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 1.79 by root, Tue Mar 25 23:00:31 2008 UTC vs.
Revision 1.102 by root, Sat Oct 10 01:48:50 2009 UTC

12#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
13# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
14#endif 14#endif
15 15
16// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
17// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
20#endif 20#endif
21 21
22// three extra for rounding, sign, and end of string
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) 23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
23 24
24#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
25#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
26#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
27#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
31#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
32#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
33#define F_ALLOW_BLESSED 0x00000400UL 34#define F_ALLOW_BLESSED 0x00000400UL
34#define F_CONV_BLESSED 0x00000800UL 35#define F_CONV_BLESSED 0x00000800UL
35#define F_RELAXED 0x00001000UL 36#define F_RELAXED 0x00001000UL
36 37#define F_ALLOW_UNKNOWN 0x00002000UL
37#define F_MAXDEPTH 0xf8000000UL
38#define S_MAXDEPTH 27
39#define F_MAXSIZE 0x01f00000UL
40#define S_MAXSIZE 20
41#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
42 39
43#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
44#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
45
46#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
47#define F_DEFAULT (9UL << S_MAXDEPTH)
48 41
49#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
50#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
51 44
52#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
67 60
68#define IN_RANGE_INC(type,val,beg,end) \ 61#define IN_RANGE_INC(type,val,beg,end) \
69 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 62 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
70 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 63 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
71 64
65#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
66
72#ifdef USE_ITHREADS 67#ifdef USE_ITHREADS
73# define JSON_SLOW 1 68# define JSON_SLOW 1
74# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
75#else 70#else
76# define JSON_SLOW 0 71# define JSON_SLOW 0
85 INCR_M_STR, // inside string 80 INCR_M_STR, // inside string
86 INCR_M_BS, // inside backslash 81 INCR_M_BS, // inside backslash
87 INCR_M_JSON // outside anything, count nesting 82 INCR_M_JSON // outside anything, count nesting
88}; 83};
89 84
90#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON) 85#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
91 86
92typedef struct { 87typedef struct {
93 U32 flags; 88 U32 flags;
89 U32 max_depth;
90 STRLEN max_size;
91
94 SV *cb_object; 92 SV *cb_object;
95 HV *cb_sk_object; 93 HV *cb_sk_object;
96 94
97 // for the incremental parser 95 // for the incremental parser
98 SV *incr_text; // the source text so far 96 SV *incr_text; // the source text so far
99 STRLEN incr_pos; // the current offset into the text 97 STRLEN incr_pos; // the current offset into the text
100 int incr_nest; // {[]}-nesting level 98 int incr_nest; // {[]}-nesting level
101 int incr_mode; 99 unsigned char incr_mode;
102} JSON; 100} JSON;
101
102INLINE void
103json_init (JSON *json)
104{
105 Zero (json, 1, JSON);
106 json->max_depth = 512;
107}
103 108
104///////////////////////////////////////////////////////////////////////////// 109/////////////////////////////////////////////////////////////////////////////
105// utility functions 110// utility functions
106 111
112INLINE SV *
113get_bool (const char *name)
114{
115 SV *sv = get_sv (name, 1);
116
117 SvREADONLY_on (sv);
118 SvREADONLY_on (SvRV (sv));
119
120 return sv;
121}
122
107INLINE void 123INLINE void
108shrink (SV *sv) 124shrink (SV *sv)
109{ 125{
110 sv_utf8_downgrade (sv, 1); 126 sv_utf8_downgrade (sv, 1);
127
111 if (SvLEN (sv) > SvCUR (sv) + 1) 128 if (SvLEN (sv) > SvCUR (sv) + 1)
112 { 129 {
113#ifdef SvPV_shrink_to_cur 130#ifdef SvPV_shrink_to_cur
114 SvPV_shrink_to_cur (sv); 131 SvPV_shrink_to_cur (sv);
115#elif defined (SvPV_renew) 132#elif defined (SvPV_renew)
159 *s++ = 0x80 | ( ch & 0x3f); 176 *s++ = 0x80 | ( ch & 0x3f);
160 177
161 return s; 178 return s;
162} 179}
163 180
181// convert offset pointer to character index, sv must be string
182static STRLEN
183ptr_to_index (SV *sv, char *offset)
184{
185 return SvUTF8 (sv)
186 ? utf8_distance (offset, SvPVX (sv))
187 : offset - SvPVX (sv);
188}
189
164///////////////////////////////////////////////////////////////////////////// 190/////////////////////////////////////////////////////////////////////////////
165// encoder 191// encoder
166 192
167// structure used for encoding JSON 193// structure used for encoding JSON
168typedef struct 194typedef struct
170 char *cur; // SvPVX (sv) + current output position 196 char *cur; // SvPVX (sv) + current output position
171 char *end; // SvEND (sv) 197 char *end; // SvEND (sv)
172 SV *sv; // result scalar 198 SV *sv; // result scalar
173 JSON json; 199 JSON json;
174 U32 indent; // indentation level 200 U32 indent; // indentation level
175 U32 maxdepth; // max. indentation/recursion level
176 UV limit; // escape character values >= this value when encoding 201 UV limit; // escape character values >= this value when encoding
177} enc_t; 202} enc_t;
178 203
179INLINE void 204INLINE void
180need (enc_t *enc, STRLEN len) 205need (enc_t *enc, STRLEN len)
181{ 206{
182 if (expect_false (enc->cur + len >= enc->end)) 207 if (expect_false (enc->cur + len >= enc->end))
183 { 208 {
184 STRLEN cur = enc->cur - SvPVX (enc->sv); 209 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
185 SvGROW (enc->sv, cur + len + 1); 210 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
186 enc->cur = SvPVX (enc->sv) + cur; 211 enc->cur = SvPVX (enc->sv) + cur;
187 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 212 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
188 } 213 }
189} 214}
190 215
265 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 290 (int)((uch - 0x10000) % 0x400 + 0xDC00));
266 enc->cur += 12; 291 enc->cur += 12;
267 } 292 }
268 else 293 else
269 { 294 {
270 static char hexdigit [16] = "0123456789abcdef";
271 need (enc, len += 5); 295 need (enc, len += 5);
272 *enc->cur++ = '\\'; 296 *enc->cur++ = '\\';
273 *enc->cur++ = 'u'; 297 *enc->cur++ = 'u';
274 *enc->cur++ = hexdigit [ uch >> 12 ]; 298 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
275 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 299 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
276 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 300 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
277 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 301 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
278 } 302 }
279 303
280 str += clen; 304 str += clen;
281 } 305 }
282 else if (enc->json.flags & F_LATIN1) 306 else if (enc->json.flags & F_LATIN1)
353static void 377static void
354encode_av (enc_t *enc, AV *av) 378encode_av (enc_t *enc, AV *av)
355{ 379{
356 int i, len = av_len (av); 380 int i, len = av_len (av);
357 381
358 if (enc->indent >= enc->maxdepth) 382 if (enc->indent >= enc->json.max_depth)
359 croak ("data structure too deep (hit recursion limit)"); 383 croak (ERR_NESTING_EXCEEDED);
360 384
361 encode_ch (enc, '['); 385 encode_ch (enc, '[');
362 386
363 if (len >= 0) 387 if (len >= 0)
364 { 388 {
439static void 463static void
440encode_hv (enc_t *enc, HV *hv) 464encode_hv (enc_t *enc, HV *hv)
441{ 465{
442 HE *he; 466 HE *he;
443 467
444 if (enc->indent >= enc->maxdepth) 468 if (enc->indent >= enc->json.max_depth)
445 croak ("data structure too deep (hit recursion limit)"); 469 croak (ERR_NESTING_EXCEEDED);
446 470
447 encode_ch (enc, '{'); 471 encode_ch (enc, '{');
448 472
449 // for canonical output we have to sort by keys first 473 // for canonical output we have to sort by keys first
450 // actually, this is mostly due to the stupid so-called 474 // actually, this is mostly due to the stupid so-called
451 // security workaround added somewhere in 5.8.x. 475 // security workaround added somewhere in 5.8.x
452 // that randomises hash orderings 476 // that randomises hash orderings
453 if (enc->json.flags & F_CANONICAL) 477 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
454 { 478 {
455 int count = hv_iterinit (hv); 479 int count = hv_iterinit (hv);
456 480
457 if (SvMAGICAL (hv)) 481 if (SvMAGICAL (hv))
458 { 482 {
631 655
632 if (len == 1 && *pv == '1') 656 if (len == 1 && *pv == '1')
633 encode_str (enc, "true", 4, 0); 657 encode_str (enc, "true", 4, 0);
634 else if (len == 1 && *pv == '0') 658 else if (len == 1 && *pv == '0')
635 encode_str (enc, "false", 5, 0); 659 encode_str (enc, "false", 5, 0);
660 else if (enc->json.flags & F_ALLOW_UNKNOWN)
661 encode_str (enc, "null", 4, 0);
636 else 662 else
637 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 663 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
638 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 664 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
639 } 665 }
666 else if (enc->json.flags & F_ALLOW_UNKNOWN)
667 encode_str (enc, "null", 4, 0);
640 else 668 else
641 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 669 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
642 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 670 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
643} 671}
644 672
706 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 734 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
707 } 735 }
708 } 736 }
709 else if (SvROK (sv)) 737 else if (SvROK (sv))
710 encode_rv (enc, SvRV (sv)); 738 encode_rv (enc, SvRV (sv));
711 else if (!SvOK (sv)) 739 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
712 encode_str (enc, "null", 4, 0); 740 encode_str (enc, "null", 4, 0);
713 else 741 else
714 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 742 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
715 SvPV_nolen (sv), SvFLAGS (sv)); 743 SvPV_nolen (sv), SvFLAGS (sv));
716} 744}
726 enc.json = *json; 754 enc.json = *json;
727 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 755 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
728 enc.cur = SvPVX (enc.sv); 756 enc.cur = SvPVX (enc.sv);
729 enc.end = SvEND (enc.sv); 757 enc.end = SvEND (enc.sv);
730 enc.indent = 0; 758 enc.indent = 0;
731 enc.maxdepth = DEC_DEPTH (enc.json.flags);
732 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL 759 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
733 : enc.json.flags & F_LATIN1 ? 0x000100UL 760 : enc.json.flags & F_LATIN1 ? 0x000100UL
734 : 0x110000UL; 761 : 0x110000UL;
735 762
736 SvPOK_only (enc.sv); 763 SvPOK_only (enc.sv);
737 encode_sv (&enc, scalar); 764 encode_sv (&enc, scalar);
765 encode_nl (&enc);
738 766
739 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 767 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
740 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 768 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
741 769
742 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 770 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
803 if (*dec->cur != ch) \ 831 if (*dec->cur != ch) \
804 ERR (# ch " expected"); \ 832 ERR (# ch " expected"); \
805 ++dec->cur; \ 833 ++dec->cur; \
806 SE 834 SE
807 835
808#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 836#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
809#define DEC_DEC_DEPTH --dec->depth 837#define DEC_DEC_DEPTH --dec->depth
810 838
811static SV *decode_sv (dec_t *dec); 839static SV *decode_sv (dec_t *dec);
812 840
813static signed char decode_hexdigit[256]; 841static signed char decode_hexdigit[256];
922 else if (expect_true (ch >= 0x20 && ch < 0x80)) 950 else if (expect_true (ch >= 0x20 && ch < 0x80))
923 *cur++ = ch; 951 *cur++ = ch;
924 else if (ch >= 0x80) 952 else if (ch >= 0x80)
925 { 953 {
926 STRLEN clen; 954 STRLEN clen;
927 UV uch;
928 955
929 --dec_cur; 956 --dec_cur;
930 957
931 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 958 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
932 if (clen == (STRLEN)-1) 959 if (clen == (STRLEN)-1)
933 ERR ("malformed UTF-8 character in JSON string"); 960 ERR ("malformed UTF-8 character in JSON string");
934 961
935 do 962 do
936 *cur++ = *dec_cur++; 963 *cur++ = *dec_cur++;
953 { 980 {
954 STRLEN len = cur - buf; 981 STRLEN len = cur - buf;
955 982
956 if (sv) 983 if (sv)
957 { 984 {
958 SvGROW (sv, SvCUR (sv) + len + 1); 985 STRLEN cur = SvCUR (sv);
986
987 if (SvLEN (sv) <= cur + len)
988 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
989
959 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 990 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
960 SvCUR_set (sv, SvCUR (sv) + len); 991 SvCUR_set (sv, SvCUR (sv) + len);
961 } 992 }
962 else 993 else
963 sv = newSVpvn (buf, len); 994 sv = newSVpvn (buf, len);
1054 1085
1055 // special case the rather common 1..5-digit-int case 1086 // special case the rather common 1..5-digit-int case
1056 if (*start == '-') 1087 if (*start == '-')
1057 switch (len) 1088 switch (len)
1058 { 1089 {
1059 case 2: return newSViv (-( start [1] - '0' * 1)); 1090 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1060 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1091 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1061 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1092 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1062 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1093 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1063 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); 1094 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1064 } 1095 }
1065 else 1096 else
1066 switch (len) 1097 switch (len)
1067 { 1098 {
1068 case 1: return newSViv ( start [0] - '0' * 1); 1099 case 1: return newSViv ( start [0] - '0' * 1);
1069 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1100 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1070 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1101 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1071 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1102 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1072 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); 1103 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1073 } 1104 }
1074 1105
1075 { 1106 {
1076 UV uv; 1107 UV uv;
1077 int numtype = grok_number (start, len, &uv); 1108 int numtype = grok_number (start, len, &uv);
1343 case 't': 1374 case 't':
1344 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1375 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1345 { 1376 {
1346 dec->cur += 4; 1377 dec->cur += 4;
1347#if JSON_SLOW 1378#if JSON_SLOW
1348 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1379 json_true = get_bool ("JSON::XS::true");
1349#endif 1380#endif
1350 return SvREFCNT_inc (json_true); 1381 return newSVsv (json_true);
1351 } 1382 }
1352 else 1383 else
1353 ERR ("'true' expected"); 1384 ERR ("'true' expected");
1354 1385
1355 break; 1386 break;
1357 case 'f': 1388 case 'f':
1358 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1389 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1359 { 1390 {
1360 dec->cur += 5; 1391 dec->cur += 5;
1361#if JSON_SLOW 1392#if JSON_SLOW
1362 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1393 json_false = get_bool ("JSON::XS::false");
1363#endif 1394#endif
1364 return SvREFCNT_inc (json_false); 1395 return newSVsv (json_false);
1365 } 1396 }
1366 else 1397 else
1367 ERR ("'false' expected"); 1398 ERR ("'false' expected");
1368 1399
1369 break; 1400 break;
1387fail: 1418fail:
1388 return 0; 1419 return 0;
1389} 1420}
1390 1421
1391static SV * 1422static SV *
1392decode_json (SV *string, JSON *json, STRLEN *offset_return) 1423decode_json (SV *string, JSON *json, char **offset_return)
1393{ 1424{
1394 dec_t dec; 1425 dec_t dec;
1395 STRLEN offset;
1396 SV *sv; 1426 SV *sv;
1397 1427
1428 /* work around bugs in 5.10 where manipulating magic values
1429 * will perl ignore the magic in subsequent accesses
1430 */
1398 SvGETMAGIC (string); 1431 /*SvGETMAGIC (string);*/
1432 if (SvMAGICAL (string))
1433 string = sv_2mortal (newSVsv (string));
1434
1399 SvUPGRADE (string, SVt_PV); 1435 SvUPGRADE (string, SVt_PV);
1400 1436
1401 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1437 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1438 * assertion with -DDEBUGGING, although SvCUR is documented to
1439 * return the xpv_cur field which certainly exists after upgrading.
1440 * according to nicholas clark, calling SvPOK fixes this.
1441 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1442 * and hope for the best.
1443 * Damnit, SvPV_nolen still trips over yet another assertion. This
1444 * assertion business is seriously broken, try yet another workaround
1445 * for the broken -DDEBUGGING.
1446 */
1447 {
1448#ifdef DEBUGGING
1449 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1450#else
1451 STRLEN offset = SvCUR (string);
1452#endif
1453
1454 if (offset > json->max_size && json->max_size)
1402 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1455 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1403 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1456 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1457 }
1404 1458
1405 if (json->flags & F_UTF8) 1459 if (json->flags & F_UTF8)
1406 sv_utf8_downgrade (string, 0); 1460 sv_utf8_downgrade (string, 0);
1407 else 1461 else
1408 sv_utf8_upgrade (string); 1462 sv_utf8_upgrade (string);
1409 1463
1410 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1464 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1411 1465
1412 dec.json = *json; 1466 dec.json = *json;
1413 dec.cur = SvPVX (string); 1467 dec.cur = SvPVX (string);
1414 dec.end = SvEND (string); 1468 dec.end = SvEND (string);
1415 dec.err = 0; 1469 dec.err = 0;
1416 dec.depth = 0; 1470 dec.depth = 0;
1417 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1418 1471
1419 if (dec.json.cb_object || dec.json.cb_sk_object) 1472 if (dec.json.cb_object || dec.json.cb_sk_object)
1420 dec.json.flags |= F_HOOK; 1473 dec.json.flags |= F_HOOK;
1421 1474
1422 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1475 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1423 1476
1424 decode_ws (&dec); 1477 decode_ws (&dec);
1425 sv = decode_sv (&dec); 1478 sv = decode_sv (&dec);
1479
1480 if (offset_return)
1481 *offset_return = dec.cur;
1426 1482
1427 if (!(offset_return || !sv)) 1483 if (!(offset_return || !sv))
1428 { 1484 {
1429 // check for trailing garbage 1485 // check for trailing garbage
1430 decode_ws (&dec); 1486 decode_ws (&dec);
1433 { 1489 {
1434 dec.err = "garbage after JSON object"; 1490 dec.err = "garbage after JSON object";
1435 SvREFCNT_dec (sv); 1491 SvREFCNT_dec (sv);
1436 sv = 0; 1492 sv = 0;
1437 } 1493 }
1438 }
1439
1440 if (offset_return || !sv)
1441 {
1442 offset = dec.json.flags & F_UTF8
1443 ? dec.cur - SvPVX (string)
1444 : utf8_distance (dec.cur, SvPVX (string));
1445
1446 if (offset_return)
1447 *offset_return = offset;
1448 } 1494 }
1449 1495
1450 if (!sv) 1496 if (!sv)
1451 { 1497 {
1452 SV *uni = sv_newmortal (); 1498 SV *uni = sv_newmortal ();
1458 SAVEVPTR (PL_curcop); 1504 SAVEVPTR (PL_curcop);
1459 PL_curcop = &cop; 1505 PL_curcop = &cop;
1460 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1506 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1461 LEAVE; 1507 LEAVE;
1462 1508
1463 croak ("%s, at character offset %d [\"%s\"]", 1509 croak ("%s, at character offset %d (before \"%s\")",
1464 dec.err, 1510 dec.err,
1465 (int)offset, 1511 ptr_to_index (string, dec.cur),
1466 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1512 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1467 } 1513 }
1468 1514
1469 sv = sv_2mortal (sv); 1515 sv = sv_2mortal (sv);
1470 1516
1568 self->incr_mode = INCR_M_STR; 1614 self->incr_mode = INCR_M_STR;
1569 goto incr_m_str; 1615 goto incr_m_str;
1570 1616
1571 case '[': 1617 case '[':
1572 case '{': 1618 case '{':
1573 ++self->incr_nest; 1619 if (++self->incr_nest > self->max_depth)
1620 croak (ERR_NESTING_EXCEEDED);
1574 break; 1621 break;
1575 1622
1576 case ']': 1623 case ']':
1577 case '}': 1624 case '}':
1578 if (!--self->incr_nest) 1625 if (--self->incr_nest <= 0)
1579 goto interrupt; 1626 goto interrupt;
1580 } 1627 }
1581 } 1628 }
1582 } 1629 }
1583 1630
1607 : -1; 1654 : -1;
1608 1655
1609 json_stash = gv_stashpv ("JSON::XS" , 1); 1656 json_stash = gv_stashpv ("JSON::XS" , 1);
1610 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1657 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1611 1658
1612 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1659 json_true = get_bool ("JSON::XS::true");
1613 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1660 json_false = get_bool ("JSON::XS::false");
1661
1662 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1614} 1663}
1615 1664
1616PROTOTYPES: DISABLE 1665PROTOTYPES: DISABLE
1617 1666
1618void CLONE (...) 1667void CLONE (...)
1623void new (char *klass) 1672void new (char *klass)
1624 PPCODE: 1673 PPCODE:
1625{ 1674{
1626 SV *pv = NEWSV (0, sizeof (JSON)); 1675 SV *pv = NEWSV (0, sizeof (JSON));
1627 SvPOK_only (pv); 1676 SvPOK_only (pv);
1628 Zero (SvPVX (pv), 1, JSON); 1677 json_init ((JSON *)SvPVX (pv));
1629 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1630 XPUSHs (sv_2mortal (sv_bless ( 1678 XPUSHs (sv_2mortal (sv_bless (
1631 newRV_noinc (pv), 1679 newRV_noinc (pv),
1632 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1680 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1633 ))); 1681 )));
1634} 1682}
1646 allow_nonref = F_ALLOW_NONREF 1694 allow_nonref = F_ALLOW_NONREF
1647 shrink = F_SHRINK 1695 shrink = F_SHRINK
1648 allow_blessed = F_ALLOW_BLESSED 1696 allow_blessed = F_ALLOW_BLESSED
1649 convert_blessed = F_CONV_BLESSED 1697 convert_blessed = F_CONV_BLESSED
1650 relaxed = F_RELAXED 1698 relaxed = F_RELAXED
1699 allow_unknown = F_ALLOW_UNKNOWN
1651 PPCODE: 1700 PPCODE:
1652{ 1701{
1653 if (enable) 1702 if (enable)
1654 self->flags |= ix; 1703 self->flags |= ix;
1655 else 1704 else
1670 get_allow_nonref = F_ALLOW_NONREF 1719 get_allow_nonref = F_ALLOW_NONREF
1671 get_shrink = F_SHRINK 1720 get_shrink = F_SHRINK
1672 get_allow_blessed = F_ALLOW_BLESSED 1721 get_allow_blessed = F_ALLOW_BLESSED
1673 get_convert_blessed = F_CONV_BLESSED 1722 get_convert_blessed = F_CONV_BLESSED
1674 get_relaxed = F_RELAXED 1723 get_relaxed = F_RELAXED
1724 get_allow_unknown = F_ALLOW_UNKNOWN
1675 PPCODE: 1725 PPCODE:
1676 XPUSHs (boolSV (self->flags & ix)); 1726 XPUSHs (boolSV (self->flags & ix));
1677 1727
1678void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1728void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1679 PPCODE: 1729 PPCODE:
1680{ 1730 self->max_depth = max_depth;
1681 UV log2 = 0;
1682
1683 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1684
1685 while ((1UL << log2) < max_depth)
1686 ++log2;
1687
1688 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1689
1690 XPUSHs (ST (0)); 1731 XPUSHs (ST (0));
1691}
1692 1732
1693U32 get_max_depth (JSON *self) 1733U32 get_max_depth (JSON *self)
1694 CODE: 1734 CODE:
1695 RETVAL = DEC_DEPTH (self->flags); 1735 RETVAL = self->max_depth;
1696 OUTPUT: 1736 OUTPUT:
1697 RETVAL 1737 RETVAL
1698 1738
1699void max_size (JSON *self, UV max_size = 0) 1739void max_size (JSON *self, U32 max_size = 0)
1700 PPCODE: 1740 PPCODE:
1701{ 1741 self->max_size = max_size;
1702 UV log2 = 0;
1703
1704 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1705 if (max_size == 1) max_size = 2;
1706
1707 while ((1UL << log2) < max_size)
1708 ++log2;
1709
1710 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1711
1712 XPUSHs (ST (0)); 1742 XPUSHs (ST (0));
1713}
1714 1743
1715int get_max_size (JSON *self) 1744int get_max_size (JSON *self)
1716 CODE: 1745 CODE:
1717 RETVAL = DEC_SIZE (self->flags); 1746 RETVAL = self->max_size;
1718 OUTPUT: 1747 OUTPUT:
1719 RETVAL 1748 RETVAL
1720 1749
1721void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1750void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1722 PPCODE: 1751 PPCODE:
1758 XPUSHs (decode_json (jsonstr, self, 0)); 1787 XPUSHs (decode_json (jsonstr, self, 0));
1759 1788
1760void decode_prefix (JSON *self, SV *jsonstr) 1789void decode_prefix (JSON *self, SV *jsonstr)
1761 PPCODE: 1790 PPCODE:
1762{ 1791{
1763 STRLEN offset; 1792 char *offset;
1764 EXTEND (SP, 2); 1793 EXTEND (SP, 2);
1765 PUSHs (decode_json (jsonstr, self, &offset)); 1794 PUSHs (decode_json (jsonstr, self, &offset));
1766 PUSHs (sv_2mortal (newSVuv (offset))); 1795 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1767} 1796}
1768 1797
1769void incr_parse (JSON *self, SV *jsonstr = 0) 1798void incr_parse (JSON *self, SV *jsonstr = 0)
1770 PPCODE: 1799 PPCODE:
1771{ 1800{
1773 self->incr_text = newSVpvn ("", 0); 1802 self->incr_text = newSVpvn ("", 0);
1774 1803
1775 // append data, if any 1804 // append data, if any
1776 if (jsonstr) 1805 if (jsonstr)
1777 { 1806 {
1778 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 1807 if (SvUTF8 (jsonstr))
1779 { 1808 {
1809 if (!SvUTF8 (self->incr_text))
1810 {
1780 /* utf-8-ness differs, need to upgrade */ 1811 /* utf-8-ness differs, need to upgrade */
1781 sv_utf8_upgrade (self->incr_text); 1812 sv_utf8_upgrade (self->incr_text);
1782 1813
1783 if (self->incr_pos) 1814 if (self->incr_pos)
1784 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) 1815 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1785 - (U8 *)SvPVX (self->incr_text); 1816 - (U8 *)SvPVX (self->incr_text);
1817 }
1786 } 1818 }
1819 else if (SvUTF8 (self->incr_text))
1820 sv_utf8_upgrade (jsonstr);
1787 1821
1788 { 1822 {
1789 STRLEN len; 1823 STRLEN len;
1790 const char *str = SvPV (jsonstr, len); 1824 const char *str = SvPV (jsonstr, len);
1791 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 1825 STRLEN cur = SvCUR (self->incr_text);
1826
1827 if (SvLEN (self->incr_text) <= cur + len)
1828 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1829
1792 Move (str, SvEND (self->incr_text), len, char); 1830 Move (str, SvEND (self->incr_text), len, char);
1793 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 1831 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1794 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 1832 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1795 } 1833 }
1796 } 1834 }
1797 1835
1798 if (GIMME_V != G_VOID) 1836 if (GIMME_V != G_VOID)
1799 do 1837 do
1800 { 1838 {
1801 STRLEN offset; 1839 char *offset;
1802
1803 incr_parse (self);
1804 1840
1805 if (!INCR_DONE (self)) 1841 if (!INCR_DONE (self))
1842 {
1843 incr_parse (self);
1844
1845 if (self->incr_pos > self->max_size && self->max_size)
1846 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1847 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1848
1849 if (!INCR_DONE (self))
1806 break; 1850 break;
1851 }
1807 1852
1808 XPUSHs (decode_json (self->incr_text, self, &offset)); 1853 XPUSHs (decode_json (self->incr_text, self, &offset));
1809 1854
1810 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1811 self->incr_pos -= offset; 1855 self->incr_pos -= offset - SvPVX (self->incr_text);
1812 self->incr_nest = 0; 1856 self->incr_nest = 0;
1813 self->incr_mode = 0; 1857 self->incr_mode = 0;
1858
1859 sv_chop (self->incr_text, offset);
1814 } 1860 }
1815 while (GIMME_V == G_ARRAY); 1861 while (GIMME_V == G_ARRAY);
1816} 1862}
1817 1863
1818SV *incr_text (JSON *self) 1864SV *incr_text (JSON *self)
1819 ATTRS: lvalue 1865 ATTRS: lvalue
1820 CODE: 1866 CODE:
1821{ 1867{
1822 if (self->incr_pos) 1868 if (self->incr_pos)
1823 croak ("incr_text can only be called after a successful incr_parse call in scalar context"); 1869 croak ("incr_text can not be called when the incremental parser already started parsing");
1824 1870
1825 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; 1871 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1826} 1872}
1827 OUTPUT: 1873 OUTPUT:
1828 RETVAL 1874 RETVAL
1829 1875
1830void incr_skip (JSON *self) 1876void incr_skip (JSON *self)
1831 CODE: 1877 CODE:
1832{ 1878{
1833 if (!self->incr_pos || !INCR_DONE (self)) 1879 if (self->incr_pos)
1834 croak ("incr_text can only be called after an unsuccessful incr_parse call in scalar context");//D 1880 {
1835
1836 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos); 1881 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1882 self->incr_pos = 0;
1883 self->incr_nest = 0;
1884 self->incr_mode = 0;
1885 }
1886}
1887
1888void incr_reset (JSON *self)
1889 CODE:
1890{
1891 SvREFCNT_dec (self->incr_text);
1892 self->incr_text = 0;
1837 self->incr_pos = 0; 1893 self->incr_pos = 0;
1838 self->incr_nest = 0; 1894 self->incr_nest = 0;
1839 self->incr_mode = 0; 1895 self->incr_mode = 0;
1840} 1896}
1841 1897
1851 ALIAS: 1907 ALIAS:
1852 to_json_ = 0 1908 to_json_ = 0
1853 encode_json = F_UTF8 1909 encode_json = F_UTF8
1854 PPCODE: 1910 PPCODE:
1855{ 1911{
1856 JSON json = { F_DEFAULT | ix }; 1912 JSON json;
1913 json_init (&json);
1914 json.flags |= ix;
1857 XPUSHs (encode_json (scalar, &json)); 1915 XPUSHs (encode_json (scalar, &json));
1858} 1916}
1859 1917
1860void decode_json (SV *jsonstr) 1918void decode_json (SV *jsonstr)
1861 ALIAS: 1919 ALIAS:
1862 from_json_ = 0 1920 from_json_ = 0
1863 decode_json = F_UTF8 1921 decode_json = F_UTF8
1864 PPCODE: 1922 PPCODE:
1865{ 1923{
1866 JSON json = { F_DEFAULT | ix }; 1924 JSON json;
1925 json_init (&json);
1926 json.flags |= ix;
1867 XPUSHs (decode_json (jsonstr, &json, 0)); 1927 XPUSHs (decode_json (jsonstr, &json, 0));
1868} 1928}
1869 1929
1870 1930

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines