from C<L</is_utf8_string_flags>> only in that the latter
1501 returns FALSE if the final few bytes of the string don't form a complete code
1506 #define is_utf8_fixed_width_buf_flags(s, len, flags) \
1507 is_utf8_fixed_width_buf_loclen_flags(s, len, 0, 0, flags)
1511 =for apidoc is_utf8_fixed_width_buf_loc_flags
1513 Like C<L</is_utf8_fixed_width_buf_flags>> but stores the location of the
1514 failure in the C<ep> pointer. If the function returns TRUE, C<*ep> will point
1515 to the beginning of any partial character at the end of the buffer; if there is
1516 no partial character C<*ep> will contain C<s>+C<len>.
1518 See also C<L</is_utf8_fixed_width_buf_loclen_flags>>.
1523 #define is_utf8_fixed_width_buf_loc_flags(s, len, loc, flags) \
1524 is_utf8_fixed_width_buf_loclen_flags(s, len, loc, 0, flags)
1528 =for apidoc is_utf8_fixed_width_buf_loclen_flags
1530 Like C<L</is_utf8_fixed_width_buf_loc_flags>> but stores the number of
1531 complete, valid characters found in the C<el> pointer.
1536 PERL_STATIC_INLINE bool
1537 S_is_utf8_fixed_width_buf_loclen_flags(const U8 * const s,
1543 const U8 * maybe_partial;
1545 PERL_ARGS_ASSERT_IS_UTF8_FIXED_WIDTH_BUF_LOCLEN_FLAGS;
1548 ep = &maybe_partial;
1551 /* If it's entirely valid, return that; otherwise see if the only error is
1552 * that the final few bytes are for a partial character */
1553 return is_utf8_string_loclen_flags(s, len, ep, el, flags)
1554 || is_utf8_valid_partial_char_flags(*ep, s + len, flags);
1557 /* ------------------------------- perl.h ----------------------------- */
1560 =head1 Miscellaneous Functions
1562 =for apidoc AiR|bool|is_safe_syscall|const char *pv|STRLEN len|const char *what|const char *op_name
1564 Test that the given C<pv> doesn't contain any internal C<NUL> characters.
1565 If it does, set C<errno> to C<ENOENT>, optionally warn, and return FALSE.
1567 Return TRUE if the name is safe.
1569 Used by the C<IS_SAFE_SYSCALL()> macro.
1574 PERL_STATIC_INLINE bool
1575 S_is_safe_syscall(pTHX_ const char *pv, STRLEN len, const char *what, const char *op_name) {
1576 /* While the Windows CE API provides only UCS-16 (or UTF-16) APIs
1577 * perl itself uses xce*() functions which accept 8-bit strings.
1580 PERL_ARGS_ASSERT_IS_SAFE_SYSCALL;
1584 if (UNLIKELY((null_at = (char *)memchr(pv, 0, len-1)) != NULL)) {
1585 SETERRNO(ENOENT, LIB_INVARG);
1586 Perl_ck_warner(aTHX_ packWARN(WARN_SYSCALLS),
1587 "Invalid \\0 character in %s for %s: %s\\0%s",
1588 what, op_name, pv, null_at+1);
1598 Return true if the supplied filename has a newline character
1599 immediately before the first (hopefully only) NUL.
1601 My original look at this incorrectly used the len from SvPV(), but
1602 that's incorrect, since we allow for a NUL in pv[len-1].
1604 So instead, strlen() and work from there.
1606 This allow for the user reading a filename, forgetting to chomp it,
1609 open my $foo, "$file\0";
1615 PERL_STATIC_INLINE bool
1616 S_should_warn_nl(const char *pv) {
1619 PERL_ARGS_ASSERT_SHOULD_WARN_NL;
1623 return len > 0 && pv[len-1] == '\n';
1628 /* ------------------ pp.c, regcomp.c, toke.c, universal.c ------------ */
1630 #define MAX_CHARSET_NAME_LENGTH 2
1632 PERL_STATIC_INLINE const char *
1633 get_regex_charset_name(const U32 flags, STRLEN* const lenp)
1635 /* Returns a string that corresponds to the name of the regex character set
1636 * given by 'flags', and *lenp is set the length of that string, which
1637 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
1640 switch (get_regex_charset(flags)) {
1641 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
1642 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
1643 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
1644 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
1645 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
1647 return ASCII_MORE_RESTRICT_PAT_MODS;
1649 /* The NOT_REACHED; hides an assert() which has a rather complex
1650 * definition in perl.h. */
1651 NOT_REACHED; /* NOTREACHED */
1652 return "?"; /* Unknown */
1657 Return false if any get magic is on the SV other than taint magic.
1661 PERL_STATIC_INLINE bool
1662 S_sv_only_taint_gmagic(SV *sv) {
1663 MAGIC *mg = SvMAGIC(sv);
1665 PERL_ARGS_ASSERT_SV_ONLY_TAINT_GMAGIC;
1668 if (mg->mg_type != PERL_MAGIC_taint
1669 && !(mg->mg_flags & MGf_GSKIP)
1670 && mg->mg_virtual->svt_get) {
1673 mg = mg->mg_moremagic;
1679 /* ------------------ cop.h ------------------------------------------- */
1682 /* Enter a block. Push a new base context and return its address. */
1684 PERL_STATIC_INLINE PERL_CONTEXT *
1685 S_cx_pushblock(pTHX_ U8 type, U8 gimme, SV** sp, I32 saveix)
1689 PERL_ARGS_ASSERT_CX_PUSHBLOCK;
1694 cx->blk_gimme = gimme;
1695 cx->blk_oldsaveix = saveix;
1696 cx->blk_oldsp = (I32)(sp - PL_stack_base);
1697 cx->blk_oldcop = PL_curcop;
1698 cx->blk_oldmarksp = (I32)(PL_markstack_ptr - PL_markstack);
1699 cx->blk_oldscopesp = PL_scopestack_ix;
1700 cx->blk_oldpm = PL_curpm;
1701 cx->blk_old_tmpsfloor = PL_tmps_floor;
1703 PL_tmps_floor = PL_tmps_ix;
1704 CX_DEBUG(cx, "PUSH");
1709 /* Exit a block (RETURN and LAST). */
1711 PERL_STATIC_INLINE void
1712 S_cx_popblock(pTHX_ PERL_CONTEXT *cx)
1714 PERL_ARGS_ASSERT_CX_POPBLOCK;
1716 CX_DEBUG(cx, "POP");
1717 /* these 3 are common to cx_popblock and cx_topblock */
1718 PL_markstack_ptr = PL_markstack + cx->blk_oldmarksp;
1719 PL_scopestack_ix = cx->blk_oldscopesp;
1720 PL_curpm = cx->blk_oldpm;
1722 /* LEAVE_SCOPE() should have made this true. /(?{})/ cheats
1723 * and leaves a CX entry lying around for repeated use, so
1724 * skip for multicall */ \
1725 assert( (CxTYPE(cx) == CXt_SUB && CxMULTICALL(cx))
1726 || PL_savestack_ix == cx->blk_oldsaveix);
1727 PL_curcop = cx->blk_oldcop;
1728 PL_tmps_floor = cx->blk_old_tmpsfloor;
1731 /* Continue a block elsewhere (e.g. NEXT, REDO, GOTO).
1732 * Whereas cx_popblock() restores the state to the point just before
1733 * cx_pushblock() was called, cx_topblock() restores it to the point just
1734 * *after* cx_pushblock() was called. */
1736 PERL_STATIC_INLINE void
1737 S_cx_topblock(pTHX_ PERL_CONTEXT *cx)
1739 PERL_ARGS_ASSERT_CX_TOPBLOCK;
1741 CX_DEBUG(cx, "TOP");
1742 /* these 3 are common to cx_popblock and cx_topblock */
1743 PL_markstack_ptr = PL_markstack + cx->blk_oldmarksp;
1744 PL_scopestack_ix = cx->blk_oldscopesp;
1745 PL_curpm = cx->blk_oldpm;
1747 PL_stack_sp = PL_stack_base + cx->blk_oldsp;
1751 PERL_STATIC_INLINE void
1752 S_cx_pushsub(pTHX_ PERL_CONTEXT *cx, CV *cv, OP *retop, bool hasargs)
1754 U8 phlags = CX_PUSHSUB_GET_LVALUE_MASK(Perl_was_lvalue_sub);
1756 PERL_ARGS_ASSERT_CX_PUSHSUB;
1758 PERL_DTRACE_PROBE_ENTRY(cv);
1759 cx->blk_sub.cv = cv;
1760 cx->blk_sub.olddepth = CvDEPTH(cv);
1761 cx->blk_sub.prevcomppad = PL_comppad;
1762 cx->cx_type |= (hasargs) ? CXp_HASARGS : 0;
1763 cx->blk_sub.retop = retop;
1764 SvREFCNT_inc_simple_void_NN(cv);
1765 cx->blk_u16 = PL_op->op_private & (phlags|OPpDEREF);
1769 /* subsets of cx_popsub() */
1771 PERL_STATIC_INLINE void
1772 S_cx_popsub_common(pTHX_ PERL_CONTEXT *cx)
1776 PERL_ARGS_ASSERT_CX_POPSUB_COMMON;
1777 assert(CxTYPE(cx) == CXt_SUB);
1779 PL_comppad = cx->blk_sub.prevcomppad;
1780 PL_curpad = LIKELY(PL_comppad) ? AvARRAY(PL_comppad) : NULL;
1781 cv = cx->blk_sub.cv;
1782 CvDEPTH(cv) = cx->blk_sub.olddepth;
1783 cx->blk_sub.cv = NULL;
1788 /* handle the @_ part of leaving a sub */
1790 PERL_STATIC_INLINE void
1791 S_cx_popsub_args(pTHX_ PERL_CONTEXT *cx)
1795 PERL_ARGS_ASSERT_CX_POPSUB_ARGS;
1796 assert(CxTYPE(cx) == CXt_SUB);
1797 assert(AvARRAY(MUTABLE_AV(
1798 PadlistARRAY(CvPADLIST(cx->blk_sub.cv))[
1799 CvDEPTH(cx->blk_sub.cv)])) == PL_curpad);
1801 CX_POP_SAVEARRAY(cx);
1802 av = MUTABLE_AV(PAD_SVl(0));
1803 if (UNLIKELY(AvREAL(av)))
1804 /* abandon @_ if it got reified */
1805 clear_defarray(av, 0);
1812 PERL_STATIC_INLINE void
1813 S_cx_popsub(pTHX_ PERL_CONTEXT *cx)
1815 PERL_ARGS_ASSERT_CX_POPSUB;
1816 assert(CxTYPE(cx) == CXt_SUB);
1818 PERL_DTRACE_PROBE_RETURN(cx->blk_sub.cv);
1822 cx_popsub_common(cx);
1826 PERL_STATIC_INLINE void
1827 S_cx_pushformat(pTHX_ PERL_CONTEXT *cx, CV *cv, OP *retop, GV *gv)
1829 PERL_ARGS_ASSERT_CX_PUSHFORMAT;
1831 cx->blk_format.cv = cv;
1832 cx->blk_format.retop = retop;
1833 cx->blk_format.gv = gv;
1834 cx->blk_format.dfoutgv = PL_defoutgv;
1835 cx->blk_format.prevcomppad = PL_comppad;
1838 SvREFCNT_inc_simple_void_NN(cv);
1840 SvREFCNT_inc_void(cx->blk_format.dfoutgv);
1844 PERL_STATIC_INLINE void
1845 S_cx_popformat(pTHX_ PERL_CONTEXT *cx)
1850 PERL_ARGS_ASSERT_CX_POPFORMAT;
1851 assert(CxTYPE(cx) == CXt_FORMAT);
1853 dfout = cx->blk_format.dfoutgv;
1855 cx->blk_format.dfoutgv = NULL;
1856 SvREFCNT_dec_NN(dfout);
1858 PL_comppad = cx->blk_format.prevcomppad;
1859 PL_curpad = LIKELY(PL_comppad) ? AvARRAY(PL_comppad) : NULL;
1860 cv = cx->blk_format.cv;
1861 cx->blk_format.cv = NULL;
1863 SvREFCNT_dec_NN(cv);
1867 PERL_STATIC_INLINE void
1868 S_cx_pusheval(pTHX_ PERL_CONTEXT *cx, OP *retop, SV *namesv)
1870 PERL_ARGS_ASSERT_CX_PUSHEVAL;
1872 cx->blk_eval.retop = retop;
1873 cx->blk_eval.old_namesv = namesv;
1874 cx->blk_eval.old_eval_root = PL_eval_root;
1875 cx->blk_eval.cur_text = PL_parser ? PL_parser->linestr : NULL;
1876 cx->blk_eval.cv = NULL; /* later set by doeval_compile() */
1877 cx->blk_eval.cur_top_env = PL_top_env;
1879 assert(!(PL_in_eval & ~ 0x3F));
1880 assert(!(PL_op->op_type & ~0x1FF));
1881 cx->blk_u16 = (PL_in_eval & 0x3F) | ((U16)PL_op->op_type << 7);
1885 PERL_STATIC_INLINE void
1886 S_cx_popeval(pTHX_ PERL_CONTEXT *cx)
1890 PERL_ARGS_ASSERT_CX_POPEVAL;
1891 assert(CxTYPE(cx) == CXt_EVAL);
1893 PL_in_eval = CxOLD_IN_EVAL(cx);
1894 assert(!(PL_in_eval & 0xc0));
1895 PL_eval_root = cx->blk_eval.old_eval_root;
1896 sv = cx->blk_eval.cur_text;
1897 if (sv && CxEVAL_TXT_REFCNTED(cx)) {
1898 cx->blk_eval.cur_text = NULL;
1899 SvREFCNT_dec_NN(sv);
1902 sv = cx->blk_eval.old_namesv;
1904 cx->blk_eval.old_namesv = NULL;
1905 SvREFCNT_dec_NN(sv);
1910 /* push a plain loop, i.e.
1912 * while (cond) { block }
1913 * for (init;cond;continue) { block }
1914 * This loop can be last/redo'ed etc.
1917 PERL_STATIC_INLINE void
1918 S_cx_pushloop_plain(pTHX_ PERL_CONTEXT *cx)
1920 PERL_ARGS_ASSERT_CX_PUSHLOOP_PLAIN;
1921 cx->blk_loop.my_op = cLOOP;
1925 /* push a true for loop, i.e.
1926 * for var (list) { block }
1929 PERL_STATIC_INLINE void
1930 S_cx_pushloop_for(pTHX_ PERL_CONTEXT *cx, void *itervarp, SV* itersave)
1932 PERL_ARGS_ASSERT_CX_PUSHLOOP_FOR;
1934 /* this one line is common with cx_pushloop_plain */
1935 cx->blk_loop.my_op = cLOOP;
1937 cx->blk_loop.itervar_u.svp = (SV**)itervarp;
1938 cx->blk_loop.itersave = itersave;
1940 cx->blk_loop.oldcomppad = PL_comppad;
1945 /* pop all loop types, including plain */
1947 PERL_STATIC_INLINE void
1948 S_cx_poploop(pTHX_ PERL_CONTEXT *cx)
1950 PERL_ARGS_ASSERT_CX_POPLOOP;
1952 assert(CxTYPE_is_LOOP(cx));
1953 if ( CxTYPE(cx) == CXt_LOOP_ARY
1954 || CxTYPE(cx) == CXt_LOOP_LAZYSV)
1956 /* Free ary or cur. This assumes that state_u.ary.ary
1957 * aligns with state_u.lazysv.cur. See cx_dup() */
1958 SV *sv = cx->blk_loop.state_u.lazysv.cur;
1959 cx->blk_loop.state_u.lazysv.cur = NULL;
1960 SvREFCNT_dec_NN(sv);
1961 if (CxTYPE(cx) == CXt_LOOP_LAZYSV) {
1962 sv = cx->blk_loop.state_u.lazysv.end;
1963 cx->blk_loop.state_u.lazysv.end = NULL;
1964 SvREFCNT_dec_NN(sv);
1967 if (cx->cx_type & (CXp_FOR_PAD|CXp_FOR_GV)) {
1969 SV **svp = (cx)->blk_loop.itervar_u.svp;
1970 if ((cx->cx_type & CXp_FOR_GV))
1971 svp = &GvSV((GV*)svp);
1973 *svp = cx->blk_loop.itersave;
1974 cx->blk_loop.itersave = NULL;
1975 SvREFCNT_dec(cursv);
1980 PERL_STATIC_INLINE void
1981 S_cx_pushwhen(pTHX_ PERL_CONTEXT *cx)
1983 PERL_ARGS_ASSERT_CX_PUSHWHEN;
1985 cx->blk_givwhen.leave_op = cLOGOP->op_other;
1989 PERL_STATIC_INLINE void
1990 S_cx_popwhen(pTHX_ PERL_CONTEXT *cx)
1992 PERL_ARGS_ASSERT_CX_POPWHEN;
1993 assert(CxTYPE(cx) == CXt_WHEN);
1995 PERL_UNUSED_ARG(cx);
1996 PERL_UNUSED_CONTEXT;
1997 /* currently NOOP */
2001 PERL_STATIC_INLINE void
2002 S_cx_pushgiven(pTHX_ PERL_CONTEXT *cx, SV *orig_defsv)
2004 PERL_ARGS_ASSERT_CX_PUSHGIVEN;
2006 cx->blk_givwhen.leave_op = cLOGOP->op_other;
2007 cx->blk_givwhen.defsv_save = orig_defsv;
2011 PERL_STATIC_INLINE void
2012 S_cx_popgiven(pTHX_ PERL_CONTEXT *cx)
2016 PERL_ARGS_ASSERT_CX_POPGIVEN;
2017 assert(CxTYPE(cx) == CXt_GIVEN);
2019 sv = GvSV(PL_defgv);
2020 GvSV(PL_defgv) = cx->blk_givwhen.defsv_save;
2021 cx->blk_givwhen.defsv_save = NULL;
2025 /* ------------------ util.h ------------------------------------------- */
2028 =head1 Miscellaneous Functions
2032 Returns true if the leading C<len> bytes of the strings C<s1> and C<s2> are the
2034 case-insensitively; false otherwise. Uppercase and lowercase ASCII range bytes
2035 match themselves and their opposite case counterparts. Non-cased and non-ASCII
2036 range bytes match only themselves.
2041 PERL_STATIC_INLINE I32
2042 Perl_foldEQ(const char *s1, const char *s2, I32 len)
2044 const U8 *a = (const U8 *)s1;
2045 const U8 *b = (const U8 *)s2;
2047 PERL_ARGS_ASSERT_FOLDEQ;
2052 if (*a != *b && *a != PL_fold[*b])
2059 PERL_STATIC_INLINE I32
2060 Perl_foldEQ_latin1(const char *s1, const char *s2, I32 len)
2062 /* Compare non-utf8 using Unicode (Latin1) semantics. Does not work on
2063 * MICRO_SIGN, LATIN_SMALL_LETTER_SHARP_S, nor
2064 * LATIN_SMALL_LETTER_Y_WITH_DIAERESIS, and does not check for these. Nor
2065 * does it check that the strings each have at least 'len' characters */
2067 const U8 *a = (const U8 *)s1;
2068 const U8 *b = (const U8 *)s2;
2070 PERL_ARGS_ASSERT_FOLDEQ_LATIN1;
2075 if (*a != *b && *a != PL_fold_latin1[*b]) {
2084 =for apidoc foldEQ_locale
2086 Returns true if the leading C<len> bytes of the strings C<s1> and C<s2> are the
2087 same case-insensitively in the current locale; false otherwise.
2092 PERL_STATIC_INLINE I32
2093 Perl_foldEQ_locale(const char *s1, const char *s2, I32 len)
2096 const U8 *a = (const U8 *)s1;
2097 const U8 *b = (const U8 *)s2;
2099 PERL_ARGS_ASSERT_FOLDEQ_LOCALE;
2104 if (*a != *b && *a != PL_fold_locale[*b])
2111 #if ! defined (HAS_MEMRCHR) && (defined(PERL_CORE) || defined(PERL_EXT))
2113 PERL_STATIC_INLINE void *
2114 S_my_memrchr(const char * s, const char c, const STRLEN len)
2116 /* memrchr(), since many platforms lack it */
2118 const char * t = s + len - 1;
2120 PERL_ARGS_ASSERT_MY_MEMRCHR;
2135 * ex: set ts=8 sts=4 sw=4 et: