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 1.15 by root, Sun Mar 25 02:32:40 2007 UTC vs.
Revision 1.26 by root, Fri Apr 6 21:17:09 2007 UTC

3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include "assert.h"
6#include "string.h" 6#include "string.h"
7#include "stdlib.h" 7#include "stdlib.h"
8#include "stdio.h"
8 9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
13
9#define F_ASCII 0x00000001 14#define F_ASCII 0x00000001UL
10#define F_UTF8 0x00000002 15#define F_UTF8 0x00000002UL
11#define F_INDENT 0x00000004 16#define F_INDENT 0x00000004UL
12#define F_CANONICAL 0x00000008 17#define F_CANONICAL 0x00000008UL
13#define F_SPACE_BEFORE 0x00000010 18#define F_SPACE_BEFORE 0x00000010UL
14#define F_SPACE_AFTER 0x00000020 19#define F_SPACE_AFTER 0x00000020UL
15#define F_ALLOW_NONREF 0x00000080 20#define F_ALLOW_NONREF 0x00000080UL
16#define F_SHRINK 0x00000100 21#define F_SHRINK 0x00000100UL
22#define F_MAXDEPTH 0xf8000000UL
23#define S_MAXDEPTH 27
24
25#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
26
27// F_SELFCONVERT? <=> to_json/toJson
28// F_BLESSED? <=> { $__class__$ => }
17 29
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 30#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 31#define F_DEFAULT (9UL << S_MAXDEPTH)
20 32
21#define INIT_SIZE 32 // initial scalar size to be allocated 33#define INIT_SIZE 32 // initial scalar size to be allocated
22#define INDENT_STEP 3 // spaces per indentation level 34#define INDENT_STEP 3 // spaces per indentation level
23 35
24#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character 36#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
69 { 81 {
70 *clen = 2; 82 *clen = 2;
71 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 83 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
72 } 84 }
73 else 85 else
86 {
87 *clen = (STRLEN)-1;
74 return (UV)-1; 88 return (UV)-1;
89 }
75} 90}
76 91
77///////////////////////////////////////////////////////////////////////////// 92/////////////////////////////////////////////////////////////////////////////
78// encoder 93// encoder
79 94
81typedef struct 96typedef struct
82{ 97{
83 char *cur; // SvPVX (sv) + current output position 98 char *cur; // SvPVX (sv) + current output position
84 char *end; // SvEND (sv) 99 char *end; // SvEND (sv)
85 SV *sv; // result scalar 100 SV *sv; // result scalar
86 UV flags; // F_* 101 U32 flags; // F_*
87 int indent; // indentation level 102 U32 indent; // indentation level
88 int max_depth; // max. recursion level 103 U32 maxdepth; // max. indentation/recursion level
89} enc_t; 104} enc_t;
90 105
91static void 106static void
92need (enc_t *enc, STRLEN len) 107need (enc_t *enc, STRLEN len)
93{ 108{
261static void 276static void
262encode_av (enc_t *enc, AV *av) 277encode_av (enc_t *enc, AV *av)
263{ 278{
264 int i, len = av_len (av); 279 int i, len = av_len (av);
265 280
281 if (enc->indent >= enc->maxdepth)
282 croak ("data structure too deep (hit recursion limit)");
283
266 encode_ch (enc, '['); encode_nl (enc); 284 encode_ch (enc, '['); encode_nl (enc);
267 ++enc->indent; 285 ++enc->indent;
268 286
269 for (i = 0; i <= len; ++i) 287 for (i = 0; i <= len; ++i)
270 { 288 {
335 353
336static void 354static void
337encode_hv (enc_t *enc, HV *hv) 355encode_hv (enc_t *enc, HV *hv)
338{ 356{
339 int count, i; 357 int count, i;
358
359 if (enc->indent >= enc->maxdepth)
360 croak ("data structure too deep (hit recursion limit)");
340 361
341 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 362 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
342 363
343 if ((count = hv_iterinit (hv))) 364 if ((count = hv_iterinit (hv)))
344 { 365 {
392 413
393 encode_nl (enc); 414 encode_nl (enc);
394 } 415 }
395 else 416 else
396 { 417 {
397 SV *sv;
398 HE *he = hv_iternext (hv); 418 HE *he = hv_iternext (hv);
399 419
400 for (;;) 420 for (;;)
401 { 421 {
402 encode_indent (enc); 422 encode_indent (enc);
411 encode_nl (enc); 431 encode_nl (enc);
412 } 432 }
413 } 433 }
414 434
415 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 435 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
436}
437
438// encode objects, arrays and special \0=false and \1=true values.
439static void
440encode_rv (enc_t *enc, SV *sv)
441{
442 svtype svt;
443
444 SvGETMAGIC (sv);
445 svt = SvTYPE (sv);
446
447 if (svt == SVt_PVHV)
448 encode_hv (enc, (HV *)sv);
449 else if (svt == SVt_PVAV)
450 encode_av (enc, (AV *)sv);
451 else if (svt < SVt_PVAV)
452 {
453 if (SvNIOK (sv) && SvIV (sv) == 0)
454 encode_str (enc, "false", 5, 0);
455 else if (SvNIOK (sv) && SvIV (sv) == 1)
456 encode_str (enc, "true", 4, 0);
457 else
458 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
459 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
460 }
461 else
462 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
463 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416} 464}
417 465
418static void 466static void
419encode_sv (enc_t *enc, SV *sv) 467encode_sv (enc_t *enc, SV *sv)
420{ 468{
441 SvIsUV(sv) 489 SvIsUV(sv)
442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 490 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 491 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
444 } 492 }
445 else if (SvROK (sv)) 493 else if (SvROK (sv))
446 { 494 encode_rv (enc, SvRV (sv));
447 SV *rv = SvRV (sv);
448
449 if (enc->indent >= enc->max_depth)
450 croak ("data structure too deep (hit recursion limit)");
451
452 switch (SvTYPE (rv))
453 {
454 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
455 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
456
457 default:
458 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
459 SvPV_nolen (sv));
460 }
461 }
462 else if (!SvOK (sv)) 495 else if (!SvOK (sv))
463 encode_str (enc, "null", 4, 0); 496 encode_str (enc, "null", 4, 0);
464 else 497 else
465 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 498 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466 SvPV_nolen (sv), SvFLAGS (sv)); 499 SvPV_nolen (sv), SvFLAGS (sv));
467} 500}
468 501
469static SV * 502static SV *
470encode_json (SV *scalar, UV flags) 503encode_json (SV *scalar, U32 flags)
471{ 504{
505 enc_t enc;
506
472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 507 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 508 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474 509
475 enc_t enc;
476 enc.flags = flags; 510 enc.flags = flags;
477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 511 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478 enc.cur = SvPVX (enc.sv); 512 enc.cur = SvPVX (enc.sv);
479 enc.end = SvEND (enc.sv); 513 enc.end = SvEND (enc.sv);
480 enc.indent = 0; 514 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL; 515 enc.maxdepth = DEC_DEPTH (flags);
482 516
483 SvPOK_only (enc.sv); 517 SvPOK_only (enc.sv);
484 encode_sv (&enc, scalar); 518 encode_sv (&enc, scalar);
485 519
486 if (!(flags & (F_ASCII | F_UTF8))) 520 if (!(flags & (F_ASCII | F_UTF8)))
501typedef struct 535typedef struct
502{ 536{
503 char *cur; // current parser pointer 537 char *cur; // current parser pointer
504 char *end; // end of input string 538 char *end; // end of input string
505 const char *err; // parse error, if != 0 539 const char *err; // parse error, if != 0
506 UV flags; // F_* 540 U32 flags; // F_*
541 U32 depth; // recursion depth
542 U32 maxdepth; // recursion depth limit
507} dec_t; 543} dec_t;
508 544
509static void 545static void
510decode_ws (dec_t *dec) 546decode_ws (dec_t *dec)
511{ 547{
520 ++dec->cur; 556 ++dec->cur;
521 } 557 }
522} 558}
523 559
524#define ERR(reason) SB dec->err = reason; goto fail; SE 560#define ERR(reason) SB dec->err = reason; goto fail; SE
561
525#define EXPECT_CH(ch) SB \ 562#define EXPECT_CH(ch) SB \
526 if (*dec->cur != ch) \ 563 if (*dec->cur != ch) \
527 ERR (# ch " expected"); \ 564 ERR (# ch " expected"); \
528 ++dec->cur; \ 565 ++dec->cur; \
529 SE 566 SE
567
568#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
569#define DEC_DEC_DEPTH --dec->depth
530 570
531static SV *decode_sv (dec_t *dec); 571static SV *decode_sv (dec_t *dec);
532 572
533static signed char decode_hexdigit[256]; 573static signed char decode_hexdigit[256];
534 574
636 } 676 }
637 else if (ch >= 0x20 && ch <= 0x7f) 677 else if (ch >= 0x20 && ch <= 0x7f)
638 *cur++ = ch; 678 *cur++ = ch;
639 else if (ch >= 0x80) 679 else if (ch >= 0x80)
640 { 680 {
681 STRLEN clen;
682 UV uch;
683
641 --dec->cur; 684 --dec->cur;
642 685
643 STRLEN clen;
644 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 686 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
645 if (clen == (STRLEN)-1) 687 if (clen == (STRLEN)-1)
646 ERR ("malformed UTF-8 character in JSON string"); 688 ERR ("malformed UTF-8 character in JSON string");
647 689
648 do 690 do
649 {
650 *cur++ = *dec->cur++; 691 *cur++ = *dec->cur++;
651 }
652 while (--clen); 692 while (--clen);
653 693
654 utf8 = 1; 694 utf8 = 1;
655 } 695 }
656 else if (!ch)
657 ERR ("unexpected end of string while parsing json string");
658 else 696 else
697 {
698 --dec->cur;
699
700 if (!ch)
701 ERR ("unexpected end of string while parsing JSON string");
702 else
659 ERR ("invalid character encountered"); 703 ERR ("invalid character encountered while parsing JSON string");
660 704 }
661 } 705 }
662 while (cur < buf + SHORT_STRING_LEN); 706 while (cur < buf + SHORT_STRING_LEN);
663 707
708 {
664 STRLEN len = cur - buf; 709 STRLEN len = cur - buf;
665 710
666 if (sv) 711 if (sv)
667 { 712 {
668 SvGROW (sv, SvCUR (sv) + len + 1); 713 SvGROW (sv, SvCUR (sv) + len + 1);
669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 714 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670 SvCUR_set (sv, SvCUR (sv) + len); 715 SvCUR_set (sv, SvCUR (sv) + len);
671 } 716 }
672 else 717 else
673 sv = newSVpvn (buf, len); 718 sv = newSVpvn (buf, len);
719 }
674 } 720 }
675 while (*dec->cur != '"'); 721 while (*dec->cur != '"');
676 722
677 ++dec->cur; 723 ++dec->cur;
678 724
778static SV * 824static SV *
779decode_av (dec_t *dec) 825decode_av (dec_t *dec)
780{ 826{
781 AV *av = newAV (); 827 AV *av = newAV ();
782 828
829 DEC_INC_DEPTH;
783 decode_ws (dec); 830 decode_ws (dec);
831
784 if (*dec->cur == ']') 832 if (*dec->cur == ']')
785 ++dec->cur; 833 ++dec->cur;
786 else 834 else
787 for (;;) 835 for (;;)
788 { 836 {
806 ERR (", or ] expected while parsing array"); 854 ERR (", or ] expected while parsing array");
807 855
808 ++dec->cur; 856 ++dec->cur;
809 } 857 }
810 858
859 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)av); 860 return newRV_noinc ((SV *)av);
812 861
813fail: 862fail:
814 SvREFCNT_dec (av); 863 SvREFCNT_dec (av);
864 DEC_DEC_DEPTH;
815 return 0; 865 return 0;
816} 866}
817 867
818static SV * 868static SV *
819decode_hv (dec_t *dec) 869decode_hv (dec_t *dec)
820{ 870{
821 HV *hv = newHV (); 871 HV *hv = newHV ();
822 872
873 DEC_INC_DEPTH;
823 decode_ws (dec); 874 decode_ws (dec);
875
824 if (*dec->cur == '}') 876 if (*dec->cur == '}')
825 ++dec->cur; 877 ++dec->cur;
826 else 878 else
827 for (;;) 879 for (;;)
828 { 880 {
841 { 893 {
842 SvREFCNT_dec (key); 894 SvREFCNT_dec (key);
843 goto fail; 895 goto fail;
844 } 896 }
845 897
846 //TODO: optimise
847 hv_store_ent (hv, key, value, 0); 898 hv_store_ent (hv, key, value, 0);
899 SvREFCNT_dec (key);
848 900
849 decode_ws (dec); 901 decode_ws (dec);
850 902
851 if (*dec->cur == '}') 903 if (*dec->cur == '}')
852 { 904 {
858 ERR (", or } expected while parsing object/hash"); 910 ERR (", or } expected while parsing object/hash");
859 911
860 ++dec->cur; 912 ++dec->cur;
861 } 913 }
862 914
915 DEC_DEC_DEPTH;
863 return newRV_noinc ((SV *)hv); 916 return newRV_noinc ((SV *)hv);
864 917
865fail: 918fail:
866 SvREFCNT_dec (hv); 919 SvREFCNT_dec (hv);
920 DEC_DEC_DEPTH;
867 return 0; 921 return 0;
868} 922}
869 923
870static SV * 924static SV *
871decode_sv (dec_t *dec) 925decode_sv (dec_t *dec)
914 ERR ("'null' expected"); 968 ERR ("'null' expected");
915 969
916 break; 970 break;
917 971
918 default: 972 default:
919 ERR ("malformed json string, neither array, object, number, string or atom"); 973 ERR ("malformed JSON string, neither array, object, number, string or atom");
920 break; 974 break;
921 } 975 }
922 976
923fail: 977fail:
924 return 0; 978 return 0;
925} 979}
926 980
927static SV * 981static SV *
928decode_json (SV *string, UV flags) 982decode_json (SV *string, U32 flags)
929{ 983{
984 dec_t dec;
930 SV *sv; 985 SV *sv;
986
987 SvUPGRADE (string, SVt_PV);
931 988
932 if (flags & F_UTF8) 989 if (flags & F_UTF8)
933 sv_utf8_downgrade (string, 0); 990 sv_utf8_downgrade (string, 0);
934 else 991 else
935 sv_utf8_upgrade (string); 992 sv_utf8_upgrade (string);
936 993
937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 994 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938 995
939 dec_t dec;
940 dec.flags = flags; 996 dec.flags = flags;
941 dec.cur = SvPVX (string); 997 dec.cur = SvPVX (string);
942 dec.end = SvEND (string); 998 dec.end = SvEND (string);
943 dec.err = 0; 999 dec.err = 0;
1000 dec.depth = 0;
1001 dec.maxdepth = DEC_DEPTH (dec.flags);
944 1002
1003 *dec.end = 0; // this should basically be a nop, too, but make sure its there
945 sv = decode_sv (&dec); 1004 sv = decode_sv (&dec);
946 1005
947 if (!sv) 1006 if (!sv)
948 { 1007 {
949 IV offset = dec.flags & F_UTF8 1008 IV offset = dec.flags & F_UTF8
958 SAVEVPTR (PL_curcop); 1017 SAVEVPTR (PL_curcop);
959 PL_curcop = &cop; 1018 PL_curcop = &cop;
960 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1019 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961 LEAVE; 1020 LEAVE;
962 1021
963 croak ("%s, at character offset %d (%s)", 1022 croak ("%s, at character offset %d [\"%s\"]",
964 dec.err, 1023 dec.err,
965 (int)offset, 1024 (int)offset,
966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1025 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967 } 1026 }
968 1027
982BOOT: 1041BOOT:
983{ 1042{
984 int i; 1043 int i;
985 1044
986 memset (decode_hexdigit, 0xff, 256); 1045 memset (decode_hexdigit, 0xff, 256);
1046
987 for (i = 10; i--; ) 1047 for (i = 0; i < 256; ++i)
988 decode_hexdigit ['0' + i] = i; 1048 decode_hexdigit [i] =
989 1049 i >= '0' && i <= '9' ? i - '0'
990 for (i = 7; i--; ) 1050 : i >= 'a' && i <= 'f' ? i - 'a' + 10
991 { 1051 : i >= 'A' && i <= 'F' ? i - 'A' + 10
992 decode_hexdigit ['a' + i] = 10 + i; 1052 : -1;
993 decode_hexdigit ['A' + i] = 10 + i;
994 }
995 1053
996 json_stash = gv_stashpv ("JSON::XS", 1); 1054 json_stash = gv_stashpv ("JSON::XS", 1);
997} 1055}
998 1056
999PROTOTYPES: DISABLE 1057PROTOTYPES: DISABLE
1026 RETVAL = newSVsv (self); 1084 RETVAL = newSVsv (self);
1027} 1085}
1028 OUTPUT: 1086 OUTPUT:
1029 RETVAL 1087 RETVAL
1030 1088
1089SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1090 CODE:
1091{
1092 UV *uv = SvJSON (self);
1093 UV log2 = 0;
1094
1095 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1096
1097 while ((1UL << log2) < max_depth)
1098 ++log2;
1099
1100 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1101
1102 RETVAL = newSVsv (self);
1103}
1104 OUTPUT:
1105 RETVAL
1106
1031void encode (SV *self, SV *scalar) 1107void encode (SV *self, SV *scalar)
1032 PPCODE: 1108 PPCODE:
1033 XPUSHs (encode_json (scalar, *SvJSON (self))); 1109 XPUSHs (encode_json (scalar, *SvJSON (self)));
1034 1110
1035void decode (SV *self, SV *jsonstr) 1111void decode (SV *self, SV *jsonstr)
1040 1116
1041void to_json (SV *scalar) 1117void to_json (SV *scalar)
1042 ALIAS: 1118 ALIAS:
1043 objToJson = 0 1119 objToJson = 0
1044 PPCODE: 1120 PPCODE:
1045 XPUSHs (encode_json (scalar, F_UTF8)); 1121 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1046 1122
1047void from_json (SV *jsonstr) 1123void from_json (SV *jsonstr)
1048 ALIAS: 1124 ALIAS:
1049 jsonToObj = 0 1125 jsonToObj = 0
1050 PPCODE: 1126 PPCODE:
1051 XPUSHs (decode_json (jsonstr, F_UTF8)); 1127 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
1052 1128

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines