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 1.81 by root, Wed Mar 26 01:40:43 2008 UTC vs.
Revision 1.98 by root, Fri Jul 17 14:33:45 2009 UTC

31#define F_ALLOW_NONREF 0x00000100UL 31#define F_ALLOW_NONREF 0x00000100UL
32#define F_SHRINK 0x00000200UL 32#define F_SHRINK 0x00000200UL
33#define F_ALLOW_BLESSED 0x00000400UL 33#define F_ALLOW_BLESSED 0x00000400UL
34#define F_CONV_BLESSED 0x00000800UL 34#define F_CONV_BLESSED 0x00000800UL
35#define F_RELAXED 0x00001000UL 35#define F_RELAXED 0x00001000UL
36 36#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 37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
42 38
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 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
47#define F_DEFAULT (9UL << S_MAXDEPTH)
48 40
49#define INIT_SIZE 32 // initial scalar size to be allocated 41#define INIT_SIZE 32 // initial scalar size to be allocated
50#define INDENT_STEP 3 // spaces per indentation level 42#define INDENT_STEP 3 // spaces per indentation level
51 43
52#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
67 59
68#define IN_RANGE_INC(type,val,beg,end) \ 60#define IN_RANGE_INC(type,val,beg,end) \
69 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 61 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
70 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 62 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
71 63
64#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
65
72#ifdef USE_ITHREADS 66#ifdef USE_ITHREADS
73# define JSON_SLOW 1 67# define JSON_SLOW 1
74# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 68# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
75#else 69#else
76# define JSON_SLOW 0 70# define JSON_SLOW 0
85 INCR_M_STR, // inside string 79 INCR_M_STR, // inside string
86 INCR_M_BS, // inside backslash 80 INCR_M_BS, // inside backslash
87 INCR_M_JSON // outside anything, count nesting 81 INCR_M_JSON // outside anything, count nesting
88}; 82};
89 83
90#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON) 84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
91 85
92typedef struct { 86typedef struct {
93 U32 flags; 87 U32 flags;
88 U32 max_depth;
89 STRLEN max_size;
90
94 SV *cb_object; 91 SV *cb_object;
95 HV *cb_sk_object; 92 HV *cb_sk_object;
96 93
97 // for the incremental parser 94 // for the incremental parser
98 SV *incr_text; // the source text so far 95 SV *incr_text; // the source text so far
99 STRLEN incr_pos; // the current offset into the text 96 STRLEN incr_pos; // the current offset into the text
100 int incr_nest; // {[]}-nesting level 97 int incr_nest; // {[]}-nesting level
101 int incr_mode; 98 unsigned char incr_mode;
102} JSON; 99} JSON;
100
101INLINE void
102json_init (JSON *json)
103{
104 Zero (json, 1, JSON);
105 json->max_depth = 512;
106}
103 107
104///////////////////////////////////////////////////////////////////////////// 108/////////////////////////////////////////////////////////////////////////////
105// utility functions 109// utility functions
106 110
111INLINE SV *
112get_bool (const char *name)
113{
114 SV *sv = get_sv (name, 1);
115
116 SvREADONLY_on (sv);
117 SvREADONLY_on (SvRV (sv));
118
119 return sv;
120}
121
107INLINE void 122INLINE void
108shrink (SV *sv) 123shrink (SV *sv)
109{ 124{
110 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
126
111 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
112 { 128 {
113#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
114 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
115#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
159 *s++ = 0x80 | ( ch & 0x3f); 175 *s++ = 0x80 | ( ch & 0x3f);
160 176
161 return s; 177 return s;
162} 178}
163 179
180// convert offset pointer to character index, sv must be string
181static STRLEN
182ptr_to_index (SV *sv, char *offset)
183{
184 return SvUTF8 (sv)
185 ? utf8_distance (offset, SvPVX (sv))
186 : offset - SvPVX (sv);
187}
188
164///////////////////////////////////////////////////////////////////////////// 189/////////////////////////////////////////////////////////////////////////////
165// encoder 190// encoder
166 191
167// structure used for encoding JSON 192// structure used for encoding JSON
168typedef struct 193typedef struct
170 char *cur; // SvPVX (sv) + current output position 195 char *cur; // SvPVX (sv) + current output position
171 char *end; // SvEND (sv) 196 char *end; // SvEND (sv)
172 SV *sv; // result scalar 197 SV *sv; // result scalar
173 JSON json; 198 JSON json;
174 U32 indent; // indentation level 199 U32 indent; // indentation level
175 U32 maxdepth; // max. indentation/recursion level
176 UV limit; // escape character values >= this value when encoding 200 UV limit; // escape character values >= this value when encoding
177} enc_t; 201} enc_t;
178 202
179INLINE void 203INLINE void
180need (enc_t *enc, STRLEN len) 204need (enc_t *enc, STRLEN len)
181{ 205{
182 if (expect_false (enc->cur + len >= enc->end)) 206 if (expect_false (enc->cur + len >= enc->end))
183 { 207 {
184 STRLEN cur = enc->cur - SvPVX (enc->sv); 208 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
185 SvGROW (enc->sv, cur + len + 1); 209 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
186 enc->cur = SvPVX (enc->sv) + cur; 210 enc->cur = SvPVX (enc->sv) + cur;
187 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 211 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
188 } 212 }
189} 213}
190 214
265 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 289 (int)((uch - 0x10000) % 0x400 + 0xDC00));
266 enc->cur += 12; 290 enc->cur += 12;
267 } 291 }
268 else 292 else
269 { 293 {
270 static char hexdigit [16] = "0123456789abcdef";
271 need (enc, len += 5); 294 need (enc, len += 5);
272 *enc->cur++ = '\\'; 295 *enc->cur++ = '\\';
273 *enc->cur++ = 'u'; 296 *enc->cur++ = 'u';
274 *enc->cur++ = hexdigit [ uch >> 12 ]; 297 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
275 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 298 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
276 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 299 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
277 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 300 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
278 } 301 }
279 302
280 str += clen; 303 str += clen;
281 } 304 }
282 else if (enc->json.flags & F_LATIN1) 305 else if (enc->json.flags & F_LATIN1)
353static void 376static void
354encode_av (enc_t *enc, AV *av) 377encode_av (enc_t *enc, AV *av)
355{ 378{
356 int i, len = av_len (av); 379 int i, len = av_len (av);
357 380
358 if (enc->indent >= enc->maxdepth) 381 if (enc->indent >= enc->json.max_depth)
359 croak ("data structure too deep (hit recursion limit)"); 382 croak (ERR_NESTING_EXCEEDED);
360 383
361 encode_ch (enc, '['); 384 encode_ch (enc, '[');
362 385
363 if (len >= 0) 386 if (len >= 0)
364 { 387 {
439static void 462static void
440encode_hv (enc_t *enc, HV *hv) 463encode_hv (enc_t *enc, HV *hv)
441{ 464{
442 HE *he; 465 HE *he;
443 466
444 if (enc->indent >= enc->maxdepth) 467 if (enc->indent >= enc->json.max_depth)
445 croak ("data structure too deep (hit recursion limit)"); 468 croak (ERR_NESTING_EXCEEDED);
446 469
447 encode_ch (enc, '{'); 470 encode_ch (enc, '{');
448 471
449 // for canonical output we have to sort by keys first 472 // for canonical output we have to sort by keys first
450 // actually, this is mostly due to the stupid so-called 473 // actually, this is mostly due to the stupid so-called
451 // security workaround added somewhere in 5.8.x. 474 // security workaround added somewhere in 5.8.x
452 // that randomises hash orderings 475 // that randomises hash orderings
453 if (enc->json.flags & F_CANONICAL) 476 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
454 { 477 {
455 int count = hv_iterinit (hv); 478 int count = hv_iterinit (hv);
456 479
457 if (SvMAGICAL (hv)) 480 if (SvMAGICAL (hv))
458 { 481 {
631 654
632 if (len == 1 && *pv == '1') 655 if (len == 1 && *pv == '1')
633 encode_str (enc, "true", 4, 0); 656 encode_str (enc, "true", 4, 0);
634 else if (len == 1 && *pv == '0') 657 else if (len == 1 && *pv == '0')
635 encode_str (enc, "false", 5, 0); 658 encode_str (enc, "false", 5, 0);
659 else if (enc->json.flags & F_ALLOW_UNKNOWN)
660 encode_str (enc, "null", 4, 0);
636 else 661 else
637 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 662 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
638 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 663 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
639 } 664 }
665 else if (enc->json.flags & F_ALLOW_UNKNOWN)
666 encode_str (enc, "null", 4, 0);
640 else 667 else
641 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 668 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
642 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 669 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
643} 670}
644 671
706 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 733 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
707 } 734 }
708 } 735 }
709 else if (SvROK (sv)) 736 else if (SvROK (sv))
710 encode_rv (enc, SvRV (sv)); 737 encode_rv (enc, SvRV (sv));
711 else if (!SvOK (sv)) 738 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
712 encode_str (enc, "null", 4, 0); 739 encode_str (enc, "null", 4, 0);
713 else 740 else
714 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 741 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
715 SvPV_nolen (sv), SvFLAGS (sv)); 742 SvPV_nolen (sv), SvFLAGS (sv));
716} 743}
726 enc.json = *json; 753 enc.json = *json;
727 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 754 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
728 enc.cur = SvPVX (enc.sv); 755 enc.cur = SvPVX (enc.sv);
729 enc.end = SvEND (enc.sv); 756 enc.end = SvEND (enc.sv);
730 enc.indent = 0; 757 enc.indent = 0;
731 enc.maxdepth = DEC_DEPTH (enc.json.flags);
732 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL 758 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
733 : enc.json.flags & F_LATIN1 ? 0x000100UL 759 : enc.json.flags & F_LATIN1 ? 0x000100UL
734 : 0x110000UL; 760 : 0x110000UL;
735 761
736 SvPOK_only (enc.sv); 762 SvPOK_only (enc.sv);
803 if (*dec->cur != ch) \ 829 if (*dec->cur != ch) \
804 ERR (# ch " expected"); \ 830 ERR (# ch " expected"); \
805 ++dec->cur; \ 831 ++dec->cur; \
806 SE 832 SE
807 833
808#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 834#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
809#define DEC_DEC_DEPTH --dec->depth 835#define DEC_DEC_DEPTH --dec->depth
810 836
811static SV *decode_sv (dec_t *dec); 837static SV *decode_sv (dec_t *dec);
812 838
813static signed char decode_hexdigit[256]; 839static signed char decode_hexdigit[256];
953 { 979 {
954 STRLEN len = cur - buf; 980 STRLEN len = cur - buf;
955 981
956 if (sv) 982 if (sv)
957 { 983 {
958 SvGROW (sv, SvCUR (sv) + len + 1); 984 STRLEN cur = SvCUR (sv);
985
986 if (SvLEN (sv) <= cur + len)
987 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
988
959 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 989 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
960 SvCUR_set (sv, SvCUR (sv) + len); 990 SvCUR_set (sv, SvCUR (sv) + len);
961 } 991 }
962 else 992 else
963 sv = newSVpvn (buf, len); 993 sv = newSVpvn (buf, len);
1054 1084
1055 // special case the rather common 1..5-digit-int case 1085 // special case the rather common 1..5-digit-int case
1056 if (*start == '-') 1086 if (*start == '-')
1057 switch (len) 1087 switch (len)
1058 { 1088 {
1059 case 2: return newSViv (-( start [1] - '0' * 1)); 1089 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1060 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1090 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)); 1091 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)); 1092 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)); 1093 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1064 } 1094 }
1065 else 1095 else
1066 switch (len) 1096 switch (len)
1067 { 1097 {
1068 case 1: return newSViv ( start [0] - '0' * 1); 1098 case 1: return newSViv ( start [0] - '0' * 1);
1069 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1099 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); 1100 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); 1101 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); 1102 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1073 } 1103 }
1074 1104
1075 { 1105 {
1076 UV uv; 1106 UV uv;
1077 int numtype = grok_number (start, len, &uv); 1107 int numtype = grok_number (start, len, &uv);
1343 case 't': 1373 case 't':
1344 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1374 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1345 { 1375 {
1346 dec->cur += 4; 1376 dec->cur += 4;
1347#if JSON_SLOW 1377#if JSON_SLOW
1348 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1378 json_true = get_bool ("JSON::XS::true");
1349#endif 1379#endif
1350 return SvREFCNT_inc (json_true); 1380 return newSVsv (json_true);
1351 } 1381 }
1352 else 1382 else
1353 ERR ("'true' expected"); 1383 ERR ("'true' expected");
1354 1384
1355 break; 1385 break;
1357 case 'f': 1387 case 'f':
1358 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1388 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1359 { 1389 {
1360 dec->cur += 5; 1390 dec->cur += 5;
1361#if JSON_SLOW 1391#if JSON_SLOW
1362 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1392 json_false = get_bool ("JSON::XS::false");
1363#endif 1393#endif
1364 return SvREFCNT_inc (json_false); 1394 return newSVsv (json_false);
1365 } 1395 }
1366 else 1396 else
1367 ERR ("'false' expected"); 1397 ERR ("'false' expected");
1368 1398
1369 break; 1399 break;
1387fail: 1417fail:
1388 return 0; 1418 return 0;
1389} 1419}
1390 1420
1391static SV * 1421static SV *
1392decode_json (SV *string, JSON *json, STRLEN *offset_return) 1422decode_json (SV *string, JSON *json, char **offset_return)
1393{ 1423{
1394 dec_t dec; 1424 dec_t dec;
1395 STRLEN offset;
1396 SV *sv; 1425 SV *sv;
1397 1426
1427 /* work around bugs in 5.10 where manipulating magic values
1428 * will perl ignore the magic in subsequent accesses
1429 */
1398 SvGETMAGIC (string); 1430 /*SvGETMAGIC (string);*/
1431 if (SvMAGICAL (string))
1432 string = sv_2mortal (newSVsv (string));
1433
1399 SvUPGRADE (string, SVt_PV); 1434 SvUPGRADE (string, SVt_PV);
1400 1435
1401 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1436 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1437 * assertion with -DDEBUGGING, although SvCUR is documented to
1438 * return the xpv_cur field which certainly exists after upgrading.
1439 * according to nicholas clark, calling SvPOK fixes this.
1440 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1441 * and hope for the best.
1442 * Damnit, SvPV_nolen still trips over yet another assertion. This
1443 * assertion business is seriously broken, try yet another workaround
1444 * for the broken -DDEBUGGING.
1445 */
1446 {
1447#ifdef DEBUGGING
1448 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1449#else
1450 STRLEN offset = SvCUR (string);
1451#endif
1452
1453 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", 1454 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)); 1455 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1456 }
1404 1457
1405 if (json->flags & F_UTF8) 1458 if (json->flags & F_UTF8)
1406 sv_utf8_downgrade (string, 0); 1459 sv_utf8_downgrade (string, 0);
1407 else 1460 else
1408 sv_utf8_upgrade (string); 1461 sv_utf8_upgrade (string);
1409 1462
1410 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1463 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1411 1464
1412 dec.json = *json; 1465 dec.json = *json;
1413 dec.cur = SvPVX (string); 1466 dec.cur = SvPVX (string);
1414 dec.end = SvEND (string); 1467 dec.end = SvEND (string);
1415 dec.err = 0; 1468 dec.err = 0;
1416 dec.depth = 0; 1469 dec.depth = 0;
1417 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1418 1470
1419 if (dec.json.cb_object || dec.json.cb_sk_object) 1471 if (dec.json.cb_object || dec.json.cb_sk_object)
1420 dec.json.flags |= F_HOOK; 1472 dec.json.flags |= F_HOOK;
1421 1473
1422 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1474 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1423 1475
1424 decode_ws (&dec); 1476 decode_ws (&dec);
1425 sv = decode_sv (&dec); 1477 sv = decode_sv (&dec);
1478
1479 if (offset_return)
1480 *offset_return = dec.cur;
1426 1481
1427 if (!(offset_return || !sv)) 1482 if (!(offset_return || !sv))
1428 { 1483 {
1429 // check for trailing garbage 1484 // check for trailing garbage
1430 decode_ws (&dec); 1485 decode_ws (&dec);
1433 { 1488 {
1434 dec.err = "garbage after JSON object"; 1489 dec.err = "garbage after JSON object";
1435 SvREFCNT_dec (sv); 1490 SvREFCNT_dec (sv);
1436 sv = 0; 1491 sv = 0;
1437 } 1492 }
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 } 1493 }
1449 1494
1450 if (!sv) 1495 if (!sv)
1451 { 1496 {
1452 SV *uni = sv_newmortal (); 1497 SV *uni = sv_newmortal ();
1458 SAVEVPTR (PL_curcop); 1503 SAVEVPTR (PL_curcop);
1459 PL_curcop = &cop; 1504 PL_curcop = &cop;
1460 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1505 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1461 LEAVE; 1506 LEAVE;
1462 1507
1463 croak ("%s, at character offset %d [\"%s\"]", 1508 croak ("%s, at character offset %d (before \"%s\")",
1464 dec.err, 1509 dec.err,
1465 (int)offset, 1510 ptr_to_index (string, dec.cur),
1466 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1511 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1467 } 1512 }
1468 1513
1469 sv = sv_2mortal (sv); 1514 sv = sv_2mortal (sv);
1470 1515
1568 self->incr_mode = INCR_M_STR; 1613 self->incr_mode = INCR_M_STR;
1569 goto incr_m_str; 1614 goto incr_m_str;
1570 1615
1571 case '[': 1616 case '[':
1572 case '{': 1617 case '{':
1573 ++self->incr_nest; 1618 if (++self->incr_nest > self->max_depth)
1619 croak (ERR_NESTING_EXCEEDED);
1574 break; 1620 break;
1575 1621
1576 case ']': 1622 case ']':
1577 case '}': 1623 case '}':
1578 if (!--self->incr_nest) 1624 if (--self->incr_nest <= 0)
1579 goto interrupt; 1625 goto interrupt;
1580 } 1626 }
1581 } 1627 }
1582 } 1628 }
1583 1629
1607 : -1; 1653 : -1;
1608 1654
1609 json_stash = gv_stashpv ("JSON::XS" , 1); 1655 json_stash = gv_stashpv ("JSON::XS" , 1);
1610 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1656 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1611 1657
1612 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1658 json_true = get_bool ("JSON::XS::true");
1613 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1659 json_false = get_bool ("JSON::XS::false");
1614} 1660}
1615 1661
1616PROTOTYPES: DISABLE 1662PROTOTYPES: DISABLE
1617 1663
1618void CLONE (...) 1664void CLONE (...)
1623void new (char *klass) 1669void new (char *klass)
1624 PPCODE: 1670 PPCODE:
1625{ 1671{
1626 SV *pv = NEWSV (0, sizeof (JSON)); 1672 SV *pv = NEWSV (0, sizeof (JSON));
1627 SvPOK_only (pv); 1673 SvPOK_only (pv);
1628 Zero (SvPVX (pv), 1, JSON); 1674 json_init ((JSON *)SvPVX (pv));
1629 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1630 XPUSHs (sv_2mortal (sv_bless ( 1675 XPUSHs (sv_2mortal (sv_bless (
1631 newRV_noinc (pv), 1676 newRV_noinc (pv),
1632 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1677 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1633 ))); 1678 )));
1634} 1679}
1646 allow_nonref = F_ALLOW_NONREF 1691 allow_nonref = F_ALLOW_NONREF
1647 shrink = F_SHRINK 1692 shrink = F_SHRINK
1648 allow_blessed = F_ALLOW_BLESSED 1693 allow_blessed = F_ALLOW_BLESSED
1649 convert_blessed = F_CONV_BLESSED 1694 convert_blessed = F_CONV_BLESSED
1650 relaxed = F_RELAXED 1695 relaxed = F_RELAXED
1696 allow_unknown = F_ALLOW_UNKNOWN
1651 PPCODE: 1697 PPCODE:
1652{ 1698{
1653 if (enable) 1699 if (enable)
1654 self->flags |= ix; 1700 self->flags |= ix;
1655 else 1701 else
1670 get_allow_nonref = F_ALLOW_NONREF 1716 get_allow_nonref = F_ALLOW_NONREF
1671 get_shrink = F_SHRINK 1717 get_shrink = F_SHRINK
1672 get_allow_blessed = F_ALLOW_BLESSED 1718 get_allow_blessed = F_ALLOW_BLESSED
1673 get_convert_blessed = F_CONV_BLESSED 1719 get_convert_blessed = F_CONV_BLESSED
1674 get_relaxed = F_RELAXED 1720 get_relaxed = F_RELAXED
1721 get_allow_unknown = F_ALLOW_UNKNOWN
1675 PPCODE: 1722 PPCODE:
1676 XPUSHs (boolSV (self->flags & ix)); 1723 XPUSHs (boolSV (self->flags & ix));
1677 1724
1678void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1725void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1679 PPCODE: 1726 PPCODE:
1680{ 1727 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)); 1728 XPUSHs (ST (0));
1691}
1692 1729
1693U32 get_max_depth (JSON *self) 1730U32 get_max_depth (JSON *self)
1694 CODE: 1731 CODE:
1695 RETVAL = DEC_DEPTH (self->flags); 1732 RETVAL = self->max_depth;
1696 OUTPUT: 1733 OUTPUT:
1697 RETVAL 1734 RETVAL
1698 1735
1699void max_size (JSON *self, UV max_size = 0) 1736void max_size (JSON *self, U32 max_size = 0)
1700 PPCODE: 1737 PPCODE:
1701{ 1738 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)); 1739 XPUSHs (ST (0));
1713}
1714 1740
1715int get_max_size (JSON *self) 1741int get_max_size (JSON *self)
1716 CODE: 1742 CODE:
1717 RETVAL = DEC_SIZE (self->flags); 1743 RETVAL = self->max_size;
1718 OUTPUT: 1744 OUTPUT:
1719 RETVAL 1745 RETVAL
1720 1746
1721void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1747void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1722 PPCODE: 1748 PPCODE:
1758 XPUSHs (decode_json (jsonstr, self, 0)); 1784 XPUSHs (decode_json (jsonstr, self, 0));
1759 1785
1760void decode_prefix (JSON *self, SV *jsonstr) 1786void decode_prefix (JSON *self, SV *jsonstr)
1761 PPCODE: 1787 PPCODE:
1762{ 1788{
1763 STRLEN offset; 1789 char *offset;
1764 EXTEND (SP, 2); 1790 EXTEND (SP, 2);
1765 PUSHs (decode_json (jsonstr, self, &offset)); 1791 PUSHs (decode_json (jsonstr, self, &offset));
1766 PUSHs (sv_2mortal (newSVuv (offset))); 1792 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1767} 1793}
1768 1794
1769void incr_parse (JSON *self, SV *jsonstr = 0) 1795void incr_parse (JSON *self, SV *jsonstr = 0)
1770 PPCODE: 1796 PPCODE:
1771{ 1797{
1773 self->incr_text = newSVpvn ("", 0); 1799 self->incr_text = newSVpvn ("", 0);
1774 1800
1775 // append data, if any 1801 // append data, if any
1776 if (jsonstr) 1802 if (jsonstr)
1777 { 1803 {
1778 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 1804 if (SvUTF8 (jsonstr))
1779 { 1805 {
1806 if (!SvUTF8 (self->incr_text))
1807 {
1780 /* utf-8-ness differs, need to upgrade */ 1808 /* utf-8-ness differs, need to upgrade */
1781 sv_utf8_upgrade (self->incr_text); 1809 sv_utf8_upgrade (self->incr_text);
1782 1810
1783 if (self->incr_pos) 1811 if (self->incr_pos)
1784 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) 1812 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1785 - (U8 *)SvPVX (self->incr_text); 1813 - (U8 *)SvPVX (self->incr_text);
1814 }
1786 } 1815 }
1816 else if (SvUTF8 (self->incr_text))
1817 sv_utf8_upgrade (jsonstr);
1787 1818
1788 { 1819 {
1789 STRLEN len; 1820 STRLEN len;
1790 const char *str = SvPV (jsonstr, len); 1821 const char *str = SvPV (jsonstr, len);
1791 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 1822 STRLEN cur = SvCUR (self->incr_text);
1823
1824 if (SvLEN (self->incr_text) <= cur + len)
1825 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1826
1792 Move (str, SvEND (self->incr_text), len, char); 1827 Move (str, SvEND (self->incr_text), len, char);
1793 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 1828 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 1829 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1795 } 1830 }
1796 } 1831 }
1797 1832
1798 if (GIMME_V != G_VOID) 1833 if (GIMME_V != G_VOID)
1799 do 1834 do
1800 { 1835 {
1801 STRLEN offset; 1836 char *offset;
1802
1803 incr_parse (self);
1804 1837
1805 if (!INCR_DONE (self)) 1838 if (!INCR_DONE (self))
1839 {
1840 incr_parse (self);
1841
1842 if (self->incr_pos > self->max_size && self->max_size)
1843 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1844 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1845
1846 if (!INCR_DONE (self))
1806 break; 1847 break;
1848 }
1807 1849
1808 XPUSHs (decode_json (self->incr_text, self, &offset)); 1850 XPUSHs (decode_json (self->incr_text, self, &offset));
1809 1851
1810 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1811 self->incr_pos -= offset; 1852 self->incr_pos -= offset - SvPVX (self->incr_text);
1812 self->incr_nest = 0; 1853 self->incr_nest = 0;
1813 self->incr_mode = 0; 1854 self->incr_mode = 0;
1855
1856 sv_chop (self->incr_text, offset);
1814 } 1857 }
1815 while (GIMME_V == G_ARRAY); 1858 while (GIMME_V == G_ARRAY);
1816} 1859}
1817 1860
1818SV *incr_text (JSON *self) 1861SV *incr_text (JSON *self)
1819 ATTRS: lvalue 1862 ATTRS: lvalue
1820 CODE: 1863 CODE:
1821{ 1864{
1822 if (self->incr_pos) 1865 if (self->incr_pos)
1823 croak ("incr_text can only be called after a successful incr_parse call in scalar context"); 1866 croak ("incr_text can not be called when the incremental parser already started parsing");
1824 1867
1825 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; 1868 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1826} 1869}
1827 OUTPUT: 1870 OUTPUT:
1828 RETVAL 1871 RETVAL
1837 self->incr_nest = 0; 1880 self->incr_nest = 0;
1838 self->incr_mode = 0; 1881 self->incr_mode = 0;
1839 } 1882 }
1840} 1883}
1841 1884
1885void incr_reset (JSON *self)
1886 CODE:
1887{
1888 SvREFCNT_dec (self->incr_text);
1889 self->incr_text = 0;
1890 self->incr_pos = 0;
1891 self->incr_nest = 0;
1892 self->incr_mode = 0;
1893}
1894
1842void DESTROY (JSON *self) 1895void DESTROY (JSON *self)
1843 CODE: 1896 CODE:
1844 SvREFCNT_dec (self->cb_sk_object); 1897 SvREFCNT_dec (self->cb_sk_object);
1845 SvREFCNT_dec (self->cb_object); 1898 SvREFCNT_dec (self->cb_object);
1846 SvREFCNT_dec (self->incr_text); 1899 SvREFCNT_dec (self->incr_text);
1851 ALIAS: 1904 ALIAS:
1852 to_json_ = 0 1905 to_json_ = 0
1853 encode_json = F_UTF8 1906 encode_json = F_UTF8
1854 PPCODE: 1907 PPCODE:
1855{ 1908{
1856 JSON json = { F_DEFAULT | ix }; 1909 JSON json;
1910 json_init (&json);
1911 json.flags |= ix;
1857 XPUSHs (encode_json (scalar, &json)); 1912 XPUSHs (encode_json (scalar, &json));
1858} 1913}
1859 1914
1860void decode_json (SV *jsonstr) 1915void decode_json (SV *jsonstr)
1861 ALIAS: 1916 ALIAS:
1862 from_json_ = 0 1917 from_json_ = 0
1863 decode_json = F_UTF8 1918 decode_json = F_UTF8
1864 PPCODE: 1919 PPCODE:
1865{ 1920{
1866 JSON json = { F_DEFAULT | ix }; 1921 JSON json;
1922 json_init (&json);
1923 json.flags |= ix;
1867 XPUSHs (decode_json (jsonstr, &json, 0)); 1924 XPUSHs (decode_json (jsonstr, &json, 0));
1868} 1925}
1869 1926
1870 1927

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines