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 1.24 by root, Tue Apr 3 23:59:04 2007 UTC vs.
Revision 1.31 by root, Wed May 9 16:33:53 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"
9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
8 13
9#define F_ASCII 0x00000001UL 14#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002UL 16#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004UL 17#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008UL 18#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010UL 19#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020UL 20#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080UL 21#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100UL 22#define F_SHRINK 0x00000200UL
17#define F_MAXDEPTH 0xf8000000UL 23#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27 24#define S_MAXDEPTH 27
19 25
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21 27
26#define F_DEFAULT (9UL << S_MAXDEPTH) 32#define F_DEFAULT (9UL << S_MAXDEPTH)
27 33
28#define INIT_SIZE 32 // initial scalar size to be allocated 34#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level 35#define INDENT_STEP 3 // spaces per indentation level
30 36
31#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32#define SHORT_STRING_LEN 512 // special-case strings of up to this size 37#define SHORT_STRING_LEN 512 // special-case strings of up to this size
33 38
34#define SB do { 39#define SB do {
35#define SE } while (0) 40#define SE } while (0)
36 41
104 if (enc->cur + len >= enc->end) 109 if (enc->cur + len >= enc->end)
105 { 110 {
106 STRLEN cur = enc->cur - SvPVX (enc->sv); 111 STRLEN cur = enc->cur - SvPVX (enc->sv);
107 SvGROW (enc->sv, cur + len + 1); 112 SvGROW (enc->sv, cur + len + 1);
108 enc->cur = SvPVX (enc->sv) + cur; 113 enc->cur = SvPVX (enc->sv) + cur;
109 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 114 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
110 } 115 }
111} 116}
112 117
113static void 118static void
114encode_ch (enc_t *enc, char ch) 119encode_ch (enc_t *enc, char ch)
176 } 181 }
177 182
178 if (uch > 0x10FFFFUL) 183 if (uch > 0x10FFFFUL)
179 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 184 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
180 185
181 if (uch < 0x80 || enc->flags & F_ASCII) 186 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
182 { 187 {
183 if (uch > 0xFFFFUL) 188 if (uch > 0xFFFFUL)
184 { 189 {
185 need (enc, len += 11); 190 need (enc, len += 11);
186 sprintf (enc->cur, "\\u%04x\\u%04x", 191 sprintf (enc->cur, "\\u%04x\\u%04x",
200 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 205 *enc->cur++ = hexdigit [(uch >> 0) & 15];
201 } 206 }
202 207
203 str += clen; 208 str += clen;
204 } 209 }
210 else if (enc->flags & F_LATIN1)
211 {
212 *enc->cur++ = uch;
213 str += clen;
214 }
205 else if (is_utf8) 215 else if (is_utf8)
206 { 216 {
207 need (enc, len += clen); 217 need (enc, len += clen);
208 do 218 do
209 { 219 {
211 } 221 }
212 while (--clen); 222 while (--clen);
213 } 223 }
214 else 224 else
215 { 225 {
216 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 226 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
217 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 227 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
218 ++str; 228 ++str;
219 } 229 }
220 } 230 }
221 } 231 }
408 418
409 encode_nl (enc); 419 encode_nl (enc);
410 } 420 }
411 else 421 else
412 { 422 {
413 SV *sv;
414 HE *he = hv_iternext (hv); 423 HE *he = hv_iternext (hv);
415 424
416 for (;;) 425 for (;;)
417 { 426 {
418 encode_indent (enc); 427 encode_indent (enc);
433 442
434// encode objects, arrays and special \0=false and \1=true values. 443// encode objects, arrays and special \0=false and \1=true values.
435static void 444static void
436encode_rv (enc_t *enc, SV *sv) 445encode_rv (enc_t *enc, SV *sv)
437{ 446{
447 svtype svt;
448
438 SvGETMAGIC (sv); 449 SvGETMAGIC (sv);
439
440 svtype svt = SvTYPE (sv); 450 svt = SvTYPE (sv);
441 451
442 if (svt == SVt_PVHV) 452 if (svt == SVt_PVHV)
443 encode_hv (enc, (HV *)sv); 453 encode_hv (enc, (HV *)sv);
444 else if (svt == SVt_PVAV) 454 else if (svt == SVt_PVAV)
445 encode_av (enc, (AV *)sv); 455 encode_av (enc, (AV *)sv);
495} 505}
496 506
497static SV * 507static SV *
498encode_json (SV *scalar, U32 flags) 508encode_json (SV *scalar, U32 flags)
499{ 509{
510 enc_t enc;
511
500 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 512 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
501 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 513 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
502 514
503 enc_t enc;
504 enc.flags = flags; 515 enc.flags = flags;
505 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 516 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
506 enc.cur = SvPVX (enc.sv); 517 enc.cur = SvPVX (enc.sv);
507 enc.end = SvEND (enc.sv); 518 enc.end = SvEND (enc.sv);
508 enc.indent = 0; 519 enc.indent = 0;
509 enc.maxdepth = DEC_DEPTH (flags); 520 enc.maxdepth = DEC_DEPTH (flags);
510 521
511 SvPOK_only (enc.sv); 522 SvPOK_only (enc.sv);
512 encode_sv (&enc, scalar); 523 encode_sv (&enc, scalar);
513 524
525 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
526 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
527
514 if (!(flags & (F_ASCII | F_UTF8))) 528 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
515 SvUTF8_on (enc.sv); 529 SvUTF8_on (enc.sv);
516
517 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
518 530
519 if (enc.flags & F_SHRINK) 531 if (enc.flags & F_SHRINK)
520 shrink (enc.sv); 532 shrink (enc.sv);
521 533
522 return enc.sv; 534 return enc.sv;
594 SV *sv = 0; 606 SV *sv = 0;
595 int utf8 = 0; 607 int utf8 = 0;
596 608
597 do 609 do
598 { 610 {
599 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 611 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
600 char *cur = buf; 612 char *cur = buf;
601 613
602 do 614 do
603 { 615 {
604 unsigned char ch = *(unsigned char *)dec->cur++; 616 unsigned char ch = *(unsigned char *)dec->cur++;
670 } 682 }
671 else if (ch >= 0x20 && ch <= 0x7f) 683 else if (ch >= 0x20 && ch <= 0x7f)
672 *cur++ = ch; 684 *cur++ = ch;
673 else if (ch >= 0x80) 685 else if (ch >= 0x80)
674 { 686 {
687 STRLEN clen;
688 UV uch;
689
675 --dec->cur; 690 --dec->cur;
676 691
677 STRLEN clen;
678 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 692 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
679 if (clen == (STRLEN)-1) 693 if (clen == (STRLEN)-1)
680 ERR ("malformed UTF-8 character in JSON string"); 694 ERR ("malformed UTF-8 character in JSON string");
681 695
682 do 696 do
683 *cur++ = *dec->cur++; 697 *cur++ = *dec->cur++;
695 ERR ("invalid character encountered while parsing JSON string"); 709 ERR ("invalid character encountered while parsing JSON string");
696 } 710 }
697 } 711 }
698 while (cur < buf + SHORT_STRING_LEN); 712 while (cur < buf + SHORT_STRING_LEN);
699 713
714 {
700 STRLEN len = cur - buf; 715 STRLEN len = cur - buf;
701 716
702 if (sv) 717 if (sv)
703 { 718 {
704 SvGROW (sv, SvCUR (sv) + len + 1); 719 SvGROW (sv, SvCUR (sv) + len + 1);
705 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 720 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
706 SvCUR_set (sv, SvCUR (sv) + len); 721 SvCUR_set (sv, SvCUR (sv) + len);
707 } 722 }
708 else 723 else
709 sv = newSVpvn (buf, len); 724 sv = newSVpvn (buf, len);
725 }
710 } 726 }
711 while (*dec->cur != '"'); 727 while (*dec->cur != '"');
712 728
713 ++dec->cur; 729 ++dec->cur;
714 730
967fail: 983fail:
968 return 0; 984 return 0;
969} 985}
970 986
971static SV * 987static SV *
972decode_json (SV *string, U32 flags) 988decode_json (SV *string, U32 flags, UV *offset_return)
973{ 989{
990 dec_t dec;
991 UV offset;
974 SV *sv; 992 SV *sv;
975 993
994 SvGETMAGIC (string);
976 SvUPGRADE (string, SVt_PV); 995 SvUPGRADE (string, SVt_PV);
977 996
978 if (flags & F_UTF8) 997 if (flags & F_UTF8)
979 sv_utf8_downgrade (string, 0); 998 sv_utf8_downgrade (string, 0);
980 else 999 else
981 sv_utf8_upgrade (string); 1000 sv_utf8_upgrade (string);
982 1001
983 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1002 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
984 1003
985 dec_t dec;
986 dec.flags = flags; 1004 dec.flags = flags;
987 dec.cur = SvPVX (string); 1005 dec.cur = SvPVX (string);
988 dec.end = SvEND (string); 1006 dec.end = SvEND (string);
989 dec.err = 0; 1007 dec.err = 0;
990 dec.depth = 0; 1008 dec.depth = 0;
991 dec.maxdepth = DEC_DEPTH (dec.flags); 1009 dec.maxdepth = DEC_DEPTH (dec.flags);
992 1010
993 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1011 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
994 sv = decode_sv (&dec); 1012 sv = decode_sv (&dec);
995 1013
1014 if (offset_return || !sv)
1015 {
1016 offset = dec.flags & F_UTF8
1017 ? dec.cur - SvPVX (string)
1018 : utf8_distance (dec.cur, SvPVX (string));
1019
1020 if (offset_return)
1021 *offset_return = offset;
1022 }
1023 else
1024 {
1025 // check for trailing garbage
1026 decode_ws (&dec);
1027
1028 if (*dec.cur)
1029 {
1030 dec.err = "garbage after JSON object";
1031 SvREFCNT_dec (sv);
1032 sv = 0;
1033 }
1034 }
1035
996 if (!sv) 1036 if (!sv)
997 { 1037 {
998 IV offset = dec.flags & F_UTF8
999 ? dec.cur - SvPVX (string)
1000 : utf8_distance (dec.cur, SvPVX (string));
1001 SV *uni = sv_newmortal (); 1038 SV *uni = sv_newmortal ();
1002 1039
1003 // horrible hack to silence warning inside pv_uni_display 1040 // horrible hack to silence warning inside pv_uni_display
1004 COP cop = *PL_curcop; 1041 COP cop = *PL_curcop;
1005 cop.cop_warnings = pWARN_NONE; 1042 cop.cop_warnings = pWARN_NONE;
1053 RETVAL 1090 RETVAL
1054 1091
1055SV *ascii (SV *self, int enable = 1) 1092SV *ascii (SV *self, int enable = 1)
1056 ALIAS: 1093 ALIAS:
1057 ascii = F_ASCII 1094 ascii = F_ASCII
1095 latin1 = F_LATIN1
1058 utf8 = F_UTF8 1096 utf8 = F_UTF8
1059 indent = F_INDENT 1097 indent = F_INDENT
1060 canonical = F_CANONICAL 1098 canonical = F_CANONICAL
1061 space_before = F_SPACE_BEFORE 1099 space_before = F_SPACE_BEFORE
1062 space_after = F_SPACE_AFTER 1100 space_after = F_SPACE_AFTER
1074 RETVAL = newSVsv (self); 1112 RETVAL = newSVsv (self);
1075} 1113}
1076 OUTPUT: 1114 OUTPUT:
1077 RETVAL 1115 RETVAL
1078 1116
1079SV *max_depth (SV *self, int max_depth = 0x80000000UL) 1117SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1080 CODE: 1118 CODE:
1081{ 1119{
1082 UV *uv = SvJSON (self); 1120 UV *uv = SvJSON (self);
1083 UV log2 = 0; 1121 UV log2 = 0;
1084 1122
1098 PPCODE: 1136 PPCODE:
1099 XPUSHs (encode_json (scalar, *SvJSON (self))); 1137 XPUSHs (encode_json (scalar, *SvJSON (self)));
1100 1138
1101void decode (SV *self, SV *jsonstr) 1139void decode (SV *self, SV *jsonstr)
1102 PPCODE: 1140 PPCODE:
1103 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1141 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1142
1143void decode_prefix (SV *self, SV *jsonstr)
1144 PPCODE:
1145{
1146 UV offset;
1147 EXTEND (SP, 2);
1148 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1149 PUSHs (sv_2mortal (newSVuv (offset)));
1150}
1104 1151
1105PROTOTYPES: ENABLE 1152PROTOTYPES: ENABLE
1106 1153
1107void to_json (SV *scalar) 1154void to_json (SV *scalar)
1108 ALIAS: 1155 ALIAS:
1112 1159
1113void from_json (SV *jsonstr) 1160void from_json (SV *jsonstr)
1114 ALIAS: 1161 ALIAS:
1115 jsonToObj = 0 1162 jsonToObj = 0
1116 PPCODE: 1163 PPCODE:
1117 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1164 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1118 1165

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines