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 1.115 by root, Fri Dec 21 05:16:19 2012 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
46
47#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
53 48
54#define SB do { 49#define SB do {
55#define SE } while (0) 50#define SE } while (0)
56 51
57#if __GNUC__ >= 3 52#if __GNUC__ >= 3
67 62
68#define IN_RANGE_INC(type,val,beg,end) \ 63#define IN_RANGE_INC(type,val,beg,end) \
69 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 64 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
70 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 65 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
71 66
67#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
68
72#ifdef USE_ITHREADS 69#ifdef USE_ITHREADS
73# define JSON_SLOW 1 70# define JSON_SLOW 1
74# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 71# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
75#else 72#else
76# define JSON_SLOW 0 73# define JSON_SLOW 0
77# define JSON_STASH json_stash 74# define JSON_STASH json_stash
78#endif 75#endif
79 76
77// the amount of HEs to allocate on the stack, when sorting keys
78#define STACK_HES 64
79
80static HV *json_stash, *json_boolean_stash; // JSON::XS:: 80static HV *json_stash, *json_boolean_stash; // JSON::XS::
81static SV *json_true, *json_false; 81static SV *json_true, *json_false;
82 82
83enum { 83enum {
84 INCR_M_WS = 0, // initial whitespace skipping, must be 0 84 INCR_M_WS = 0, // initial whitespace skipping, must be 0
85 INCR_M_STR, // inside string 85 INCR_M_STR, // inside string
86 INCR_M_BS, // inside backslash 86 INCR_M_BS, // inside backslash
87 INCR_M_C0, // inside comment in initial whitespace sequence
88 INCR_M_C1, // inside comment in other places
87 INCR_M_JSON // outside anything, count nesting 89 INCR_M_JSON // outside anything, count nesting
88}; 90};
89 91
90#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON) 92#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
91 93
92typedef struct { 94typedef struct {
93 U32 flags; 95 U32 flags;
96 U32 max_depth;
97 STRLEN max_size;
98
94 SV *cb_object; 99 SV *cb_object;
95 HV *cb_sk_object; 100 HV *cb_sk_object;
96 101
97 // for the incremental parser 102 // for the incremental parser
98 SV *incr_text; // the source text so far 103 SV *incr_text; // the source text so far
99 STRLEN incr_pos; // the current offset into the text 104 STRLEN incr_pos; // the current offset into the text
100 int incr_nest; // {[]}-nesting level 105 int incr_nest; // {[]}-nesting level
101 int incr_mode; 106 unsigned char incr_mode;
102} JSON; 107} JSON;
108
109INLINE void
110json_init (JSON *json)
111{
112 Zero (json, 1, JSON);
113 json->max_depth = 512;
114}
103 115
104///////////////////////////////////////////////////////////////////////////// 116/////////////////////////////////////////////////////////////////////////////
105// utility functions 117// utility functions
106 118
107INLINE SV * 119INLINE SV *
117 129
118INLINE void 130INLINE void
119shrink (SV *sv) 131shrink (SV *sv)
120{ 132{
121 sv_utf8_downgrade (sv, 1); 133 sv_utf8_downgrade (sv, 1);
134
122 if (SvLEN (sv) > SvCUR (sv) + 1) 135 if (SvLEN (sv) > SvCUR (sv) + 1)
123 { 136 {
124#ifdef SvPV_shrink_to_cur 137#ifdef SvPV_shrink_to_cur
125 SvPV_shrink_to_cur (sv); 138 SvPV_shrink_to_cur (sv);
126#elif defined (SvPV_renew) 139#elif defined (SvPV_renew)
170 *s++ = 0x80 | ( ch & 0x3f); 183 *s++ = 0x80 | ( ch & 0x3f);
171 184
172 return s; 185 return s;
173} 186}
174 187
188// convert offset pointer to character index, sv must be string
189static STRLEN
190ptr_to_index (SV *sv, char *offset)
191{
192 return SvUTF8 (sv)
193 ? utf8_distance (offset, SvPVX (sv))
194 : offset - SvPVX (sv);
195}
196
197/////////////////////////////////////////////////////////////////////////////
198// fp hell
199
200// scan a group of digits, and a trailing exponent
201static void
202json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
203{
204 UV uaccum = 0;
205 int eaccum = 0;
206
207 // if we recurse too deep, skip all remaining digits
208 // to avoid a stack overflow attack
209 if (expect_false (--maxdepth <= 0))
210 while (((U8)*s - '0') < 10)
211 ++s;
212
213 for (;;)
214 {
215 U8 dig = (U8)*s - '0';
216
217 if (expect_false (dig >= 10))
218 {
219 if (dig == (U8)((U8)'.' - (U8)'0'))
220 {
221 ++s;
222 json_atof_scan1 (s, accum, expo, 1, maxdepth);
223 }
224 else if ((dig | ' ') == 'e' - '0')
225 {
226 int exp2 = 0;
227 int neg = 0;
228
229 ++s;
230
231 if (*s == '-')
232 {
233 ++s;
234 neg = 1;
235 }
236 else if (*s == '+')
237 ++s;
238
239 while ((dig = (U8)*s - '0') < 10)
240 exp2 = exp2 * 10 + *s++ - '0';
241
242 *expo += neg ? -exp2 : exp2;
243 }
244
245 break;
246 }
247
248 ++s;
249
250 uaccum = uaccum * 10 + dig;
251 ++eaccum;
252
253 // if we have too many digits, then recurse for more
254 // we actually do this for rather few digits
255 if (uaccum >= (UV_MAX - 9) / 10)
256 {
257 if (postdp) *expo -= eaccum;
258 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
259 if (postdp) *expo += eaccum;
260
261 break;
262 }
263 }
264
265 // this relies greatly on the quality of the pow ()
266 // implementation of the platform, but a good
267 // implementation is hard to beat.
268 // (IEEE 754 conformant ones are required to be exact)
269 if (postdp) *expo -= eaccum;
270 *accum += uaccum * Perl_pow (10., *expo);
271 *expo += eaccum;
272}
273
274static NV
275json_atof (const char *s)
276{
277 NV accum = 0.;
278 int expo = 0;
279 int neg = 0;
280
281 if (*s == '-')
282 {
283 ++s;
284 neg = 1;
285 }
286
287 // a recursion depth of ten gives us >>500 bits
288 json_atof_scan1 (s, &accum, &expo, 0, 10);
289
290 return neg ? -accum : accum;
291}
175///////////////////////////////////////////////////////////////////////////// 292/////////////////////////////////////////////////////////////////////////////
176// encoder 293// encoder
177 294
178// structure used for encoding JSON 295// structure used for encoding JSON
179typedef struct 296typedef struct
181 char *cur; // SvPVX (sv) + current output position 298 char *cur; // SvPVX (sv) + current output position
182 char *end; // SvEND (sv) 299 char *end; // SvEND (sv)
183 SV *sv; // result scalar 300 SV *sv; // result scalar
184 JSON json; 301 JSON json;
185 U32 indent; // indentation level 302 U32 indent; // indentation level
186 U32 maxdepth; // max. indentation/recursion level
187 UV limit; // escape character values >= this value when encoding 303 UV limit; // escape character values >= this value when encoding
188} enc_t; 304} enc_t;
189 305
190INLINE void 306INLINE void
191need (enc_t *enc, STRLEN len) 307need (enc_t *enc, STRLEN len)
192{ 308{
193 if (expect_false (enc->cur + len >= enc->end)) 309 if (expect_false (enc->cur + len >= enc->end))
194 { 310 {
195 STRLEN cur = enc->cur - SvPVX (enc->sv); 311 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
196 SvGROW (enc->sv, cur + len + 1); 312 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
197 enc->cur = SvPVX (enc->sv) + cur; 313 enc->cur = SvPVX (enc->sv) + cur;
198 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 314 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
199 } 315 }
200} 316}
201 317
276 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 392 (int)((uch - 0x10000) % 0x400 + 0xDC00));
277 enc->cur += 12; 393 enc->cur += 12;
278 } 394 }
279 else 395 else
280 { 396 {
281 static char hexdigit [16] = "0123456789abcdef";
282 need (enc, len += 5); 397 need (enc, len += 5);
283 *enc->cur++ = '\\'; 398 *enc->cur++ = '\\';
284 *enc->cur++ = 'u'; 399 *enc->cur++ = 'u';
285 *enc->cur++ = hexdigit [ uch >> 12 ]; 400 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
286 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 401 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
287 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 402 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
288 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 403 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
289 } 404 }
290 405
291 str += clen; 406 str += clen;
292 } 407 }
293 else if (enc->json.flags & F_LATIN1) 408 else if (enc->json.flags & F_LATIN1)
364static void 479static void
365encode_av (enc_t *enc, AV *av) 480encode_av (enc_t *enc, AV *av)
366{ 481{
367 int i, len = av_len (av); 482 int i, len = av_len (av);
368 483
369 if (enc->indent >= enc->maxdepth) 484 if (enc->indent >= enc->json.max_depth)
370 croak ("data structure too deep (hit recursion limit)"); 485 croak (ERR_NESTING_EXCEEDED);
371 486
372 encode_ch (enc, '['); 487 encode_ch (enc, '[');
373 488
374 if (len >= 0) 489 if (len >= 0)
375 { 490 {
450static void 565static void
451encode_hv (enc_t *enc, HV *hv) 566encode_hv (enc_t *enc, HV *hv)
452{ 567{
453 HE *he; 568 HE *he;
454 569
455 if (enc->indent >= enc->maxdepth) 570 if (enc->indent >= enc->json.max_depth)
456 croak ("data structure too deep (hit recursion limit)"); 571 croak (ERR_NESTING_EXCEEDED);
457 572
458 encode_ch (enc, '{'); 573 encode_ch (enc, '{');
459 574
460 // for canonical output we have to sort by keys first 575 // for canonical output we have to sort by keys first
461 // actually, this is mostly due to the stupid so-called 576 // actually, this is mostly due to the stupid so-called
462 // security workaround added somewhere in 5.8.x. 577 // security workaround added somewhere in 5.8.x
463 // that randomises hash orderings 578 // that randomises hash orderings
464 if (enc->json.flags & F_CANONICAL) 579 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
465 { 580 {
466 int count = hv_iterinit (hv); 581 int count = hv_iterinit (hv);
467 582
468 if (SvMAGICAL (hv)) 583 if (SvMAGICAL (hv))
469 { 584 {
480 } 595 }
481 596
482 if (count) 597 if (count)
483 { 598 {
484 int i, fast = 1; 599 int i, fast = 1;
485#if defined(__BORLANDC__) || defined(_MSC_VER) 600 HE *hes_stack [STACK_HES];
486 HE **hes = _alloca (count * sizeof (HE)); 601 HE **hes = hes_stack;
487#else 602
488 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 603 // allocate larger arrays on the heap
489#endif 604 if (count > STACK_HES)
605 {
606 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
607 hes = (HE **)SvPVX (sv);
608 }
490 609
491 i = 0; 610 i = 0;
492 while ((he = hv_iternext (hv))) 611 while ((he = hv_iternext (hv)))
493 { 612 {
494 hes [i++] = he; 613 hes [i++] = he;
642 761
643 if (len == 1 && *pv == '1') 762 if (len == 1 && *pv == '1')
644 encode_str (enc, "true", 4, 0); 763 encode_str (enc, "true", 4, 0);
645 else if (len == 1 && *pv == '0') 764 else if (len == 1 && *pv == '0')
646 encode_str (enc, "false", 5, 0); 765 encode_str (enc, "false", 5, 0);
766 else if (enc->json.flags & F_ALLOW_UNKNOWN)
767 encode_str (enc, "null", 4, 0);
647 else 768 else
648 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 769 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
649 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 770 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
650 } 771 }
772 else if (enc->json.flags & F_ALLOW_UNKNOWN)
773 encode_str (enc, "null", 4, 0);
651 else 774 else
652 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 775 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
653 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 776 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
654} 777}
655 778
709 } 832 }
710 else 833 else
711 { 834 {
712 // large integer, use the (rather slow) snprintf way. 835 // large integer, use the (rather slow) snprintf way.
713 need (enc, IVUV_MAXCHARS); 836 need (enc, IVUV_MAXCHARS);
714 enc->cur += 837 enc->cur +=
715 SvIsUV(sv) 838 SvIsUV(sv)
716 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) 839 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
717 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 840 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
718 } 841 }
719 } 842 }
720 else if (SvROK (sv)) 843 else if (SvROK (sv))
721 encode_rv (enc, SvRV (sv)); 844 encode_rv (enc, SvRV (sv));
722 else if (!SvOK (sv)) 845 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
723 encode_str (enc, "null", 4, 0); 846 encode_str (enc, "null", 4, 0);
724 else 847 else
725 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 848 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
726 SvPV_nolen (sv), SvFLAGS (sv)); 849 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
727} 850}
728 851
729static SV * 852static SV *
730encode_json (SV *scalar, JSON *json) 853encode_json (SV *scalar, JSON *json)
731{ 854{
737 enc.json = *json; 860 enc.json = *json;
738 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 861 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
739 enc.cur = SvPVX (enc.sv); 862 enc.cur = SvPVX (enc.sv);
740 enc.end = SvEND (enc.sv); 863 enc.end = SvEND (enc.sv);
741 enc.indent = 0; 864 enc.indent = 0;
742 enc.maxdepth = DEC_DEPTH (enc.json.flags);
743 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL 865 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
744 : enc.json.flags & F_LATIN1 ? 0x000100UL 866 : enc.json.flags & F_LATIN1 ? 0x000100UL
745 : 0x110000UL; 867 : 0x110000UL;
746 868
747 SvPOK_only (enc.sv); 869 SvPOK_only (enc.sv);
748 encode_sv (&enc, scalar); 870 encode_sv (&enc, scalar);
871 encode_nl (&enc);
749 872
750 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 873 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
751 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 874 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
752 875
753 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 876 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
814 if (*dec->cur != ch) \ 937 if (*dec->cur != ch) \
815 ERR (# ch " expected"); \ 938 ERR (# ch " expected"); \
816 ++dec->cur; \ 939 ++dec->cur; \
817 SE 940 SE
818 941
819#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 942#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
820#define DEC_DEC_DEPTH --dec->depth 943#define DEC_DEC_DEPTH --dec->depth
821 944
822static SV *decode_sv (dec_t *dec); 945static SV *decode_sv (dec_t *dec);
823 946
824static signed char decode_hexdigit[256]; 947static signed char decode_hexdigit[256];
933 else if (expect_true (ch >= 0x20 && ch < 0x80)) 1056 else if (expect_true (ch >= 0x20 && ch < 0x80))
934 *cur++ = ch; 1057 *cur++ = ch;
935 else if (ch >= 0x80) 1058 else if (ch >= 0x80)
936 { 1059 {
937 STRLEN clen; 1060 STRLEN clen;
938 UV uch;
939 1061
940 --dec_cur; 1062 --dec_cur;
941 1063
942 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1064 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
943 if (clen == (STRLEN)-1) 1065 if (clen == (STRLEN)-1)
944 ERR ("malformed UTF-8 character in JSON string"); 1066 ERR ("malformed UTF-8 character in JSON string");
945 1067
946 do 1068 do
947 *cur++ = *dec_cur++; 1069 *cur++ = *dec_cur++;
964 { 1086 {
965 STRLEN len = cur - buf; 1087 STRLEN len = cur - buf;
966 1088
967 if (sv) 1089 if (sv)
968 { 1090 {
969 SvGROW (sv, SvCUR (sv) + len + 1); 1091 STRLEN cur = SvCUR (sv);
1092
1093 if (SvLEN (sv) <= cur + len)
1094 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1095
970 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1096 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
971 SvCUR_set (sv, SvCUR (sv) + len); 1097 SvCUR_set (sv, SvCUR (sv) + len);
972 } 1098 }
973 else 1099 else
974 sv = newSVpvn (buf, len); 1100 sv = newSVpvn (buf, len);
1065 1191
1066 // special case the rather common 1..5-digit-int case 1192 // special case the rather common 1..5-digit-int case
1067 if (*start == '-') 1193 if (*start == '-')
1068 switch (len) 1194 switch (len)
1069 { 1195 {
1070 case 2: return newSViv (-( start [1] - '0' * 1)); 1196 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1071 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1197 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)); 1198 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)); 1199 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)); 1200 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1075 } 1201 }
1076 else 1202 else
1077 switch (len) 1203 switch (len)
1078 { 1204 {
1079 case 1: return newSViv ( start [0] - '0' * 1); 1205 case 1: return newSViv ( start [0] - '0' * 1);
1080 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1206 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); 1207 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); 1208 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); 1209 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1084 } 1210 }
1085 1211
1086 { 1212 {
1087 UV uv; 1213 UV uv;
1088 int numtype = grok_number (start, len, &uv); 1214 int numtype = grok_number (start, len, &uv);
1097 } 1223 }
1098 1224
1099 len -= *start == '-' ? 1 : 0; 1225 len -= *start == '-' ? 1 : 0;
1100 1226
1101 // does not fit into IV or UV, try NV 1227 // does not fit into IV or UV, try NV
1102 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1228 if (len <= NV_DIG)
1103 #if defined (LDBL_DIG)
1104 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1105 #endif
1106 )
1107 // fits into NV without loss of precision 1229 // fits into NV without loss of precision
1108 return newSVnv (Atof (start)); 1230 return newSVnv (json_atof (start));
1109 1231
1110 // everything else fails, convert it to a string 1232 // everything else fails, convert it to a string
1111 return newSVpvn (start, dec->cur - start); 1233 return newSVpvn (start, dec->cur - start);
1112 } 1234 }
1113 1235
1114 // loss of precision here 1236 // loss of precision here
1115 return newSVnv (Atof (start)); 1237 return newSVnv (json_atof (start));
1116 1238
1117fail: 1239fail:
1118 return 0; 1240 return 0;
1119} 1241}
1120 1242
1289 dSP; 1411 dSP;
1290 int count; 1412 int count;
1291 1413
1292 ENTER; SAVETMPS; PUSHMARK (SP); 1414 ENTER; SAVETMPS; PUSHMARK (SP);
1293 XPUSHs (HeVAL (he)); 1415 XPUSHs (HeVAL (he));
1416 sv_2mortal (sv);
1294 1417
1295 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1418 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1296 1419
1297 if (count == 1) 1420 if (count == 1)
1298 { 1421 {
1299 sv = newSVsv (POPs); 1422 sv = newSVsv (POPs);
1300 FREETMPS; LEAVE; 1423 FREETMPS; LEAVE;
1301 return sv; 1424 return sv;
1302 } 1425 }
1303 1426
1427 SvREFCNT_inc (sv);
1304 FREETMPS; LEAVE; 1428 FREETMPS; LEAVE;
1305 } 1429 }
1306 } 1430 }
1307 1431
1308 if (dec->json.cb_object) 1432 if (dec->json.cb_object)
1398fail: 1522fail:
1399 return 0; 1523 return 0;
1400} 1524}
1401 1525
1402static SV * 1526static SV *
1403decode_json (SV *string, JSON *json, STRLEN *offset_return) 1527decode_json (SV *string, JSON *json, char **offset_return)
1404{ 1528{
1405 dec_t dec; 1529 dec_t dec;
1406 STRLEN offset;
1407 SV *sv; 1530 SV *sv;
1408 1531
1532 /* work around bugs in 5.10 where manipulating magic values
1533 * will perl ignore the magic in subsequent accesses.
1534 * also make a copy of non-PV values, to get them into a clean
1535 * state (SvPV should do that, but it's buggy, see below).
1536 */
1409 SvGETMAGIC (string); 1537 /*SvGETMAGIC (string);*/
1538 if (SvMAGICAL (string) || !SvPOK (string))
1539 string = sv_2mortal (newSVsv (string));
1540
1410 SvUPGRADE (string, SVt_PV); 1541 SvUPGRADE (string, SVt_PV);
1411 1542
1412 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1543 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1544 * assertion with -DDEBUGGING, although SvCUR is documented to
1545 * return the xpv_cur field which certainly exists after upgrading.
1546 * according to nicholas clark, calling SvPOK fixes this.
1547 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1548 * and hope for the best.
1549 * Damnit, SvPV_nolen still trips over yet another assertion. This
1550 * assertion business is seriously broken, try yet another workaround
1551 * for the broken -DDEBUGGING.
1552 */
1553 {
1554#ifdef DEBUGGING
1555 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1556#else
1557 STRLEN offset = SvCUR (string);
1558#endif
1559
1560 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", 1561 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)); 1562 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1563 }
1415 1564
1416 if (json->flags & F_UTF8) 1565 if (DECODE_WANTS_OCTETS (json))
1417 sv_utf8_downgrade (string, 0); 1566 sv_utf8_downgrade (string, 0);
1418 else 1567 else
1419 sv_utf8_upgrade (string); 1568 sv_utf8_upgrade (string);
1420 1569
1421 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1570 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1422 1571
1423 dec.json = *json; 1572 dec.json = *json;
1424 dec.cur = SvPVX (string); 1573 dec.cur = SvPVX (string);
1425 dec.end = SvEND (string); 1574 dec.end = SvEND (string);
1426 dec.err = 0; 1575 dec.err = 0;
1427 dec.depth = 0; 1576 dec.depth = 0;
1428 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1429 1577
1430 if (dec.json.cb_object || dec.json.cb_sk_object) 1578 if (dec.json.cb_object || dec.json.cb_sk_object)
1431 dec.json.flags |= F_HOOK; 1579 dec.json.flags |= F_HOOK;
1432 1580
1433 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1581 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1434 1582
1435 decode_ws (&dec); 1583 decode_ws (&dec);
1436 sv = decode_sv (&dec); 1584 sv = decode_sv (&dec);
1585
1586 if (offset_return)
1587 *offset_return = dec.cur;
1437 1588
1438 if (!(offset_return || !sv)) 1589 if (!(offset_return || !sv))
1439 { 1590 {
1440 // check for trailing garbage 1591 // check for trailing garbage
1441 decode_ws (&dec); 1592 decode_ws (&dec);
1444 { 1595 {
1445 dec.err = "garbage after JSON object"; 1596 dec.err = "garbage after JSON object";
1446 SvREFCNT_dec (sv); 1597 SvREFCNT_dec (sv);
1447 sv = 0; 1598 sv = 0;
1448 } 1599 }
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 } 1600 }
1460 1601
1461 if (!sv) 1602 if (!sv)
1462 { 1603 {
1463 SV *uni = sv_newmortal (); 1604 SV *uni = sv_newmortal ();
1469 SAVEVPTR (PL_curcop); 1610 SAVEVPTR (PL_curcop);
1470 PL_curcop = &cop; 1611 PL_curcop = &cop;
1471 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1612 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1472 LEAVE; 1613 LEAVE;
1473 1614
1474 croak ("%s, at character offset %d [\"%s\"]", 1615 croak ("%s, at character offset %d (before \"%s\")",
1475 dec.err, 1616 dec.err,
1476 (int)offset, 1617 (int)ptr_to_index (string, dec.cur),
1477 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1618 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1478 } 1619 }
1479 1620
1480 sv = sv_2mortal (sv); 1621 sv = sv_2mortal (sv);
1481 1622
1490 1631
1491static void 1632static void
1492incr_parse (JSON *self) 1633incr_parse (JSON *self)
1493{ 1634{
1494 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1635 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1636
1637 // the state machine here is a bit convoluted and could be simplified a lot
1638 // but this would make it slower, so...
1495 1639
1496 for (;;) 1640 for (;;)
1497 { 1641 {
1498 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D 1642 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1499 switch (self->incr_mode) 1643 switch (self->incr_mode)
1500 { 1644 {
1501 // only used for intiial whitespace skipping 1645 // only used for initial whitespace skipping
1502 case INCR_M_WS: 1646 case INCR_M_WS:
1503 for (;;) 1647 for (;;)
1504 { 1648 {
1505 if (*p > 0x20) 1649 if (*p > 0x20)
1506 { 1650 {
1651 if (*p == '#')
1652 {
1653 self->incr_mode = INCR_M_C0;
1654 goto incr_m_c;
1655 }
1656 else
1657 {
1507 self->incr_mode = INCR_M_JSON; 1658 self->incr_mode = INCR_M_JSON;
1508 goto incr_m_json; 1659 goto incr_m_json;
1660 }
1509 } 1661 }
1510 else if (!*p) 1662 else if (!*p)
1511 goto interrupt; 1663 goto interrupt;
1512 1664
1513 ++p; 1665 ++p;
1519 goto interrupt; 1671 goto interrupt;
1520 1672
1521 ++p; 1673 ++p;
1522 self->incr_mode = INCR_M_STR; 1674 self->incr_mode = INCR_M_STR;
1523 goto incr_m_str; 1675 goto incr_m_str;
1676
1677 // inside #-style comments
1678 case INCR_M_C0:
1679 case INCR_M_C1:
1680 incr_m_c:
1681 for (;;)
1682 {
1683 if (*p == '\n')
1684 {
1685 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1686 break;
1687 }
1688 else if (!*p)
1689 goto interrupt;
1690
1691 ++p;
1692 }
1693
1694 break;
1524 1695
1525 // inside a string 1696 // inside a string
1526 case INCR_M_STR: 1697 case INCR_M_STR:
1527 incr_m_str: 1698 incr_m_str:
1528 for (;;) 1699 for (;;)
1579 self->incr_mode = INCR_M_STR; 1750 self->incr_mode = INCR_M_STR;
1580 goto incr_m_str; 1751 goto incr_m_str;
1581 1752
1582 case '[': 1753 case '[':
1583 case '{': 1754 case '{':
1584 ++self->incr_nest; 1755 if (++self->incr_nest > self->max_depth)
1756 croak (ERR_NESTING_EXCEEDED);
1585 break; 1757 break;
1586 1758
1587 case ']': 1759 case ']':
1588 case '}': 1760 case '}':
1589 if (!--self->incr_nest) 1761 if (--self->incr_nest <= 0)
1590 goto interrupt; 1762 goto interrupt;
1763 break;
1764
1765 case '#':
1766 self->incr_mode = INCR_M_C1;
1767 goto incr_m_c;
1591 } 1768 }
1592 } 1769 }
1593 } 1770 }
1594 1771
1595 modechange: 1772 modechange:
1596 ; 1773 ;
1597 } 1774 }
1598 1775
1599interrupt: 1776interrupt:
1600 self->incr_pos = p - SvPVX (self->incr_text); 1777 self->incr_pos = p - SvPVX (self->incr_text);
1778 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1601 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D 1779 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1602} 1780}
1603 1781
1604///////////////////////////////////////////////////////////////////////////// 1782/////////////////////////////////////////////////////////////////////////////
1605// XS interface functions 1783// XS interface functions
1620 json_stash = gv_stashpv ("JSON::XS" , 1); 1798 json_stash = gv_stashpv ("JSON::XS" , 1);
1621 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1799 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1622 1800
1623 json_true = get_bool ("JSON::XS::true"); 1801 json_true = get_bool ("JSON::XS::true");
1624 json_false = get_bool ("JSON::XS::false"); 1802 json_false = get_bool ("JSON::XS::false");
1803
1804 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1625} 1805}
1626 1806
1627PROTOTYPES: DISABLE 1807PROTOTYPES: DISABLE
1628 1808
1629void CLONE (...) 1809void CLONE (...)
1632 json_boolean_stash = 0; 1812 json_boolean_stash = 0;
1633 1813
1634void new (char *klass) 1814void new (char *klass)
1635 PPCODE: 1815 PPCODE:
1636{ 1816{
1637 SV *pv = NEWSV (0, sizeof (JSON)); 1817 SV *pv = NEWSV (0, sizeof (JSON));
1638 SvPOK_only (pv); 1818 SvPOK_only (pv);
1639 Zero (SvPVX (pv), 1, JSON); 1819 json_init ((JSON *)SvPVX (pv));
1640 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1641 XPUSHs (sv_2mortal (sv_bless ( 1820 XPUSHs (sv_2mortal (sv_bless (
1642 newRV_noinc (pv), 1821 newRV_noinc (pv),
1643 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1822 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1644 ))); 1823 )));
1645} 1824}
1657 allow_nonref = F_ALLOW_NONREF 1836 allow_nonref = F_ALLOW_NONREF
1658 shrink = F_SHRINK 1837 shrink = F_SHRINK
1659 allow_blessed = F_ALLOW_BLESSED 1838 allow_blessed = F_ALLOW_BLESSED
1660 convert_blessed = F_CONV_BLESSED 1839 convert_blessed = F_CONV_BLESSED
1661 relaxed = F_RELAXED 1840 relaxed = F_RELAXED
1841 allow_unknown = F_ALLOW_UNKNOWN
1662 PPCODE: 1842 PPCODE:
1663{ 1843{
1664 if (enable) 1844 if (enable)
1665 self->flags |= ix; 1845 self->flags |= ix;
1666 else 1846 else
1681 get_allow_nonref = F_ALLOW_NONREF 1861 get_allow_nonref = F_ALLOW_NONREF
1682 get_shrink = F_SHRINK 1862 get_shrink = F_SHRINK
1683 get_allow_blessed = F_ALLOW_BLESSED 1863 get_allow_blessed = F_ALLOW_BLESSED
1684 get_convert_blessed = F_CONV_BLESSED 1864 get_convert_blessed = F_CONV_BLESSED
1685 get_relaxed = F_RELAXED 1865 get_relaxed = F_RELAXED
1866 get_allow_unknown = F_ALLOW_UNKNOWN
1686 PPCODE: 1867 PPCODE:
1687 XPUSHs (boolSV (self->flags & ix)); 1868 XPUSHs (boolSV (self->flags & ix));
1688 1869
1689void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1870void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1690 PPCODE: 1871 PPCODE:
1691{ 1872 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)); 1873 XPUSHs (ST (0));
1702}
1703 1874
1704U32 get_max_depth (JSON *self) 1875U32 get_max_depth (JSON *self)
1705 CODE: 1876 CODE:
1706 RETVAL = DEC_DEPTH (self->flags); 1877 RETVAL = self->max_depth;
1707 OUTPUT: 1878 OUTPUT:
1708 RETVAL 1879 RETVAL
1709 1880
1710void max_size (JSON *self, UV max_size = 0) 1881void max_size (JSON *self, U32 max_size = 0)
1711 PPCODE: 1882 PPCODE:
1712{ 1883 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)); 1884 XPUSHs (ST (0));
1724}
1725 1885
1726int get_max_size (JSON *self) 1886int get_max_size (JSON *self)
1727 CODE: 1887 CODE:
1728 RETVAL = DEC_SIZE (self->flags); 1888 RETVAL = self->max_size;
1729 OUTPUT: 1889 OUTPUT:
1730 RETVAL 1890 RETVAL
1731 1891
1732void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1892void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1733 PPCODE: 1893 PPCODE:
1739} 1899}
1740 1900
1741void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 1901void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1742 PPCODE: 1902 PPCODE:
1743{ 1903{
1744 if (!self->cb_sk_object) 1904 if (!self->cb_sk_object)
1745 self->cb_sk_object = newHV (); 1905 self->cb_sk_object = newHV ();
1746 1906
1747 if (SvOK (cb)) 1907 if (SvOK (cb))
1748 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 1908 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1749 else 1909 else
1760 XPUSHs (ST (0)); 1920 XPUSHs (ST (0));
1761} 1921}
1762 1922
1763void encode (JSON *self, SV *scalar) 1923void encode (JSON *self, SV *scalar)
1764 PPCODE: 1924 PPCODE:
1765 XPUSHs (encode_json (scalar, self)); 1925 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
1926 XPUSHs (scalar);
1766 1927
1767void decode (JSON *self, SV *jsonstr) 1928void decode (JSON *self, SV *jsonstr)
1768 PPCODE: 1929 PPCODE:
1769 XPUSHs (decode_json (jsonstr, self, 0)); 1930 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
1931 XPUSHs (jsonstr);
1770 1932
1771void decode_prefix (JSON *self, SV *jsonstr) 1933void decode_prefix (JSON *self, SV *jsonstr)
1772 PPCODE: 1934 PPCODE:
1773{ 1935{
1774 STRLEN offset; 1936 SV *sv;
1937 char *offset;
1938 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1775 EXTEND (SP, 2); 1939 EXTEND (SP, 2);
1776 PUSHs (decode_json (jsonstr, self, &offset)); 1940 PUSHs (sv);
1777 PUSHs (sv_2mortal (newSVuv (offset))); 1941 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1778} 1942}
1779 1943
1780void incr_parse (JSON *self, SV *jsonstr = 0) 1944void incr_parse (JSON *self, SV *jsonstr = 0)
1781 PPCODE: 1945 PPCODE:
1782{ 1946{
1783 if (!self->incr_text) 1947 if (!self->incr_text)
1784 self->incr_text = newSVpvn ("", 0); 1948 self->incr_text = newSVpvn ("", 0);
1949
1950 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
1951 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
1952 if (DECODE_WANTS_OCTETS (self))
1953 {
1954 if (self->incr_pos)
1955 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
1956 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
1957
1958 sv_utf8_downgrade (self->incr_text, 0);
1959 }
1960 else
1961 {
1962 sv_utf8_upgrade (self->incr_text);
1963
1964 if (self->incr_pos)
1965 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1966 - (U8 *)SvPVX (self->incr_text);
1967 }
1785 1968
1786 // append data, if any 1969 // append data, if any
1787 if (jsonstr) 1970 if (jsonstr)
1788 { 1971 {
1972 /* make sure both strings have same encoding */
1789 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 1973 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1974 if (SvUTF8 (jsonstr))
1975 sv_utf8_downgrade (jsonstr, 0);
1790 { 1976 else
1791 /* utf-8-ness differs, need to upgrade */
1792 sv_utf8_upgrade (self->incr_text); 1977 sv_utf8_upgrade (jsonstr);
1793 1978
1794 if (self->incr_pos) 1979 /* and then just blindly append */
1795 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1796 - (U8 *)SvPVX (self->incr_text);
1797 }
1798
1799 { 1980 {
1800 STRLEN len; 1981 STRLEN len;
1801 const char *str = SvPV (jsonstr, len); 1982 const char *str = SvPV (jsonstr, len);
1802 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 1983 STRLEN cur = SvCUR (self->incr_text);
1984
1985 if (SvLEN (self->incr_text) <= cur + len)
1986 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1987
1803 Move (str, SvEND (self->incr_text), len, char); 1988 Move (str, SvEND (self->incr_text), len, char);
1804 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 1989 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 1990 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1806 } 1991 }
1807 } 1992 }
1808 1993
1809 if (GIMME_V != G_VOID) 1994 if (GIMME_V != G_VOID)
1810 do 1995 do
1811 { 1996 {
1997 SV *sv;
1812 STRLEN offset; 1998 char *offset;
1813 1999
1814 if (!INCR_DONE (self)) 2000 if (!INCR_DONE (self))
1815 { 2001 {
1816 incr_parse (self); 2002 incr_parse (self);
2003
2004 if (self->incr_pos > self->max_size && self->max_size)
2005 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2006 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2007
1817 if (!INCR_DONE (self)) 2008 if (!INCR_DONE (self))
2009 {
2010 // as an optimisation, do not accumulate white space in the incr buffer
2011 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2012 {
2013 self->incr_pos = 0;
2014 SvCUR_set (self->incr_text, 0);
2015 }
2016
1818 break; 2017 break;
2018 }
1819 } 2019 }
1820 2020
1821 XPUSHs (decode_json (self->incr_text, self, &offset)); 2021 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2022 XPUSHs (sv);
1822 2023
1823 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1824 self->incr_pos -= offset; 2024 self->incr_pos -= offset - SvPVX (self->incr_text);
1825 self->incr_nest = 0; 2025 self->incr_nest = 0;
1826 self->incr_mode = 0; 2026 self->incr_mode = 0;
2027
2028 sv_chop (self->incr_text, offset);
1827 } 2029 }
1828 while (GIMME_V == G_ARRAY); 2030 while (GIMME_V == G_ARRAY);
1829} 2031}
1830 2032
1831SV *incr_text (JSON *self) 2033SV *incr_text (JSON *self)
1850 self->incr_nest = 0; 2052 self->incr_nest = 0;
1851 self->incr_mode = 0; 2053 self->incr_mode = 0;
1852 } 2054 }
1853} 2055}
1854 2056
2057void incr_reset (JSON *self)
2058 CODE:
2059{
2060 SvREFCNT_dec (self->incr_text);
2061 self->incr_text = 0;
2062 self->incr_pos = 0;
2063 self->incr_nest = 0;
2064 self->incr_mode = 0;
2065}
2066
1855void DESTROY (JSON *self) 2067void DESTROY (JSON *self)
1856 CODE: 2068 CODE:
1857 SvREFCNT_dec (self->cb_sk_object); 2069 SvREFCNT_dec (self->cb_sk_object);
1858 SvREFCNT_dec (self->cb_object); 2070 SvREFCNT_dec (self->cb_object);
1859 SvREFCNT_dec (self->incr_text); 2071 SvREFCNT_dec (self->incr_text);
1864 ALIAS: 2076 ALIAS:
1865 to_json_ = 0 2077 to_json_ = 0
1866 encode_json = F_UTF8 2078 encode_json = F_UTF8
1867 PPCODE: 2079 PPCODE:
1868{ 2080{
1869 JSON json = { F_DEFAULT | ix }; 2081 JSON json;
1870 XPUSHs (encode_json (scalar, &json)); 2082 json_init (&json);
2083 json.flags |= ix;
2084 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2085 XPUSHs (scalar);
1871} 2086}
1872 2087
1873void decode_json (SV *jsonstr) 2088void decode_json (SV *jsonstr)
1874 ALIAS: 2089 ALIAS:
1875 from_json_ = 0 2090 from_json_ = 0
1876 decode_json = F_UTF8 2091 decode_json = F_UTF8
1877 PPCODE: 2092 PPCODE:
1878{ 2093{
1879 JSON json = { F_DEFAULT | ix }; 2094 JSON json;
1880 XPUSHs (decode_json (jsonstr, &json, 0)); 2095 json_init (&json);
2096 json.flags |= ix;
2097 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2098 XPUSHs (jsonstr);
1881} 2099}
1882 2100
1883

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines