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 1.50 by root, Mon Jul 2 00:29:38 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, *cb_sk_object;
68} JSON;
54 69
55///////////////////////////////////////////////////////////////////////////// 70/////////////////////////////////////////////////////////////////////////////
56// utility functions 71// utility functions
57 72
58static UV * 73inline 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) 74shrink (SV *sv)
69{ 75{
70 sv_utf8_downgrade (sv, 1); 76 sv_utf8_downgrade (sv, 1);
71 if (SvLEN (sv) > SvCUR (sv) + 1) 77 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 78 {
107typedef struct 113typedef struct
108{ 114{
109 char *cur; // SvPVX (sv) + current output position 115 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 116 char *end; // SvEND (sv)
111 SV *sv; // result scalar 117 SV *sv; // result scalar
112 U32 flags; // F_* 118 JSON json;
113 U32 indent; // indentation level 119 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 120 U32 maxdepth; // max. indentation/recursion level
115} enc_t; 121} enc_t;
116 122
117inline void 123inline void
178 STRLEN clen; 184 STRLEN clen;
179 UV uch; 185 UV uch;
180 186
181 if (is_utf8) 187 if (is_utf8)
182 { 188 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen); 189 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1) 190 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 191 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 } 192 }
188 else 193 else
192 } 197 }
193 198
194 if (uch > 0x10FFFFUL) 199 if (uch > 0x10FFFFUL)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196 201
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) 202 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
198 { 203 {
199 if (uch > 0xFFFFUL) 204 if (uch > 0xFFFFUL)
200 { 205 {
201 need (enc, len += 11); 206 need (enc, len += 11);
202 sprintf (enc->cur, "\\u%04x\\u%04x", 207 sprintf (enc->cur, "\\u%04x\\u%04x",
216 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 221 *enc->cur++ = hexdigit [(uch >> 0) & 15];
217 } 222 }
218 223
219 str += clen; 224 str += clen;
220 } 225 }
221 else if (enc->flags & F_LATIN1) 226 else if (enc->json.flags & F_LATIN1)
222 { 227 {
223 *enc->cur++ = uch; 228 *enc->cur++ = uch;
224 str += clen; 229 str += clen;
225 } 230 }
226 else if (is_utf8) 231 else if (is_utf8)
247} 252}
248 253
249inline void 254inline void
250encode_indent (enc_t *enc) 255encode_indent (enc_t *enc)
251{ 256{
252 if (enc->flags & F_INDENT) 257 if (enc->json.flags & F_INDENT)
253 { 258 {
254 int spaces = enc->indent * INDENT_STEP; 259 int spaces = enc->indent * INDENT_STEP;
255 260
256 need (enc, spaces); 261 need (enc, spaces);
257 memset (enc->cur, ' ', spaces); 262 memset (enc->cur, ' ', spaces);
267} 272}
268 273
269inline void 274inline void
270encode_nl (enc_t *enc) 275encode_nl (enc_t *enc)
271{ 276{
272 if (enc->flags & F_INDENT) 277 if (enc->json.flags & F_INDENT)
273 { 278 {
274 need (enc, 1); 279 need (enc, 1);
275 encode_ch (enc, '\n'); 280 encode_ch (enc, '\n');
276 } 281 }
277} 282}
279inline void 284inline void
280encode_comma (enc_t *enc) 285encode_comma (enc_t *enc)
281{ 286{
282 encode_ch (enc, ','); 287 encode_ch (enc, ',');
283 288
284 if (enc->flags & F_INDENT) 289 if (enc->json.flags & F_INDENT)
285 encode_nl (enc); 290 encode_nl (enc);
286 else if (enc->flags & F_SPACE_AFTER) 291 else if (enc->json.flags & F_SPACE_AFTER)
287 encode_space (enc); 292 encode_space (enc);
288} 293}
289 294
290static void encode_sv (enc_t *enc, SV *sv); 295static void encode_sv (enc_t *enc, SV *sv);
291 296
334 else 339 else
335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336 341
337 encode_ch (enc, '"'); 342 encode_ch (enc, '"');
338 343
339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 344 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340 encode_ch (enc, ':'); 345 encode_ch (enc, ':');
341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 346 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342 encode_sv (enc, HeVAL (he)); 347 encode_sv (enc, HeVAL (he));
343} 348}
344 349
345// compare hash entries, used when all keys are bytestrings 350// compare hash entries, used when all keys are bytestrings
346static int 351static int
381 { 386 {
382 // for canonical output we have to sort by keys first 387 // for canonical output we have to sort by keys first
383 // actually, this is mostly due to the stupid so-called 388 // actually, this is mostly due to the stupid so-called
384 // security workaround added somewhere in 5.8.x. 389 // security workaround added somewhere in 5.8.x.
385 // that randomises hash orderings 390 // that randomises hash orderings
386 if (enc->flags & F_CANONICAL) 391 if (enc->json.flags & F_CANONICAL)
387 { 392 {
388 int fast = 1; 393 int fast = 1;
389 HE *he; 394 HE *he;
390#if defined(__BORLANDC__) || defined(_MSC_VER) 395#if defined(__BORLANDC__) || defined(_MSC_VER)
391 HE **hes = _alloca (count * sizeof (HE)); 396 HE **hes = _alloca (count * sizeof (HE));
463 svtype svt; 468 svtype svt;
464 469
465 SvGETMAGIC (sv); 470 SvGETMAGIC (sv);
466 svt = SvTYPE (sv); 471 svt = SvTYPE (sv);
467 472
473 if (expect_false (SvOBJECT (sv)))
474 {
475 if (SvSTASH (sv) == json_boolean_stash)
476 {
477 if (SvIV (sv) == 0)
478 encode_str (enc, "false", 5, 0);
479 else
480 encode_str (enc, "true", 4, 0);
481 }
482 else
483 {
484#if 0
485 if (0 && sv_derived_from (rv, "JSON::Literal"))
486 {
487 // not yet
488 }
489#endif
490 if (enc->json.flags & F_CONV_BLESSED)
491 {
492 // we re-bless the reference to get overload and other niceties right
493 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
494
495 if (to_json)
496 {
497 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
498 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
499
500 // calling with G_SCALAR ensures that we always get a 1 reutrn value
501 // check anyways.
502 PUTBACK;
503 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
504 SPAGAIN;
505
506 encode_sv (enc, POPs);
507
508 FREETMPS; LEAVE;
509 }
510 else if (enc->json.flags & F_ALLOW_BLESSED)
511 encode_str (enc, "null", 4, 0);
512 else
513 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
514 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
515 }
516 else if (enc->json.flags & F_ALLOW_BLESSED)
517 encode_str (enc, "null", 4, 0);
518 else
519 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
520 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
521 }
522 }
468 if (svt == SVt_PVHV) 523 else if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv); 524 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV) 525 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv); 526 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV) 527 else if (svt < SVt_PVAV)
473 { 528 {
474 if (SvNIOK (sv) && SvIV (sv) == 0) 529 STRLEN len = 0;
530 char *pv = svt ? SvPV (sv, len) : 0;
531
532 if (len == 1 && *pv == '0')
475 encode_str (enc, "false", 5, 0); 533 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1) 534 else if (len == 1 && *pv == '1')
477 encode_str (enc, "true", 4, 0); 535 encode_str (enc, "true", 4, 0);
478 else 536 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 537 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 538 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 } 539 }
552 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 610 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
553 SvPV_nolen (sv), SvFLAGS (sv)); 611 SvPV_nolen (sv), SvFLAGS (sv));
554} 612}
555 613
556static SV * 614static SV *
557encode_json (SV *scalar, U32 flags) 615encode_json (SV *scalar, JSON *json)
558{ 616{
559 enc_t enc; 617 enc_t enc;
560 618
561 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 619 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
562 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 620 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
563 621
564 enc.flags = flags; 622 enc.json = *json;
565 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 623 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
566 enc.cur = SvPVX (enc.sv); 624 enc.cur = SvPVX (enc.sv);
567 enc.end = SvEND (enc.sv); 625 enc.end = SvEND (enc.sv);
568 enc.indent = 0; 626 enc.indent = 0;
569 enc.maxdepth = DEC_DEPTH (flags); 627 enc.maxdepth = DEC_DEPTH (enc.json.flags);
570 628
571 SvPOK_only (enc.sv); 629 SvPOK_only (enc.sv);
572 encode_sv (&enc, scalar); 630 encode_sv (&enc, scalar);
573 631
574 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 632 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
575 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 633 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
576 634
577 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 635 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
578 SvUTF8_on (enc.sv); 636 SvUTF8_on (enc.sv);
579 637
580 if (enc.flags & F_SHRINK) 638 if (enc.json.flags & F_SHRINK)
581 shrink (enc.sv); 639 shrink (enc.sv);
582 640
583 return enc.sv; 641 return enc.sv;
584} 642}
585 643
590typedef struct 648typedef struct
591{ 649{
592 char *cur; // current parser pointer 650 char *cur; // current parser pointer
593 char *end; // end of input string 651 char *end; // end of input string
594 const char *err; // parse error, if != 0 652 const char *err; // parse error, if != 0
595 U32 flags; // F_* 653 JSON json;
596 U32 depth; // recursion depth 654 U32 depth; // recursion depth
597 U32 maxdepth; // recursion depth limit 655 U32 maxdepth; // recursion depth limit
598} dec_t; 656} dec_t;
599 657
600inline void 658inline void
631decode_4hex (dec_t *dec) 689decode_4hex (dec_t *dec)
632{ 690{
633 signed char d1, d2, d3, d4; 691 signed char d1, d2, d3, d4;
634 unsigned char *cur = (unsigned char *)dec->cur; 692 unsigned char *cur = (unsigned char *)dec->cur;
635 693
636 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); 694 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"); 695 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"); 696 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"); 697 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
640 698
641 dec->cur += 4; 699 dec->cur += 4;
642 700
643 return ((UV)d1) << 12 701 return ((UV)d1) << 12
644 | ((UV)d2) << 8 702 | ((UV)d2) << 8
867 { 925 {
868 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 926 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
869 if (*start == '-') 927 if (*start == '-')
870 switch (dec->cur - start) 928 switch (dec->cur - start)
871 { 929 {
872 case 2: return newSViv (-( start [1] - '0' )); 930 case 2: return newSViv (-( start [1] - '0' * 1));
873 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 931 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)); 932 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)); 933 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
876 } 934 }
877 else 935 else
878 switch (dec->cur - start) 936 switch (dec->cur - start)
879 { 937 {
880 case 1: return newSViv ( start [0] - '0' ); 938 case 1: return newSViv ( start [0] - '0' * 1);
881 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 939 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); 940 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); 941 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
884 } 942 }
885 943
892 if (uv < (UV)IV_MIN) 950 if (uv < (UV)IV_MIN)
893 return newSViv (-(IV)uv); 951 return newSViv (-(IV)uv);
894 } 952 }
895 else 953 else
896 return newSVuv (uv); 954 return newSVuv (uv);
955
956 // here would likely be the place for bigint support
897 } 957 }
898 } 958 }
899 959
960 // if we ever support bigint or bigfloat, this is the place for bigfloat
900 return newSVnv (Atof (start)); 961 return newSVnv (Atof (start));
901 962
902fail: 963fail:
903 return 0; 964 return 0;
904} 965}
948} 1009}
949 1010
950static SV * 1011static SV *
951decode_hv (dec_t *dec) 1012decode_hv (dec_t *dec)
952{ 1013{
1014 SV *sv;
953 HV *hv = newHV (); 1015 HV *hv = newHV ();
954 1016
955 DEC_INC_DEPTH; 1017 DEC_INC_DEPTH;
956 decode_ws (dec); 1018 decode_ws (dec);
957 1019
958 if (*dec->cur == '}') 1020 if (*dec->cur == '}')
959 ++dec->cur; 1021 ++dec->cur;
960 else 1022 else
961 for (;;) 1023 for (;;)
962 { 1024 {
963 SV *key, *value;
964
965 decode_ws (dec); EXPECT_CH ('"'); 1025 decode_ws (dec); EXPECT_CH ('"');
966 1026
967 key = decode_str (dec); 1027 // heuristic: assume that
968 if (!key) 1028 // a) decode_str + hv_store_ent are abysmally slow.
969 goto fail; 1029 // b) most hash keys are short, simple ascii text.
1030 // => try to "fast-match" such strings to avoid
1031 // the overhead of decode_str + hv_store_ent.
1032 {
1033 SV *value;
1034 char *p = dec->cur;
1035 char *e = p + 24; // only try up to 24 bytes
970 1036
971 decode_ws (dec); EXPECT_CH (':'); 1037 for (;;)
972
973 value = decode_sv (dec);
974 if (!value)
975 { 1038 {
1039 // the >= 0x80 is true on most architectures
1040 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1041 {
1042 // slow path, back up and use decode_str
1043 SV *key = decode_str (dec);
1044 if (!key)
1045 goto fail;
1046
1047 decode_ws (dec); EXPECT_CH (':');
1048
1049 value = decode_sv (dec);
1050 if (!value)
1051 {
1052 SvREFCNT_dec (key);
1053 goto fail;
1054 }
1055
1056 hv_store_ent (hv, key, value, 0);
976 SvREFCNT_dec (key); 1057 SvREFCNT_dec (key);
1058
1059 break;
1060 }
1061 else if (*p == '"')
1062 {
1063 // fast path, got a simple key
1064 char *key = dec->cur;
1065 int len = p - key;
1066 dec->cur = p + 1;
1067
1068 decode_ws (dec); EXPECT_CH (':');
1069
1070 value = decode_sv (dec);
1071 if (!value)
977 goto fail; 1072 goto fail;
1073
1074 hv_store (hv, key, len, value, 0);
1075
1076 break;
1077 }
1078
1079 ++p;
978 } 1080 }
979 1081 }
980 hv_store_ent (hv, key, value, 0);
981 SvREFCNT_dec (key);
982 1082
983 decode_ws (dec); 1083 decode_ws (dec);
984 1084
985 if (*dec->cur == '}') 1085 if (*dec->cur == '}')
986 { 1086 {
993 1093
994 ++dec->cur; 1094 ++dec->cur;
995 } 1095 }
996 1096
997 DEC_DEC_DEPTH; 1097 DEC_DEC_DEPTH;
1098 sv = newRV_noinc ((SV *)hv);
1099
1100 // check filter callbacks
1101 if (dec->json.flags & F_HOOK)
1102 {
1103 ENTER; SAVETMPS;
1104
1105 if (HvKEYS (hv) == 1 && dec->json.cb_sk_object)
1106 {
1107 int count;
1108
1109 dSP; PUSHMARK (SP);
1110 XPUSHs (sv_2mortal (sv));
1111
1112 PUTBACK; count = call_sv (dec->json.cb_sk_object, G_ARRAY); SPAGAIN;
1113
1114 if (count == 1)
1115 sv = newSVsv (POPs);
1116 else
1117 SvREFCNT_inc (sv);
1118 }
1119
1120 if (dec->json.cb_object)
1121 {
1122 int count;
1123
1124 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
1125 XPUSHs (sv_2mortal (sv));
1126
1127 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1128
1129 if (count == 1)
1130 sv = newSVsv (POPs);
1131 else
1132 SvREFCNT_inc (sv);
1133 }
1134
1135 FREETMPS; LEAVE;
1136 }
1137
998 return newRV_noinc ((SV *)hv); 1138 return newRV_noinc ((SV *)hv);
999 1139
1000fail: 1140fail:
1001 SvREFCNT_dec (hv); 1141 SvREFCNT_dec (hv);
1002 DEC_DEC_DEPTH; 1142 DEC_DEC_DEPTH;
1005 1145
1006static SV * 1146static SV *
1007decode_sv (dec_t *dec) 1147decode_sv (dec_t *dec)
1008{ 1148{
1009 decode_ws (dec); 1149 decode_ws (dec);
1150
1151 // the beauty of JSON: you need exactly one character lookahead
1152 // to parse anything.
1010 switch (*dec->cur) 1153 switch (*dec->cur)
1011 { 1154 {
1012 case '"': ++dec->cur; return decode_str (dec); 1155 case '"': ++dec->cur; return decode_str (dec);
1013 case '[': ++dec->cur; return decode_av (dec); 1156 case '[': ++dec->cur; return decode_av (dec);
1014 case '{': ++dec->cur; return decode_hv (dec); 1157 case '{': ++dec->cur; return decode_hv (dec);
1020 1163
1021 case 't': 1164 case 't':
1022 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1165 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1023 { 1166 {
1024 dec->cur += 4; 1167 dec->cur += 4;
1025 return newSViv (1); 1168 return SvREFCNT_inc (json_true);
1026 } 1169 }
1027 else 1170 else
1028 ERR ("'true' expected"); 1171 ERR ("'true' expected");
1029 1172
1030 break; 1173 break;
1031 1174
1032 case 'f': 1175 case 'f':
1033 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1176 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1034 { 1177 {
1035 dec->cur += 5; 1178 dec->cur += 5;
1036 return newSViv (0); 1179 return SvREFCNT_inc (json_false);
1037 } 1180 }
1038 else 1181 else
1039 ERR ("'false' expected"); 1182 ERR ("'false' expected");
1040 1183
1041 break; 1184 break;
1059fail: 1202fail:
1060 return 0; 1203 return 0;
1061} 1204}
1062 1205
1063static SV * 1206static SV *
1064decode_json (SV *string, U32 flags, UV *offset_return) 1207decode_json (SV *string, JSON *json, UV *offset_return)
1065{ 1208{
1066 dec_t dec; 1209 dec_t dec;
1067 UV offset; 1210 UV offset;
1068 SV *sv; 1211 SV *sv;
1069 1212
1070 SvGETMAGIC (string); 1213 SvGETMAGIC (string);
1071 SvUPGRADE (string, SVt_PV); 1214 SvUPGRADE (string, SVt_PV);
1072 1215
1216 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1217 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1218 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1219
1073 if (flags & F_UTF8) 1220 if (json->flags & F_UTF8)
1074 sv_utf8_downgrade (string, 0); 1221 sv_utf8_downgrade (string, 0);
1075 else 1222 else
1076 sv_utf8_upgrade (string); 1223 sv_utf8_upgrade (string);
1077 1224
1078 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1225 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1079 1226
1080 dec.flags = flags; 1227 dec.json = *json;
1081 dec.cur = SvPVX (string); 1228 dec.cur = SvPVX (string);
1082 dec.end = SvEND (string); 1229 dec.end = SvEND (string);
1083 dec.err = 0; 1230 dec.err = 0;
1084 dec.depth = 0; 1231 dec.depth = 0;
1085 dec.maxdepth = DEC_DEPTH (dec.flags); 1232 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1233
1234 if (dec.json.cb_object || dec.json.cb_sk_object)
1235 dec.json.flags |= F_HOOK;
1086 1236
1087 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1237 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1088 sv = decode_sv (&dec); 1238 sv = decode_sv (&dec);
1089 1239
1090 if (!(offset_return || !sv)) 1240 if (!(offset_return || !sv))
1100 } 1250 }
1101 } 1251 }
1102 1252
1103 if (offset_return || !sv) 1253 if (offset_return || !sv)
1104 { 1254 {
1105 offset = dec.flags & F_UTF8 1255 offset = dec.json.flags & F_UTF8
1106 ? dec.cur - SvPVX (string) 1256 ? dec.cur - SvPVX (string)
1107 : utf8_distance (dec.cur, SvPVX (string)); 1257 : utf8_distance (dec.cur, SvPVX (string));
1108 1258
1109 if (offset_return) 1259 if (offset_return)
1110 *offset_return = offset; 1260 *offset_return = offset;
1129 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1279 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1130 } 1280 }
1131 1281
1132 sv = sv_2mortal (sv); 1282 sv = sv_2mortal (sv);
1133 1283
1134 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1284 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)"); 1285 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1136 1286
1137 return sv; 1287 return sv;
1138} 1288}
1139 1289
1143MODULE = JSON::XS PACKAGE = JSON::XS 1293MODULE = JSON::XS PACKAGE = JSON::XS
1144 1294
1145BOOT: 1295BOOT:
1146{ 1296{
1147 int i; 1297 int i;
1148
1149 memset (decode_hexdigit, 0xff, 256);
1150 1298
1151 for (i = 0; i < 256; ++i) 1299 for (i = 0; i < 256; ++i)
1152 decode_hexdigit [i] = 1300 decode_hexdigit [i] =
1153 i >= '0' && i <= '9' ? i - '0' 1301 i >= '0' && i <= '9' ? i - '0'
1154 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1302 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1155 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1303 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1156 : -1; 1304 : -1;
1157 1305
1158 json_stash = gv_stashpv ("JSON::XS", 1); 1306 json_stash = gv_stashpv ("JSON::XS" , 1);
1307 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1308
1309 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1310 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1159} 1311}
1160 1312
1161PROTOTYPES: DISABLE 1313PROTOTYPES: DISABLE
1162 1314
1163SV *new (char *dummy) 1315void new (char *klass)
1164 CODE: 1316 PPCODE:
1165 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1317{
1166 OUTPUT: 1318 SV *pv = NEWSV (0, sizeof (JSON));
1167 RETVAL 1319 SvPOK_only (pv);
1320 Zero (SvPVX (pv), 1, JSON);
1321 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1322 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1323}
1168 1324
1169SV *ascii (SV *self, int enable = 1) 1325void ascii (JSON *self, int enable = 1)
1170 ALIAS: 1326 ALIAS:
1171 ascii = F_ASCII 1327 ascii = F_ASCII
1172 latin1 = F_LATIN1 1328 latin1 = F_LATIN1
1173 utf8 = F_UTF8 1329 utf8 = F_UTF8
1174 indent = F_INDENT 1330 indent = F_INDENT
1175 canonical = F_CANONICAL 1331 canonical = F_CANONICAL
1176 space_before = F_SPACE_BEFORE 1332 space_before = F_SPACE_BEFORE
1177 space_after = F_SPACE_AFTER 1333 space_after = F_SPACE_AFTER
1178 pretty = F_PRETTY 1334 pretty = F_PRETTY
1179 allow_nonref = F_ALLOW_NONREF 1335 allow_nonref = F_ALLOW_NONREF
1180 shrink = F_SHRINK 1336 shrink = F_SHRINK
1337 allow_blessed = F_ALLOW_BLESSED
1338 convert_blessed = F_CONV_BLESSED
1181 CODE: 1339 PPCODE:
1182{ 1340{
1183 UV *uv = SvJSON (self);
1184 if (enable) 1341 if (enable)
1185 *uv |= ix; 1342 self->flags |= ix;
1186 else 1343 else
1187 *uv &= ~ix; 1344 self->flags &= ~ix;
1188 1345
1189 RETVAL = newSVsv (self); 1346 XPUSHs (ST (0));
1190} 1347}
1191 OUTPUT:
1192 RETVAL
1193 1348
1194SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1349void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1195 CODE: 1350 PPCODE:
1196{ 1351{
1197 UV *uv = SvJSON (self);
1198 UV log2 = 0; 1352 UV log2 = 0;
1199 1353
1200 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1354 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1201 1355
1202 while ((1UL << log2) < max_depth) 1356 while ((1UL << log2) < max_depth)
1203 ++log2; 1357 ++log2;
1204 1358
1205 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1359 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1206 1360
1207 RETVAL = newSVsv (self); 1361 XPUSHs (ST (0));
1208} 1362}
1209 OUTPUT:
1210 RETVAL
1211 1363
1212void encode (SV *self, SV *scalar) 1364void max_size (JSON *self, UV max_size = 0)
1213 PPCODE: 1365 PPCODE:
1214 XPUSHs (encode_json (scalar, *SvJSON (self))); 1366{
1367 UV log2 = 0;
1215 1368
1216void decode (SV *self, SV *jsonstr) 1369 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1370 if (max_size == 1) max_size = 2;
1371
1372 while ((1UL << log2) < max_size)
1373 ++log2;
1374
1375 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1376
1377 XPUSHs (ST (0));
1378}
1379
1380void filter_json_objects (JSON *self, SV *cb = &PL_sv_undef)
1381 ALIAS:
1382 filter_sk_json_objects = 1
1217 PPCODE: 1383 PPCODE:
1384{
1385 if (!SvOK (cb))
1386 cb = 0;
1387
1388 switch (ix)
1389 {
1390 case 0: self->cb_object = cb; break;
1391 case 1: self->cb_sk_object = cb; break;
1392 }
1393
1394 XPUSHs (ST (0));
1395}
1396
1397void encode (JSON *self, SV *scalar)
1398 PPCODE:
1399 XPUSHs (encode_json (scalar, self));
1400
1401void decode (JSON *self, SV *jsonstr)
1402 PPCODE:
1218 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1403 XPUSHs (decode_json (jsonstr, self, 0));
1219 1404
1220void decode_prefix (SV *self, SV *jsonstr) 1405void decode_prefix (JSON *self, SV *jsonstr)
1221 PPCODE: 1406 PPCODE:
1222{ 1407{
1223 UV offset; 1408 UV offset;
1224 EXTEND (SP, 2); 1409 EXTEND (SP, 2);
1225 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 1410 PUSHs (decode_json (jsonstr, self, &offset));
1226 PUSHs (sv_2mortal (newSVuv (offset))); 1411 PUSHs (sv_2mortal (newSVuv (offset)));
1227} 1412}
1228 1413
1229PROTOTYPES: ENABLE 1414PROTOTYPES: ENABLE
1230 1415
1231void to_json (SV *scalar) 1416void to_json (SV *scalar)
1232 ALIAS:
1233 objToJson = 0
1234 PPCODE: 1417 PPCODE:
1418{
1419 JSON json = { F_DEFAULT | F_UTF8 };
1235 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1420 XPUSHs (encode_json (scalar, &json));
1421}
1236 1422
1237void from_json (SV *jsonstr) 1423void from_json (SV *jsonstr)
1238 ALIAS:
1239 jsonToObj = 0
1240 PPCODE: 1424 PPCODE:
1425{
1426 JSON json = { F_DEFAULT | F_UTF8 };
1241 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); 1427 XPUSHs (decode_json (jsonstr, &json, 0));
1428}
1242 1429

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines