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 1.122 by root, Tue Oct 29 15:55:49 2013 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::TO_JSON 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++;
… …
1075 1226
1076 // special case the rather common 1..5-digit-int case 1227 // special case the rather common 1..5-digit-int case
1077 if (*start == '-') 1228 if (*start == '-')
1078 switch (len) 1229 switch (len)
1079 { 1230 {
1080 case 2: return newSViv (-( start [1] - '0' * 1)); 1231 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1081 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1232 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)); 1233 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)); 1234 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)); 1235 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1085 } 1236 }
1086 else 1237 else
1087 switch (len) 1238 switch (len)
1088 { 1239 {
1089 case 1: return newSViv ( start [0] - '0' * 1); 1240 case 1: return newSViv ( start [0] - '0' * 1);
1090 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1241 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); 1242 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); 1243 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); 1244 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1094 } 1245 }
1095 1246
1096 { 1247 {
1097 UV uv; 1248 UV uv;
1098 int numtype = grok_number (start, len, &uv); 1249 int numtype = grok_number (start, len, &uv);
… …
1107 } 1258 }
1108 1259
1109 len -= *start == '-' ? 1 : 0; 1260 len -= *start == '-' ? 1 : 0;
1110 1261
1111 // does not fit into IV or UV, try NV 1262 // does not fit into IV or UV, try NV
1112 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1263 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 1264 // fits into NV without loss of precision
1118 return newSVnv (Atof (start)); 1265 return newSVnv (json_atof (start));
1119 1266
1120 // everything else fails, convert it to a string 1267 // everything else fails, convert it to a string
1121 return newSVpvn (start, dec->cur - start); 1268 return newSVpvn (start, dec->cur - start);
1122 } 1269 }
1123 1270
1124 // loss of precision here 1271 // loss of precision here
1125 return newSVnv (Atof (start)); 1272 return newSVnv (json_atof (start));
1126 1273
1127fail: 1274fail:
1128 return 0; 1275 return 0;
1129} 1276}
1130 1277
… …
1154 if (*dec->cur == ']') 1301 if (*dec->cur == ']')
1155 { 1302 {
1156 ++dec->cur; 1303 ++dec->cur;
1157 break; 1304 break;
1158 } 1305 }
1159 1306
1160 if (*dec->cur != ',') 1307 if (*dec->cur != ',')
1161 ERR (", or ] expected while parsing array"); 1308 ERR (", or ] expected while parsing array");
1162 1309
1163 ++dec->cur; 1310 ++dec->cur;
1164 1311
… …
1299 dSP; 1446 dSP;
1300 int count; 1447 int count;
1301 1448
1302 ENTER; SAVETMPS; PUSHMARK (SP); 1449 ENTER; SAVETMPS; PUSHMARK (SP);
1303 XPUSHs (HeVAL (he)); 1450 XPUSHs (HeVAL (he));
1451 sv_2mortal (sv);
1304 1452
1305 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1453 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1306 1454
1307 if (count == 1) 1455 if (count == 1)
1308 { 1456 {
1309 sv = newSVsv (POPs); 1457 sv = newSVsv (POPs);
1310 FREETMPS; LEAVE; 1458 FREETMPS; LEAVE;
1311 return sv; 1459 return sv;
1312 } 1460 }
1313 1461
1462 SvREFCNT_inc (sv);
1314 FREETMPS; LEAVE; 1463 FREETMPS; LEAVE;
1315 } 1464 }
1316 } 1465 }
1317 1466
1318 if (dec->json.cb_object) 1467 if (dec->json.cb_object)
… …
1344 DEC_DEC_DEPTH; 1493 DEC_DEC_DEPTH;
1345 return 0; 1494 return 0;
1346} 1495}
1347 1496
1348static SV * 1497static SV *
1498decode_tag (dec_t *dec)
1499{
1500 SV *tag = 0;
1501 SV *val = 0;
1502
1503 if (!(dec->json.flags & F_ALLOW_TAGS))
1504 ERR ("malformed JSON string, neither array, object, number, string or atom");
1505
1506 ++dec->cur;
1507
1508 decode_ws (dec);
1509
1510 tag = decode_sv (dec);
1511 if (!tag)
1512 goto fail;
1513
1514 if (!SvPOK (tag))
1515 ERR ("malformed JSON string, (tag) must be a string");
1516
1517 decode_ws (dec);
1518
1519 if (*dec->cur != ')')
1520 ERR (") expected after tag");
1521
1522 ++dec->cur;
1523
1524 decode_ws (dec);
1525
1526 val = decode_sv (dec);
1527 if (!val)
1528 goto fail;
1529
1530 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1531 ERR ("malformed JSON string, tag value must be an array");
1532
1533 {
1534 AV *av = (AV *)SvRV (val);
1535 int i, len = av_len (av) + 1;
1536 HV *stash = gv_stashsv (tag, 0);
1537 SV *sv;
1538
1539 if (!stash)
1540 ERR ("cannot decode perl-object (package does not exist)");
1541
1542 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1543
1544 if (!method)
1545 ERR ("cannot decode perl-object (package does not have a THAW method)");
1546
1547 dSP;
1548
1549 ENTER; SAVETMPS; PUSHMARK (SP);
1550 EXTEND (SP, len + 2);
1551 // we re-bless the reference to get overload and other niceties right
1552 PUSHs (tag);
1553 PUSHs (sv_json);
1554
1555 for (i = 0; i < len; ++i)
1556 PUSHs (*av_fetch (av, i, 1));
1557
1558 PUTBACK;
1559 call_sv ((SV *)GvCV (method), G_SCALAR);
1560 SPAGAIN;
1561
1562 SvREFCNT_dec (tag);
1563 SvREFCNT_dec (val);
1564 sv = SvREFCNT_inc (POPs);
1565
1566 PUTBACK;
1567
1568 FREETMPS; LEAVE;
1569
1570 return sv;
1571 }
1572
1573fail:
1574 SvREFCNT_dec (tag);
1575 SvREFCNT_dec (val);
1576 return 0;
1577}
1578
1579static SV *
1349decode_sv (dec_t *dec) 1580decode_sv (dec_t *dec)
1350{ 1581{
1351 // the beauty of JSON: you need exactly one character lookahead 1582 // the beauty of JSON: you need exactly one character lookahead
1352 // to parse everything. 1583 // to parse everything.
1353 switch (*dec->cur) 1584 switch (*dec->cur)
1354 { 1585 {
1355 case '"': ++dec->cur; return decode_str (dec); 1586 case '"': ++dec->cur; return decode_str (dec);
1356 case '[': ++dec->cur; return decode_av (dec); 1587 case '[': ++dec->cur; return decode_av (dec);
1357 case '{': ++dec->cur; return decode_hv (dec); 1588 case '{': ++dec->cur; return decode_hv (dec);
1589 case '(': return decode_tag (dec);
1358 1590
1359 case '-': 1591 case '-':
1360 case '0': case '1': case '2': case '3': case '4': 1592 case '0': case '1': case '2': case '3': case '4':
1361 case '5': case '6': case '7': case '8': case '9': 1593 case '5': case '6': case '7': case '8': case '9':
1362 return decode_num (dec); 1594 return decode_num (dec);
… …
1364 case 't': 1596 case 't':
1365 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1597 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1366 { 1598 {
1367 dec->cur += 4; 1599 dec->cur += 4;
1368#if JSON_SLOW 1600#if JSON_SLOW
1369 json_true = get_bool ("JSON::XS::true"); 1601 types_true = get_bool ("Types::Serialiser::true");
1370#endif 1602#endif
1371 return newSVsv (json_true); 1603 return newSVsv (types_true);
1372 } 1604 }
1373 else 1605 else
1374 ERR ("'true' expected"); 1606 ERR ("'true' expected");
1375 1607
1376 break; 1608 break;
… …
1378 case 'f': 1610 case 'f':
1379 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1611 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1380 { 1612 {
1381 dec->cur += 5; 1613 dec->cur += 5;
1382#if JSON_SLOW 1614#if JSON_SLOW
1383 json_false = get_bool ("JSON::XS::false"); 1615 types_false = get_bool ("Types::Serialiser::false");
1384#endif 1616#endif
1385 return newSVsv (json_false); 1617 return newSVsv (types_false);
1386 } 1618 }
1387 else 1619 else
1388 ERR ("'false' expected"); 1620 ERR ("'false' expected");
1389 1621
1390 break; 1622 break;
… …
1399 ERR ("'null' expected"); 1631 ERR ("'null' expected");
1400 1632
1401 break; 1633 break;
1402 1634
1403 default: 1635 default:
1404 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1636 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1405 break; 1637 break;
1406 } 1638 }
1407 1639
1408fail: 1640fail:
1409 return 0; 1641 return 0;
1410} 1642}
1411 1643
1412static SV * 1644static SV *
1413decode_json (SV *string, JSON *json, STRLEN *offset_return) 1645decode_json (SV *string, JSON *json, char **offset_return)
1414{ 1646{
1415 dec_t dec; 1647 dec_t dec;
1416 STRLEN offset;
1417 SV *sv; 1648 SV *sv;
1418 1649
1419 /* work around bugs in 5.10 where manipulating magic values 1650 /* work around bugs in 5.10 where manipulating magic values
1420 * will perl ignore the magic in subsequent accesses 1651 * makes perl ignore the magic in subsequent accesses.
1652 * also make a copy of non-PV values, to get them into a clean
1653 * state (SvPV should do that, but it's buggy, see below).
1421 */ 1654 */
1422 /*SvGETMAGIC (string);*/ 1655 /*SvGETMAGIC (string);*/
1423 if (SvMAGICAL (string)) 1656 if (SvMAGICAL (string) || !SvPOK (string))
1424 string = sv_2mortal (newSVsv (string)); 1657 string = sv_2mortal (newSVsv (string));
1425 1658
1426 SvUPGRADE (string, SVt_PV); 1659 SvUPGRADE (string, SVt_PV);
1427 1660
1428 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1661 /* work around a bug in perl 5.10, which causes SvCUR to fail an
… …
1433 * and hope for the best. 1666 * and hope for the best.
1434 * Damnit, SvPV_nolen still trips over yet another assertion. This 1667 * Damnit, SvPV_nolen still trips over yet another assertion. This
1435 * assertion business is seriously broken, try yet another workaround 1668 * assertion business is seriously broken, try yet another workaround
1436 * for the broken -DDEBUGGING. 1669 * for the broken -DDEBUGGING.
1437 */ 1670 */
1671 {
1438#ifdef DEBUGGING 1672#ifdef DEBUGGING
1439 offset = SvOK (string) ? sv_len (string) : 0; 1673 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1440#else 1674#else
1441 offset = SvCUR (string); 1675 STRLEN offset = SvCUR (string);
1442#endif 1676#endif
1443 1677
1444 if (offset > json->max_size && json->max_size) 1678 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", 1679 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); 1680 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1681 }
1447 1682
1448 if (json->flags & F_UTF8) 1683 if (DECODE_WANTS_OCTETS (json))
1449 sv_utf8_downgrade (string, 0); 1684 sv_utf8_downgrade (string, 0);
1450 else 1685 else
1451 sv_utf8_upgrade (string); 1686 sv_utf8_upgrade (string);
1452 1687
1453 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1688 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
… …
1463 1698
1464 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1699 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1465 1700
1466 decode_ws (&dec); 1701 decode_ws (&dec);
1467 sv = decode_sv (&dec); 1702 sv = decode_sv (&dec);
1703
1704 if (offset_return)
1705 *offset_return = dec.cur;
1468 1706
1469 if (!(offset_return || !sv)) 1707 if (!(offset_return || !sv))
1470 { 1708 {
1471 // check for trailing garbage 1709 // check for trailing garbage
1472 decode_ws (&dec); 1710 decode_ws (&dec);
… …
1475 { 1713 {
1476 dec.err = "garbage after JSON object"; 1714 dec.err = "garbage after JSON object";
1477 SvREFCNT_dec (sv); 1715 SvREFCNT_dec (sv);
1478 sv = 0; 1716 sv = 0;
1479 } 1717 }
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 } 1718 }
1491 1719
1492 if (!sv) 1720 if (!sv)
1493 { 1721 {
1494 SV *uni = sv_newmortal (); 1722 SV *uni = sv_newmortal ();
… …
1500 SAVEVPTR (PL_curcop); 1728 SAVEVPTR (PL_curcop);
1501 PL_curcop = &cop; 1729 PL_curcop = &cop;
1502 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1730 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1503 LEAVE; 1731 LEAVE;
1504 1732
1505 croak ("%s, at character offset %d [\"%s\"]", 1733 croak ("%s, at character offset %d (before \"%s\")",
1506 dec.err, 1734 dec.err,
1507 (int)offset, 1735 (int)ptr_to_index (string, dec.cur),
1508 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1736 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1509 } 1737 }
1510 1738
1511 sv = sv_2mortal (sv); 1739 sv = sv_2mortal (sv);
1512 1740
… …
1521 1749
1522static void 1750static void
1523incr_parse (JSON *self) 1751incr_parse (JSON *self)
1524{ 1752{
1525 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1753 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1754
1755 // the state machine here is a bit convoluted and could be simplified a lot
1756 // but this would make it slower, so...
1526 1757
1527 for (;;) 1758 for (;;)
1528 { 1759 {
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 1760 //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) 1761 switch (self->incr_mode)
1531 { 1762 {
1532 // only used for intiial whitespace skipping 1763 // only used for initial whitespace skipping
1533 case INCR_M_WS: 1764 case INCR_M_WS:
1534 for (;;) 1765 for (;;)
1535 { 1766 {
1536 if (*p > 0x20) 1767 if (*p > 0x20)
1537 { 1768 {
1769 if (*p == '#')
1770 {
1771 self->incr_mode = INCR_M_C0;
1772 goto incr_m_c;
1773 }
1774 else
1775 {
1538 self->incr_mode = INCR_M_JSON; 1776 self->incr_mode = INCR_M_JSON;
1539 goto incr_m_json; 1777 goto incr_m_json;
1778 }
1540 } 1779 }
1541 else if (!*p) 1780 else if (!*p)
1542 goto interrupt; 1781 goto interrupt;
1543 1782
1544 ++p; 1783 ++p;
… …
1550 goto interrupt; 1789 goto interrupt;
1551 1790
1552 ++p; 1791 ++p;
1553 self->incr_mode = INCR_M_STR; 1792 self->incr_mode = INCR_M_STR;
1554 goto incr_m_str; 1793 goto incr_m_str;
1794
1795 // inside #-style comments
1796 case INCR_M_C0:
1797 case INCR_M_C1:
1798 incr_m_c:
1799 for (;;)
1800 {
1801 if (*p == '\n')
1802 {
1803 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1804 break;
1805 }
1806 else if (!*p)
1807 goto interrupt;
1808
1809 ++p;
1810 }
1811
1812 break;
1555 1813
1556 // inside a string 1814 // inside a string
1557 case INCR_M_STR: 1815 case INCR_M_STR:
1558 incr_m_str: 1816 incr_m_str:
1559 for (;;) 1817 for (;;)
… …
1610 self->incr_mode = INCR_M_STR; 1868 self->incr_mode = INCR_M_STR;
1611 goto incr_m_str; 1869 goto incr_m_str;
1612 1870
1613 case '[': 1871 case '[':
1614 case '{': 1872 case '{':
1873 case '(':
1615 if (++self->incr_nest > self->max_depth) 1874 if (++self->incr_nest > self->max_depth)
1616 croak (ERR_NESTING_EXCEEDED); 1875 croak (ERR_NESTING_EXCEEDED);
1617 break; 1876 break;
1618 1877
1619 case ']': 1878 case ']':
1620 case '}': 1879 case '}':
1621 if (--self->incr_nest <= 0) 1880 if (--self->incr_nest <= 0)
1622 goto interrupt; 1881 goto interrupt;
1882 break;
1883
1884 case ')':
1885 --self->incr_nest;
1886 break;
1887
1888 case '#':
1889 self->incr_mode = INCR_M_C1;
1890 goto incr_m_c;
1623 } 1891 }
1624 } 1892 }
1625 } 1893 }
1626 1894
1627 modechange: 1895 modechange:
1628 ; 1896 ;
1629 } 1897 }
1630 1898
1631interrupt: 1899interrupt:
1632 self->incr_pos = p - SvPVX (self->incr_text); 1900 self->incr_pos = p - SvPVX (self->incr_text);
1901 //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 1902 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1634} 1903}
1635 1904
1636///////////////////////////////////////////////////////////////////////////// 1905/////////////////////////////////////////////////////////////////////////////
1637// XS interface functions 1906// XS interface functions
… …
1647 i >= '0' && i <= '9' ? i - '0' 1916 i >= '0' && i <= '9' ? i - '0'
1648 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1917 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1649 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1918 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1650 : -1; 1919 : -1;
1651 1920
1652 json_stash = gv_stashpv ("JSON::XS" , 1); 1921 json_stash = gv_stashpv ("JSON::XS" , 1);
1653 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1922 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1654 1923
1655 json_true = get_bool ("JSON::XS::true"); 1924 types_true = get_bool ("Types::Serialiser::true");
1656 json_false = get_bool ("JSON::XS::false"); 1925 types_false = get_bool ("Types::Serialiser::false");
1926
1927 sv_json = newSVpv ("JSON", 0);
1928 SvREADONLY_on (sv_json);
1929
1930 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1657} 1931}
1658 1932
1659PROTOTYPES: DISABLE 1933PROTOTYPES: DISABLE
1660 1934
1661void CLONE (...) 1935void CLONE (...)
1662 CODE: 1936 CODE:
1663 json_stash = 0; 1937 json_stash = 0;
1664 json_boolean_stash = 0; 1938 types_boolean_stash = 0;
1665 1939
1666void new (char *klass) 1940void new (char *klass)
1667 PPCODE: 1941 PPCODE:
1668{ 1942{
1669 SV *pv = NEWSV (0, sizeof (JSON)); 1943 SV *pv = NEWSV (0, sizeof (JSON));
1670 SvPOK_only (pv); 1944 SvPOK_only (pv);
1671 json_init ((JSON *)SvPVX (pv)); 1945 json_init ((JSON *)SvPVX (pv));
1672 XPUSHs (sv_2mortal (sv_bless ( 1946 XPUSHs (sv_2mortal (sv_bless (
1673 newRV_noinc (pv), 1947 newRV_noinc (pv),
1674 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1948 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
… …
1689 shrink = F_SHRINK 1963 shrink = F_SHRINK
1690 allow_blessed = F_ALLOW_BLESSED 1964 allow_blessed = F_ALLOW_BLESSED
1691 convert_blessed = F_CONV_BLESSED 1965 convert_blessed = F_CONV_BLESSED
1692 relaxed = F_RELAXED 1966 relaxed = F_RELAXED
1693 allow_unknown = F_ALLOW_UNKNOWN 1967 allow_unknown = F_ALLOW_UNKNOWN
1968 allow_tags = F_ALLOW_TAGS
1694 PPCODE: 1969 PPCODE:
1695{ 1970{
1696 if (enable) 1971 if (enable)
1697 self->flags |= ix; 1972 self->flags |= ix;
1698 else 1973 else
… …
1714 get_shrink = F_SHRINK 1989 get_shrink = F_SHRINK
1715 get_allow_blessed = F_ALLOW_BLESSED 1990 get_allow_blessed = F_ALLOW_BLESSED
1716 get_convert_blessed = F_CONV_BLESSED 1991 get_convert_blessed = F_CONV_BLESSED
1717 get_relaxed = F_RELAXED 1992 get_relaxed = F_RELAXED
1718 get_allow_unknown = F_ALLOW_UNKNOWN 1993 get_allow_unknown = F_ALLOW_UNKNOWN
1994 get_allow_tags = F_ALLOW_TAGS
1719 PPCODE: 1995 PPCODE:
1720 XPUSHs (boolSV (self->flags & ix)); 1996 XPUSHs (boolSV (self->flags & ix));
1721 1997
1722void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 1998void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1723 PPCODE: 1999 PPCODE:
… …
1751} 2027}
1752 2028
1753void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2029void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1754 PPCODE: 2030 PPCODE:
1755{ 2031{
1756 if (!self->cb_sk_object) 2032 if (!self->cb_sk_object)
1757 self->cb_sk_object = newHV (); 2033 self->cb_sk_object = newHV ();
1758 2034
1759 if (SvOK (cb)) 2035 if (SvOK (cb))
1760 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2036 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1761 else 2037 else
… …
1772 XPUSHs (ST (0)); 2048 XPUSHs (ST (0));
1773} 2049}
1774 2050
1775void encode (JSON *self, SV *scalar) 2051void encode (JSON *self, SV *scalar)
1776 PPCODE: 2052 PPCODE:
1777 XPUSHs (encode_json (scalar, self)); 2053 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2054 XPUSHs (scalar);
1778 2055
1779void decode (JSON *self, SV *jsonstr) 2056void decode (JSON *self, SV *jsonstr)
1780 PPCODE: 2057 PPCODE:
1781 XPUSHs (decode_json (jsonstr, self, 0)); 2058 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2059 XPUSHs (jsonstr);
1782 2060
1783void decode_prefix (JSON *self, SV *jsonstr) 2061void decode_prefix (JSON *self, SV *jsonstr)
1784 PPCODE: 2062 PPCODE:
1785{ 2063{
1786 STRLEN offset; 2064 SV *sv;
2065 char *offset;
2066 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1787 EXTEND (SP, 2); 2067 EXTEND (SP, 2);
1788 PUSHs (decode_json (jsonstr, self, &offset)); 2068 PUSHs (sv);
1789 PUSHs (sv_2mortal (newSVuv (offset))); 2069 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1790} 2070}
1791 2071
1792void incr_parse (JSON *self, SV *jsonstr = 0) 2072void incr_parse (JSON *self, SV *jsonstr = 0)
1793 PPCODE: 2073 PPCODE:
1794{ 2074{
1795 if (!self->incr_text) 2075 if (!self->incr_text)
1796 self->incr_text = newSVpvn ("", 0); 2076 self->incr_text = newSVpvn ("", 0);
2077
2078 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2079 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2080 if (DECODE_WANTS_OCTETS (self))
2081 {
2082 if (self->incr_pos)
2083 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2084 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2085
2086 sv_utf8_downgrade (self->incr_text, 0);
2087 }
2088 else
2089 {
2090 sv_utf8_upgrade (self->incr_text);
2091
2092 if (self->incr_pos)
2093 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2094 - (U8 *)SvPVX (self->incr_text);
2095 }
1797 2096
1798 // append data, if any 2097 // append data, if any
1799 if (jsonstr) 2098 if (jsonstr)
1800 { 2099 {
2100 /* make sure both strings have same encoding */
1801 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 2101 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2102 if (SvUTF8 (jsonstr))
2103 sv_utf8_downgrade (jsonstr, 0);
1802 { 2104 else
1803 /* utf-8-ness differs, need to upgrade */
1804 sv_utf8_upgrade (self->incr_text); 2105 sv_utf8_upgrade (jsonstr);
1805 2106
1806 if (self->incr_pos) 2107 /* 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 { 2108 {
1812 STRLEN len; 2109 STRLEN len;
1813 const char *str = SvPV (jsonstr, len); 2110 const char *str = SvPV (jsonstr, len);
1814 STRLEN cur = SvCUR (self->incr_text); 2111 STRLEN cur = SvCUR (self->incr_text);
1815 2112
… …
1823 } 2120 }
1824 2121
1825 if (GIMME_V != G_VOID) 2122 if (GIMME_V != G_VOID)
1826 do 2123 do
1827 { 2124 {
2125 SV *sv;
1828 STRLEN offset; 2126 char *offset;
1829 2127
1830 if (!INCR_DONE (self)) 2128 if (!INCR_DONE (self))
1831 { 2129 {
1832 incr_parse (self); 2130 incr_parse (self);
1833 2131
1834 if (self->incr_pos > self->max_size && self->max_size) 2132 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", 2133 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); 2134 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1837 2135
1838 if (!INCR_DONE (self)) 2136 if (!INCR_DONE (self))
2137 {
2138 // as an optimisation, do not accumulate white space in the incr buffer
2139 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2140 {
2141 self->incr_pos = 0;
2142 SvCUR_set (self->incr_text, 0);
2143 }
2144
1839 break; 2145 break;
2146 }
1840 } 2147 }
1841 2148
1842 XPUSHs (decode_json (self->incr_text, self, &offset)); 2149 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2150 XPUSHs (sv);
1843 2151
1844 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1845 self->incr_pos -= offset; 2152 self->incr_pos -= offset - SvPVX (self->incr_text);
1846 self->incr_nest = 0; 2153 self->incr_nest = 0;
1847 self->incr_mode = 0; 2154 self->incr_mode = 0;
2155
2156 sv_chop (self->incr_text, offset);
1848 } 2157 }
1849 while (GIMME_V == G_ARRAY); 2158 while (GIMME_V == G_ARRAY);
1850} 2159}
1851 2160
1852SV *incr_text (JSON *self) 2161SV *incr_text (JSON *self)
… …
1890 SvREFCNT_dec (self->incr_text); 2199 SvREFCNT_dec (self->incr_text);
1891 2200
1892PROTOTYPES: ENABLE 2201PROTOTYPES: ENABLE
1893 2202
1894void encode_json (SV *scalar) 2203void encode_json (SV *scalar)
1895 ALIAS:
1896 to_json_ = 0
1897 encode_json = F_UTF8
1898 PPCODE: 2204 PPCODE:
1899{ 2205{
1900 JSON json; 2206 JSON json;
1901 json_init (&json); 2207 json_init (&json);
1902 json.flags |= ix; 2208 json.flags |= F_UTF8;
1903 XPUSHs (encode_json (scalar, &json)); 2209 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2210 XPUSHs (scalar);
1904} 2211}
1905 2212
1906void decode_json (SV *jsonstr) 2213void decode_json (SV *jsonstr)
1907 ALIAS:
1908 from_json_ = 0
1909 decode_json = F_UTF8
1910 PPCODE: 2214 PPCODE:
1911{ 2215{
1912 JSON json; 2216 JSON json;
1913 json_init (&json); 2217 json_init (&json);
1914 json.flags |= ix; 2218 json.flags |= F_UTF8;
1915 XPUSHs (decode_json (jsonstr, &json, 0)); 2219 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2220 XPUSHs (jsonstr);
1916} 2221}
1917 2222
1918

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines