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 1.25 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
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 {
82 *clen = (STRLEN)-1;
83 return (UV)-1;
84 }
85}
86
58///////////////////////////////////////////////////////////////////////////// 87/////////////////////////////////////////////////////////////////////////////
59// encoder 88// encoder
60 89
61// structure used for encoding JSON 90// structure used for encoding JSON
62typedef struct 91typedef struct
63{ 92{
64 char *cur; // SvPVX (sv) + current output position 93 char *cur; // SvPVX (sv) + current output position
65 char *end; // SvEND (sv) 94 char *end; // SvEND (sv)
66 SV *sv; // result scalar 95 SV *sv; // result scalar
67 UV flags; // F_* 96 U32 flags; // F_*
68 int indent; // indentation level 97 U32 indent; // indentation level
69 int max_depth; // max. recursion level 98 U32 maxdepth; // max. indentation/recursion level
70} enc_t; 99} enc_t;
71 100
72static void 101static void
73need (enc_t *enc, STRLEN len) 102need (enc_t *enc, STRLEN len)
74{ 103{
… …
133 STRLEN clen; 162 STRLEN clen;
134 UV uch; 163 UV uch;
135 164
136 if (is_utf8) 165 if (is_utf8)
137 { 166 {
138 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 167 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
168 uch = decode_utf8 (str, end - str, &clen);
139 if (clen == (STRLEN)-1) 169 if (clen == (STRLEN)-1)
140 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 170 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
141 } 171 }
142 else 172 else
143 { 173 {
… …
241static void 271static void
242encode_av (enc_t *enc, AV *av) 272encode_av (enc_t *enc, AV *av)
243{ 273{
244 int i, len = av_len (av); 274 int i, len = av_len (av);
245 275
276 if (enc->indent >= enc->maxdepth)
277 croak ("data structure too deep (hit recursion limit)");
278
246 encode_ch (enc, '['); encode_nl (enc); 279 encode_ch (enc, '['); encode_nl (enc);
247 ++enc->indent; 280 ++enc->indent;
248 281
249 for (i = 0; i <= len; ++i) 282 for (i = 0; i <= len; ++i)
250 { 283 {
… …
315 348
316static void 349static void
317encode_hv (enc_t *enc, HV *hv) 350encode_hv (enc_t *enc, HV *hv)
318{ 351{
319 int count, i; 352 int count, i;
353
354 if (enc->indent >= enc->maxdepth)
355 croak ("data structure too deep (hit recursion limit)");
320 356
321 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 357 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
322 358
323 if ((count = hv_iterinit (hv))) 359 if ((count = hv_iterinit (hv)))
324 { 360 {
… …
372 408
373 encode_nl (enc); 409 encode_nl (enc);
374 } 410 }
375 else 411 else
376 { 412 {
377 SV *sv;
378 HE *he = hv_iternext (hv); 413 HE *he = hv_iternext (hv);
379 414
380 for (;;) 415 for (;;)
381 { 416 {
382 encode_indent (enc); 417 encode_indent (enc);
… …
391 encode_nl (enc); 426 encode_nl (enc);
392 } 427 }
393 } 428 }
394 429
395 --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))));
396} 459}
397 460
398static void 461static void
399encode_sv (enc_t *enc, SV *sv) 462encode_sv (enc_t *enc, SV *sv)
400{ 463{
… …
421 SvIsUV(sv) 484 SvIsUV(sv)
422 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 485 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
423 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 486 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
424 } 487 }
425 else if (SvROK (sv)) 488 else if (SvROK (sv))
426 { 489 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)) 490 else if (!SvOK (sv))
443 encode_str (enc, "null", 4, 0); 491 encode_str (enc, "null", 4, 0);
444 else 492 else
445 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",
446 SvPV_nolen (sv), SvFLAGS (sv)); 494 SvPV_nolen (sv), SvFLAGS (sv));
447} 495}
448 496
449static SV * 497static SV *
450encode_json (SV *scalar, UV flags) 498encode_json (SV *scalar, U32 flags)
451{ 499{
500 enc_t enc;
501
452 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 502 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
453 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)");
454 504
455 enc_t enc;
456 enc.flags = flags; 505 enc.flags = flags;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 506 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 507 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 508 enc.end = SvEND (enc.sv);
460 enc.indent = 0; 509 enc.indent = 0;
461 enc.max_depth = 0x7fffffffUL; 510 enc.maxdepth = DEC_DEPTH (flags);
462 511
463 SvPOK_only (enc.sv); 512 SvPOK_only (enc.sv);
464 encode_sv (&enc, scalar); 513 encode_sv (&enc, scalar);
465 514
466 if (!(flags & (F_ASCII | F_UTF8))) 515 if (!(flags & (F_ASCII | F_UTF8)))
… …
481typedef struct 530typedef struct
482{ 531{
483 char *cur; // current parser pointer 532 char *cur; // current parser pointer
484 char *end; // end of input string 533 char *end; // end of input string
485 const char *err; // parse error, if != 0 534 const char *err; // parse error, if != 0
486 UV flags; // F_* 535 U32 flags; // F_*
536 U32 depth; // recursion depth
537 U32 maxdepth; // recursion depth limit
487} dec_t; 538} dec_t;
488 539
489static void 540static void
490decode_ws (dec_t *dec) 541decode_ws (dec_t *dec)
491{ 542{
… …
500 ++dec->cur; 551 ++dec->cur;
501 } 552 }
502} 553}
503 554
504#define ERR(reason) SB dec->err = reason; goto fail; SE 555#define ERR(reason) SB dec->err = reason; goto fail; SE
556
505#define EXPECT_CH(ch) SB \ 557#define EXPECT_CH(ch) SB \
506 if (*dec->cur != ch) \ 558 if (*dec->cur != ch) \
507 ERR (# ch " expected"); \ 559 ERR (# ch " expected"); \
508 ++dec->cur; \ 560 ++dec->cur; \
509 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
510 565
511static SV *decode_sv (dec_t *dec); 566static SV *decode_sv (dec_t *dec);
512 567
513static signed char decode_hexdigit[256]; 568static signed char decode_hexdigit[256];
514 569
… …
616 } 671 }
617 else if (ch >= 0x20 && ch <= 0x7f) 672 else if (ch >= 0x20 && ch <= 0x7f)
618 *cur++ = ch; 673 *cur++ = ch;
619 else if (ch >= 0x80) 674 else if (ch >= 0x80)
620 { 675 {
676 STRLEN clen;
677 UV uch;
678
621 --dec->cur; 679 --dec->cur;
622 680
623 STRLEN clen;
624 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 681 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
625 if (clen == (STRLEN)-1) 682 if (clen == (STRLEN)-1)
626 ERR ("malformed UTF-8 character in JSON string"); 683 ERR ("malformed UTF-8 character in JSON string");
627 684
628 do 685 do
629 {
630 *cur++ = *dec->cur++; 686 *cur++ = *dec->cur++;
631 }
632 while (--clen); 687 while (--clen);
633 688
634 utf8 = 1; 689 utf8 = 1;
635 } 690 }
636 else if (!ch)
637 ERR ("unexpected end of string while parsing json string");
638 else 691 else
692 {
693 --dec->cur;
694
695 if (!ch)
696 ERR ("unexpected end of string while parsing JSON string");
697 else
639 ERR ("invalid character encountered"); 698 ERR ("invalid character encountered while parsing JSON string");
640 699 }
641 } 700 }
642 while (cur < buf + SHORT_STRING_LEN); 701 while (cur < buf + SHORT_STRING_LEN);
643 702
703 {
644 STRLEN len = cur - buf; 704 STRLEN len = cur - buf;
645 705
646 if (sv) 706 if (sv)
647 { 707 {
648 SvGROW (sv, SvCUR (sv) + len + 1); 708 SvGROW (sv, SvCUR (sv) + len + 1);
649 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 709 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
650 SvCUR_set (sv, SvCUR (sv) + len); 710 SvCUR_set (sv, SvCUR (sv) + len);
651 } 711 }
652 else 712 else
653 sv = newSVpvn (buf, len); 713 sv = newSVpvn (buf, len);
714 }
654 } 715 }
655 while (*dec->cur != '"'); 716 while (*dec->cur != '"');
656 717
657 ++dec->cur; 718 ++dec->cur;
658 719
… …
758static SV * 819static SV *
759decode_av (dec_t *dec) 820decode_av (dec_t *dec)
760{ 821{
761 AV *av = newAV (); 822 AV *av = newAV ();
762 823
824 DEC_INC_DEPTH;
763 decode_ws (dec); 825 decode_ws (dec);
826
764 if (*dec->cur == ']') 827 if (*dec->cur == ']')
765 ++dec->cur; 828 ++dec->cur;
766 else 829 else
767 for (;;) 830 for (;;)
768 { 831 {
… …
786 ERR (", or ] expected while parsing array"); 849 ERR (", or ] expected while parsing array");
787 850
788 ++dec->cur; 851 ++dec->cur;
789 } 852 }
790 853
854 DEC_DEC_DEPTH;
791 return newRV_noinc ((SV *)av); 855 return newRV_noinc ((SV *)av);
792 856
793fail: 857fail:
794 SvREFCNT_dec (av); 858 SvREFCNT_dec (av);
859 DEC_DEC_DEPTH;
795 return 0; 860 return 0;
796} 861}
797 862
798static SV * 863static SV *
799decode_hv (dec_t *dec) 864decode_hv (dec_t *dec)
800{ 865{
801 HV *hv = newHV (); 866 HV *hv = newHV ();
802 867
868 DEC_INC_DEPTH;
803 decode_ws (dec); 869 decode_ws (dec);
870
804 if (*dec->cur == '}') 871 if (*dec->cur == '}')
805 ++dec->cur; 872 ++dec->cur;
806 else 873 else
807 for (;;) 874 for (;;)
808 { 875 {
… …
821 { 888 {
822 SvREFCNT_dec (key); 889 SvREFCNT_dec (key);
823 goto fail; 890 goto fail;
824 } 891 }
825 892
826 //TODO: optimise
827 hv_store_ent (hv, key, value, 0); 893 hv_store_ent (hv, key, value, 0);
894 SvREFCNT_dec (key);
828 895
829 decode_ws (dec); 896 decode_ws (dec);
830 897
831 if (*dec->cur == '}') 898 if (*dec->cur == '}')
832 { 899 {
… …
838 ERR (", or } expected while parsing object/hash"); 905 ERR (", or } expected while parsing object/hash");
839 906
840 ++dec->cur; 907 ++dec->cur;
841 } 908 }
842 909
910 DEC_DEC_DEPTH;
843 return newRV_noinc ((SV *)hv); 911 return newRV_noinc ((SV *)hv);
844 912
845fail: 913fail:
846 SvREFCNT_dec (hv); 914 SvREFCNT_dec (hv);
915 DEC_DEC_DEPTH;
847 return 0; 916 return 0;
848} 917}
849 918
850static SV * 919static SV *
851decode_sv (dec_t *dec) 920decode_sv (dec_t *dec)
… …
894 ERR ("'null' expected"); 963 ERR ("'null' expected");
895 964
896 break; 965 break;
897 966
898 default: 967 default:
899 ERR ("malformed json string, neither array, object, number, string or atom"); 968 ERR ("malformed JSON string, neither array, object, number, string or atom");
900 break; 969 break;
901 } 970 }
902 971
903fail: 972fail:
904 return 0; 973 return 0;
905} 974}
906 975
907static SV * 976static SV *
908decode_json (SV *string, UV flags) 977decode_json (SV *string, U32 flags)
909{ 978{
979 dec_t dec;
910 SV *sv; 980 SV *sv;
981
982 SvUPGRADE (string, SVt_PV);
911 983
912 if (flags & F_UTF8) 984 if (flags & F_UTF8)
913 sv_utf8_downgrade (string, 0); 985 sv_utf8_downgrade (string, 0);
914 else 986 else
915 sv_utf8_upgrade (string); 987 sv_utf8_upgrade (string);
916 988
917 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 989 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
918 990
919 dec_t dec;
920 dec.flags = flags; 991 dec.flags = flags;
921 dec.cur = SvPVX (string); 992 dec.cur = SvPVX (string);
922 dec.end = SvEND (string); 993 dec.end = SvEND (string);
923 dec.err = 0; 994 dec.err = 0;
995 dec.depth = 0;
996 dec.maxdepth = DEC_DEPTH (dec.flags);
924 997
998 *dec.end = 0; // this should basically be a nop, too, but make sure its there
925 sv = decode_sv (&dec); 999 sv = decode_sv (&dec);
926 1000
927 if (!sv) 1001 if (!sv)
928 { 1002 {
929 IV offset = dec.flags & F_UTF8 1003 IV offset = dec.flags & F_UTF8
… …
938 SAVEVPTR (PL_curcop); 1012 SAVEVPTR (PL_curcop);
939 PL_curcop = &cop; 1013 PL_curcop = &cop;
940 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);
941 LEAVE; 1015 LEAVE;
942 1016
943 croak ("%s, at character offset %d (%s)", 1017 croak ("%s, at character offset %d [\"%s\"]",
944 dec.err, 1018 dec.err,
945 (int)offset, 1019 (int)offset,
946 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1020 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
947 } 1021 }
948 1022
… …
962BOOT: 1036BOOT:
963{ 1037{
964 int i; 1038 int i;
965 1039
966 memset (decode_hexdigit, 0xff, 256); 1040 memset (decode_hexdigit, 0xff, 256);
1041
967 for (i = 10; i--; ) 1042 for (i = 0; i < 256; ++i)
968 decode_hexdigit ['0' + i] = i; 1043 decode_hexdigit [i] =
969 1044 i >= '0' && i <= '9' ? i - '0'
970 for (i = 7; i--; ) 1045 : i >= 'a' && i <= 'f' ? i - 'a' + 10
971 { 1046 : i >= 'A' && i <= 'F' ? i - 'A' + 10
972 decode_hexdigit ['a' + i] = 10 + i; 1047 : -1;
973 decode_hexdigit ['A' + i] = 10 + i;
974 }
975 1048
976 json_stash = gv_stashpv ("JSON::XS", 1); 1049 json_stash = gv_stashpv ("JSON::XS", 1);
977} 1050}
978 1051
979PROTOTYPES: DISABLE 1052PROTOTYPES: DISABLE
… …
1006 RETVAL = newSVsv (self); 1079 RETVAL = newSVsv (self);
1007} 1080}
1008 OUTPUT: 1081 OUTPUT:
1009 RETVAL 1082 RETVAL
1010 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
1011void encode (SV *self, SV *scalar) 1102void encode (SV *self, SV *scalar)
1012 PPCODE: 1103 PPCODE:
1013 XPUSHs (encode_json (scalar, *SvJSON (self))); 1104 XPUSHs (encode_json (scalar, *SvJSON (self)));
1014 1105
1015void decode (SV *self, SV *jsonstr) 1106void decode (SV *self, SV *jsonstr)
… …
1017 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1108 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
1018 1109
1019PROTOTYPES: ENABLE 1110PROTOTYPES: ENABLE
1020 1111
1021void to_json (SV *scalar) 1112void to_json (SV *scalar)
1113 ALIAS:
1114 objToJson = 0
1022 PPCODE: 1115 PPCODE:
1023 XPUSHs (encode_json (scalar, F_UTF8)); 1116 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1024 1117
1025void from_json (SV *jsonstr) 1118void from_json (SV *jsonstr)
1119 ALIAS:
1120 jsonToObj = 0
1026 PPCODE: 1121 PPCODE:
1027 XPUSHs (decode_json (jsonstr, F_UTF8)); 1122 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
1028 1123

Diff Legend

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