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 1.25 by root, Fri Apr 6 20:37:22 2007 UTC vs.
Revision 1.33 by root, Wed May 23 20:26:40 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 }
… …
362 // actually, this is mostly due to the stupid so-called 372 // actually, this is mostly due to the stupid so-called
363 // security workaround added somewhere in 5.8.x. 373 // security workaround added somewhere in 5.8.x.
364 // that randomises hash orderings 374 // that randomises hash orderings
365 if (enc->flags & F_CANONICAL) 375 if (enc->flags & F_CANONICAL)
366 { 376 {
367 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
368 int fast = 1; 377 int fast = 1;
378 HE *he;
379#if WIN32
380 HE **hes = _alloca (count * sizeof (HE));
381#else
382 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
383#endif
369 384
370 i = 0; 385 i = 0;
371 while ((he = hv_iternext (hv))) 386 while ((he = hv_iternext (hv)))
372 { 387 {
373 hes [i++] = he; 388 hes [i++] = he;
… …
510 enc.maxdepth = DEC_DEPTH (flags); 525 enc.maxdepth = DEC_DEPTH (flags);
511 526
512 SvPOK_only (enc.sv); 527 SvPOK_only (enc.sv);
513 encode_sv (&enc, scalar); 528 encode_sv (&enc, scalar);
514 529
530 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
531 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
532
515 if (!(flags & (F_ASCII | F_UTF8))) 533 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
516 SvUTF8_on (enc.sv); 534 SvUTF8_on (enc.sv);
517
518 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
519 535
520 if (enc.flags & F_SHRINK) 536 if (enc.flags & F_SHRINK)
521 shrink (enc.sv); 537 shrink (enc.sv);
522 538
523 return enc.sv; 539 return enc.sv;
… …
595 SV *sv = 0; 611 SV *sv = 0;
596 int utf8 = 0; 612 int utf8 = 0;
597 613
598 do 614 do
599 { 615 {
600 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 616 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
601 char *cur = buf; 617 char *cur = buf;
602 618
603 do 619 do
604 { 620 {
605 unsigned char ch = *(unsigned char *)dec->cur++; 621 unsigned char ch = *(unsigned char *)dec->cur++;
… …
972fail: 988fail:
973 return 0; 989 return 0;
974} 990}
975 991
976static SV * 992static SV *
977decode_json (SV *string, U32 flags) 993decode_json (SV *string, U32 flags, UV *offset_return)
978{ 994{
979 dec_t dec; 995 dec_t dec;
996 UV offset;
980 SV *sv; 997 SV *sv;
981 998
999 SvGETMAGIC (string);
982 SvUPGRADE (string, SVt_PV); 1000 SvUPGRADE (string, SVt_PV);
983 1001
984 if (flags & F_UTF8) 1002 if (flags & F_UTF8)
985 sv_utf8_downgrade (string, 0); 1003 sv_utf8_downgrade (string, 0);
986 else 1004 else
… …
993 dec.end = SvEND (string); 1011 dec.end = SvEND (string);
994 dec.err = 0; 1012 dec.err = 0;
995 dec.depth = 0; 1013 dec.depth = 0;
996 dec.maxdepth = DEC_DEPTH (dec.flags); 1014 dec.maxdepth = DEC_DEPTH (dec.flags);
997 1015
998 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1016 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
999 sv = decode_sv (&dec); 1017 sv = decode_sv (&dec);
1000 1018
1019 if (!(offset_return || !sv))
1020 {
1021 // check for trailing garbage
1022 decode_ws (&dec);
1023
1024 if (*dec.cur)
1025 {
1026 dec.err = "garbage after JSON object";
1027 SvREFCNT_dec (sv);
1028 sv = 0;
1029 }
1030 }
1031
1032 if (offset_return || !sv)
1033 {
1034 offset = dec.flags & F_UTF8
1035 ? dec.cur - SvPVX (string)
1036 : utf8_distance (dec.cur, SvPVX (string));
1037
1038 if (offset_return)
1039 *offset_return = offset;
1040 }
1041
1001 if (!sv) 1042 if (!sv)
1002 { 1043 {
1003 IV offset = dec.flags & F_UTF8
1004 ? dec.cur - SvPVX (string)
1005 : utf8_distance (dec.cur, SvPVX (string));
1006 SV *uni = sv_newmortal (); 1044 SV *uni = sv_newmortal ();
1007 1045
1008 // horrible hack to silence warning inside pv_uni_display 1046 // horrible hack to silence warning inside pv_uni_display
1009 COP cop = *PL_curcop; 1047 COP cop = *PL_curcop;
1010 cop.cop_warnings = pWARN_NONE; 1048 cop.cop_warnings = pWARN_NONE;
… …
1058 RETVAL 1096 RETVAL
1059 1097
1060SV *ascii (SV *self, int enable = 1) 1098SV *ascii (SV *self, int enable = 1)
1061 ALIAS: 1099 ALIAS:
1062 ascii = F_ASCII 1100 ascii = F_ASCII
1101 latin1 = F_LATIN1
1063 utf8 = F_UTF8 1102 utf8 = F_UTF8
1064 indent = F_INDENT 1103 indent = F_INDENT
1065 canonical = F_CANONICAL 1104 canonical = F_CANONICAL
1066 space_before = F_SPACE_BEFORE 1105 space_before = F_SPACE_BEFORE
1067 space_after = F_SPACE_AFTER 1106 space_after = F_SPACE_AFTER
… …
1103 PPCODE: 1142 PPCODE:
1104 XPUSHs (encode_json (scalar, *SvJSON (self))); 1143 XPUSHs (encode_json (scalar, *SvJSON (self)));
1105 1144
1106void decode (SV *self, SV *jsonstr) 1145void decode (SV *self, SV *jsonstr)
1107 PPCODE: 1146 PPCODE:
1108 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1147 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1148
1149void decode_prefix (SV *self, SV *jsonstr)
1150 PPCODE:
1151{
1152 UV offset;
1153 EXTEND (SP, 2);
1154 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1155 PUSHs (sv_2mortal (newSVuv (offset)));
1156}
1109 1157
1110PROTOTYPES: ENABLE 1158PROTOTYPES: ENABLE
1111 1159
1112void to_json (SV *scalar) 1160void to_json (SV *scalar)
1113 ALIAS: 1161 ALIAS:
… …
1117 1165
1118void from_json (SV *jsonstr) 1166void from_json (SV *jsonstr)
1119 ALIAS: 1167 ALIAS:
1120 jsonToObj = 0 1168 jsonToObj = 0
1121 PPCODE: 1169 PPCODE:
1122 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1170 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1123 1171

Diff Legend

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