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 1.12 by root, Sat Mar 24 22:10:08 2007 UTC vs.
Revision 1.21 by root, Thu Mar 29 02:45:49 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 (12UL << 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
25#define SHORT_STRING_LEN 256 // special-case strings of up to this size 32#define SHORT_STRING_LEN 512 // special-case strings of up to this size
26 33
27#define SB do { 34#define SB do {
28#define SE } while (0) 35#define SE } while (0)
29 36
30static HV *json_stash; // JSON::XS:: 37static HV *json_stash; // JSON::XS::
53 SvPV_renew (sv, SvCUR (sv) + 1); 60 SvPV_renew (sv, SvCUR (sv) + 1);
54#endif 61#endif
55 } 62 }
56} 63}
57 64
65// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints
70static UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{
73 if (s[0] > 0xdf || s[0] < 0xc2)
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 {
77 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 }
80 else
81 return (UV)-1;
82}
83
58///////////////////////////////////////////////////////////////////////////// 84/////////////////////////////////////////////////////////////////////////////
59// encoder 85// encoder
60 86
61// structure used for encoding JSON 87// structure used for encoding JSON
62typedef struct 88typedef struct
63{ 89{
64 char *cur; // SvPVX (sv) + current output position 90 char *cur; // SvPVX (sv) + current output position
65 char *end; // SvEND (sv) 91 char *end; // SvEND (sv)
66 SV *sv; // result scalar 92 SV *sv; // result scalar
67 UV flags; // F_* 93 U32 flags; // F_*
68 int indent; // indentation level 94 U32 indent; // indentation level
69 int max_depth; // max. recursion level 95 U32 maxdepth; // max. indentation/recursion level
70} enc_t; 96} enc_t;
71 97
72static void 98static void
73need (enc_t *enc, STRLEN len) 99need (enc_t *enc, STRLEN len)
74{ 100{
133 STRLEN clen; 159 STRLEN clen;
134 UV uch; 160 UV uch;
135 161
136 if (is_utf8) 162 if (is_utf8)
137 { 163 {
138 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen);
139 if (clen == (STRLEN)-1) 166 if (clen == (STRLEN)-1)
140 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 167 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
141 } 168 }
142 else 169 else
143 { 170 {
241static void 268static void
242encode_av (enc_t *enc, AV *av) 269encode_av (enc_t *enc, AV *av)
243{ 270{
244 int i, len = av_len (av); 271 int i, len = av_len (av);
245 272
273 if (enc->indent >= enc->maxdepth)
274 croak ("data structure too deep (hit recursion limit)");
275
246 encode_ch (enc, '['); encode_nl (enc); 276 encode_ch (enc, '['); encode_nl (enc);
247 ++enc->indent; 277 ++enc->indent;
248 278
249 for (i = 0; i <= len; ++i) 279 for (i = 0; i <= len; ++i)
250 { 280 {
315 345
316static void 346static void
317encode_hv (enc_t *enc, HV *hv) 347encode_hv (enc_t *enc, HV *hv)
318{ 348{
319 int count, i; 349 int count, i;
350
351 if (enc->indent >= enc->maxdepth)
352 croak ("data structure too deep (hit recursion limit)");
320 353
321 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 354 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
322 355
323 if ((count = hv_iterinit (hv))) 356 if ((count = hv_iterinit (hv)))
324 { 357 {
393 } 426 }
394 427
395 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 428 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
396} 429}
397 430
431// encode objects, arrays and special \0=false and \1=true values.
432static void
433encode_rv (enc_t *enc, SV *sv)
434{
435 SvGETMAGIC (sv);
436
437 svtype svt = SvTYPE (sv);
438
439 if (svt == SVt_PVHV)
440 encode_hv (enc, (HV *)sv);
441 else if (svt == SVt_PVAV)
442 encode_av (enc, (AV *)sv);
443 else if (svt < SVt_PVAV)
444 {
445 if (SvNIOK (sv) && SvIV (sv) == 0)
446 encode_str (enc, "false", 5, 0);
447 else if (SvNIOK (sv) && SvIV (sv) == 1)
448 encode_str (enc, "true", 4, 0);
449 else
450 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
451 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
452 }
453 else
454 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
455 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
456}
457
398static void 458static void
399encode_sv (enc_t *enc, SV *sv) 459encode_sv (enc_t *enc, SV *sv)
400{ 460{
401 SvGETMAGIC (sv); 461 SvGETMAGIC (sv);
402 462
421 SvIsUV(sv) 481 SvIsUV(sv)
422 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 482 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
423 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 483 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
424 } 484 }
425 else if (SvROK (sv)) 485 else if (SvROK (sv))
426 { 486 encode_rv (enc, SvRV (sv));
427 SV *rv = SvRV (sv);
428
429 if (enc->indent >= enc->max_depth)
430 croak ("data structure too deep (hit recursion limit)");
431
432 switch (SvTYPE (rv))
433 {
434 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
435 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
436
437 default:
438 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
439 SvPV_nolen (sv));
440 }
441 }
442 else if (!SvOK (sv)) 487 else if (!SvOK (sv))
443 encode_str (enc, "null", 4, 0); 488 encode_str (enc, "null", 4, 0);
444 else 489 else
445 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 490 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
446 SvPV_nolen (sv), SvFLAGS (sv)); 491 SvPV_nolen (sv), SvFLAGS (sv));
447} 492}
448 493
449static SV * 494static SV *
450encode_json (SV *scalar, UV flags) 495encode_json (SV *scalar, U32 flags)
451{ 496{
452 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 497 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
453 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 498 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
454 499
455 enc_t enc; 500 enc_t enc;
456 enc.flags = flags; 501 enc.flags = flags;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 502 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 503 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 504 enc.end = SvEND (enc.sv);
460 enc.indent = 0; 505 enc.indent = 0;
461 enc.max_depth = 0x7fffffffUL; 506 enc.maxdepth = DEC_DEPTH (flags);
462 507
463 SvPOK_only (enc.sv); 508 SvPOK_only (enc.sv);
464 encode_sv (&enc, scalar); 509 encode_sv (&enc, scalar);
465 510
466 if (!(flags & (F_ASCII | F_UTF8))) 511 if (!(flags & (F_ASCII | F_UTF8)))
481typedef struct 526typedef struct
482{ 527{
483 char *cur; // current parser pointer 528 char *cur; // current parser pointer
484 char *end; // end of input string 529 char *end; // end of input string
485 const char *err; // parse error, if != 0 530 const char *err; // parse error, if != 0
486 UV flags; // F_* 531 U32 flags; // F_*
532 U32 depth; // recursion depth
533 U32 maxdepth; // recursion depth limit
487} dec_t; 534} dec_t;
488 535
489static void 536static void
490decode_ws (dec_t *dec) 537decode_ws (dec_t *dec)
491{ 538{
500 ++dec->cur; 547 ++dec->cur;
501 } 548 }
502} 549}
503 550
504#define ERR(reason) SB dec->err = reason; goto fail; SE 551#define ERR(reason) SB dec->err = reason; goto fail; SE
552
505#define EXPECT_CH(ch) SB \ 553#define EXPECT_CH(ch) SB \
506 if (*dec->cur != ch) \ 554 if (*dec->cur != ch) \
507 ERR (# ch " expected"); \ 555 ERR (# ch " expected"); \
508 ++dec->cur; \ 556 ++dec->cur; \
509 SE 557 SE
558
559#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
560#define DEC_DEC_DEPTH --dec->depth
510 561
511static SV *decode_sv (dec_t *dec); 562static SV *decode_sv (dec_t *dec);
512 563
513static signed char decode_hexdigit[256]; 564static signed char decode_hexdigit[256];
514 565
619 else if (ch >= 0x80) 670 else if (ch >= 0x80)
620 { 671 {
621 --dec->cur; 672 --dec->cur;
622 673
623 STRLEN clen; 674 STRLEN clen;
624 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 675 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
625 if (clen == (STRLEN)-1) 676 if (clen == (STRLEN)-1)
626 ERR ("malformed UTF-8 character in JSON string"); 677 ERR ("malformed UTF-8 character in JSON string");
627 678
628 do 679 do
629 { 680 {
758static SV * 809static SV *
759decode_av (dec_t *dec) 810decode_av (dec_t *dec)
760{ 811{
761 AV *av = newAV (); 812 AV *av = newAV ();
762 813
814 DEC_INC_DEPTH;
763 decode_ws (dec); 815 decode_ws (dec);
816
764 if (*dec->cur == ']') 817 if (*dec->cur == ']')
765 ++dec->cur; 818 ++dec->cur;
766 else 819 else
767 for (;;) 820 for (;;)
768 { 821 {
786 ERR (", or ] expected while parsing array"); 839 ERR (", or ] expected while parsing array");
787 840
788 ++dec->cur; 841 ++dec->cur;
789 } 842 }
790 843
844 DEC_DEC_DEPTH;
791 return newRV_noinc ((SV *)av); 845 return newRV_noinc ((SV *)av);
792 846
793fail: 847fail:
794 SvREFCNT_dec (av); 848 SvREFCNT_dec (av);
849 DEC_DEC_DEPTH;
795 return 0; 850 return 0;
796} 851}
797 852
798static SV * 853static SV *
799decode_hv (dec_t *dec) 854decode_hv (dec_t *dec)
800{ 855{
801 HV *hv = newHV (); 856 HV *hv = newHV ();
802 857
858 DEC_INC_DEPTH;
803 decode_ws (dec); 859 decode_ws (dec);
860
804 if (*dec->cur == '}') 861 if (*dec->cur == '}')
805 ++dec->cur; 862 ++dec->cur;
806 else 863 else
807 for (;;) 864 for (;;)
808 { 865 {
821 { 878 {
822 SvREFCNT_dec (key); 879 SvREFCNT_dec (key);
823 goto fail; 880 goto fail;
824 } 881 }
825 882
826 //TODO: optimise
827 hv_store_ent (hv, key, value, 0); 883 hv_store_ent (hv, key, value, 0);
884 SvREFCNT_dec (key);
828 885
829 decode_ws (dec); 886 decode_ws (dec);
830 887
831 if (*dec->cur == '}') 888 if (*dec->cur == '}')
832 { 889 {
838 ERR (", or } expected while parsing object/hash"); 895 ERR (", or } expected while parsing object/hash");
839 896
840 ++dec->cur; 897 ++dec->cur;
841 } 898 }
842 899
900 DEC_DEC_DEPTH;
843 return newRV_noinc ((SV *)hv); 901 return newRV_noinc ((SV *)hv);
844 902
845fail: 903fail:
846 SvREFCNT_dec (hv); 904 SvREFCNT_dec (hv);
905 DEC_DEC_DEPTH;
847 return 0; 906 return 0;
848} 907}
849 908
850static SV * 909static SV *
851decode_sv (dec_t *dec) 910decode_sv (dec_t *dec)
903fail: 962fail:
904 return 0; 963 return 0;
905} 964}
906 965
907static SV * 966static SV *
908decode_json (SV *string, UV flags) 967decode_json (SV *string, U32 flags)
909{ 968{
910 SV *sv; 969 SV *sv;
911 970
912 if (flags & F_UTF8) 971 if (flags & F_UTF8)
913 sv_utf8_downgrade (string, 0); 972 sv_utf8_downgrade (string, 0);
915 sv_utf8_upgrade (string); 974 sv_utf8_upgrade (string);
916 975
917 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 976 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
918 977
919 dec_t dec; 978 dec_t dec;
920 dec.flags = flags; 979 dec.flags = flags;
921 dec.cur = SvPVX (string); 980 dec.cur = SvPVX (string);
922 dec.end = SvEND (string); 981 dec.end = SvEND (string);
923 dec.err = 0; 982 dec.err = 0;
983 dec.depth = 0;
984 dec.maxdepth = DEC_DEPTH (dec.flags);
924 985
986 *dec.end = 0; // this should basically be a nop, too, but make sure its there
925 sv = decode_sv (&dec); 987 sv = decode_sv (&dec);
926 988
927 if (!sv) 989 if (!sv)
928 { 990 {
929 IV offset = dec.flags & F_UTF8 991 IV offset = dec.flags & F_UTF8
962BOOT: 1024BOOT:
963{ 1025{
964 int i; 1026 int i;
965 1027
966 memset (decode_hexdigit, 0xff, 256); 1028 memset (decode_hexdigit, 0xff, 256);
1029
967 for (i = 10; i--; ) 1030 for (i = 0; i < 256; ++i)
968 decode_hexdigit ['0' + i] = i; 1031 decode_hexdigit [i] =
969 1032 i >= '0' && i <= '9' ? i - '0'
970 for (i = 7; i--; ) 1033 : i >= 'a' && i <= 'f' ? i - 'a' + 10
971 { 1034 : i >= 'A' && i <= 'F' ? i - 'A' + 10
972 decode_hexdigit ['a' + i] = 10 + i; 1035 : -1;
973 decode_hexdigit ['A' + i] = 10 + i;
974 }
975 1036
976 json_stash = gv_stashpv ("JSON::XS", 1); 1037 json_stash = gv_stashpv ("JSON::XS", 1);
977} 1038}
978 1039
979PROTOTYPES: DISABLE 1040PROTOTYPES: DISABLE
1006 RETVAL = newSVsv (self); 1067 RETVAL = newSVsv (self);
1007} 1068}
1008 OUTPUT: 1069 OUTPUT:
1009 RETVAL 1070 RETVAL
1010 1071
1072SV *max_depth (SV *self, int max_depth = 0x80000000UL)
1073 CODE:
1074{
1075 UV *uv = SvJSON (self);
1076 UV log2 = 0;
1077
1078 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1079
1080 while ((1UL << log2) < max_depth)
1081 ++log2;
1082
1083 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1084
1085 RETVAL = newSVsv (self);
1086}
1087 OUTPUT:
1088 RETVAL
1089
1011void encode (SV *self, SV *scalar) 1090void encode (SV *self, SV *scalar)
1012 PPCODE: 1091 PPCODE:
1013 XPUSHs (encode_json (scalar, *SvJSON (self))); 1092 XPUSHs (encode_json (scalar, *SvJSON (self)));
1014 1093
1015void decode (SV *self, SV *jsonstr) 1094void decode (SV *self, SV *jsonstr)
1017 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1096 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
1018 1097
1019PROTOTYPES: ENABLE 1098PROTOTYPES: ENABLE
1020 1099
1021void to_json (SV *scalar) 1100void to_json (SV *scalar)
1101 ALIAS:
1102 objToJson = 0
1022 PPCODE: 1103 PPCODE:
1023 XPUSHs (encode_json (scalar, F_UTF8)); 1104 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1024 1105
1025void from_json (SV *jsonstr) 1106void from_json (SV *jsonstr)
1107 ALIAS:
1108 jsonToObj = 0
1026 PPCODE: 1109 PPCODE:
1027 XPUSHs (decode_json (jsonstr, F_UTF8)); 1110 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
1028 1111

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines