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 1.14 by root, Sat Mar 24 22:55:43 2007 UTC vs.
Revision 1.25 by root, Fri Apr 6 20:37:22 2007 UTC

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 8
9#define F_ASCII 0x00000001 9#define F_ASCII 0x00000001UL
10#define F_UTF8 0x00000002 10#define F_UTF8 0x00000002UL
11#define F_INDENT 0x00000004 11#define F_INDENT 0x00000004UL
12#define F_CANONICAL 0x00000008 12#define F_CANONICAL 0x00000008UL
13#define F_SPACE_BEFORE 0x00000010 13#define F_SPACE_BEFORE 0x00000010UL
14#define F_SPACE_AFTER 0x00000020 14#define F_SPACE_AFTER 0x00000020UL
15#define F_ALLOW_NONREF 0x00000080 15#define F_ALLOW_NONREF 0x00000080UL
16#define F_SHRINK 0x00000100 16#define F_SHRINK 0x00000100UL
17#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27
19
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
17 24
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 26#define F_DEFAULT (9UL << S_MAXDEPTH)
20 27
21#define INIT_SIZE 32 // initial scalar size to be allocated 28#define INIT_SIZE 32 // initial scalar size to be allocated
22#define INDENT_STEP 3 // spaces per indentation level 29#define INDENT_STEP 3 // spaces per indentation level
23 30
24#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character 31#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
69 { 76 {
70 *clen = 2; 77 *clen = 2;
71 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
72 } 79 }
73 else 80 else
81 {
82 *clen = (STRLEN)-1;
74 return (UV)-1; 83 return (UV)-1;
84 }
75} 85}
76 86
77///////////////////////////////////////////////////////////////////////////// 87/////////////////////////////////////////////////////////////////////////////
78// encoder 88// encoder
79 89
81typedef struct 91typedef struct
82{ 92{
83 char *cur; // SvPVX (sv) + current output position 93 char *cur; // SvPVX (sv) + current output position
84 char *end; // SvEND (sv) 94 char *end; // SvEND (sv)
85 SV *sv; // result scalar 95 SV *sv; // result scalar
86 UV flags; // F_* 96 U32 flags; // F_*
87 int indent; // indentation level 97 U32 indent; // indentation level
88 int max_depth; // max. recursion level 98 U32 maxdepth; // max. indentation/recursion level
89} enc_t; 99} enc_t;
90 100
91static void 101static void
92need (enc_t *enc, STRLEN len) 102need (enc_t *enc, STRLEN len)
93{ 103{
261static void 271static void
262encode_av (enc_t *enc, AV *av) 272encode_av (enc_t *enc, AV *av)
263{ 273{
264 int i, len = av_len (av); 274 int i, len = av_len (av);
265 275
276 if (enc->indent >= enc->maxdepth)
277 croak ("data structure too deep (hit recursion limit)");
278
266 encode_ch (enc, '['); encode_nl (enc); 279 encode_ch (enc, '['); encode_nl (enc);
267 ++enc->indent; 280 ++enc->indent;
268 281
269 for (i = 0; i <= len; ++i) 282 for (i = 0; i <= len; ++i)
270 { 283 {
335 348
336static void 349static void
337encode_hv (enc_t *enc, HV *hv) 350encode_hv (enc_t *enc, HV *hv)
338{ 351{
339 int count, i; 352 int count, i;
353
354 if (enc->indent >= enc->maxdepth)
355 croak ("data structure too deep (hit recursion limit)");
340 356
341 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 357 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
342 358
343 if ((count = hv_iterinit (hv))) 359 if ((count = hv_iterinit (hv)))
344 { 360 {
392 408
393 encode_nl (enc); 409 encode_nl (enc);
394 } 410 }
395 else 411 else
396 { 412 {
397 SV *sv;
398 HE *he = hv_iternext (hv); 413 HE *he = hv_iternext (hv);
399 414
400 for (;;) 415 for (;;)
401 { 416 {
402 encode_indent (enc); 417 encode_indent (enc);
411 encode_nl (enc); 426 encode_nl (enc);
412 } 427 }
413 } 428 }
414 429
415 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 430 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
431}
432
433// encode objects, arrays and special \0=false and \1=true values.
434static void
435encode_rv (enc_t *enc, SV *sv)
436{
437 svtype svt;
438
439 SvGETMAGIC (sv);
440 svt = SvTYPE (sv);
441
442 if (svt == SVt_PVHV)
443 encode_hv (enc, (HV *)sv);
444 else if (svt == SVt_PVAV)
445 encode_av (enc, (AV *)sv);
446 else if (svt < SVt_PVAV)
447 {
448 if (SvNIOK (sv) && SvIV (sv) == 0)
449 encode_str (enc, "false", 5, 0);
450 else if (SvNIOK (sv) && SvIV (sv) == 1)
451 encode_str (enc, "true", 4, 0);
452 else
453 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
454 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
455 }
456 else
457 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
458 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416} 459}
417 460
418static void 461static void
419encode_sv (enc_t *enc, SV *sv) 462encode_sv (enc_t *enc, SV *sv)
420{ 463{
441 SvIsUV(sv) 484 SvIsUV(sv)
442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 485 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 486 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
444 } 487 }
445 else if (SvROK (sv)) 488 else if (SvROK (sv))
446 { 489 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)) 490 else if (!SvOK (sv))
463 encode_str (enc, "null", 4, 0); 491 encode_str (enc, "null", 4, 0);
464 else 492 else
465 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 493 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466 SvPV_nolen (sv), SvFLAGS (sv)); 494 SvPV_nolen (sv), SvFLAGS (sv));
467} 495}
468 496
469static SV * 497static SV *
470encode_json (SV *scalar, UV flags) 498encode_json (SV *scalar, U32 flags)
471{ 499{
500 enc_t enc;
501
472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 502 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 503 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474 504
475 enc_t enc;
476 enc.flags = flags; 505 enc.flags = flags;
477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 506 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478 enc.cur = SvPVX (enc.sv); 507 enc.cur = SvPVX (enc.sv);
479 enc.end = SvEND (enc.sv); 508 enc.end = SvEND (enc.sv);
480 enc.indent = 0; 509 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL; 510 enc.maxdepth = DEC_DEPTH (flags);
482 511
483 SvPOK_only (enc.sv); 512 SvPOK_only (enc.sv);
484 encode_sv (&enc, scalar); 513 encode_sv (&enc, scalar);
485 514
486 if (!(flags & (F_ASCII | F_UTF8))) 515 if (!(flags & (F_ASCII | F_UTF8)))
501typedef struct 530typedef struct
502{ 531{
503 char *cur; // current parser pointer 532 char *cur; // current parser pointer
504 char *end; // end of input string 533 char *end; // end of input string
505 const char *err; // parse error, if != 0 534 const char *err; // parse error, if != 0
506 UV flags; // F_* 535 U32 flags; // F_*
536 U32 depth; // recursion depth
537 U32 maxdepth; // recursion depth limit
507} dec_t; 538} dec_t;
508 539
509static void 540static void
510decode_ws (dec_t *dec) 541decode_ws (dec_t *dec)
511{ 542{
520 ++dec->cur; 551 ++dec->cur;
521 } 552 }
522} 553}
523 554
524#define ERR(reason) SB dec->err = reason; goto fail; SE 555#define ERR(reason) SB dec->err = reason; goto fail; SE
556
525#define EXPECT_CH(ch) SB \ 557#define EXPECT_CH(ch) SB \
526 if (*dec->cur != ch) \ 558 if (*dec->cur != ch) \
527 ERR (# ch " expected"); \ 559 ERR (# ch " expected"); \
528 ++dec->cur; \ 560 ++dec->cur; \
529 SE 561 SE
562
563#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
564#define DEC_DEC_DEPTH --dec->depth
530 565
531static SV *decode_sv (dec_t *dec); 566static SV *decode_sv (dec_t *dec);
532 567
533static signed char decode_hexdigit[256]; 568static signed char decode_hexdigit[256];
534 569
636 } 671 }
637 else if (ch >= 0x20 && ch <= 0x7f) 672 else if (ch >= 0x20 && ch <= 0x7f)
638 *cur++ = ch; 673 *cur++ = ch;
639 else if (ch >= 0x80) 674 else if (ch >= 0x80)
640 { 675 {
676 STRLEN clen;
677 UV uch;
678
641 --dec->cur; 679 --dec->cur;
642 680
643 STRLEN clen;
644 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 681 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
645 if (clen == (STRLEN)-1) 682 if (clen == (STRLEN)-1)
646 ERR ("malformed UTF-8 character in JSON string"); 683 ERR ("malformed UTF-8 character in JSON string");
647 684
648 do 685 do
649 {
650 *cur++ = *dec->cur++; 686 *cur++ = *dec->cur++;
651 }
652 while (--clen); 687 while (--clen);
653 688
654 utf8 = 1; 689 utf8 = 1;
655 } 690 }
656 else if (!ch)
657 ERR ("unexpected end of string while parsing json string");
658 else 691 else
692 {
693 --dec->cur;
694
695 if (!ch)
696 ERR ("unexpected end of string while parsing JSON string");
697 else
659 ERR ("invalid character encountered"); 698 ERR ("invalid character encountered while parsing JSON string");
660 699 }
661 } 700 }
662 while (cur < buf + SHORT_STRING_LEN); 701 while (cur < buf + SHORT_STRING_LEN);
663 702
703 {
664 STRLEN len = cur - buf; 704 STRLEN len = cur - buf;
665 705
666 if (sv) 706 if (sv)
667 { 707 {
668 SvGROW (sv, SvCUR (sv) + len + 1); 708 SvGROW (sv, SvCUR (sv) + len + 1);
669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 709 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670 SvCUR_set (sv, SvCUR (sv) + len); 710 SvCUR_set (sv, SvCUR (sv) + len);
671 } 711 }
672 else 712 else
673 sv = newSVpvn (buf, len); 713 sv = newSVpvn (buf, len);
714 }
674 } 715 }
675 while (*dec->cur != '"'); 716 while (*dec->cur != '"');
676 717
677 ++dec->cur; 718 ++dec->cur;
678 719
778static SV * 819static SV *
779decode_av (dec_t *dec) 820decode_av (dec_t *dec)
780{ 821{
781 AV *av = newAV (); 822 AV *av = newAV ();
782 823
824 DEC_INC_DEPTH;
783 decode_ws (dec); 825 decode_ws (dec);
826
784 if (*dec->cur == ']') 827 if (*dec->cur == ']')
785 ++dec->cur; 828 ++dec->cur;
786 else 829 else
787 for (;;) 830 for (;;)
788 { 831 {
806 ERR (", or ] expected while parsing array"); 849 ERR (", or ] expected while parsing array");
807 850
808 ++dec->cur; 851 ++dec->cur;
809 } 852 }
810 853
854 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)av); 855 return newRV_noinc ((SV *)av);
812 856
813fail: 857fail:
814 SvREFCNT_dec (av); 858 SvREFCNT_dec (av);
859 DEC_DEC_DEPTH;
815 return 0; 860 return 0;
816} 861}
817 862
818static SV * 863static SV *
819decode_hv (dec_t *dec) 864decode_hv (dec_t *dec)
820{ 865{
821 HV *hv = newHV (); 866 HV *hv = newHV ();
822 867
868 DEC_INC_DEPTH;
823 decode_ws (dec); 869 decode_ws (dec);
870
824 if (*dec->cur == '}') 871 if (*dec->cur == '}')
825 ++dec->cur; 872 ++dec->cur;
826 else 873 else
827 for (;;) 874 for (;;)
828 { 875 {
841 { 888 {
842 SvREFCNT_dec (key); 889 SvREFCNT_dec (key);
843 goto fail; 890 goto fail;
844 } 891 }
845 892
846 //TODO: optimise
847 hv_store_ent (hv, key, value, 0); 893 hv_store_ent (hv, key, value, 0);
894 SvREFCNT_dec (key);
848 895
849 decode_ws (dec); 896 decode_ws (dec);
850 897
851 if (*dec->cur == '}') 898 if (*dec->cur == '}')
852 { 899 {
858 ERR (", or } expected while parsing object/hash"); 905 ERR (", or } expected while parsing object/hash");
859 906
860 ++dec->cur; 907 ++dec->cur;
861 } 908 }
862 909
910 DEC_DEC_DEPTH;
863 return newRV_noinc ((SV *)hv); 911 return newRV_noinc ((SV *)hv);
864 912
865fail: 913fail:
866 SvREFCNT_dec (hv); 914 SvREFCNT_dec (hv);
915 DEC_DEC_DEPTH;
867 return 0; 916 return 0;
868} 917}
869 918
870static SV * 919static SV *
871decode_sv (dec_t *dec) 920decode_sv (dec_t *dec)
914 ERR ("'null' expected"); 963 ERR ("'null' expected");
915 964
916 break; 965 break;
917 966
918 default: 967 default:
919 ERR ("malformed json string, neither array, object, number, string or atom"); 968 ERR ("malformed JSON string, neither array, object, number, string or atom");
920 break; 969 break;
921 } 970 }
922 971
923fail: 972fail:
924 return 0; 973 return 0;
925} 974}
926 975
927static SV * 976static SV *
928decode_json (SV *string, UV flags) 977decode_json (SV *string, U32 flags)
929{ 978{
979 dec_t dec;
930 SV *sv; 980 SV *sv;
981
982 SvUPGRADE (string, SVt_PV);
931 983
932 if (flags & F_UTF8) 984 if (flags & F_UTF8)
933 sv_utf8_downgrade (string, 0); 985 sv_utf8_downgrade (string, 0);
934 else 986 else
935 sv_utf8_upgrade (string); 987 sv_utf8_upgrade (string);
936 988
937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 989 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938 990
939 dec_t dec;
940 dec.flags = flags; 991 dec.flags = flags;
941 dec.cur = SvPVX (string); 992 dec.cur = SvPVX (string);
942 dec.end = SvEND (string); 993 dec.end = SvEND (string);
943 dec.err = 0; 994 dec.err = 0;
995 dec.depth = 0;
996 dec.maxdepth = DEC_DEPTH (dec.flags);
944 997
998 *dec.end = 0; // this should basically be a nop, too, but make sure its there
945 sv = decode_sv (&dec); 999 sv = decode_sv (&dec);
946 1000
947 if (!sv) 1001 if (!sv)
948 { 1002 {
949 IV offset = dec.flags & F_UTF8 1003 IV offset = dec.flags & F_UTF8
958 SAVEVPTR (PL_curcop); 1012 SAVEVPTR (PL_curcop);
959 PL_curcop = &cop; 1013 PL_curcop = &cop;
960 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1014 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961 LEAVE; 1015 LEAVE;
962 1016
963 croak ("%s, at character offset %d (%s)", 1017 croak ("%s, at character offset %d [\"%s\"]",
964 dec.err, 1018 dec.err,
965 (int)offset, 1019 (int)offset,
966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1020 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967 } 1021 }
968 1022
982BOOT: 1036BOOT:
983{ 1037{
984 int i; 1038 int i;
985 1039
986 memset (decode_hexdigit, 0xff, 256); 1040 memset (decode_hexdigit, 0xff, 256);
1041
987 for (i = 10; i--; ) 1042 for (i = 0; i < 256; ++i)
988 decode_hexdigit ['0' + i] = i; 1043 decode_hexdigit [i] =
989 1044 i >= '0' && i <= '9' ? i - '0'
990 for (i = 7; i--; ) 1045 : i >= 'a' && i <= 'f' ? i - 'a' + 10
991 { 1046 : i >= 'A' && i <= 'F' ? i - 'A' + 10
992 decode_hexdigit ['a' + i] = 10 + i; 1047 : -1;
993 decode_hexdigit ['A' + i] = 10 + i;
994 }
995 1048
996 json_stash = gv_stashpv ("JSON::XS", 1); 1049 json_stash = gv_stashpv ("JSON::XS", 1);
997} 1050}
998 1051
999PROTOTYPES: DISABLE 1052PROTOTYPES: DISABLE
1026 RETVAL = newSVsv (self); 1079 RETVAL = newSVsv (self);
1027} 1080}
1028 OUTPUT: 1081 OUTPUT:
1029 RETVAL 1082 RETVAL
1030 1083
1084SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1085 CODE:
1086{
1087 UV *uv = SvJSON (self);
1088 UV log2 = 0;
1089
1090 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1091
1092 while ((1UL << log2) < max_depth)
1093 ++log2;
1094
1095 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1096
1097 RETVAL = newSVsv (self);
1098}
1099 OUTPUT:
1100 RETVAL
1101
1031void encode (SV *self, SV *scalar) 1102void encode (SV *self, SV *scalar)
1032 PPCODE: 1103 PPCODE:
1033 XPUSHs (encode_json (scalar, *SvJSON (self))); 1104 XPUSHs (encode_json (scalar, *SvJSON (self)));
1034 1105
1035void decode (SV *self, SV *jsonstr) 1106void decode (SV *self, SV *jsonstr)
1037 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1108 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
1038 1109
1039PROTOTYPES: ENABLE 1110PROTOTYPES: ENABLE
1040 1111
1041void to_json (SV *scalar) 1112void to_json (SV *scalar)
1113 ALIAS:
1114 objToJson = 0
1042 PPCODE: 1115 PPCODE:
1043 XPUSHs (encode_json (scalar, F_UTF8)); 1116 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1044 1117
1045void from_json (SV *jsonstr) 1118void from_json (SV *jsonstr)
1119 ALIAS:
1120 jsonToObj = 0
1046 PPCODE: 1121 PPCODE:
1047 XPUSHs (decode_json (jsonstr, F_UTF8)); 1122 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
1048 1123

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines