ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/JSON-XS/XS.xs
(Generate patch)

Comparing JSON-XS/XS.xs (file contents):
Revision 1.30 by root, Wed May 9 16:10:37 2007 UTC vs.
Revision 1.43 by root, Sat Jun 23 23:49:29 2007 UTC

… …
7#include "stdlib.h" 7#include "stdlib.h"
8#include "stdio.h" 8#include "stdio.h"
9 9
10#if defined(__BORLANDC__) || defined(_MSC_VER) 10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
13
14// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13
12#endif 18#endif
13 19
14#define F_ASCII 0x00000001UL 20#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL 21#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 22#define F_UTF8 0x00000004UL
… …
32#define F_DEFAULT (9UL << S_MAXDEPTH) 38#define F_DEFAULT (9UL << S_MAXDEPTH)
33 39
34#define INIT_SIZE 32 // initial scalar size to be allocated 40#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level 41#define INDENT_STEP 3 // spaces per indentation level
36 42
37#define SHORT_STRING_LEN 512 // special-case strings of up to this size 43#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
38 44
39#define SB do { 45#define SB do {
40#define SE } while (0) 46#define SE } while (0)
41 47
48#if __GNUC__ >= 3
49# define expect(expr,value) __builtin_expect ((expr),(value))
50# define inline inline
51#else
52# define expect(expr,value) (expr)
53# define inline static
54#endif
55
56#define expect_false(expr) expect ((expr) != 0, 0)
57#define expect_true(expr) expect ((expr) != 0, 1)
58
42static HV *json_stash; // JSON::XS:: 59static HV *json_stash; // JSON::XS::
60static SV *json_true, *json_false;
43 61
44///////////////////////////////////////////////////////////////////////////// 62/////////////////////////////////////////////////////////////////////////////
45// utility functions 63// utility functions
46 64
47static UV * 65static UV *
… …
70// decode an utf-8 character and return it, or (UV)-1 in 88// decode an utf-8 character and return it, or (UV)-1 in
71// case of an error. 89// case of an error.
72// we special-case "safe" characters from U+80 .. U+7FF, 90// we special-case "safe" characters from U+80 .. U+7FF,
73// but use the very good perl function to parse anything else. 91// but use the very good perl function to parse anything else.
74// note that we never call this function for a ascii codepoints 92// note that we never call this function for a ascii codepoints
75static UV 93inline UV
76decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
77{ 95{
78 if (s[0] > 0xdf || s[0] < 0xc2) 96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
79 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
80 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
81 { 99 {
82 *clen = 2; 100 *clen = 2;
83 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
… …
101 U32 flags; // F_* 119 U32 flags; // F_*
102 U32 indent; // indentation level 120 U32 indent; // indentation level
103 U32 maxdepth; // max. indentation/recursion level 121 U32 maxdepth; // max. indentation/recursion level
104} enc_t; 122} enc_t;
105 123
106static void 124inline void
107need (enc_t *enc, STRLEN len) 125need (enc_t *enc, STRLEN len)
108{ 126{
109 if (enc->cur + len >= enc->end) 127 if (expect_false (enc->cur + len >= enc->end))
110 { 128 {
111 STRLEN cur = enc->cur - SvPVX (enc->sv); 129 STRLEN cur = enc->cur - SvPVX (enc->sv);
112 SvGROW (enc->sv, cur + len + 1); 130 SvGROW (enc->sv, cur + len + 1);
113 enc->cur = SvPVX (enc->sv) + cur; 131 enc->cur = SvPVX (enc->sv) + cur;
114 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
115 } 133 }
116} 134}
117 135
118static void 136inline void
119encode_ch (enc_t *enc, char ch) 137encode_ch (enc_t *enc, char ch)
120{ 138{
121 need (enc, 1); 139 need (enc, 1);
122 *enc->cur++ = ch; 140 *enc->cur++ = ch;
123} 141}
… …
131 149
132 while (str < end) 150 while (str < end)
133 { 151 {
134 unsigned char ch = *(unsigned char *)str; 152 unsigned char ch = *(unsigned char *)str;
135 153
136 if (ch >= 0x20 && ch < 0x80) // most common case 154 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
137 { 155 {
138 if (ch == '"') // but with slow exceptions 156 if (expect_false (ch == '"')) // but with slow exceptions
139 { 157 {
140 need (enc, len += 1); 158 need (enc, len += 1);
141 *enc->cur++ = '\\'; 159 *enc->cur++ = '\\';
142 *enc->cur++ = '"'; 160 *enc->cur++ = '"';
143 } 161 }
144 else if (ch == '\\') 162 else if (expect_false (ch == '\\'))
145 { 163 {
146 need (enc, len += 1); 164 need (enc, len += 1);
147 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
148 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
149 } 167 }
… …
167 STRLEN clen; 185 STRLEN clen;
168 UV uch; 186 UV uch;
169 187
170 if (is_utf8) 188 if (is_utf8)
171 { 189 {
172 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
173 uch = decode_utf8 (str, end - str, &clen); 190 uch = decode_utf8 (str, end - str, &clen);
174 if (clen == (STRLEN)-1) 191 if (clen == (STRLEN)-1)
175 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 192 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
176 } 193 }
177 else 194 else
… …
233 250
234 --len; 251 --len;
235 } 252 }
236} 253}
237 254
238static void 255inline void
239encode_indent (enc_t *enc) 256encode_indent (enc_t *enc)
240{ 257{
241 if (enc->flags & F_INDENT) 258 if (enc->flags & F_INDENT)
242 { 259 {
243 int spaces = enc->indent * INDENT_STEP; 260 int spaces = enc->indent * INDENT_STEP;
… …
246 memset (enc->cur, ' ', spaces); 263 memset (enc->cur, ' ', spaces);
247 enc->cur += spaces; 264 enc->cur += spaces;
248 } 265 }
249} 266}
250 267
251static void 268inline void
252encode_space (enc_t *enc) 269encode_space (enc_t *enc)
253{ 270{
254 need (enc, 1); 271 need (enc, 1);
255 encode_ch (enc, ' '); 272 encode_ch (enc, ' ');
256} 273}
257 274
258static void 275inline void
259encode_nl (enc_t *enc) 276encode_nl (enc_t *enc)
260{ 277{
261 if (enc->flags & F_INDENT) 278 if (enc->flags & F_INDENT)
262 { 279 {
263 need (enc, 1); 280 need (enc, 1);
264 encode_ch (enc, '\n'); 281 encode_ch (enc, '\n');
265 } 282 }
266} 283}
267 284
268static void 285inline void
269encode_comma (enc_t *enc) 286encode_comma (enc_t *enc)
270{ 287{
271 encode_ch (enc, ','); 288 encode_ch (enc, ',');
272 289
273 if (enc->flags & F_INDENT) 290 if (enc->flags & F_INDENT)
… …
372 // actually, this is mostly due to the stupid so-called 389 // actually, this is mostly due to the stupid so-called
373 // security workaround added somewhere in 5.8.x. 390 // security workaround added somewhere in 5.8.x.
374 // that randomises hash orderings 391 // that randomises hash orderings
375 if (enc->flags & F_CANONICAL) 392 if (enc->flags & F_CANONICAL)
376 { 393 {
377 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
378 int fast = 1; 394 int fast = 1;
395 HE *he;
396#if defined(__BORLANDC__) || defined(_MSC_VER)
397 HE **hes = _alloca (count * sizeof (HE));
398#else
399 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
400#endif
379 401
380 i = 0; 402 i = 0;
381 while ((he = hv_iternext (hv))) 403 while ((he = hv_iternext (hv)))
382 { 404 {
383 hes [i++] = he; 405 hes [i++] = he;
… …
481 encode_str (enc, str, len, SvUTF8 (sv)); 503 encode_str (enc, str, len, SvUTF8 (sv));
482 encode_ch (enc, '"'); 504 encode_ch (enc, '"');
483 } 505 }
484 else if (SvNOKp (sv)) 506 else if (SvNOKp (sv))
485 { 507 {
508 // trust that perl will do the right thing w.r.t. JSON syntax.
486 need (enc, NV_DIG + 32); 509 need (enc, NV_DIG + 32);
487 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 510 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
488 enc->cur += strlen (enc->cur); 511 enc->cur += strlen (enc->cur);
489 } 512 }
490 else if (SvIOKp (sv)) 513 else if (SvIOKp (sv))
491 { 514 {
492 need (enc, 64); 515 // we assume we can always read an IV as a UV
516 if (SvUV (sv) & ~(UV)0x7fff)
517 {
518 // large integer, use the (rather slow) snprintf way.
519 need (enc, sizeof (UV) * 3);
493 enc->cur += 520 enc->cur +=
494 SvIsUV(sv) 521 SvIsUV(sv)
495 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 522 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
496 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 523 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
524 }
525 else
526 {
527 // optimise the "small number case"
528 // code will likely be branchless and use only a single multiplication
529 I32 i = SvIV (sv);
530 U32 u;
531 char digit, nz = 0;
532
533 need (enc, 6);
534
535 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
536 u = i < 0 ? -i : i;
537
538 // convert to 4.28 fixed-point representation
539 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
540
541 // now output digit by digit, each time masking out the integer part
542 // and multiplying by 5 while moving the decimal point one to the right,
543 // resulting in a net multiplication by 10.
544 // we always write the digit to memory but conditionally increment
545 // the pointer, to ease the usage of conditional move instructions.
546 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
547 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
548 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
549 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
550 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
551 }
497 } 552 }
498 else if (SvROK (sv)) 553 else if (SvROK (sv))
499 encode_rv (enc, SvRV (sv)); 554 encode_rv (enc, SvRV (sv));
500 else if (!SvOK (sv)) 555 else if (!SvOK (sv))
501 encode_str (enc, "null", 4, 0); 556 encode_str (enc, "null", 4, 0);
… …
546 U32 flags; // F_* 601 U32 flags; // F_*
547 U32 depth; // recursion depth 602 U32 depth; // recursion depth
548 U32 maxdepth; // recursion depth limit 603 U32 maxdepth; // recursion depth limit
549} dec_t; 604} dec_t;
550 605
551static void 606inline void
552decode_ws (dec_t *dec) 607decode_ws (dec_t *dec)
553{ 608{
554 for (;;) 609 for (;;)
555 { 610 {
556 char ch = *dec->cur; 611 char ch = *dec->cur;
… …
582decode_4hex (dec_t *dec) 637decode_4hex (dec_t *dec)
583{ 638{
584 signed char d1, d2, d3, d4; 639 signed char d1, d2, d3, d4;
585 unsigned char *cur = (unsigned char *)dec->cur; 640 unsigned char *cur = (unsigned char *)dec->cur;
586 641
587 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 642 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
588 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 643 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
589 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 644 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
590 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 645 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
591 646
592 dec->cur += 4; 647 dec->cur += 4;
593 648
594 return ((UV)d1) << 12 649 return ((UV)d1) << 12
595 | ((UV)d2) << 8 650 | ((UV)d2) << 8
… …
603static SV * 658static SV *
604decode_str (dec_t *dec) 659decode_str (dec_t *dec)
605{ 660{
606 SV *sv = 0; 661 SV *sv = 0;
607 int utf8 = 0; 662 int utf8 = 0;
663 char *dec_cur = dec->cur;
608 664
609 do 665 do
610 { 666 {
611 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; 667 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
612 char *cur = buf; 668 char *cur = buf;
613 669
614 do 670 do
615 { 671 {
616 unsigned char ch = *(unsigned char *)dec->cur++; 672 unsigned char ch = *(unsigned char *)dec_cur++;
617 673
618 if (ch == '"') 674 if (expect_false (ch == '"'))
619 { 675 {
620 --dec->cur; 676 --dec_cur;
621 break; 677 break;
622 } 678 }
623 else if (ch == '\\') 679 else if (expect_false (ch == '\\'))
624 { 680 {
625 switch (*dec->cur) 681 switch (*dec_cur)
626 { 682 {
627 case '\\': 683 case '\\':
628 case '/': 684 case '/':
629 case '"': *cur++ = *dec->cur++; break; 685 case '"': *cur++ = *dec_cur++; break;
630 686
631 case 'b': ++dec->cur; *cur++ = '\010'; break; 687 case 'b': ++dec_cur; *cur++ = '\010'; break;
632 case 't': ++dec->cur; *cur++ = '\011'; break; 688 case 't': ++dec_cur; *cur++ = '\011'; break;
633 case 'n': ++dec->cur; *cur++ = '\012'; break; 689 case 'n': ++dec_cur; *cur++ = '\012'; break;
634 case 'f': ++dec->cur; *cur++ = '\014'; break; 690 case 'f': ++dec_cur; *cur++ = '\014'; break;
635 case 'r': ++dec->cur; *cur++ = '\015'; break; 691 case 'r': ++dec_cur; *cur++ = '\015'; break;
636 692
637 case 'u': 693 case 'u':
638 { 694 {
639 UV lo, hi; 695 UV lo, hi;
640 ++dec->cur; 696 ++dec_cur;
641 697
698 dec->cur = dec_cur;
642 hi = decode_4hex (dec); 699 hi = decode_4hex (dec);
700 dec_cur = dec->cur;
643 if (hi == (UV)-1) 701 if (hi == (UV)-1)
644 goto fail; 702 goto fail;
645 703
646 // possibly a surrogate pair 704 // possibly a surrogate pair
647 if (hi >= 0xd800) 705 if (hi >= 0xd800)
648 if (hi < 0xdc00) 706 if (hi < 0xdc00)
649 { 707 {
650 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 708 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
651 ERR ("missing low surrogate character in surrogate pair"); 709 ERR ("missing low surrogate character in surrogate pair");
652 710
653 dec->cur += 2; 711 dec_cur += 2;
654 712
713 dec->cur = dec_cur;
655 lo = decode_4hex (dec); 714 lo = decode_4hex (dec);
715 dec_cur = dec->cur;
656 if (lo == (UV)-1) 716 if (lo == (UV)-1)
657 goto fail; 717 goto fail;
658 718
659 if (lo < 0xdc00 || lo >= 0xe000) 719 if (lo < 0xdc00 || lo >= 0xe000)
660 ERR ("surrogate pair expected"); 720 ERR ("surrogate pair expected");
… …
674 *cur++ = hi; 734 *cur++ = hi;
675 } 735 }
676 break; 736 break;
677 737
678 default: 738 default:
679 --dec->cur; 739 --dec_cur;
680 ERR ("illegal backslash escape sequence in string"); 740 ERR ("illegal backslash escape sequence in string");
681 } 741 }
682 } 742 }
683 else if (ch >= 0x20 && ch <= 0x7f) 743 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
684 *cur++ = ch; 744 *cur++ = ch;
685 else if (ch >= 0x80) 745 else if (ch >= 0x80)
686 { 746 {
687 STRLEN clen; 747 STRLEN clen;
688 UV uch; 748 UV uch;
689 749
690 --dec->cur; 750 --dec_cur;
691 751
692 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 752 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
693 if (clen == (STRLEN)-1) 753 if (clen == (STRLEN)-1)
694 ERR ("malformed UTF-8 character in JSON string"); 754 ERR ("malformed UTF-8 character in JSON string");
695 755
696 do 756 do
697 *cur++ = *dec->cur++; 757 *cur++ = *dec_cur++;
698 while (--clen); 758 while (--clen);
699 759
700 utf8 = 1; 760 utf8 = 1;
701 } 761 }
702 else 762 else
703 { 763 {
704 --dec->cur; 764 --dec_cur;
705 765
706 if (!ch) 766 if (!ch)
707 ERR ("unexpected end of string while parsing JSON string"); 767 ERR ("unexpected end of string while parsing JSON string");
708 else 768 else
709 ERR ("invalid character encountered while parsing JSON string"); 769 ERR ("invalid character encountered while parsing JSON string");
… …
722 } 782 }
723 else 783 else
724 sv = newSVpvn (buf, len); 784 sv = newSVpvn (buf, len);
725 } 785 }
726 } 786 }
727 while (*dec->cur != '"'); 787 while (*dec_cur != '"');
728 788
729 ++dec->cur; 789 ++dec_cur;
730 790
731 if (sv) 791 if (sv)
732 { 792 {
733 SvPOK_only (sv); 793 SvPOK_only (sv);
734 *SvEND (sv) = 0; 794 *SvEND (sv) = 0;
… …
737 SvUTF8_on (sv); 797 SvUTF8_on (sv);
738 } 798 }
739 else 799 else
740 sv = newSVpvn ("", 0); 800 sv = newSVpvn ("", 0);
741 801
802 dec->cur = dec_cur;
742 return sv; 803 return sv;
743 804
744fail: 805fail:
806 dec->cur = dec_cur;
745 return 0; 807 return 0;
746} 808}
747 809
748static SV * 810static SV *
749decode_num (dec_t *dec) 811decode_num (dec_t *dec)
… …
807 is_nv = 1; 869 is_nv = 1;
808 } 870 }
809 871
810 if (!is_nv) 872 if (!is_nv)
811 { 873 {
812 UV uv; 874 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
813 int numtype = grok_number (start, dec->cur - start, &uv); 875 if (*start == '-')
814 if (numtype & IS_NUMBER_IN_UV) 876 switch (dec->cur - start)
815 if (numtype & IS_NUMBER_NEG)
816 { 877 {
817 if (uv < (UV)IV_MIN) 878 case 2: return newSViv (-( start [1] - '0' * 1));
818 return newSViv (-(IV)uv); 879 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
880 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
881 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
819 } 882 }
883 else
884 switch (dec->cur - start)
885 {
886 case 1: return newSViv ( start [0] - '0' * 1);
887 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
888 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
889 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
890 }
891
892 {
893 UV uv;
894 int numtype = grok_number (start, dec->cur - start, &uv);
895 if (numtype & IS_NUMBER_IN_UV)
896 if (numtype & IS_NUMBER_NEG)
897 {
898 if (uv < (UV)IV_MIN)
899 return newSViv (-(IV)uv);
900 }
820 else 901 else
821 return newSVuv (uv); 902 return newSVuv (uv);
903
904 // here would likely be the place for bigint support
822 } 905 }
906 }
823 907
908 // if we ever support bigint or bigfloat, this is the place for bigfloat
824 return newSVnv (Atof (start)); 909 return newSVnv (Atof (start));
825 910
826fail: 911fail:
827 return 0; 912 return 0;
828} 913}
… …
929 1014
930static SV * 1015static SV *
931decode_sv (dec_t *dec) 1016decode_sv (dec_t *dec)
932{ 1017{
933 decode_ws (dec); 1018 decode_ws (dec);
1019
1020 // the beauty of JSON: you need exactly one character lookahead
1021 // to parse anything.
934 switch (*dec->cur) 1022 switch (*dec->cur)
935 { 1023 {
936 case '"': ++dec->cur; return decode_str (dec); 1024 case '"': ++dec->cur; return decode_str (dec);
937 case '[': ++dec->cur; return decode_av (dec); 1025 case '[': ++dec->cur; return decode_av (dec);
938 case '{': ++dec->cur; return decode_hv (dec); 1026 case '{': ++dec->cur; return decode_hv (dec);
… …
944 1032
945 case 't': 1033 case 't':
946 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1034 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
947 { 1035 {
948 dec->cur += 4; 1036 dec->cur += 4;
949 return newSViv (1); 1037 return SvREFCNT_inc (json_true);
950 } 1038 }
951 else 1039 else
952 ERR ("'true' expected"); 1040 ERR ("'true' expected");
953 1041
954 break; 1042 break;
955 1043
956 case 'f': 1044 case 'f':
957 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1045 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
958 { 1046 {
959 dec->cur += 5; 1047 dec->cur += 5;
960 return newSViv (0); 1048 return SvREFCNT_inc (json_false);
961 } 1049 }
962 else 1050 else
963 ERR ("'false' expected"); 1051 ERR ("'false' expected");
964 1052
965 break; 1053 break;
… …
983fail: 1071fail:
984 return 0; 1072 return 0;
985} 1073}
986 1074
987static SV * 1075static SV *
988decode_json (SV *string, U32 flags) 1076decode_json (SV *string, U32 flags, UV *offset_return)
989{ 1077{
990 dec_t dec; 1078 dec_t dec;
1079 UV offset;
991 SV *sv; 1080 SV *sv;
992 1081
993 SvGETMAGIC (string); 1082 SvGETMAGIC (string);
994 SvUPGRADE (string, SVt_PV); 1083 SvUPGRADE (string, SVt_PV);
995 1084
… …
1005 dec.end = SvEND (string); 1094 dec.end = SvEND (string);
1006 dec.err = 0; 1095 dec.err = 0;
1007 dec.depth = 0; 1096 dec.depth = 0;
1008 dec.maxdepth = DEC_DEPTH (dec.flags); 1097 dec.maxdepth = DEC_DEPTH (dec.flags);
1009 1098
1010 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1099 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1011 sv = decode_sv (&dec); 1100 sv = decode_sv (&dec);
1012 1101
1102 if (!(offset_return || !sv))
1103 {
1104 // check for trailing garbage
1105 decode_ws (&dec);
1106
1107 if (*dec.cur)
1108 {
1109 dec.err = "garbage after JSON object";
1110 SvREFCNT_dec (sv);
1111 sv = 0;
1112 }
1113 }
1114
1115 if (offset_return || !sv)
1116 {
1117 offset = dec.flags & F_UTF8
1118 ? dec.cur - SvPVX (string)
1119 : utf8_distance (dec.cur, SvPVX (string));
1120
1121 if (offset_return)
1122 *offset_return = offset;
1123 }
1124
1013 if (!sv) 1125 if (!sv)
1014 { 1126 {
1015 IV offset = dec.flags & F_UTF8
1016 ? dec.cur - SvPVX (string)
1017 : utf8_distance (dec.cur, SvPVX (string));
1018 SV *uni = sv_newmortal (); 1127 SV *uni = sv_newmortal ();
1019 1128
1020 // horrible hack to silence warning inside pv_uni_display 1129 // horrible hack to silence warning inside pv_uni_display
1021 COP cop = *PL_curcop; 1130 COP cop = *PL_curcop;
1022 cop.cop_warnings = pWARN_NONE; 1131 cop.cop_warnings = pWARN_NONE;
… …
1047 1156
1048BOOT: 1157BOOT:
1049{ 1158{
1050 int i; 1159 int i;
1051 1160
1052 memset (decode_hexdigit, 0xff, 256);
1053
1054 for (i = 0; i < 256; ++i) 1161 for (i = 0; i < 256; ++i)
1055 decode_hexdigit [i] = 1162 decode_hexdigit [i] =
1056 i >= '0' && i <= '9' ? i - '0' 1163 i >= '0' && i <= '9' ? i - '0'
1057 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1164 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1058 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1165 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1059 : -1; 1166 : -1;
1060 1167
1061 json_stash = gv_stashpv ("JSON::XS", 1); 1168 json_stash = gv_stashpv ("JSON::XS", 1);
1169
1170 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1171 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1062} 1172}
1063 1173
1064PROTOTYPES: DISABLE 1174PROTOTYPES: DISABLE
1065 1175
1066SV *new (char *dummy) 1176SV *new (char *dummy)
… …
1116 PPCODE: 1226 PPCODE:
1117 XPUSHs (encode_json (scalar, *SvJSON (self))); 1227 XPUSHs (encode_json (scalar, *SvJSON (self)));
1118 1228
1119void decode (SV *self, SV *jsonstr) 1229void decode (SV *self, SV *jsonstr)
1120 PPCODE: 1230 PPCODE:
1121 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1231 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1232
1233void decode_prefix (SV *self, SV *jsonstr)
1234 PPCODE:
1235{
1236 UV offset;
1237 EXTEND (SP, 2);
1238 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1239 PUSHs (sv_2mortal (newSVuv (offset)));
1240}
1122 1241
1123PROTOTYPES: ENABLE 1242PROTOTYPES: ENABLE
1124 1243
1125void to_json (SV *scalar) 1244void to_json (SV *scalar)
1126 ALIAS: 1245 ALIAS:
… …
1130 1249
1131void from_json (SV *jsonstr) 1250void from_json (SV *jsonstr)
1132 ALIAS: 1251 ALIAS:
1133 jsonToObj = 0 1252 jsonToObj = 0
1134 PPCODE: 1253 PPCODE:
1135 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1254 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1136 1255

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines