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 1.39 by root, Mon Jun 11 03:42:57 2007 UTC vs.
Revision 1.55 by root, Mon Jul 23 22:57:40 2007 UTC

9 9
10#if defined(__BORLANDC__) || defined(_MSC_VER) 10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 12#endif
13 13
14// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13
18#endif
19
14#define F_ASCII 0x00000001UL 20#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL 21#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 22#define F_UTF8 0x00000004UL
17#define F_INDENT 0x00000008UL 23#define F_INDENT 0x00000008UL
18#define F_CANONICAL 0x00000010UL 24#define F_CANONICAL 0x00000010UL
19#define F_SPACE_BEFORE 0x00000020UL 25#define F_SPACE_BEFORE 0x00000020UL
20#define F_SPACE_AFTER 0x00000040UL 26#define F_SPACE_AFTER 0x00000040UL
21#define F_ALLOW_NONREF 0x00000100UL 27#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL 28#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL
23#define F_MAXDEPTH 0xf8000000UL 31#define F_MAXDEPTH 0xf8000000UL
24#define S_MAXDEPTH 27 32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL
34#define S_MAXSIZE 20
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
25 36
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27 38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 39
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH) 41#define F_DEFAULT (9UL << S_MAXDEPTH)
33 42
34#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
48#endif 57#endif
49 58
50#define expect_false(expr) expect ((expr) != 0, 0) 59#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1) 60#define expect_true(expr) expect ((expr) != 0, 1)
52 61
53static HV *json_stash; // JSON::XS:: 62static HV *json_stash, *json_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false;
64
65typedef struct {
66 U32 flags;
67 SV *cb_object;
68 HV *cb_sk_object;
69} JSON;
54 70
55///////////////////////////////////////////////////////////////////////////// 71/////////////////////////////////////////////////////////////////////////////
56// utility functions 72// utility functions
57 73
58static UV * 74inline void
59SvJSON (SV *sv)
60{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
62 croak ("object is not of type JSON::XS");
63
64 return &SvUVX (SvRV (sv));
65}
66
67static void
68shrink (SV *sv) 75shrink (SV *sv)
69{ 76{
70 sv_utf8_downgrade (sv, 1); 77 sv_utf8_downgrade (sv, 1);
71 if (SvLEN (sv) > SvCUR (sv) + 1) 78 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 79 {
107typedef struct 114typedef struct
108{ 115{
109 char *cur; // SvPVX (sv) + current output position 116 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 117 char *end; // SvEND (sv)
111 SV *sv; // result scalar 118 SV *sv; // result scalar
112 U32 flags; // F_* 119 JSON json;
113 U32 indent; // indentation level 120 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 121 U32 maxdepth; // max. indentation/recursion level
115} enc_t; 122} enc_t;
116 123
117inline void 124inline void
178 STRLEN clen; 185 STRLEN clen;
179 UV uch; 186 UV uch;
180 187
181 if (is_utf8) 188 if (is_utf8)
182 { 189 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen); 190 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1) 191 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 192 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 } 193 }
188 else 194 else
192 } 198 }
193 199
194 if (uch > 0x10FFFFUL) 200 if (uch > 0x10FFFFUL)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 201 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196 202
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) 203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
198 { 204 {
199 if (uch > 0xFFFFUL) 205 if (uch > 0xFFFFUL)
200 { 206 {
201 need (enc, len += 11); 207 need (enc, len += 11);
202 sprintf (enc->cur, "\\u%04x\\u%04x", 208 sprintf (enc->cur, "\\u%04x\\u%04x",
216 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 222 *enc->cur++ = hexdigit [(uch >> 0) & 15];
217 } 223 }
218 224
219 str += clen; 225 str += clen;
220 } 226 }
221 else if (enc->flags & F_LATIN1) 227 else if (enc->json.flags & F_LATIN1)
222 { 228 {
223 *enc->cur++ = uch; 229 *enc->cur++ = uch;
224 str += clen; 230 str += clen;
225 } 231 }
226 else if (is_utf8) 232 else if (is_utf8)
247} 253}
248 254
249inline void 255inline void
250encode_indent (enc_t *enc) 256encode_indent (enc_t *enc)
251{ 257{
252 if (enc->flags & F_INDENT) 258 if (enc->json.flags & F_INDENT)
253 { 259 {
254 int spaces = enc->indent * INDENT_STEP; 260 int spaces = enc->indent * INDENT_STEP;
255 261
256 need (enc, spaces); 262 need (enc, spaces);
257 memset (enc->cur, ' ', spaces); 263 memset (enc->cur, ' ', spaces);
267} 273}
268 274
269inline void 275inline void
270encode_nl (enc_t *enc) 276encode_nl (enc_t *enc)
271{ 277{
272 if (enc->flags & F_INDENT) 278 if (enc->json.flags & F_INDENT)
273 { 279 {
274 need (enc, 1); 280 need (enc, 1);
275 encode_ch (enc, '\n'); 281 encode_ch (enc, '\n');
276 } 282 }
277} 283}
279inline void 285inline void
280encode_comma (enc_t *enc) 286encode_comma (enc_t *enc)
281{ 287{
282 encode_ch (enc, ','); 288 encode_ch (enc, ',');
283 289
284 if (enc->flags & F_INDENT) 290 if (enc->json.flags & F_INDENT)
285 encode_nl (enc); 291 encode_nl (enc);
286 else if (enc->flags & F_SPACE_AFTER) 292 else if (enc->json.flags & F_SPACE_AFTER)
287 encode_space (enc); 293 encode_space (enc);
288} 294}
289 295
290static void encode_sv (enc_t *enc, SV *sv); 296static void encode_sv (enc_t *enc, SV *sv);
291 297
300 encode_ch (enc, '['); encode_nl (enc); 306 encode_ch (enc, '['); encode_nl (enc);
301 ++enc->indent; 307 ++enc->indent;
302 308
303 for (i = 0; i <= len; ++i) 309 for (i = 0; i <= len; ++i)
304 { 310 {
311 SV **svp = av_fetch (av, i, 0);
312
305 encode_indent (enc); 313 encode_indent (enc);
306 encode_sv (enc, *av_fetch (av, i, 0)); 314
315 if (svp)
316 encode_sv (enc, *svp);
317 else
318 encode_str (enc, "null", 4, 0);
307 319
308 if (i < len) 320 if (i < len)
309 encode_comma (enc); 321 encode_comma (enc);
310 } 322 }
311 323
334 else 346 else
335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 347 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336 348
337 encode_ch (enc, '"'); 349 encode_ch (enc, '"');
338 350
339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 351 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340 encode_ch (enc, ':'); 352 encode_ch (enc, ':');
341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 353 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342 encode_sv (enc, HeVAL (he)); 354 encode_sv (enc, HeVAL (he));
343} 355}
344 356
345// compare hash entries, used when all keys are bytestrings 357// compare hash entries, used when all keys are bytestrings
346static int 358static int
381 { 393 {
382 // for canonical output we have to sort by keys first 394 // for canonical output we have to sort by keys first
383 // actually, this is mostly due to the stupid so-called 395 // actually, this is mostly due to the stupid so-called
384 // security workaround added somewhere in 5.8.x. 396 // security workaround added somewhere in 5.8.x.
385 // that randomises hash orderings 397 // that randomises hash orderings
386 if (enc->flags & F_CANONICAL) 398 if (enc->json.flags & F_CANONICAL)
387 { 399 {
388 int fast = 1; 400 int fast = 1;
389 HE *he; 401 HE *he;
390#if defined(__BORLANDC__) || defined(_MSC_VER) 402#if defined(__BORLANDC__) || defined(_MSC_VER)
391 HE **hes = _alloca (count * sizeof (HE)); 403 HE **hes = _alloca (count * sizeof (HE));
463 svtype svt; 475 svtype svt;
464 476
465 SvGETMAGIC (sv); 477 SvGETMAGIC (sv);
466 svt = SvTYPE (sv); 478 svt = SvTYPE (sv);
467 479
480 if (expect_false (SvOBJECT (sv)))
481 {
482 if (SvSTASH (sv) == json_boolean_stash)
483 {
484 if (SvIV (sv))
485 encode_str (enc, "true", 4, 0);
486 else
487 encode_str (enc, "false", 5, 0);
488 }
489 else
490 {
491#if 0
492 if (0 && sv_derived_from (rv, "JSON::Literal"))
493 {
494 // not yet
495 }
496#endif
497 if (enc->json.flags & F_CONV_BLESSED)
498 {
499 // we re-bless the reference to get overload and other niceties right
500 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
501
502 if (to_json)
503 {
504 int count;
505 dSP;
506
507 ENTER; SAVETMPS; PUSHMARK (SP);
508 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
509
510 // calling with G_SCALAR ensures that we always get a 1 return value
511 PUTBACK;
512 call_sv ((SV *)GvCV (to_json), G_SCALAR);
513 SPAGAIN;
514
515 // catch this surprisingly common error
516 if (SvROK (TOPs) && SvRV (TOPs) == sv)
517 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
518
519 sv = POPs;
520 PUTBACK;
521
522 encode_sv (enc, sv);
523
524 FREETMPS; LEAVE;
525 }
526 else if (enc->json.flags & F_ALLOW_BLESSED)
527 encode_str (enc, "null", 4, 0);
528 else
529 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
530 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
531 }
532 else if (enc->json.flags & F_ALLOW_BLESSED)
533 encode_str (enc, "null", 4, 0);
534 else
535 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
536 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
537 }
538 }
468 if (svt == SVt_PVHV) 539 else if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv); 540 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV) 541 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv); 542 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV) 543 else if (svt < SVt_PVAV)
473 { 544 {
474 if (SvNIOK (sv) && SvIV (sv) == 0) 545 STRLEN len = 0;
546 char *pv = svt ? SvPV (sv, len) : 0;
547
548 if (len == 1 && *pv == '1')
549 encode_str (enc, "true", 4, 0);
550 else if (len == 1 && *pv == '0')
475 encode_str (enc, "false", 5, 0); 551 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1)
477 encode_str (enc, "true", 4, 0);
478 else 552 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 553 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 554 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 } 555 }
482 else 556 else
552 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 626 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
553 SvPV_nolen (sv), SvFLAGS (sv)); 627 SvPV_nolen (sv), SvFLAGS (sv));
554} 628}
555 629
556static SV * 630static SV *
557encode_json (SV *scalar, U32 flags) 631encode_json (SV *scalar, JSON *json)
558{ 632{
559 enc_t enc; 633 enc_t enc;
560 634
561 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 635 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
562 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 636 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
563 637
564 enc.flags = flags; 638 enc.json = *json;
565 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 639 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
566 enc.cur = SvPVX (enc.sv); 640 enc.cur = SvPVX (enc.sv);
567 enc.end = SvEND (enc.sv); 641 enc.end = SvEND (enc.sv);
568 enc.indent = 0; 642 enc.indent = 0;
569 enc.maxdepth = DEC_DEPTH (flags); 643 enc.maxdepth = DEC_DEPTH (enc.json.flags);
570 644
571 SvPOK_only (enc.sv); 645 SvPOK_only (enc.sv);
572 encode_sv (&enc, scalar); 646 encode_sv (&enc, scalar);
573 647
574 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 648 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
575 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 649 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
576 650
577 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 651 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
578 SvUTF8_on (enc.sv); 652 SvUTF8_on (enc.sv);
579 653
580 if (enc.flags & F_SHRINK) 654 if (enc.json.flags & F_SHRINK)
581 shrink (enc.sv); 655 shrink (enc.sv);
582 656
583 return enc.sv; 657 return enc.sv;
584} 658}
585 659
590typedef struct 664typedef struct
591{ 665{
592 char *cur; // current parser pointer 666 char *cur; // current parser pointer
593 char *end; // end of input string 667 char *end; // end of input string
594 const char *err; // parse error, if != 0 668 const char *err; // parse error, if != 0
595 U32 flags; // F_* 669 JSON json;
596 U32 depth; // recursion depth 670 U32 depth; // recursion depth
597 U32 maxdepth; // recursion depth limit 671 U32 maxdepth; // recursion depth limit
598} dec_t; 672} dec_t;
599 673
600inline void 674inline void
631decode_4hex (dec_t *dec) 705decode_4hex (dec_t *dec)
632{ 706{
633 signed char d1, d2, d3, d4; 707 signed char d1, d2, d3, d4;
634 unsigned char *cur = (unsigned char *)dec->cur; 708 unsigned char *cur = (unsigned char *)dec->cur;
635 709
636 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); 710 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
637 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); 711 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
638 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); 712 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
639 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); 713 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
640 714
641 dec->cur += 4; 715 dec->cur += 4;
642 716
643 return ((UV)d1) << 12 717 return ((UV)d1) << 12
644 | ((UV)d2) << 8 718 | ((UV)d2) << 8
867 { 941 {
868 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 942 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
869 if (*start == '-') 943 if (*start == '-')
870 switch (dec->cur - start) 944 switch (dec->cur - start)
871 { 945 {
872 case 2: return newSViv (-( start [1] - '0' )); 946 case 2: return newSViv (-( start [1] - '0' * 1));
873 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 947 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
874 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 948 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
875 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 949 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
876 } 950 }
877 else 951 else
878 switch (dec->cur - start) 952 switch (dec->cur - start)
879 { 953 {
880 case 1: return newSViv ( start [0] - '0' ); 954 case 1: return newSViv ( start [0] - '0' * 1);
881 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 955 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
882 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 956 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
883 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 957 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
884 } 958 }
885 959
892 if (uv < (UV)IV_MIN) 966 if (uv < (UV)IV_MIN)
893 return newSViv (-(IV)uv); 967 return newSViv (-(IV)uv);
894 } 968 }
895 else 969 else
896 return newSVuv (uv); 970 return newSVuv (uv);
971
972 // here would likely be the place for bigint support
897 } 973 }
898 } 974 }
899 975
976 // if we ever support bigint or bigfloat, this is the place for bigfloat
900 return newSVnv (Atof (start)); 977 return newSVnv (Atof (start));
901 978
902fail: 979fail:
903 return 0; 980 return 0;
904} 981}
948} 1025}
949 1026
950static SV * 1027static SV *
951decode_hv (dec_t *dec) 1028decode_hv (dec_t *dec)
952{ 1029{
1030 SV *sv;
953 HV *hv = newHV (); 1031 HV *hv = newHV ();
954 1032
955 DEC_INC_DEPTH; 1033 DEC_INC_DEPTH;
956 decode_ws (dec); 1034 decode_ws (dec);
957 1035
958 if (*dec->cur == '}') 1036 if (*dec->cur == '}')
959 ++dec->cur; 1037 ++dec->cur;
960 else 1038 else
961 for (;;) 1039 for (;;)
962 { 1040 {
963 SV *key, *value;
964
965 decode_ws (dec); EXPECT_CH ('"'); 1041 decode_ws (dec); EXPECT_CH ('"');
966 1042
967 key = decode_str (dec); 1043 // heuristic: assume that
968 if (!key) 1044 // a) decode_str + hv_store_ent are abysmally slow.
969 goto fail; 1045 // b) most hash keys are short, simple ascii text.
1046 // => try to "fast-match" such strings to avoid
1047 // the overhead of decode_str + hv_store_ent.
1048 {
1049 SV *value;
1050 char *p = dec->cur;
1051 char *e = p + 24; // only try up to 24 bytes
970 1052
971 decode_ws (dec); EXPECT_CH (':'); 1053 for (;;)
972
973 value = decode_sv (dec);
974 if (!value)
975 { 1054 {
1055 // the >= 0x80 is true on most architectures
1056 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1057 {
1058 // slow path, back up and use decode_str
1059 SV *key = decode_str (dec);
1060 if (!key)
1061 goto fail;
1062
1063 decode_ws (dec); EXPECT_CH (':');
1064
1065 value = decode_sv (dec);
1066 if (!value)
1067 {
1068 SvREFCNT_dec (key);
1069 goto fail;
1070 }
1071
1072 hv_store_ent (hv, key, value, 0);
976 SvREFCNT_dec (key); 1073 SvREFCNT_dec (key);
1074
1075 break;
1076 }
1077 else if (*p == '"')
1078 {
1079 // fast path, got a simple key
1080 char *key = dec->cur;
1081 int len = p - key;
1082 dec->cur = p + 1;
1083
1084 decode_ws (dec); EXPECT_CH (':');
1085
1086 value = decode_sv (dec);
1087 if (!value)
977 goto fail; 1088 goto fail;
1089
1090 hv_store (hv, key, len, value, 0);
1091
1092 break;
1093 }
1094
1095 ++p;
978 } 1096 }
979 1097 }
980 hv_store_ent (hv, key, value, 0);
981 SvREFCNT_dec (key);
982 1098
983 decode_ws (dec); 1099 decode_ws (dec);
984 1100
985 if (*dec->cur == '}') 1101 if (*dec->cur == '}')
986 { 1102 {
993 1109
994 ++dec->cur; 1110 ++dec->cur;
995 } 1111 }
996 1112
997 DEC_DEC_DEPTH; 1113 DEC_DEC_DEPTH;
998 return newRV_noinc ((SV *)hv); 1114 sv = newRV_noinc ((SV *)hv);
1115
1116 // check filter callbacks
1117 if (dec->json.flags & F_HOOK)
1118 {
1119 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1120 {
1121 HE *cb, *he;
1122
1123 hv_iterinit (hv);
1124 he = hv_iternext (hv);
1125 hv_iterinit (hv);
1126
1127 // the next line creates a mortal sv each time its called.
1128 // might want to optimise this for common cases.
1129 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1130
1131 if (cb)
1132 {
1133 dSP;
1134 int count;
1135
1136 ENTER; SAVETMPS; PUSHMARK (SP);
1137 XPUSHs (HeVAL (he));
1138
1139 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1140
1141 if (count == 1)
1142 {
1143 sv = newSVsv (POPs);
1144 FREETMPS; LEAVE;
1145 return sv;
1146 }
1147
1148 FREETMPS; LEAVE;
1149 }
1150 }
1151
1152 if (dec->json.cb_object)
1153 {
1154 dSP;
1155 int count;
1156
1157 ENTER; SAVETMPS; PUSHMARK (SP);
1158 XPUSHs (sv_2mortal (sv));
1159
1160 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1161
1162 if (count == 1)
1163 {
1164 sv = newSVsv (POPs);
1165 FREETMPS; LEAVE;
1166 return sv;
1167 }
1168
1169 SvREFCNT_inc (sv);
1170 FREETMPS; LEAVE;
1171 }
1172 }
1173
1174 return sv;
999 1175
1000fail: 1176fail:
1001 SvREFCNT_dec (hv); 1177 SvREFCNT_dec (hv);
1002 DEC_DEC_DEPTH; 1178 DEC_DEC_DEPTH;
1003 return 0; 1179 return 0;
1005 1181
1006static SV * 1182static SV *
1007decode_sv (dec_t *dec) 1183decode_sv (dec_t *dec)
1008{ 1184{
1009 decode_ws (dec); 1185 decode_ws (dec);
1186
1187 // the beauty of JSON: you need exactly one character lookahead
1188 // to parse anything.
1010 switch (*dec->cur) 1189 switch (*dec->cur)
1011 { 1190 {
1012 case '"': ++dec->cur; return decode_str (dec); 1191 case '"': ++dec->cur; return decode_str (dec);
1013 case '[': ++dec->cur; return decode_av (dec); 1192 case '[': ++dec->cur; return decode_av (dec);
1014 case '{': ++dec->cur; return decode_hv (dec); 1193 case '{': ++dec->cur; return decode_hv (dec);
1020 1199
1021 case 't': 1200 case 't':
1022 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1201 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1023 { 1202 {
1024 dec->cur += 4; 1203 dec->cur += 4;
1025 return newSViv (1); 1204 return SvREFCNT_inc (json_true);
1026 } 1205 }
1027 else 1206 else
1028 ERR ("'true' expected"); 1207 ERR ("'true' expected");
1029 1208
1030 break; 1209 break;
1031 1210
1032 case 'f': 1211 case 'f':
1033 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1212 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1034 { 1213 {
1035 dec->cur += 5; 1214 dec->cur += 5;
1036 return newSViv (0); 1215 return SvREFCNT_inc (json_false);
1037 } 1216 }
1038 else 1217 else
1039 ERR ("'false' expected"); 1218 ERR ("'false' expected");
1040 1219
1041 break; 1220 break;
1059fail: 1238fail:
1060 return 0; 1239 return 0;
1061} 1240}
1062 1241
1063static SV * 1242static SV *
1064decode_json (SV *string, U32 flags, UV *offset_return) 1243decode_json (SV *string, JSON *json, UV *offset_return)
1065{ 1244{
1066 dec_t dec; 1245 dec_t dec;
1067 UV offset; 1246 UV offset;
1068 SV *sv; 1247 SV *sv;
1069 1248
1070 SvGETMAGIC (string); 1249 SvGETMAGIC (string);
1071 SvUPGRADE (string, SVt_PV); 1250 SvUPGRADE (string, SVt_PV);
1072 1251
1252 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1253 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1254 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1255
1073 if (flags & F_UTF8) 1256 if (json->flags & F_UTF8)
1074 sv_utf8_downgrade (string, 0); 1257 sv_utf8_downgrade (string, 0);
1075 else 1258 else
1076 sv_utf8_upgrade (string); 1259 sv_utf8_upgrade (string);
1077 1260
1078 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1261 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1079 1262
1080 dec.flags = flags; 1263 dec.json = *json;
1081 dec.cur = SvPVX (string); 1264 dec.cur = SvPVX (string);
1082 dec.end = SvEND (string); 1265 dec.end = SvEND (string);
1083 dec.err = 0; 1266 dec.err = 0;
1084 dec.depth = 0; 1267 dec.depth = 0;
1085 dec.maxdepth = DEC_DEPTH (dec.flags); 1268 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1269
1270 if (dec.json.cb_object || dec.json.cb_sk_object)
1271 dec.json.flags |= F_HOOK;
1086 1272
1087 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1273 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1088 sv = decode_sv (&dec); 1274 sv = decode_sv (&dec);
1089 1275
1090 if (!(offset_return || !sv)) 1276 if (!(offset_return || !sv))
1100 } 1286 }
1101 } 1287 }
1102 1288
1103 if (offset_return || !sv) 1289 if (offset_return || !sv)
1104 { 1290 {
1105 offset = dec.flags & F_UTF8 1291 offset = dec.json.flags & F_UTF8
1106 ? dec.cur - SvPVX (string) 1292 ? dec.cur - SvPVX (string)
1107 : utf8_distance (dec.cur, SvPVX (string)); 1293 : utf8_distance (dec.cur, SvPVX (string));
1108 1294
1109 if (offset_return) 1295 if (offset_return)
1110 *offset_return = offset; 1296 *offset_return = offset;
1129 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1315 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1130 } 1316 }
1131 1317
1132 sv = sv_2mortal (sv); 1318 sv = sv_2mortal (sv);
1133 1319
1134 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1320 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1135 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1321 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1136 1322
1137 return sv; 1323 return sv;
1138} 1324}
1139 1325
1143MODULE = JSON::XS PACKAGE = JSON::XS 1329MODULE = JSON::XS PACKAGE = JSON::XS
1144 1330
1145BOOT: 1331BOOT:
1146{ 1332{
1147 int i; 1333 int i;
1148
1149 memset (decode_hexdigit, 0xff, 256);
1150 1334
1151 for (i = 0; i < 256; ++i) 1335 for (i = 0; i < 256; ++i)
1152 decode_hexdigit [i] = 1336 decode_hexdigit [i] =
1153 i >= '0' && i <= '9' ? i - '0' 1337 i >= '0' && i <= '9' ? i - '0'
1154 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1338 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1155 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1339 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1156 : -1; 1340 : -1;
1157 1341
1158 json_stash = gv_stashpv ("JSON::XS", 1); 1342 json_stash = gv_stashpv ("JSON::XS" , 1);
1343 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1344
1345 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1346 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1159} 1347}
1160 1348
1161PROTOTYPES: DISABLE 1349PROTOTYPES: DISABLE
1162 1350
1163SV *new (char *dummy) 1351void new (char *klass)
1164 CODE: 1352 PPCODE:
1165 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1353{
1166 OUTPUT: 1354 SV *pv = NEWSV (0, sizeof (JSON));
1167 RETVAL 1355 SvPOK_only (pv);
1356 Zero (SvPVX (pv), 1, JSON);
1357 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1358 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1359}
1168 1360
1169SV *ascii (SV *self, int enable = 1) 1361void ascii (JSON *self, int enable = 1)
1170 ALIAS: 1362 ALIAS:
1171 ascii = F_ASCII 1363 ascii = F_ASCII
1172 latin1 = F_LATIN1 1364 latin1 = F_LATIN1
1173 utf8 = F_UTF8 1365 utf8 = F_UTF8
1174 indent = F_INDENT 1366 indent = F_INDENT
1175 canonical = F_CANONICAL 1367 canonical = F_CANONICAL
1176 space_before = F_SPACE_BEFORE 1368 space_before = F_SPACE_BEFORE
1177 space_after = F_SPACE_AFTER 1369 space_after = F_SPACE_AFTER
1178 pretty = F_PRETTY 1370 pretty = F_PRETTY
1179 allow_nonref = F_ALLOW_NONREF 1371 allow_nonref = F_ALLOW_NONREF
1180 shrink = F_SHRINK 1372 shrink = F_SHRINK
1373 allow_blessed = F_ALLOW_BLESSED
1374 convert_blessed = F_CONV_BLESSED
1181 CODE: 1375 PPCODE:
1182{ 1376{
1183 UV *uv = SvJSON (self);
1184 if (enable) 1377 if (enable)
1185 *uv |= ix; 1378 self->flags |= ix;
1186 else 1379 else
1187 *uv &= ~ix; 1380 self->flags &= ~ix;
1188 1381
1189 RETVAL = newSVsv (self); 1382 XPUSHs (ST (0));
1190} 1383}
1191 OUTPUT:
1192 RETVAL
1193 1384
1194SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1385void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1195 CODE: 1386 PPCODE:
1196{ 1387{
1197 UV *uv = SvJSON (self);
1198 UV log2 = 0; 1388 UV log2 = 0;
1199 1389
1200 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1390 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1201 1391
1202 while ((1UL << log2) < max_depth) 1392 while ((1UL << log2) < max_depth)
1203 ++log2; 1393 ++log2;
1204 1394
1205 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1395 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1206 1396
1207 RETVAL = newSVsv (self); 1397 XPUSHs (ST (0));
1208} 1398}
1209 OUTPUT:
1210 RETVAL
1211 1399
1212void encode (SV *self, SV *scalar) 1400void max_size (JSON *self, UV max_size = 0)
1213 PPCODE: 1401 PPCODE:
1214 XPUSHs (encode_json (scalar, *SvJSON (self))); 1402{
1403 UV log2 = 0;
1215 1404
1216void decode (SV *self, SV *jsonstr) 1405 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1406 if (max_size == 1) max_size = 2;
1407
1408 while ((1UL << log2) < max_size)
1409 ++log2;
1410
1411 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1412
1413 XPUSHs (ST (0));
1414}
1415
1416void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1217 PPCODE: 1417 PPCODE:
1418{
1419 SvREFCNT_dec (self->cb_object);
1420 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1421
1422 XPUSHs (ST (0));
1423}
1424
1425void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1426 PPCODE:
1427{
1428 if (!self->cb_sk_object)
1429 self->cb_sk_object = newHV ();
1430
1431 if (SvOK (cb))
1432 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1433 else
1434 {
1435 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1436
1437 if (!HvKEYS (self->cb_sk_object))
1438 {
1439 SvREFCNT_dec (self->cb_sk_object);
1440 self->cb_sk_object = 0;
1441 }
1442 }
1443
1444 XPUSHs (ST (0));
1445}
1446
1447void encode (JSON *self, SV *scalar)
1448 PPCODE:
1449 XPUSHs (encode_json (scalar, self));
1450
1451void decode (JSON *self, SV *jsonstr)
1452 PPCODE:
1218 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1453 XPUSHs (decode_json (jsonstr, self, 0));
1219 1454
1220void decode_prefix (SV *self, SV *jsonstr) 1455void decode_prefix (JSON *self, SV *jsonstr)
1221 PPCODE: 1456 PPCODE:
1222{ 1457{
1223 UV offset; 1458 UV offset;
1224 EXTEND (SP, 2); 1459 EXTEND (SP, 2);
1225 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 1460 PUSHs (decode_json (jsonstr, self, &offset));
1226 PUSHs (sv_2mortal (newSVuv (offset))); 1461 PUSHs (sv_2mortal (newSVuv (offset)));
1227} 1462}
1228 1463
1464void DESTROY (JSON *self)
1465 CODE:
1466 SvREFCNT_dec (self->cb_sk_object);
1467 SvREFCNT_dec (self->cb_object);
1468
1229PROTOTYPES: ENABLE 1469PROTOTYPES: ENABLE
1230 1470
1231void to_json (SV *scalar) 1471void to_json (SV *scalar)
1232 ALIAS:
1233 objToJson = 0
1234 PPCODE: 1472 PPCODE:
1473{
1474 JSON json = { F_DEFAULT | F_UTF8 };
1235 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1475 XPUSHs (encode_json (scalar, &json));
1476}
1236 1477
1237void from_json (SV *jsonstr) 1478void from_json (SV *jsonstr)
1238 ALIAS:
1239 jsonToObj = 0
1240 PPCODE: 1479 PPCODE:
1480{
1481 JSON json = { F_DEFAULT | F_UTF8 };
1241 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); 1482 XPUSHs (decode_json (jsonstr, &json, 0));
1483}
1242 1484

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines