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 1.83 by root, Wed Mar 26 22:54:38 2008 UTC vs.
Revision 1.101 by root, Tue Sep 8 18:00:03 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#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) 22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
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
107INLINE SV * 111INLINE SV *
117 121
118INLINE void 122INLINE void
119shrink (SV *sv) 123shrink (SV *sv)
120{ 124{
121 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
126
122 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
123 { 128 {
124#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
125 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
126#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
170 *s++ = 0x80 | ( ch & 0x3f); 175 *s++ = 0x80 | ( ch & 0x3f);
171 176
172 return s; 177 return s;
173} 178}
174 179
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
175///////////////////////////////////////////////////////////////////////////// 189/////////////////////////////////////////////////////////////////////////////
176// encoder 190// encoder
177 191
178// structure used for encoding JSON 192// structure used for encoding JSON
179typedef struct 193typedef struct
181 char *cur; // SvPVX (sv) + current output position 195 char *cur; // SvPVX (sv) + current output position
182 char *end; // SvEND (sv) 196 char *end; // SvEND (sv)
183 SV *sv; // result scalar 197 SV *sv; // result scalar
184 JSON json; 198 JSON json;
185 U32 indent; // indentation level 199 U32 indent; // indentation level
186 U32 maxdepth; // max. indentation/recursion level
187 UV limit; // escape character values >= this value when encoding 200 UV limit; // escape character values >= this value when encoding
188} enc_t; 201} enc_t;
189 202
190INLINE void 203INLINE void
191need (enc_t *enc, STRLEN len) 204need (enc_t *enc, STRLEN len)
192{ 205{
193 if (expect_false (enc->cur + len >= enc->end)) 206 if (expect_false (enc->cur + len >= enc->end))
194 { 207 {
195 STRLEN cur = enc->cur - SvPVX (enc->sv); 208 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
196 SvGROW (enc->sv, cur + len + 1); 209 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
197 enc->cur = SvPVX (enc->sv) + cur; 210 enc->cur = SvPVX (enc->sv) + cur;
198 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 211 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
199 } 212 }
200} 213}
201 214
276 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 289 (int)((uch - 0x10000) % 0x400 + 0xDC00));
277 enc->cur += 12; 290 enc->cur += 12;
278 } 291 }
279 else 292 else
280 { 293 {
281 static char hexdigit [16] = "0123456789abcdef";
282 need (enc, len += 5); 294 need (enc, len += 5);
283 *enc->cur++ = '\\'; 295 *enc->cur++ = '\\';
284 *enc->cur++ = 'u'; 296 *enc->cur++ = 'u';
285 *enc->cur++ = hexdigit [ uch >> 12 ]; 297 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
286 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 298 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
287 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 299 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
288 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 300 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
289 } 301 }
290 302
291 str += clen; 303 str += clen;
292 } 304 }
293 else if (enc->json.flags & F_LATIN1) 305 else if (enc->json.flags & F_LATIN1)
364static void 376static void
365encode_av (enc_t *enc, AV *av) 377encode_av (enc_t *enc, AV *av)
366{ 378{
367 int i, len = av_len (av); 379 int i, len = av_len (av);
368 380
369 if (enc->indent >= enc->maxdepth) 381 if (enc->indent >= enc->json.max_depth)
370 croak ("data structure too deep (hit recursion limit)"); 382 croak (ERR_NESTING_EXCEEDED);
371 383
372 encode_ch (enc, '['); 384 encode_ch (enc, '[');
373 385
374 if (len >= 0) 386 if (len >= 0)
375 { 387 {
450static void 462static void
451encode_hv (enc_t *enc, HV *hv) 463encode_hv (enc_t *enc, HV *hv)
452{ 464{
453 HE *he; 465 HE *he;
454 466
455 if (enc->indent >= enc->maxdepth) 467 if (enc->indent >= enc->json.max_depth)
456 croak ("data structure too deep (hit recursion limit)"); 468 croak (ERR_NESTING_EXCEEDED);
457 469
458 encode_ch (enc, '{'); 470 encode_ch (enc, '{');
459 471
460 // for canonical output we have to sort by keys first 472 // for canonical output we have to sort by keys first
461 // actually, this is mostly due to the stupid so-called 473 // actually, this is mostly due to the stupid so-called
462 // security workaround added somewhere in 5.8.x. 474 // security workaround added somewhere in 5.8.x
463 // that randomises hash orderings 475 // that randomises hash orderings
464 if (enc->json.flags & F_CANONICAL) 476 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
465 { 477 {
466 int count = hv_iterinit (hv); 478 int count = hv_iterinit (hv);
467 479
468 if (SvMAGICAL (hv)) 480 if (SvMAGICAL (hv))
469 { 481 {
642 654
643 if (len == 1 && *pv == '1') 655 if (len == 1 && *pv == '1')
644 encode_str (enc, "true", 4, 0); 656 encode_str (enc, "true", 4, 0);
645 else if (len == 1 && *pv == '0') 657 else if (len == 1 && *pv == '0')
646 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);
647 else 661 else
648 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",
649 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 663 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
650 } 664 }
665 else if (enc->json.flags & F_ALLOW_UNKNOWN)
666 encode_str (enc, "null", 4, 0);
651 else 667 else
652 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",
653 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 669 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
654} 670}
655 671
717 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 733 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
718 } 734 }
719 } 735 }
720 else if (SvROK (sv)) 736 else if (SvROK (sv))
721 encode_rv (enc, SvRV (sv)); 737 encode_rv (enc, SvRV (sv));
722 else if (!SvOK (sv)) 738 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
723 encode_str (enc, "null", 4, 0); 739 encode_str (enc, "null", 4, 0);
724 else 740 else
725 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",
726 SvPV_nolen (sv), SvFLAGS (sv)); 742 SvPV_nolen (sv), SvFLAGS (sv));
727} 743}
737 enc.json = *json; 753 enc.json = *json;
738 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 754 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
739 enc.cur = SvPVX (enc.sv); 755 enc.cur = SvPVX (enc.sv);
740 enc.end = SvEND (enc.sv); 756 enc.end = SvEND (enc.sv);
741 enc.indent = 0; 757 enc.indent = 0;
742 enc.maxdepth = DEC_DEPTH (enc.json.flags);
743 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL 758 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
744 : enc.json.flags & F_LATIN1 ? 0x000100UL 759 : enc.json.flags & F_LATIN1 ? 0x000100UL
745 : 0x110000UL; 760 : 0x110000UL;
746 761
747 SvPOK_only (enc.sv); 762 SvPOK_only (enc.sv);
748 encode_sv (&enc, scalar); 763 encode_sv (&enc, scalar);
764 encode_nl (&enc);
749 765
750 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 766 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
751 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 767 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
752 768
753 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 769 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
814 if (*dec->cur != ch) \ 830 if (*dec->cur != ch) \
815 ERR (# ch " expected"); \ 831 ERR (# ch " expected"); \
816 ++dec->cur; \ 832 ++dec->cur; \
817 SE 833 SE
818 834
819#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 835#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
820#define DEC_DEC_DEPTH --dec->depth 836#define DEC_DEC_DEPTH --dec->depth
821 837
822static SV *decode_sv (dec_t *dec); 838static SV *decode_sv (dec_t *dec);
823 839
824static signed char decode_hexdigit[256]; 840static signed char decode_hexdigit[256];
933 else if (expect_true (ch >= 0x20 && ch < 0x80)) 949 else if (expect_true (ch >= 0x20 && ch < 0x80))
934 *cur++ = ch; 950 *cur++ = ch;
935 else if (ch >= 0x80) 951 else if (ch >= 0x80)
936 { 952 {
937 STRLEN clen; 953 STRLEN clen;
938 UV uch;
939 954
940 --dec_cur; 955 --dec_cur;
941 956
942 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 957 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
943 if (clen == (STRLEN)-1) 958 if (clen == (STRLEN)-1)
944 ERR ("malformed UTF-8 character in JSON string"); 959 ERR ("malformed UTF-8 character in JSON string");
945 960
946 do 961 do
947 *cur++ = *dec_cur++; 962 *cur++ = *dec_cur++;
964 { 979 {
965 STRLEN len = cur - buf; 980 STRLEN len = cur - buf;
966 981
967 if (sv) 982 if (sv)
968 { 983 {
969 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
970 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 989 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
971 SvCUR_set (sv, SvCUR (sv) + len); 990 SvCUR_set (sv, SvCUR (sv) + len);
972 } 991 }
973 else 992 else
974 sv = newSVpvn (buf, len); 993 sv = newSVpvn (buf, len);
1065 1084
1066 // special case the rather common 1..5-digit-int case 1085 // special case the rather common 1..5-digit-int case
1067 if (*start == '-') 1086 if (*start == '-')
1068 switch (len) 1087 switch (len)
1069 { 1088 {
1070 case 2: return newSViv (-( start [1] - '0' * 1)); 1089 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1071 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1090 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1072 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));
1073 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));
1074 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));
1075 } 1094 }
1076 else 1095 else
1077 switch (len) 1096 switch (len)
1078 { 1097 {
1079 case 1: return newSViv ( start [0] - '0' * 1); 1098 case 1: return newSViv ( start [0] - '0' * 1);
1080 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1099 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1081 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);
1082 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);
1083 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);
1084 } 1103 }
1085 1104
1086 { 1105 {
1087 UV uv; 1106 UV uv;
1088 int numtype = grok_number (start, len, &uv); 1107 int numtype = grok_number (start, len, &uv);
1398fail: 1417fail:
1399 return 0; 1418 return 0;
1400} 1419}
1401 1420
1402static SV * 1421static SV *
1403decode_json (SV *string, JSON *json, STRLEN *offset_return) 1422decode_json (SV *string, JSON *json, char **offset_return)
1404{ 1423{
1405 dec_t dec; 1424 dec_t dec;
1406 STRLEN offset;
1407 SV *sv; 1425 SV *sv;
1408 1426
1427 /* work around bugs in 5.10 where manipulating magic values
1428 * will perl ignore the magic in subsequent accesses
1429 */
1409 SvGETMAGIC (string); 1430 /*SvGETMAGIC (string);*/
1431 if (SvMAGICAL (string))
1432 string = sv_2mortal (newSVsv (string));
1433
1410 SvUPGRADE (string, SVt_PV); 1434 SvUPGRADE (string, SVt_PV);
1411 1435
1412 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)
1413 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",
1414 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1455 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1456 }
1415 1457
1416 if (json->flags & F_UTF8) 1458 if (json->flags & F_UTF8)
1417 sv_utf8_downgrade (string, 0); 1459 sv_utf8_downgrade (string, 0);
1418 else 1460 else
1419 sv_utf8_upgrade (string); 1461 sv_utf8_upgrade (string);
1420 1462
1421 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1463 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1422 1464
1423 dec.json = *json; 1465 dec.json = *json;
1424 dec.cur = SvPVX (string); 1466 dec.cur = SvPVX (string);
1425 dec.end = SvEND (string); 1467 dec.end = SvEND (string);
1426 dec.err = 0; 1468 dec.err = 0;
1427 dec.depth = 0; 1469 dec.depth = 0;
1428 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1429 1470
1430 if (dec.json.cb_object || dec.json.cb_sk_object) 1471 if (dec.json.cb_object || dec.json.cb_sk_object)
1431 dec.json.flags |= F_HOOK; 1472 dec.json.flags |= F_HOOK;
1432 1473
1433 *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
1434 1475
1435 decode_ws (&dec); 1476 decode_ws (&dec);
1436 sv = decode_sv (&dec); 1477 sv = decode_sv (&dec);
1478
1479 if (offset_return)
1480 *offset_return = dec.cur;
1437 1481
1438 if (!(offset_return || !sv)) 1482 if (!(offset_return || !sv))
1439 { 1483 {
1440 // check for trailing garbage 1484 // check for trailing garbage
1441 decode_ws (&dec); 1485 decode_ws (&dec);
1444 { 1488 {
1445 dec.err = "garbage after JSON object"; 1489 dec.err = "garbage after JSON object";
1446 SvREFCNT_dec (sv); 1490 SvREFCNT_dec (sv);
1447 sv = 0; 1491 sv = 0;
1448 } 1492 }
1449 }
1450
1451 if (offset_return || !sv)
1452 {
1453 offset = dec.json.flags & F_UTF8
1454 ? dec.cur - SvPVX (string)
1455 : utf8_distance (dec.cur, SvPVX (string));
1456
1457 if (offset_return)
1458 *offset_return = offset;
1459 } 1493 }
1460 1494
1461 if (!sv) 1495 if (!sv)
1462 { 1496 {
1463 SV *uni = sv_newmortal (); 1497 SV *uni = sv_newmortal ();
1469 SAVEVPTR (PL_curcop); 1503 SAVEVPTR (PL_curcop);
1470 PL_curcop = &cop; 1504 PL_curcop = &cop;
1471 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);
1472 LEAVE; 1506 LEAVE;
1473 1507
1474 croak ("%s, at character offset %d [\"%s\"]", 1508 croak ("%s, at character offset %d (before \"%s\")",
1475 dec.err, 1509 dec.err,
1476 (int)offset, 1510 ptr_to_index (string, dec.cur),
1477 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1511 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1478 } 1512 }
1479 1513
1480 sv = sv_2mortal (sv); 1514 sv = sv_2mortal (sv);
1481 1515
1579 self->incr_mode = INCR_M_STR; 1613 self->incr_mode = INCR_M_STR;
1580 goto incr_m_str; 1614 goto incr_m_str;
1581 1615
1582 case '[': 1616 case '[':
1583 case '{': 1617 case '{':
1584 ++self->incr_nest; 1618 if (++self->incr_nest > self->max_depth)
1619 croak (ERR_NESTING_EXCEEDED);
1585 break; 1620 break;
1586 1621
1587 case ']': 1622 case ']':
1588 case '}': 1623 case '}':
1589 if (!--self->incr_nest) 1624 if (--self->incr_nest <= 0)
1590 goto interrupt; 1625 goto interrupt;
1591 } 1626 }
1592 } 1627 }
1593 } 1628 }
1594 1629
1620 json_stash = gv_stashpv ("JSON::XS" , 1); 1655 json_stash = gv_stashpv ("JSON::XS" , 1);
1621 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1656 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1622 1657
1623 json_true = get_bool ("JSON::XS::true"); 1658 json_true = get_bool ("JSON::XS::true");
1624 json_false = get_bool ("JSON::XS::false"); 1659 json_false = get_bool ("JSON::XS::false");
1660
1661 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1625} 1662}
1626 1663
1627PROTOTYPES: DISABLE 1664PROTOTYPES: DISABLE
1628 1665
1629void CLONE (...) 1666void CLONE (...)
1634void new (char *klass) 1671void new (char *klass)
1635 PPCODE: 1672 PPCODE:
1636{ 1673{
1637 SV *pv = NEWSV (0, sizeof (JSON)); 1674 SV *pv = NEWSV (0, sizeof (JSON));
1638 SvPOK_only (pv); 1675 SvPOK_only (pv);
1639 Zero (SvPVX (pv), 1, JSON); 1676 json_init ((JSON *)SvPVX (pv));
1640 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1641 XPUSHs (sv_2mortal (sv_bless ( 1677 XPUSHs (sv_2mortal (sv_bless (
1642 newRV_noinc (pv), 1678 newRV_noinc (pv),
1643 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1679 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1644 ))); 1680 )));
1645} 1681}
1657 allow_nonref = F_ALLOW_NONREF 1693 allow_nonref = F_ALLOW_NONREF
1658 shrink = F_SHRINK 1694 shrink = F_SHRINK
1659 allow_blessed = F_ALLOW_BLESSED 1695 allow_blessed = F_ALLOW_BLESSED
1660 convert_blessed = F_CONV_BLESSED 1696 convert_blessed = F_CONV_BLESSED
1661 relaxed = F_RELAXED 1697 relaxed = F_RELAXED
1698 allow_unknown = F_ALLOW_UNKNOWN
1662 PPCODE: 1699 PPCODE:
1663{ 1700{
1664 if (enable) 1701 if (enable)
1665 self->flags |= ix; 1702 self->flags |= ix;
1666 else 1703 else
1681 get_allow_nonref = F_ALLOW_NONREF 1718 get_allow_nonref = F_ALLOW_NONREF
1682 get_shrink = F_SHRINK 1719 get_shrink = F_SHRINK
1683 get_allow_blessed = F_ALLOW_BLESSED 1720 get_allow_blessed = F_ALLOW_BLESSED
1684 get_convert_blessed = F_CONV_BLESSED 1721 get_convert_blessed = F_CONV_BLESSED
1685 get_relaxed = F_RELAXED 1722 get_relaxed = F_RELAXED
1723 get_allow_unknown = F_ALLOW_UNKNOWN
1686 PPCODE: 1724 PPCODE:
1687 XPUSHs (boolSV (self->flags & ix)); 1725 XPUSHs (boolSV (self->flags & ix));
1688 1726
1689void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1727void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1690 PPCODE: 1728 PPCODE:
1691{ 1729 self->max_depth = max_depth;
1692 UV log2 = 0;
1693
1694 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1695
1696 while ((1UL << log2) < max_depth)
1697 ++log2;
1698
1699 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1700
1701 XPUSHs (ST (0)); 1730 XPUSHs (ST (0));
1702}
1703 1731
1704U32 get_max_depth (JSON *self) 1732U32 get_max_depth (JSON *self)
1705 CODE: 1733 CODE:
1706 RETVAL = DEC_DEPTH (self->flags); 1734 RETVAL = self->max_depth;
1707 OUTPUT: 1735 OUTPUT:
1708 RETVAL 1736 RETVAL
1709 1737
1710void max_size (JSON *self, UV max_size = 0) 1738void max_size (JSON *self, U32 max_size = 0)
1711 PPCODE: 1739 PPCODE:
1712{ 1740 self->max_size = max_size;
1713 UV log2 = 0;
1714
1715 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1716 if (max_size == 1) max_size = 2;
1717
1718 while ((1UL << log2) < max_size)
1719 ++log2;
1720
1721 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1722
1723 XPUSHs (ST (0)); 1741 XPUSHs (ST (0));
1724}
1725 1742
1726int get_max_size (JSON *self) 1743int get_max_size (JSON *self)
1727 CODE: 1744 CODE:
1728 RETVAL = DEC_SIZE (self->flags); 1745 RETVAL = self->max_size;
1729 OUTPUT: 1746 OUTPUT:
1730 RETVAL 1747 RETVAL
1731 1748
1732void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1749void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1733 PPCODE: 1750 PPCODE:
1769 XPUSHs (decode_json (jsonstr, self, 0)); 1786 XPUSHs (decode_json (jsonstr, self, 0));
1770 1787
1771void decode_prefix (JSON *self, SV *jsonstr) 1788void decode_prefix (JSON *self, SV *jsonstr)
1772 PPCODE: 1789 PPCODE:
1773{ 1790{
1774 STRLEN offset; 1791 char *offset;
1775 EXTEND (SP, 2); 1792 EXTEND (SP, 2);
1776 PUSHs (decode_json (jsonstr, self, &offset)); 1793 PUSHs (decode_json (jsonstr, self, &offset));
1777 PUSHs (sv_2mortal (newSVuv (offset))); 1794 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1778} 1795}
1779 1796
1780void incr_parse (JSON *self, SV *jsonstr = 0) 1797void incr_parse (JSON *self, SV *jsonstr = 0)
1781 PPCODE: 1798 PPCODE:
1782{ 1799{
1784 self->incr_text = newSVpvn ("", 0); 1801 self->incr_text = newSVpvn ("", 0);
1785 1802
1786 // append data, if any 1803 // append data, if any
1787 if (jsonstr) 1804 if (jsonstr)
1788 { 1805 {
1789 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 1806 if (SvUTF8 (jsonstr))
1790 { 1807 {
1808 if (!SvUTF8 (self->incr_text))
1809 {
1791 /* utf-8-ness differs, need to upgrade */ 1810 /* utf-8-ness differs, need to upgrade */
1792 sv_utf8_upgrade (self->incr_text); 1811 sv_utf8_upgrade (self->incr_text);
1793 1812
1794 if (self->incr_pos) 1813 if (self->incr_pos)
1795 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) 1814 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1796 - (U8 *)SvPVX (self->incr_text); 1815 - (U8 *)SvPVX (self->incr_text);
1816 }
1797 } 1817 }
1818 else if (SvUTF8 (self->incr_text))
1819 sv_utf8_upgrade (jsonstr);
1798 1820
1799 { 1821 {
1800 STRLEN len; 1822 STRLEN len;
1801 const char *str = SvPV (jsonstr, len); 1823 const char *str = SvPV (jsonstr, len);
1802 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 1824 STRLEN cur = SvCUR (self->incr_text);
1825
1826 if (SvLEN (self->incr_text) <= cur + len)
1827 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1828
1803 Move (str, SvEND (self->incr_text), len, char); 1829 Move (str, SvEND (self->incr_text), len, char);
1804 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 1830 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1805 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 1831 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1806 } 1832 }
1807 } 1833 }
1808 1834
1809 if (GIMME_V != G_VOID) 1835 if (GIMME_V != G_VOID)
1810 do 1836 do
1811 { 1837 {
1812 STRLEN offset; 1838 char *offset;
1813 1839
1814 if (!INCR_DONE (self)) 1840 if (!INCR_DONE (self))
1815 { 1841 {
1816 incr_parse (self); 1842 incr_parse (self);
1843
1844 if (self->incr_pos > self->max_size && self->max_size)
1845 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1846 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1847
1817 if (!INCR_DONE (self)) 1848 if (!INCR_DONE (self))
1818 break; 1849 break;
1819 } 1850 }
1820 1851
1821 XPUSHs (decode_json (self->incr_text, self, &offset)); 1852 XPUSHs (decode_json (self->incr_text, self, &offset));
1822 1853
1823 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1824 self->incr_pos -= offset; 1854 self->incr_pos -= offset - SvPVX (self->incr_text);
1825 self->incr_nest = 0; 1855 self->incr_nest = 0;
1826 self->incr_mode = 0; 1856 self->incr_mode = 0;
1857
1858 sv_chop (self->incr_text, offset);
1827 } 1859 }
1828 while (GIMME_V == G_ARRAY); 1860 while (GIMME_V == G_ARRAY);
1829} 1861}
1830 1862
1831SV *incr_text (JSON *self) 1863SV *incr_text (JSON *self)
1850 self->incr_nest = 0; 1882 self->incr_nest = 0;
1851 self->incr_mode = 0; 1883 self->incr_mode = 0;
1852 } 1884 }
1853} 1885}
1854 1886
1887void incr_reset (JSON *self)
1888 CODE:
1889{
1890 SvREFCNT_dec (self->incr_text);
1891 self->incr_text = 0;
1892 self->incr_pos = 0;
1893 self->incr_nest = 0;
1894 self->incr_mode = 0;
1895}
1896
1855void DESTROY (JSON *self) 1897void DESTROY (JSON *self)
1856 CODE: 1898 CODE:
1857 SvREFCNT_dec (self->cb_sk_object); 1899 SvREFCNT_dec (self->cb_sk_object);
1858 SvREFCNT_dec (self->cb_object); 1900 SvREFCNT_dec (self->cb_object);
1859 SvREFCNT_dec (self->incr_text); 1901 SvREFCNT_dec (self->incr_text);
1864 ALIAS: 1906 ALIAS:
1865 to_json_ = 0 1907 to_json_ = 0
1866 encode_json = F_UTF8 1908 encode_json = F_UTF8
1867 PPCODE: 1909 PPCODE:
1868{ 1910{
1869 JSON json = { F_DEFAULT | ix }; 1911 JSON json;
1912 json_init (&json);
1913 json.flags |= ix;
1870 XPUSHs (encode_json (scalar, &json)); 1914 XPUSHs (encode_json (scalar, &json));
1871} 1915}
1872 1916
1873void decode_json (SV *jsonstr) 1917void decode_json (SV *jsonstr)
1874 ALIAS: 1918 ALIAS:
1875 from_json_ = 0 1919 from_json_ = 0
1876 decode_json = F_UTF8 1920 decode_json = F_UTF8
1877 PPCODE: 1921 PPCODE:
1878{ 1922{
1879 JSON json = { F_DEFAULT | ix }; 1923 JSON json;
1924 json_init (&json);
1925 json.flags |= ix;
1880 XPUSHs (decode_json (jsonstr, &json, 0)); 1926 XPUSHs (decode_json (jsonstr, &json, 0));
1881} 1927}
1882 1928
1883 1929

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines