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 1.95 by root, Sun Feb 22 06:55:28 2009 UTC vs.
Revision 1.124 by root, Mon Nov 24 18:42:51 2014 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// compatibility with perl <5.18
23#ifndef HvNAMELEN_get
24# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
25#endif
26#ifndef HvNAMELEN
27# define HvNAMELEN(hv) HvNAMELEN_get (hv)
28#endif
29#ifndef HvNAMEUTF8
30# define HvNAMEUTF8(hv) 0
31#endif
32
33// three extra for rounding, sign, and end of string
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) 34#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
23 35
24#define F_ASCII 0x00000001UL 36#define F_ASCII 0x00000001UL
25#define F_LATIN1 0x00000002UL 37#define F_LATIN1 0x00000002UL
26#define F_UTF8 0x00000004UL 38#define F_UTF8 0x00000004UL
27#define F_INDENT 0x00000008UL 39#define F_INDENT 0x00000008UL
32#define F_SHRINK 0x00000200UL 44#define F_SHRINK 0x00000200UL
33#define F_ALLOW_BLESSED 0x00000400UL 45#define F_ALLOW_BLESSED 0x00000400UL
34#define F_CONV_BLESSED 0x00000800UL 46#define F_CONV_BLESSED 0x00000800UL
35#define F_RELAXED 0x00001000UL 47#define F_RELAXED 0x00001000UL
36#define F_ALLOW_UNKNOWN 0x00002000UL 48#define F_ALLOW_UNKNOWN 0x00002000UL
49#define F_ALLOW_TAGS 0x00004000UL
37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 50#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
38 51
39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 52#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
40 53
41#define INIT_SIZE 32 // initial scalar size to be allocated 54#define INIT_SIZE 32 // initial scalar size to be allocated
42#define INDENT_STEP 3 // spaces per indentation level 55#define INDENT_STEP 3 // spaces per indentation level
43 56
44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 57#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
58
59#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
45 60
46#define SB do { 61#define SB do {
47#define SE } while (0) 62#define SE } while (0)
48 63
49#if __GNUC__ >= 3 64#if __GNUC__ >= 3
69#else 84#else
70# define JSON_SLOW 0 85# define JSON_SLOW 0
71# define JSON_STASH json_stash 86# define JSON_STASH json_stash
72#endif 87#endif
73 88
89// the amount of HEs to allocate on the stack, when sorting keys
90#define STACK_HES 64
91
74static HV *json_stash, *json_boolean_stash; // JSON::XS:: 92static HV *json_stash, *types_boolean_stash; // JSON::XS::
75static SV *json_true, *json_false; 93static SV *types_true, *types_false, *sv_json;
76 94
77enum { 95enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0 96 INCR_M_WS = 0, // initial whitespace skipping, must be 0
79 INCR_M_STR, // inside string 97 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash 98 INCR_M_BS, // inside backslash
99 INCR_M_C0, // inside comment in initial whitespace sequence
100 INCR_M_C1, // inside comment in other places
81 INCR_M_JSON // outside anything, count nesting 101 INCR_M_JSON // outside anything, count nesting
82}; 102};
83 103
84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) 104#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85 105
175 *s++ = 0x80 | ( ch & 0x3f); 195 *s++ = 0x80 | ( ch & 0x3f);
176 196
177 return s; 197 return s;
178} 198}
179 199
200// convert offset pointer to character index, sv must be string
201static STRLEN
202ptr_to_index (SV *sv, char *offset)
203{
204 return SvUTF8 (sv)
205 ? utf8_distance (offset, SvPVX (sv))
206 : offset - SvPVX (sv);
207}
208
209/////////////////////////////////////////////////////////////////////////////
210// fp hell
211
212// scan a group of digits, and a trailing exponent
213static void
214json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
215{
216 UV uaccum = 0;
217 int eaccum = 0;
218
219 // if we recurse too deep, skip all remaining digits
220 // to avoid a stack overflow attack
221 if (expect_false (--maxdepth <= 0))
222 while (((U8)*s - '0') < 10)
223 ++s;
224
225 for (;;)
226 {
227 U8 dig = (U8)*s - '0';
228
229 if (expect_false (dig >= 10))
230 {
231 if (dig == (U8)((U8)'.' - (U8)'0'))
232 {
233 ++s;
234 json_atof_scan1 (s, accum, expo, 1, maxdepth);
235 }
236 else if ((dig | ' ') == 'e' - '0')
237 {
238 int exp2 = 0;
239 int neg = 0;
240
241 ++s;
242
243 if (*s == '-')
244 {
245 ++s;
246 neg = 1;
247 }
248 else if (*s == '+')
249 ++s;
250
251 while ((dig = (U8)*s - '0') < 10)
252 exp2 = exp2 * 10 + *s++ - '0';
253
254 *expo += neg ? -exp2 : exp2;
255 }
256
257 break;
258 }
259
260 ++s;
261
262 uaccum = uaccum * 10 + dig;
263 ++eaccum;
264
265 // if we have too many digits, then recurse for more
266 // we actually do this for rather few digits
267 if (uaccum >= (UV_MAX - 9) / 10)
268 {
269 if (postdp) *expo -= eaccum;
270 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
271 if (postdp) *expo += eaccum;
272
273 break;
274 }
275 }
276
277 // this relies greatly on the quality of the pow ()
278 // implementation of the platform, but a good
279 // implementation is hard to beat.
280 // (IEEE 754 conformant ones are required to be exact)
281 if (postdp) *expo -= eaccum;
282 *accum += uaccum * Perl_pow (10., *expo);
283 *expo += eaccum;
284}
285
286static NV
287json_atof (const char *s)
288{
289 NV accum = 0.;
290 int expo = 0;
291 int neg = 0;
292
293 if (*s == '-')
294 {
295 ++s;
296 neg = 1;
297 }
298
299 // a recursion depth of ten gives us >>500 bits
300 json_atof_scan1 (s, &accum, &expo, 0, 10);
301
302 return neg ? -accum : accum;
303}
180///////////////////////////////////////////////////////////////////////////// 304/////////////////////////////////////////////////////////////////////////////
181// encoder 305// encoder
182 306
183// structure used for encoding JSON 307// structure used for encoding JSON
184typedef struct 308typedef struct
371 495
372 if (enc->indent >= enc->json.max_depth) 496 if (enc->indent >= enc->json.max_depth)
373 croak (ERR_NESTING_EXCEEDED); 497 croak (ERR_NESTING_EXCEEDED);
374 498
375 encode_ch (enc, '['); 499 encode_ch (enc, '[');
376 500
377 if (len >= 0) 501 if (len >= 0)
378 { 502 {
379 encode_nl (enc); ++enc->indent; 503 encode_nl (enc); ++enc->indent;
380 504
381 for (i = 0; i <= len; ++i) 505 for (i = 0; i <= len; ++i)
393 encode_comma (enc); 517 encode_comma (enc);
394 } 518 }
395 519
396 encode_nl (enc); --enc->indent; encode_indent (enc); 520 encode_nl (enc); --enc->indent; encode_indent (enc);
397 } 521 }
398 522
399 encode_ch (enc, ']'); 523 encode_ch (enc, ']');
400} 524}
401 525
402static void 526static void
403encode_hk (enc_t *enc, HE *he) 527encode_hk (enc_t *enc, HE *he)
407 if (HeKLEN (he) == HEf_SVKEY) 531 if (HeKLEN (he) == HEf_SVKEY)
408 { 532 {
409 SV *sv = HeSVKEY (he); 533 SV *sv = HeSVKEY (he);
410 STRLEN len; 534 STRLEN len;
411 char *str; 535 char *str;
412 536
413 SvGETMAGIC (sv); 537 SvGETMAGIC (sv);
414 str = SvPV (sv, len); 538 str = SvPV (sv, len);
415 539
416 encode_str (enc, str, len, SvUTF8 (sv)); 540 encode_str (enc, str, len, SvUTF8 (sv));
417 } 541 }
462 586
463 // for canonical output we have to sort by keys first 587 // for canonical output we have to sort by keys first
464 // actually, this is mostly due to the stupid so-called 588 // actually, this is mostly due to the stupid so-called
465 // security workaround added somewhere in 5.8.x 589 // security workaround added somewhere in 5.8.x
466 // that randomises hash orderings 590 // that randomises hash orderings
467 if (enc->json.flags & F_CANONICAL) 591 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
468 { 592 {
469 int count = hv_iterinit (hv); 593 int count = hv_iterinit (hv);
470 594
471 if (SvMAGICAL (hv)) 595 if (SvMAGICAL (hv))
472 { 596 {
483 } 607 }
484 608
485 if (count) 609 if (count)
486 { 610 {
487 int i, fast = 1; 611 int i, fast = 1;
488#if defined(__BORLANDC__) || defined(_MSC_VER) 612 HE *hes_stack [STACK_HES];
489 HE **hes = _alloca (count * sizeof (HE)); 613 HE **hes = hes_stack;
490#else 614
491 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 615 // allocate larger arrays on the heap
492#endif 616 if (count > STACK_HES)
617 {
618 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
619 hes = (HE **)SvPVX (sv);
620 }
493 621
494 i = 0; 622 i = 0;
495 while ((he = hv_iternext (hv))) 623 while ((he = hv_iternext (hv)))
496 { 624 {
497 hes [i++] = he; 625 hes [i++] = he;
566// encode objects, arrays and special \0=false and \1=true values. 694// encode objects, arrays and special \0=false and \1=true values.
567static void 695static void
568encode_rv (enc_t *enc, SV *sv) 696encode_rv (enc_t *enc, SV *sv)
569{ 697{
570 svtype svt; 698 svtype svt;
699 GV *method;
571 700
572 SvGETMAGIC (sv); 701 SvGETMAGIC (sv);
573 svt = SvTYPE (sv); 702 svt = SvTYPE (sv);
574 703
575 if (expect_false (SvOBJECT (sv))) 704 if (expect_false (SvOBJECT (sv)))
576 { 705 {
577 HV *stash = !JSON_SLOW || json_boolean_stash 706 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
578 ? json_boolean_stash 707 ? types_boolean_stash
579 : gv_stashpv ("JSON::XS::Boolean", 1); 708 : gv_stashpv ("Types::Serialiser::Boolean", 1);
709 HV *stash = SvSTASH (sv);
580 710
581 if (SvSTASH (sv) == stash) 711 if (stash == boolean_stash)
582 { 712 {
583 if (SvIV (sv)) 713 if (SvIV (sv))
584 encode_str (enc, "true", 4, 0); 714 encode_str (enc, "true", 4, 0);
585 else 715 else
586 encode_str (enc, "false", 5, 0); 716 encode_str (enc, "false", 5, 0);
587 } 717 }
718 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
719 {
720 int count;
721 dSP;
722
723 ENTER; SAVETMPS; PUSHMARK (SP);
724 EXTEND (SP, 2);
725 // we re-bless the reference to get overload and other niceties right
726 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
727 PUSHs (sv_json);
728
729 PUTBACK;
730 count = call_sv ((SV *)GvCV (method), G_ARRAY);
731 SPAGAIN;
732
733 // catch this surprisingly common error
734 if (SvROK (TOPs) && SvRV (TOPs) == sv)
735 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
736
737 encode_ch (enc, '(');
738 encode_ch (enc, '"');
739 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
740 encode_ch (enc, '"');
741 encode_ch (enc, ')');
742 encode_ch (enc, '[');
743
744 while (count)
745 {
746 encode_sv (enc, SP[1 - count--]);
747
748 if (count)
749 encode_ch (enc, ',');
750 }
751
752 encode_ch (enc, ']');
753
754 PUTBACK;
755
756 FREETMPS; LEAVE;
757 }
758 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
759 {
760 dSP;
761
762 ENTER; SAVETMPS; PUSHMARK (SP);
763 // we re-bless the reference to get overload and other niceties right
764 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
765
766 // calling with G_SCALAR ensures that we always get a 1 return value
767 PUTBACK;
768 call_sv ((SV *)GvCV (method), G_SCALAR);
769 SPAGAIN;
770
771 // catch this surprisingly common error
772 if (SvROK (TOPs) && SvRV (TOPs) == sv)
773 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
774
775 sv = POPs;
776 PUTBACK;
777
778 encode_sv (enc, sv);
779
780 FREETMPS; LEAVE;
781 }
782 else if (enc->json.flags & F_ALLOW_BLESSED)
783 encode_str (enc, "null", 4, 0);
588 else 784 else
589 {
590#if 0
591 if (0 && sv_derived_from (rv, "JSON::Literal"))
592 {
593 // not yet
594 }
595#endif
596 if (enc->json.flags & F_CONV_BLESSED)
597 {
598 // we re-bless the reference to get overload and other niceties right
599 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
600
601 if (to_json)
602 {
603 dSP;
604
605 ENTER; SAVETMPS; PUSHMARK (SP);
606 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
607
608 // calling with G_SCALAR ensures that we always get a 1 return value
609 PUTBACK;
610 call_sv ((SV *)GvCV (to_json), G_SCALAR);
611 SPAGAIN;
612
613 // catch this surprisingly common error
614 if (SvROK (TOPs) && SvRV (TOPs) == sv)
615 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
616
617 sv = POPs;
618 PUTBACK;
619
620 encode_sv (enc, sv);
621
622 FREETMPS; LEAVE;
623 }
624 else if (enc->json.flags & F_ALLOW_BLESSED)
625 encode_str (enc, "null", 4, 0);
626 else
627 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
628 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
629 }
630 else if (enc->json.flags & F_ALLOW_BLESSED)
631 encode_str (enc, "null", 4, 0);
632 else
633 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 785 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
634 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 786 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
635 }
636 } 787 }
637 else if (svt == SVt_PVHV) 788 else if (svt == SVt_PVHV)
638 encode_hv (enc, (HV *)sv); 789 encode_hv (enc, (HV *)sv);
639 else if (svt == SVt_PVAV) 790 else if (svt == SVt_PVAV)
640 encode_av (enc, (AV *)sv); 791 encode_av (enc, (AV *)sv);
716 } 867 }
717 else 868 else
718 { 869 {
719 // large integer, use the (rather slow) snprintf way. 870 // large integer, use the (rather slow) snprintf way.
720 need (enc, IVUV_MAXCHARS); 871 need (enc, IVUV_MAXCHARS);
721 enc->cur += 872 enc->cur +=
722 SvIsUV(sv) 873 SvIsUV(sv)
723 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) 874 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
724 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 875 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
725 } 876 }
726 } 877 }
727 else if (SvROK (sv)) 878 else if (SvROK (sv))
728 encode_rv (enc, SvRV (sv)); 879 encode_rv (enc, SvRV (sv));
729 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 880 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
730 encode_str (enc, "null", 4, 0); 881 encode_str (enc, "null", 4, 0);
731 else 882 else
732 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 883 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
733 SvPV_nolen (sv), SvFLAGS (sv)); 884 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
734} 885}
735 886
736static SV * 887static SV *
737encode_json (SV *scalar, JSON *json) 888encode_json (SV *scalar, JSON *json)
738{ 889{
750 : enc.json.flags & F_LATIN1 ? 0x000100UL 901 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL; 902 : 0x110000UL;
752 903
753 SvPOK_only (enc.sv); 904 SvPOK_only (enc.sv);
754 encode_sv (&enc, scalar); 905 encode_sv (&enc, scalar);
906 encode_nl (&enc);
755 907
756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 908 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
757 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 909 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
758 910
759 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 911 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
939 else if (expect_true (ch >= 0x20 && ch < 0x80)) 1091 else if (expect_true (ch >= 0x20 && ch < 0x80))
940 *cur++ = ch; 1092 *cur++ = ch;
941 else if (ch >= 0x80) 1093 else if (ch >= 0x80)
942 { 1094 {
943 STRLEN clen; 1095 STRLEN clen;
944 UV uch;
945 1096
946 --dec_cur; 1097 --dec_cur;
947 1098
948 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1099 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
949 if (clen == (STRLEN)-1) 1100 if (clen == (STRLEN)-1)
950 ERR ("malformed UTF-8 character in JSON string"); 1101 ERR ("malformed UTF-8 character in JSON string");
951 1102
952 do 1103 do
953 *cur++ = *dec_cur++; 1104 *cur++ = *dec_cur++;
954 while (--clen); 1105 while (--clen);
955 1106
956 utf8 = 1; 1107 utf8 = 1;
957 } 1108 }
1109 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1110 *cur++ = ch;
958 else 1111 else
959 { 1112 {
960 --dec_cur; 1113 --dec_cur;
961 1114
962 if (!ch) 1115 if (!ch)
1075 1228
1076 // special case the rather common 1..5-digit-int case 1229 // special case the rather common 1..5-digit-int case
1077 if (*start == '-') 1230 if (*start == '-')
1078 switch (len) 1231 switch (len)
1079 { 1232 {
1080 case 2: return newSViv (-( start [1] - '0' * 1)); 1233 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1081 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1234 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1082 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1235 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1083 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1236 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1084 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); 1237 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1085 } 1238 }
1086 else 1239 else
1087 switch (len) 1240 switch (len)
1088 { 1241 {
1089 case 1: return newSViv ( start [0] - '0' * 1); 1242 case 1: return newSViv ( start [0] - '0' * 1);
1090 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1243 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1091 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1244 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1092 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1245 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1093 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); 1246 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1094 } 1247 }
1095 1248
1096 { 1249 {
1097 UV uv; 1250 UV uv;
1098 int numtype = grok_number (start, len, &uv); 1251 int numtype = grok_number (start, len, &uv);
1107 } 1260 }
1108 1261
1109 len -= *start == '-' ? 1 : 0; 1262 len -= *start == '-' ? 1 : 0;
1110 1263
1111 // does not fit into IV or UV, try NV 1264 // does not fit into IV or UV, try NV
1112 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1265 if (len <= NV_DIG)
1113 #if defined (LDBL_DIG)
1114 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1115 #endif
1116 )
1117 // fits into NV without loss of precision 1266 // fits into NV without loss of precision
1118 return newSVnv (Atof (start)); 1267 return newSVnv (json_atof (start));
1119 1268
1120 // everything else fails, convert it to a string 1269 // everything else fails, convert it to a string
1121 return newSVpvn (start, dec->cur - start); 1270 return newSVpvn (start, dec->cur - start);
1122 } 1271 }
1123 1272
1124 // loss of precision here 1273 // loss of precision here
1125 return newSVnv (Atof (start)); 1274 return newSVnv (json_atof (start));
1126 1275
1127fail: 1276fail:
1128 return 0; 1277 return 0;
1129} 1278}
1130 1279
1154 if (*dec->cur == ']') 1303 if (*dec->cur == ']')
1155 { 1304 {
1156 ++dec->cur; 1305 ++dec->cur;
1157 break; 1306 break;
1158 } 1307 }
1159 1308
1160 if (*dec->cur != ',') 1309 if (*dec->cur != ',')
1161 ERR (", or ] expected while parsing array"); 1310 ERR (", or ] expected while parsing array");
1162 1311
1163 ++dec->cur; 1312 ++dec->cur;
1164 1313
1299 dSP; 1448 dSP;
1300 int count; 1449 int count;
1301 1450
1302 ENTER; SAVETMPS; PUSHMARK (SP); 1451 ENTER; SAVETMPS; PUSHMARK (SP);
1303 XPUSHs (HeVAL (he)); 1452 XPUSHs (HeVAL (he));
1453 sv_2mortal (sv);
1304 1454
1305 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1455 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1306 1456
1307 if (count == 1) 1457 if (count == 1)
1308 { 1458 {
1309 sv = newSVsv (POPs); 1459 sv = newSVsv (POPs);
1310 FREETMPS; LEAVE; 1460 FREETMPS; LEAVE;
1311 return sv; 1461 return sv;
1312 } 1462 }
1313 1463
1464 SvREFCNT_inc (sv);
1314 FREETMPS; LEAVE; 1465 FREETMPS; LEAVE;
1315 } 1466 }
1316 } 1467 }
1317 1468
1318 if (dec->json.cb_object) 1469 if (dec->json.cb_object)
1344 DEC_DEC_DEPTH; 1495 DEC_DEC_DEPTH;
1345 return 0; 1496 return 0;
1346} 1497}
1347 1498
1348static SV * 1499static SV *
1500decode_tag (dec_t *dec)
1501{
1502 SV *tag = 0;
1503 SV *val = 0;
1504
1505 if (!(dec->json.flags & F_ALLOW_TAGS))
1506 ERR ("malformed JSON string, neither array, object, number, string or atom");
1507
1508 ++dec->cur;
1509
1510 decode_ws (dec);
1511
1512 tag = decode_sv (dec);
1513 if (!tag)
1514 goto fail;
1515
1516 if (!SvPOK (tag))
1517 ERR ("malformed JSON string, (tag) must be a string");
1518
1519 decode_ws (dec);
1520
1521 if (*dec->cur != ')')
1522 ERR (") expected after tag");
1523
1524 ++dec->cur;
1525
1526 decode_ws (dec);
1527
1528 val = decode_sv (dec);
1529 if (!val)
1530 goto fail;
1531
1532 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1533 ERR ("malformed JSON string, tag value must be an array");
1534
1535 {
1536 AV *av = (AV *)SvRV (val);
1537 int i, len = av_len (av) + 1;
1538 HV *stash = gv_stashsv (tag, 0);
1539 SV *sv;
1540
1541 if (!stash)
1542 ERR ("cannot decode perl-object (package does not exist)");
1543
1544 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1545
1546 if (!method)
1547 ERR ("cannot decode perl-object (package does not have a THAW method)");
1548
1549 dSP;
1550
1551 ENTER; SAVETMPS; PUSHMARK (SP);
1552 EXTEND (SP, len + 2);
1553 // we re-bless the reference to get overload and other niceties right
1554 PUSHs (tag);
1555 PUSHs (sv_json);
1556
1557 for (i = 0; i < len; ++i)
1558 PUSHs (*av_fetch (av, i, 1));
1559
1560 PUTBACK;
1561 call_sv ((SV *)GvCV (method), G_SCALAR);
1562 SPAGAIN;
1563
1564 SvREFCNT_dec (tag);
1565 SvREFCNT_dec (val);
1566 sv = SvREFCNT_inc (POPs);
1567
1568 PUTBACK;
1569
1570 FREETMPS; LEAVE;
1571
1572 return sv;
1573 }
1574
1575fail:
1576 SvREFCNT_dec (tag);
1577 SvREFCNT_dec (val);
1578 return 0;
1579}
1580
1581static SV *
1349decode_sv (dec_t *dec) 1582decode_sv (dec_t *dec)
1350{ 1583{
1351 // the beauty of JSON: you need exactly one character lookahead 1584 // the beauty of JSON: you need exactly one character lookahead
1352 // to parse everything. 1585 // to parse everything.
1353 switch (*dec->cur) 1586 switch (*dec->cur)
1354 { 1587 {
1355 case '"': ++dec->cur; return decode_str (dec); 1588 case '"': ++dec->cur; return decode_str (dec);
1356 case '[': ++dec->cur; return decode_av (dec); 1589 case '[': ++dec->cur; return decode_av (dec);
1357 case '{': ++dec->cur; return decode_hv (dec); 1590 case '{': ++dec->cur; return decode_hv (dec);
1591 case '(': return decode_tag (dec);
1358 1592
1359 case '-': 1593 case '-':
1360 case '0': case '1': case '2': case '3': case '4': 1594 case '0': case '1': case '2': case '3': case '4':
1361 case '5': case '6': case '7': case '8': case '9': 1595 case '5': case '6': case '7': case '8': case '9':
1362 return decode_num (dec); 1596 return decode_num (dec);
1364 case 't': 1598 case 't':
1365 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1599 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1366 { 1600 {
1367 dec->cur += 4; 1601 dec->cur += 4;
1368#if JSON_SLOW 1602#if JSON_SLOW
1369 json_true = get_bool ("JSON::XS::true"); 1603 types_true = get_bool ("Types::Serialiser::true");
1370#endif 1604#endif
1371 return newSVsv (json_true); 1605 return newSVsv (types_true);
1372 } 1606 }
1373 else 1607 else
1374 ERR ("'true' expected"); 1608 ERR ("'true' expected");
1375 1609
1376 break; 1610 break;
1378 case 'f': 1612 case 'f':
1379 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1613 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1380 { 1614 {
1381 dec->cur += 5; 1615 dec->cur += 5;
1382#if JSON_SLOW 1616#if JSON_SLOW
1383 json_false = get_bool ("JSON::XS::false"); 1617 types_false = get_bool ("Types::Serialiser::false");
1384#endif 1618#endif
1385 return newSVsv (json_false); 1619 return newSVsv (types_false);
1386 } 1620 }
1387 else 1621 else
1388 ERR ("'false' expected"); 1622 ERR ("'false' expected");
1389 1623
1390 break; 1624 break;
1399 ERR ("'null' expected"); 1633 ERR ("'null' expected");
1400 1634
1401 break; 1635 break;
1402 1636
1403 default: 1637 default:
1404 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1638 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1405 break; 1639 break;
1406 } 1640 }
1407 1641
1408fail: 1642fail:
1409 return 0; 1643 return 0;
1410} 1644}
1411 1645
1412static SV * 1646static SV *
1413decode_json (SV *string, JSON *json, STRLEN *offset_return) 1647decode_json (SV *string, JSON *json, char **offset_return)
1414{ 1648{
1415 dec_t dec; 1649 dec_t dec;
1416 STRLEN offset;
1417 SV *sv; 1650 SV *sv;
1418 1651
1419 /* work around bugs in 5.10 where manipulating magic values 1652 /* work around bugs in 5.10 where manipulating magic values
1420 * will perl ignore the magic in subsequent accesses 1653 * makes perl ignore the magic in subsequent accesses.
1654 * also make a copy of non-PV values, to get them into a clean
1655 * state (SvPV should do that, but it's buggy, see below).
1421 */ 1656 */
1422 /*SvGETMAGIC (string);*/ 1657 /*SvGETMAGIC (string);*/
1423 if (SvMAGICAL (string)) 1658 if (SvMAGICAL (string) || !SvPOK (string))
1424 string = sv_2mortal (newSVsv (string)); 1659 string = sv_2mortal (newSVsv (string));
1425 1660
1426 SvUPGRADE (string, SVt_PV); 1661 SvUPGRADE (string, SVt_PV);
1427 1662
1428 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1663 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1433 * and hope for the best. 1668 * and hope for the best.
1434 * Damnit, SvPV_nolen still trips over yet another assertion. This 1669 * Damnit, SvPV_nolen still trips over yet another assertion. This
1435 * assertion business is seriously broken, try yet another workaround 1670 * assertion business is seriously broken, try yet another workaround
1436 * for the broken -DDEBUGGING. 1671 * for the broken -DDEBUGGING.
1437 */ 1672 */
1673 {
1438#ifdef DEBUGGING 1674#ifdef DEBUGGING
1439 offset = SvOK (string) ? sv_len (string) : 0; 1675 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1440#else 1676#else
1441 offset = SvCUR (string); 1677 STRLEN offset = SvCUR (string);
1442#endif 1678#endif
1443 1679
1444 if (offset > json->max_size && json->max_size) 1680 if (offset > json->max_size && json->max_size)
1445 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1681 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1446 (unsigned long)SvCUR (string), (unsigned long)json->max_size); 1682 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1683 }
1447 1684
1448 if (json->flags & F_UTF8) 1685 if (DECODE_WANTS_OCTETS (json))
1449 sv_utf8_downgrade (string, 0); 1686 sv_utf8_downgrade (string, 0);
1450 else 1687 else
1451 sv_utf8_upgrade (string); 1688 sv_utf8_upgrade (string);
1452 1689
1453 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1690 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1463 1700
1464 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1701 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1465 1702
1466 decode_ws (&dec); 1703 decode_ws (&dec);
1467 sv = decode_sv (&dec); 1704 sv = decode_sv (&dec);
1705
1706 if (offset_return)
1707 *offset_return = dec.cur;
1468 1708
1469 if (!(offset_return || !sv)) 1709 if (!(offset_return || !sv))
1470 { 1710 {
1471 // check for trailing garbage 1711 // check for trailing garbage
1472 decode_ws (&dec); 1712 decode_ws (&dec);
1475 { 1715 {
1476 dec.err = "garbage after JSON object"; 1716 dec.err = "garbage after JSON object";
1477 SvREFCNT_dec (sv); 1717 SvREFCNT_dec (sv);
1478 sv = 0; 1718 sv = 0;
1479 } 1719 }
1480 }
1481
1482 if (offset_return || !sv)
1483 {
1484 offset = dec.json.flags & F_UTF8
1485 ? dec.cur - SvPVX (string)
1486 : utf8_distance (dec.cur, SvPVX (string));
1487
1488 if (offset_return)
1489 *offset_return = offset;
1490 } 1720 }
1491 1721
1492 if (!sv) 1722 if (!sv)
1493 { 1723 {
1494 SV *uni = sv_newmortal (); 1724 SV *uni = sv_newmortal ();
1500 SAVEVPTR (PL_curcop); 1730 SAVEVPTR (PL_curcop);
1501 PL_curcop = &cop; 1731 PL_curcop = &cop;
1502 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1732 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1503 LEAVE; 1733 LEAVE;
1504 1734
1505 croak ("%s, at character offset %d [\"%s\"]", 1735 croak ("%s, at character offset %d (before \"%s\")",
1506 dec.err, 1736 dec.err,
1507 (int)offset, 1737 (int)ptr_to_index (string, dec.cur),
1508 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1738 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1509 } 1739 }
1510 1740
1511 sv = sv_2mortal (sv); 1741 sv = sv_2mortal (sv);
1512 1742
1521 1751
1522static void 1752static void
1523incr_parse (JSON *self) 1753incr_parse (JSON *self)
1524{ 1754{
1525 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1755 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1756
1757 // the state machine here is a bit convoluted and could be simplified a lot
1758 // but this would make it slower, so...
1526 1759
1527 for (;;) 1760 for (;;)
1528 { 1761 {
1529 //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 1762 //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
1530 switch (self->incr_mode) 1763 switch (self->incr_mode)
1531 { 1764 {
1532 // only used for intiial whitespace skipping 1765 // only used for initial whitespace skipping
1533 case INCR_M_WS: 1766 case INCR_M_WS:
1534 for (;;) 1767 for (;;)
1535 { 1768 {
1536 if (*p > 0x20) 1769 if (*p > 0x20)
1537 { 1770 {
1771 if (*p == '#')
1772 {
1773 self->incr_mode = INCR_M_C0;
1774 goto incr_m_c;
1775 }
1776 else
1777 {
1538 self->incr_mode = INCR_M_JSON; 1778 self->incr_mode = INCR_M_JSON;
1539 goto incr_m_json; 1779 goto incr_m_json;
1780 }
1540 } 1781 }
1541 else if (!*p) 1782 else if (!*p)
1542 goto interrupt; 1783 goto interrupt;
1543 1784
1544 ++p; 1785 ++p;
1550 goto interrupt; 1791 goto interrupt;
1551 1792
1552 ++p; 1793 ++p;
1553 self->incr_mode = INCR_M_STR; 1794 self->incr_mode = INCR_M_STR;
1554 goto incr_m_str; 1795 goto incr_m_str;
1796
1797 // inside #-style comments
1798 case INCR_M_C0:
1799 case INCR_M_C1:
1800 incr_m_c:
1801 for (;;)
1802 {
1803 if (*p == '\n')
1804 {
1805 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1806 break;
1807 }
1808 else if (!*p)
1809 goto interrupt;
1810
1811 ++p;
1812 }
1813
1814 break;
1555 1815
1556 // inside a string 1816 // inside a string
1557 case INCR_M_STR: 1817 case INCR_M_STR:
1558 incr_m_str: 1818 incr_m_str:
1559 for (;;) 1819 for (;;)
1610 self->incr_mode = INCR_M_STR; 1870 self->incr_mode = INCR_M_STR;
1611 goto incr_m_str; 1871 goto incr_m_str;
1612 1872
1613 case '[': 1873 case '[':
1614 case '{': 1874 case '{':
1875 case '(':
1615 if (++self->incr_nest > self->max_depth) 1876 if (++self->incr_nest > self->max_depth)
1616 croak (ERR_NESTING_EXCEEDED); 1877 croak (ERR_NESTING_EXCEEDED);
1617 break; 1878 break;
1618 1879
1619 case ']': 1880 case ']':
1620 case '}': 1881 case '}':
1621 if (--self->incr_nest <= 0) 1882 if (--self->incr_nest <= 0)
1622 goto interrupt; 1883 goto interrupt;
1884 break;
1885
1886 case ')':
1887 --self->incr_nest;
1888 break;
1889
1890 case '#':
1891 self->incr_mode = INCR_M_C1;
1892 goto incr_m_c;
1623 } 1893 }
1624 } 1894 }
1625 } 1895 }
1626 1896
1627 modechange: 1897 modechange:
1628 ; 1898 ;
1629 } 1899 }
1630 1900
1631interrupt: 1901interrupt:
1632 self->incr_pos = p - SvPVX (self->incr_text); 1902 self->incr_pos = p - SvPVX (self->incr_text);
1903 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1633 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D 1904 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1634} 1905}
1635 1906
1636///////////////////////////////////////////////////////////////////////////// 1907/////////////////////////////////////////////////////////////////////////////
1637// XS interface functions 1908// XS interface functions
1647 i >= '0' && i <= '9' ? i - '0' 1918 i >= '0' && i <= '9' ? i - '0'
1648 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1919 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1649 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1920 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1650 : -1; 1921 : -1;
1651 1922
1652 json_stash = gv_stashpv ("JSON::XS" , 1); 1923 json_stash = gv_stashpv ("JSON::XS" , 1);
1653 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1924 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1654 1925
1655 json_true = get_bool ("JSON::XS::true"); 1926 types_true = get_bool ("Types::Serialiser::true");
1656 json_false = get_bool ("JSON::XS::false"); 1927 types_false = get_bool ("Types::Serialiser::false");
1928
1929 sv_json = newSVpv ("JSON", 0);
1930 SvREADONLY_on (sv_json);
1931
1932 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1657} 1933}
1658 1934
1659PROTOTYPES: DISABLE 1935PROTOTYPES: DISABLE
1660 1936
1661void CLONE (...) 1937void CLONE (...)
1662 CODE: 1938 CODE:
1663 json_stash = 0; 1939 json_stash = 0;
1664 json_boolean_stash = 0; 1940 types_boolean_stash = 0;
1665 1941
1666void new (char *klass) 1942void new (char *klass)
1667 PPCODE: 1943 PPCODE:
1668{ 1944{
1669 SV *pv = NEWSV (0, sizeof (JSON)); 1945 SV *pv = NEWSV (0, sizeof (JSON));
1670 SvPOK_only (pv); 1946 SvPOK_only (pv);
1671 json_init ((JSON *)SvPVX (pv)); 1947 json_init ((JSON *)SvPVX (pv));
1672 XPUSHs (sv_2mortal (sv_bless ( 1948 XPUSHs (sv_2mortal (sv_bless (
1673 newRV_noinc (pv), 1949 newRV_noinc (pv),
1674 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1950 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1689 shrink = F_SHRINK 1965 shrink = F_SHRINK
1690 allow_blessed = F_ALLOW_BLESSED 1966 allow_blessed = F_ALLOW_BLESSED
1691 convert_blessed = F_CONV_BLESSED 1967 convert_blessed = F_CONV_BLESSED
1692 relaxed = F_RELAXED 1968 relaxed = F_RELAXED
1693 allow_unknown = F_ALLOW_UNKNOWN 1969 allow_unknown = F_ALLOW_UNKNOWN
1970 allow_tags = F_ALLOW_TAGS
1694 PPCODE: 1971 PPCODE:
1695{ 1972{
1696 if (enable) 1973 if (enable)
1697 self->flags |= ix; 1974 self->flags |= ix;
1698 else 1975 else
1714 get_shrink = F_SHRINK 1991 get_shrink = F_SHRINK
1715 get_allow_blessed = F_ALLOW_BLESSED 1992 get_allow_blessed = F_ALLOW_BLESSED
1716 get_convert_blessed = F_CONV_BLESSED 1993 get_convert_blessed = F_CONV_BLESSED
1717 get_relaxed = F_RELAXED 1994 get_relaxed = F_RELAXED
1718 get_allow_unknown = F_ALLOW_UNKNOWN 1995 get_allow_unknown = F_ALLOW_UNKNOWN
1996 get_allow_tags = F_ALLOW_TAGS
1719 PPCODE: 1997 PPCODE:
1720 XPUSHs (boolSV (self->flags & ix)); 1998 XPUSHs (boolSV (self->flags & ix));
1721 1999
1722void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 2000void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1723 PPCODE: 2001 PPCODE:
1751} 2029}
1752 2030
1753void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2031void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1754 PPCODE: 2032 PPCODE:
1755{ 2033{
1756 if (!self->cb_sk_object) 2034 if (!self->cb_sk_object)
1757 self->cb_sk_object = newHV (); 2035 self->cb_sk_object = newHV ();
1758 2036
1759 if (SvOK (cb)) 2037 if (SvOK (cb))
1760 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2038 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1761 else 2039 else
1772 XPUSHs (ST (0)); 2050 XPUSHs (ST (0));
1773} 2051}
1774 2052
1775void encode (JSON *self, SV *scalar) 2053void encode (JSON *self, SV *scalar)
1776 PPCODE: 2054 PPCODE:
1777 XPUSHs (encode_json (scalar, self)); 2055 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2056 XPUSHs (scalar);
1778 2057
1779void decode (JSON *self, SV *jsonstr) 2058void decode (JSON *self, SV *jsonstr)
1780 PPCODE: 2059 PPCODE:
1781 XPUSHs (decode_json (jsonstr, self, 0)); 2060 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2061 XPUSHs (jsonstr);
1782 2062
1783void decode_prefix (JSON *self, SV *jsonstr) 2063void decode_prefix (JSON *self, SV *jsonstr)
1784 PPCODE: 2064 PPCODE:
1785{ 2065{
1786 STRLEN offset; 2066 SV *sv;
2067 char *offset;
2068 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1787 EXTEND (SP, 2); 2069 EXTEND (SP, 2);
1788 PUSHs (decode_json (jsonstr, self, &offset)); 2070 PUSHs (sv);
1789 PUSHs (sv_2mortal (newSVuv (offset))); 2071 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1790} 2072}
1791 2073
1792void incr_parse (JSON *self, SV *jsonstr = 0) 2074void incr_parse (JSON *self, SV *jsonstr = 0)
1793 PPCODE: 2075 PPCODE:
1794{ 2076{
1795 if (!self->incr_text) 2077 if (!self->incr_text)
1796 self->incr_text = newSVpvn ("", 0); 2078 self->incr_text = newSVpvn ("", 0);
2079
2080 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2081 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2082 if (DECODE_WANTS_OCTETS (self))
2083 {
2084 if (self->incr_pos)
2085 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2086 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2087
2088 sv_utf8_downgrade (self->incr_text, 0);
2089 }
2090 else
2091 {
2092 sv_utf8_upgrade (self->incr_text);
2093
2094 if (self->incr_pos)
2095 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2096 - (U8 *)SvPVX (self->incr_text);
2097 }
1797 2098
1798 // append data, if any 2099 // append data, if any
1799 if (jsonstr) 2100 if (jsonstr)
1800 { 2101 {
2102 /* make sure both strings have same encoding */
1801 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 2103 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2104 if (SvUTF8 (jsonstr))
2105 sv_utf8_downgrade (jsonstr, 0);
1802 { 2106 else
1803 /* utf-8-ness differs, need to upgrade */
1804 sv_utf8_upgrade (self->incr_text); 2107 sv_utf8_upgrade (jsonstr);
1805 2108
1806 if (self->incr_pos) 2109 /* and then just blindly append */
1807 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1808 - (U8 *)SvPVX (self->incr_text);
1809 }
1810
1811 { 2110 {
1812 STRLEN len; 2111 STRLEN len;
1813 const char *str = SvPV (jsonstr, len); 2112 const char *str = SvPV (jsonstr, len);
1814 STRLEN cur = SvCUR (self->incr_text); 2113 STRLEN cur = SvCUR (self->incr_text);
1815 2114
1823 } 2122 }
1824 2123
1825 if (GIMME_V != G_VOID) 2124 if (GIMME_V != G_VOID)
1826 do 2125 do
1827 { 2126 {
2127 SV *sv;
1828 STRLEN offset; 2128 char *offset;
1829 2129
1830 if (!INCR_DONE (self)) 2130 if (!INCR_DONE (self))
1831 { 2131 {
1832 incr_parse (self); 2132 incr_parse (self);
1833 2133
1834 if (self->incr_pos > self->max_size && self->max_size) 2134 if (self->incr_pos > self->max_size && self->max_size)
1835 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 2135 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1836 (unsigned long)self->incr_pos, (unsigned long)self->max_size); 2136 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1837 2137
1838 if (!INCR_DONE (self)) 2138 if (!INCR_DONE (self))
2139 {
2140 // as an optimisation, do not accumulate white space in the incr buffer
2141 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2142 {
2143 self->incr_pos = 0;
2144 SvCUR_set (self->incr_text, 0);
2145 }
2146
1839 break; 2147 break;
2148 }
1840 } 2149 }
1841 2150
1842 XPUSHs (decode_json (self->incr_text, self, &offset)); 2151 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2152 XPUSHs (sv);
1843 2153
1844 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1845 self->incr_pos -= offset; 2154 self->incr_pos -= offset - SvPVX (self->incr_text);
1846 self->incr_nest = 0; 2155 self->incr_nest = 0;
1847 self->incr_mode = 0; 2156 self->incr_mode = 0;
2157
2158 sv_chop (self->incr_text, offset);
1848 } 2159 }
1849 while (GIMME_V == G_ARRAY); 2160 while (GIMME_V == G_ARRAY);
1850} 2161}
1851 2162
1852SV *incr_text (JSON *self) 2163SV *incr_text (JSON *self)
1890 SvREFCNT_dec (self->incr_text); 2201 SvREFCNT_dec (self->incr_text);
1891 2202
1892PROTOTYPES: ENABLE 2203PROTOTYPES: ENABLE
1893 2204
1894void encode_json (SV *scalar) 2205void encode_json (SV *scalar)
1895 ALIAS:
1896 to_json_ = 0
1897 encode_json = F_UTF8
1898 PPCODE: 2206 PPCODE:
1899{ 2207{
1900 JSON json; 2208 JSON json;
1901 json_init (&json); 2209 json_init (&json);
1902 json.flags |= ix; 2210 json.flags |= F_UTF8;
1903 XPUSHs (encode_json (scalar, &json)); 2211 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2212 XPUSHs (scalar);
1904} 2213}
1905 2214
1906void decode_json (SV *jsonstr) 2215void decode_json (SV *jsonstr)
1907 ALIAS:
1908 from_json_ = 0
1909 decode_json = F_UTF8
1910 PPCODE: 2216 PPCODE:
1911{ 2217{
1912 JSON json; 2218 JSON json;
1913 json_init (&json); 2219 json_init (&json);
1914 json.flags |= ix; 2220 json.flags |= F_UTF8;
1915 XPUSHs (decode_json (jsonstr, &json, 0)); 2221 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2222 XPUSHs (jsonstr);
1916} 2223}
1917 2224
1918

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines