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 1.121 by root, Tue Oct 29 00:06:40 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// 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
… …
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#define F_ALLOW_UNKNOWN 0x00002000UL 37#define F_ALLOW_UNKNOWN 0x00002000UL
38#define F_ALLOW_TAGS 0x00004000UL
37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 39#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
38 40
39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
40 42
41#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
42#define INDENT_STEP 3 // spaces per indentation level 44#define INDENT_STEP 3 // spaces per indentation level
43 45
44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
47
48#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
45 49
46#define SB do { 50#define SB do {
47#define SE } while (0) 51#define SE } while (0)
48 52
49#if __GNUC__ >= 3 53#if __GNUC__ >= 3
… …
69#else 73#else
70# define JSON_SLOW 0 74# define JSON_SLOW 0
71# define JSON_STASH json_stash 75# define JSON_STASH json_stash
72#endif 76#endif
73 77
78// the amount of HEs to allocate on the stack, when sorting keys
79#define STACK_HES 64
80
74static HV *json_stash, *json_boolean_stash; // JSON::XS:: 81static HV *json_stash, *types_boolean_stash; // JSON::XS::
75static SV *json_true, *json_false; 82static SV *types_true, *types_false, *sv_json;
76 83
77enum { 84enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0 85 INCR_M_WS = 0, // initial whitespace skipping, must be 0
79 INCR_M_STR, // inside string 86 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash 87 INCR_M_BS, // inside backslash
88 INCR_M_C0, // inside comment in initial whitespace sequence
89 INCR_M_C1, // inside comment in other places
81 INCR_M_JSON // outside anything, count nesting 90 INCR_M_JSON // outside anything, count nesting
82}; 91};
83 92
84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) 93#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85 94
… …
175 *s++ = 0x80 | ( ch & 0x3f); 184 *s++ = 0x80 | ( ch & 0x3f);
176 185
177 return s; 186 return s;
178} 187}
179 188
189// convert offset pointer to character index, sv must be string
190static STRLEN
191ptr_to_index (SV *sv, char *offset)
192{
193 return SvUTF8 (sv)
194 ? utf8_distance (offset, SvPVX (sv))
195 : offset - SvPVX (sv);
196}
197
198/////////////////////////////////////////////////////////////////////////////
199// fp hell
200
201// scan a group of digits, and a trailing exponent
202static void
203json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
204{
205 UV uaccum = 0;
206 int eaccum = 0;
207
208 // if we recurse too deep, skip all remaining digits
209 // to avoid a stack overflow attack
210 if (expect_false (--maxdepth <= 0))
211 while (((U8)*s - '0') < 10)
212 ++s;
213
214 for (;;)
215 {
216 U8 dig = (U8)*s - '0';
217
218 if (expect_false (dig >= 10))
219 {
220 if (dig == (U8)((U8)'.' - (U8)'0'))
221 {
222 ++s;
223 json_atof_scan1 (s, accum, expo, 1, maxdepth);
224 }
225 else if ((dig | ' ') == 'e' - '0')
226 {
227 int exp2 = 0;
228 int neg = 0;
229
230 ++s;
231
232 if (*s == '-')
233 {
234 ++s;
235 neg = 1;
236 }
237 else if (*s == '+')
238 ++s;
239
240 while ((dig = (U8)*s - '0') < 10)
241 exp2 = exp2 * 10 + *s++ - '0';
242
243 *expo += neg ? -exp2 : exp2;
244 }
245
246 break;
247 }
248
249 ++s;
250
251 uaccum = uaccum * 10 + dig;
252 ++eaccum;
253
254 // if we have too many digits, then recurse for more
255 // we actually do this for rather few digits
256 if (uaccum >= (UV_MAX - 9) / 10)
257 {
258 if (postdp) *expo -= eaccum;
259 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
260 if (postdp) *expo += eaccum;
261
262 break;
263 }
264 }
265
266 // this relies greatly on the quality of the pow ()
267 // implementation of the platform, but a good
268 // implementation is hard to beat.
269 // (IEEE 754 conformant ones are required to be exact)
270 if (postdp) *expo -= eaccum;
271 *accum += uaccum * Perl_pow (10., *expo);
272 *expo += eaccum;
273}
274
275static NV
276json_atof (const char *s)
277{
278 NV accum = 0.;
279 int expo = 0;
280 int neg = 0;
281
282 if (*s == '-')
283 {
284 ++s;
285 neg = 1;
286 }
287
288 // a recursion depth of ten gives us >>500 bits
289 json_atof_scan1 (s, &accum, &expo, 0, 10);
290
291 return neg ? -accum : accum;
292}
180///////////////////////////////////////////////////////////////////////////// 293/////////////////////////////////////////////////////////////////////////////
181// encoder 294// encoder
182 295
183// structure used for encoding JSON 296// structure used for encoding JSON
184typedef struct 297typedef struct
… …
371 484
372 if (enc->indent >= enc->json.max_depth) 485 if (enc->indent >= enc->json.max_depth)
373 croak (ERR_NESTING_EXCEEDED); 486 croak (ERR_NESTING_EXCEEDED);
374 487
375 encode_ch (enc, '['); 488 encode_ch (enc, '[');
376 489
377 if (len >= 0) 490 if (len >= 0)
378 { 491 {
379 encode_nl (enc); ++enc->indent; 492 encode_nl (enc); ++enc->indent;
380 493
381 for (i = 0; i <= len; ++i) 494 for (i = 0; i <= len; ++i)
… …
393 encode_comma (enc); 506 encode_comma (enc);
394 } 507 }
395 508
396 encode_nl (enc); --enc->indent; encode_indent (enc); 509 encode_nl (enc); --enc->indent; encode_indent (enc);
397 } 510 }
398 511
399 encode_ch (enc, ']'); 512 encode_ch (enc, ']');
400} 513}
401 514
402static void 515static void
403encode_hk (enc_t *enc, HE *he) 516encode_hk (enc_t *enc, HE *he)
… …
407 if (HeKLEN (he) == HEf_SVKEY) 520 if (HeKLEN (he) == HEf_SVKEY)
408 { 521 {
409 SV *sv = HeSVKEY (he); 522 SV *sv = HeSVKEY (he);
410 STRLEN len; 523 STRLEN len;
411 char *str; 524 char *str;
412 525
413 SvGETMAGIC (sv); 526 SvGETMAGIC (sv);
414 str = SvPV (sv, len); 527 str = SvPV (sv, len);
415 528
416 encode_str (enc, str, len, SvUTF8 (sv)); 529 encode_str (enc, str, len, SvUTF8 (sv));
417 } 530 }
… …
462 575
463 // for canonical output we have to sort by keys first 576 // for canonical output we have to sort by keys first
464 // actually, this is mostly due to the stupid so-called 577 // actually, this is mostly due to the stupid so-called
465 // security workaround added somewhere in 5.8.x 578 // security workaround added somewhere in 5.8.x
466 // that randomises hash orderings 579 // that randomises hash orderings
467 if (enc->json.flags & F_CANONICAL) 580 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
468 { 581 {
469 int count = hv_iterinit (hv); 582 int count = hv_iterinit (hv);
470 583
471 if (SvMAGICAL (hv)) 584 if (SvMAGICAL (hv))
472 { 585 {
… …
483 } 596 }
484 597
485 if (count) 598 if (count)
486 { 599 {
487 int i, fast = 1; 600 int i, fast = 1;
488#if defined(__BORLANDC__) || defined(_MSC_VER) 601 HE *hes_stack [STACK_HES];
489 HE **hes = _alloca (count * sizeof (HE)); 602 HE **hes = hes_stack;
490#else 603
491 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 604 // allocate larger arrays on the heap
492#endif 605 if (count > STACK_HES)
606 {
607 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
608 hes = (HE **)SvPVX (sv);
609 }
493 610
494 i = 0; 611 i = 0;
495 while ((he = hv_iternext (hv))) 612 while ((he = hv_iternext (hv)))
496 { 613 {
497 hes [i++] = he; 614 hes [i++] = he;
… …
566// encode objects, arrays and special \0=false and \1=true values. 683// encode objects, arrays and special \0=false and \1=true values.
567static void 684static void
568encode_rv (enc_t *enc, SV *sv) 685encode_rv (enc_t *enc, SV *sv)
569{ 686{
570 svtype svt; 687 svtype svt;
688 GV *method;
571 689
572 SvGETMAGIC (sv); 690 SvGETMAGIC (sv);
573 svt = SvTYPE (sv); 691 svt = SvTYPE (sv);
574 692
575 if (expect_false (SvOBJECT (sv))) 693 if (expect_false (SvOBJECT (sv)))
576 { 694 {
577 HV *stash = !JSON_SLOW || json_boolean_stash 695 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
578 ? json_boolean_stash 696 ? types_boolean_stash
579 : gv_stashpv ("JSON::XS::Boolean", 1); 697 : gv_stashpv ("Types::Serialiser::Boolean", 1);
698 HV *stash = SvSTASH (sv);
580 699
581 if (SvSTASH (sv) == stash) 700 if (stash == boolean_stash)
582 { 701 {
583 if (SvIV (sv)) 702 if (SvIV (sv))
584 encode_str (enc, "true", 4, 0); 703 encode_str (enc, "true", 4, 0);
585 else 704 else
586 encode_str (enc, "false", 5, 0); 705 encode_str (enc, "false", 5, 0);
587 } 706 }
707 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
708 {
709 int count;
710 dSP;
711
712 ENTER; SAVETMPS; PUSHMARK (SP);
713 EXTEND (SP, 2);
714 // we re-bless the reference to get overload and other niceties right
715 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
716 PUSHs (sv_json);
717
718 PUTBACK;
719 count = call_sv ((SV *)GvCV (method), G_ARRAY);
720 SPAGAIN;
721
722 // catch this surprisingly common error
723 if (SvROK (TOPs) && SvRV (TOPs) == sv)
724 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
725
726 encode_ch (enc, '(');
727 encode_ch (enc, '"');
728 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
729 encode_ch (enc, '"');
730 encode_ch (enc, ')');
731 encode_ch (enc, '[');
732
733 while (count)
734 {
735 encode_sv (enc, SP[1 - count--]);
736
737 if (count)
738 encode_ch (enc, ',');
739 }
740
741 encode_ch (enc, ']');
742
743 PUTBACK;
744
745 FREETMPS; LEAVE;
746 }
747 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
748 {
749 dSP;
750
751 ENTER; SAVETMPS; PUSHMARK (SP);
752 // we re-bless the reference to get overload and other niceties right
753 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
754
755 // calling with G_SCALAR ensures that we always get a 1 return value
756 PUTBACK;
757 call_sv ((SV *)GvCV (method), G_SCALAR);
758 SPAGAIN;
759
760 // catch this surprisingly common error
761 if (SvROK (TOPs) && SvRV (TOPs) == sv)
762 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
763
764 sv = POPs;
765 PUTBACK;
766
767 encode_sv (enc, sv);
768
769 FREETMPS; LEAVE;
770 }
771 else if (enc->json.flags & F_ALLOW_BLESSED)
772 encode_str (enc, "null", 4, 0);
588 else 773 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", 774 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)))); 775 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
635 }
636 } 776 }
637 else if (svt == SVt_PVHV) 777 else if (svt == SVt_PVHV)
638 encode_hv (enc, (HV *)sv); 778 encode_hv (enc, (HV *)sv);
639 else if (svt == SVt_PVAV) 779 else if (svt == SVt_PVAV)
640 encode_av (enc, (AV *)sv); 780 encode_av (enc, (AV *)sv);
… …
716 } 856 }
717 else 857 else
718 { 858 {
719 // large integer, use the (rather slow) snprintf way. 859 // large integer, use the (rather slow) snprintf way.
720 need (enc, IVUV_MAXCHARS); 860 need (enc, IVUV_MAXCHARS);
721 enc->cur += 861 enc->cur +=
722 SvIsUV(sv) 862 SvIsUV(sv)
723 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) 863 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
724 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 864 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
725 } 865 }
726 } 866 }
727 else if (SvROK (sv)) 867 else if (SvROK (sv))
728 encode_rv (enc, SvRV (sv)); 868 encode_rv (enc, SvRV (sv));
729 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 869 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
730 encode_str (enc, "null", 4, 0); 870 encode_str (enc, "null", 4, 0);
731 else 871 else
732 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 872 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
733 SvPV_nolen (sv), SvFLAGS (sv)); 873 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
734} 874}
735 875
736static SV * 876static SV *
737encode_json (SV *scalar, JSON *json) 877encode_json (SV *scalar, JSON *json)
738{ 878{
… …
750 : enc.json.flags & F_LATIN1 ? 0x000100UL 890 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL; 891 : 0x110000UL;
752 892
753 SvPOK_only (enc.sv); 893 SvPOK_only (enc.sv);
754 encode_sv (&enc, scalar); 894 encode_sv (&enc, scalar);
895 encode_nl (&enc);
755 896
756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 897 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
757 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 898 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
758 899
759 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 900 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
939 else if (expect_true (ch >= 0x20 && ch < 0x80)) 1080 else if (expect_true (ch >= 0x20 && ch < 0x80))
940 *cur++ = ch; 1081 *cur++ = ch;
941 else if (ch >= 0x80) 1082 else if (ch >= 0x80)
942 { 1083 {
943 STRLEN clen; 1084 STRLEN clen;
944 UV uch;
945 1085
946 --dec_cur; 1086 --dec_cur;
947 1087
948 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1088 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
949 if (clen == (STRLEN)-1) 1089 if (clen == (STRLEN)-1)
950 ERR ("malformed UTF-8 character in JSON string"); 1090 ERR ("malformed UTF-8 character in JSON string");
951 1091
952 do 1092 do
953 *cur++ = *dec_cur++; 1093 *cur++ = *dec_cur++;
… …
1075 1215
1076 // special case the rather common 1..5-digit-int case 1216 // special case the rather common 1..5-digit-int case
1077 if (*start == '-') 1217 if (*start == '-')
1078 switch (len) 1218 switch (len)
1079 { 1219 {
1080 case 2: return newSViv (-( start [1] - '0' * 1)); 1220 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1081 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1221 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)); 1222 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)); 1223 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)); 1224 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1085 } 1225 }
1086 else 1226 else
1087 switch (len) 1227 switch (len)
1088 { 1228 {
1089 case 1: return newSViv ( start [0] - '0' * 1); 1229 case 1: return newSViv ( start [0] - '0' * 1);
1090 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1230 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); 1231 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); 1232 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); 1233 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1094 } 1234 }
1095 1235
1096 { 1236 {
1097 UV uv; 1237 UV uv;
1098 int numtype = grok_number (start, len, &uv); 1238 int numtype = grok_number (start, len, &uv);
… …
1107 } 1247 }
1108 1248
1109 len -= *start == '-' ? 1 : 0; 1249 len -= *start == '-' ? 1 : 0;
1110 1250
1111 // does not fit into IV or UV, try NV 1251 // does not fit into IV or UV, try NV
1112 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1252 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 1253 // fits into NV without loss of precision
1118 return newSVnv (Atof (start)); 1254 return newSVnv (json_atof (start));
1119 1255
1120 // everything else fails, convert it to a string 1256 // everything else fails, convert it to a string
1121 return newSVpvn (start, dec->cur - start); 1257 return newSVpvn (start, dec->cur - start);
1122 } 1258 }
1123 1259
1124 // loss of precision here 1260 // loss of precision here
1125 return newSVnv (Atof (start)); 1261 return newSVnv (json_atof (start));
1126 1262
1127fail: 1263fail:
1128 return 0; 1264 return 0;
1129} 1265}
1130 1266
… …
1154 if (*dec->cur == ']') 1290 if (*dec->cur == ']')
1155 { 1291 {
1156 ++dec->cur; 1292 ++dec->cur;
1157 break; 1293 break;
1158 } 1294 }
1159 1295
1160 if (*dec->cur != ',') 1296 if (*dec->cur != ',')
1161 ERR (", or ] expected while parsing array"); 1297 ERR (", or ] expected while parsing array");
1162 1298
1163 ++dec->cur; 1299 ++dec->cur;
1164 1300
… …
1299 dSP; 1435 dSP;
1300 int count; 1436 int count;
1301 1437
1302 ENTER; SAVETMPS; PUSHMARK (SP); 1438 ENTER; SAVETMPS; PUSHMARK (SP);
1303 XPUSHs (HeVAL (he)); 1439 XPUSHs (HeVAL (he));
1440 sv_2mortal (sv);
1304 1441
1305 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1442 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1306 1443
1307 if (count == 1) 1444 if (count == 1)
1308 { 1445 {
1309 sv = newSVsv (POPs); 1446 sv = newSVsv (POPs);
1310 FREETMPS; LEAVE; 1447 FREETMPS; LEAVE;
1311 return sv; 1448 return sv;
1312 } 1449 }
1313 1450
1451 SvREFCNT_inc (sv);
1314 FREETMPS; LEAVE; 1452 FREETMPS; LEAVE;
1315 } 1453 }
1316 } 1454 }
1317 1455
1318 if (dec->json.cb_object) 1456 if (dec->json.cb_object)
… …
1344 DEC_DEC_DEPTH; 1482 DEC_DEC_DEPTH;
1345 return 0; 1483 return 0;
1346} 1484}
1347 1485
1348static SV * 1486static SV *
1487decode_tag (dec_t *dec)
1488{
1489 SV *tag = 0;
1490 SV *val = 0;
1491
1492 if (!(dec->json.flags & F_ALLOW_TAGS))
1493 ERR ("malformed JSON string, neither array, object, number, string or atom");
1494
1495 ++dec->cur;
1496
1497 decode_ws (dec);
1498
1499 tag = decode_sv (dec);
1500 if (!tag)
1501 goto fail;
1502
1503 if (!SvPOK (tag))
1504 ERR ("malformed JSON string, (tag) must be a string");
1505
1506 decode_ws (dec);
1507
1508 if (*dec->cur != ')')
1509 ERR (") expected after tag");
1510
1511 ++dec->cur;
1512
1513 decode_ws (dec);
1514
1515 val = decode_sv (dec);
1516 if (!val)
1517 goto fail;
1518
1519 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1520 ERR ("malformed JSON string, tag value must be an array");
1521
1522 {
1523 AV *av = (AV *)SvRV (val);
1524 int i, len = av_len (av) + 1;
1525 HV *stash = gv_stashsv (tag, 0);
1526 SV *sv;
1527
1528 if (!stash)
1529 ERR ("cannot decode perl-object (package does not exist)");
1530
1531 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1532
1533 if (!method)
1534 ERR ("cannot decode perl-object (package does not have a THAW method)");
1535
1536 dSP;
1537
1538 ENTER; SAVETMPS; PUSHMARK (SP);
1539 EXTEND (SP, len + 2);
1540 // we re-bless the reference to get overload and other niceties right
1541 PUSHs (tag);
1542 PUSHs (sv_json);
1543
1544 for (i = 0; i < len; ++i)
1545 PUSHs (*av_fetch (av, i, 1));
1546
1547 PUTBACK;
1548 call_sv ((SV *)GvCV (method), G_SCALAR);
1549 SPAGAIN;
1550
1551 SvREFCNT_dec (tag);
1552 SvREFCNT_dec (val);
1553 sv = SvREFCNT_inc (POPs);
1554
1555 PUTBACK;
1556
1557 FREETMPS; LEAVE;
1558
1559 return sv;
1560 }
1561
1562fail:
1563 SvREFCNT_dec (tag);
1564 SvREFCNT_dec (val);
1565 return 0;
1566}
1567
1568static SV *
1349decode_sv (dec_t *dec) 1569decode_sv (dec_t *dec)
1350{ 1570{
1351 // the beauty of JSON: you need exactly one character lookahead 1571 // the beauty of JSON: you need exactly one character lookahead
1352 // to parse everything. 1572 // to parse everything.
1353 switch (*dec->cur) 1573 switch (*dec->cur)
1354 { 1574 {
1355 case '"': ++dec->cur; return decode_str (dec); 1575 case '"': ++dec->cur; return decode_str (dec);
1356 case '[': ++dec->cur; return decode_av (dec); 1576 case '[': ++dec->cur; return decode_av (dec);
1357 case '{': ++dec->cur; return decode_hv (dec); 1577 case '{': ++dec->cur; return decode_hv (dec);
1578 case '(': return decode_tag (dec);
1358 1579
1359 case '-': 1580 case '-':
1360 case '0': case '1': case '2': case '3': case '4': 1581 case '0': case '1': case '2': case '3': case '4':
1361 case '5': case '6': case '7': case '8': case '9': 1582 case '5': case '6': case '7': case '8': case '9':
1362 return decode_num (dec); 1583 return decode_num (dec);
… …
1364 case 't': 1585 case 't':
1365 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1586 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1366 { 1587 {
1367 dec->cur += 4; 1588 dec->cur += 4;
1368#if JSON_SLOW 1589#if JSON_SLOW
1369 json_true = get_bool ("JSON::XS::true"); 1590 types_true = get_bool ("Types::Serialiser::true");
1370#endif 1591#endif
1371 return newSVsv (json_true); 1592 return newSVsv (types_true);
1372 } 1593 }
1373 else 1594 else
1374 ERR ("'true' expected"); 1595 ERR ("'true' expected");
1375 1596
1376 break; 1597 break;
… …
1378 case 'f': 1599 case 'f':
1379 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1600 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1380 { 1601 {
1381 dec->cur += 5; 1602 dec->cur += 5;
1382#if JSON_SLOW 1603#if JSON_SLOW
1383 json_false = get_bool ("JSON::XS::false"); 1604 types_false = get_bool ("Types::Serialiser::false");
1384#endif 1605#endif
1385 return newSVsv (json_false); 1606 return newSVsv (types_false);
1386 } 1607 }
1387 else 1608 else
1388 ERR ("'false' expected"); 1609 ERR ("'false' expected");
1389 1610
1390 break; 1611 break;
… …
1399 ERR ("'null' expected"); 1620 ERR ("'null' expected");
1400 1621
1401 break; 1622 break;
1402 1623
1403 default: 1624 default:
1404 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1625 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1405 break; 1626 break;
1406 } 1627 }
1407 1628
1408fail: 1629fail:
1409 return 0; 1630 return 0;
1410} 1631}
1411 1632
1412static SV * 1633static SV *
1413decode_json (SV *string, JSON *json, STRLEN *offset_return) 1634decode_json (SV *string, JSON *json, char **offset_return)
1414{ 1635{
1415 dec_t dec; 1636 dec_t dec;
1416 STRLEN offset;
1417 SV *sv; 1637 SV *sv;
1418 1638
1419 /* work around bugs in 5.10 where manipulating magic values 1639 /* work around bugs in 5.10 where manipulating magic values
1420 * will perl ignore the magic in subsequent accesses 1640 * makes perl ignore the magic in subsequent accesses.
1641 * also make a copy of non-PV values, to get them into a clean
1642 * state (SvPV should do that, but it's buggy, see below).
1421 */ 1643 */
1422 /*SvGETMAGIC (string);*/ 1644 /*SvGETMAGIC (string);*/
1423 if (SvMAGICAL (string)) 1645 if (SvMAGICAL (string) || !SvPOK (string))
1424 string = sv_2mortal (newSVsv (string)); 1646 string = sv_2mortal (newSVsv (string));
1425 1647
1426 SvUPGRADE (string, SVt_PV); 1648 SvUPGRADE (string, SVt_PV);
1427 1649
1428 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1650 /* work around a bug in perl 5.10, which causes SvCUR to fail an
… …
1433 * and hope for the best. 1655 * and hope for the best.
1434 * Damnit, SvPV_nolen still trips over yet another assertion. This 1656 * Damnit, SvPV_nolen still trips over yet another assertion. This
1435 * assertion business is seriously broken, try yet another workaround 1657 * assertion business is seriously broken, try yet another workaround
1436 * for the broken -DDEBUGGING. 1658 * for the broken -DDEBUGGING.
1437 */ 1659 */
1660 {
1438#ifdef DEBUGGING 1661#ifdef DEBUGGING
1439 offset = SvOK (string) ? sv_len (string) : 0; 1662 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1440#else 1663#else
1441 offset = SvCUR (string); 1664 STRLEN offset = SvCUR (string);
1442#endif 1665#endif
1443 1666
1444 if (offset > json->max_size && json->max_size) 1667 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", 1668 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); 1669 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1670 }
1447 1671
1448 if (json->flags & F_UTF8) 1672 if (DECODE_WANTS_OCTETS (json))
1449 sv_utf8_downgrade (string, 0); 1673 sv_utf8_downgrade (string, 0);
1450 else 1674 else
1451 sv_utf8_upgrade (string); 1675 sv_utf8_upgrade (string);
1452 1676
1453 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1677 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
… …
1463 1687
1464 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1688 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1465 1689
1466 decode_ws (&dec); 1690 decode_ws (&dec);
1467 sv = decode_sv (&dec); 1691 sv = decode_sv (&dec);
1692
1693 if (offset_return)
1694 *offset_return = dec.cur;
1468 1695
1469 if (!(offset_return || !sv)) 1696 if (!(offset_return || !sv))
1470 { 1697 {
1471 // check for trailing garbage 1698 // check for trailing garbage
1472 decode_ws (&dec); 1699 decode_ws (&dec);
… …
1475 { 1702 {
1476 dec.err = "garbage after JSON object"; 1703 dec.err = "garbage after JSON object";
1477 SvREFCNT_dec (sv); 1704 SvREFCNT_dec (sv);
1478 sv = 0; 1705 sv = 0;
1479 } 1706 }
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 } 1707 }
1491 1708
1492 if (!sv) 1709 if (!sv)
1493 { 1710 {
1494 SV *uni = sv_newmortal (); 1711 SV *uni = sv_newmortal ();
… …
1500 SAVEVPTR (PL_curcop); 1717 SAVEVPTR (PL_curcop);
1501 PL_curcop = &cop; 1718 PL_curcop = &cop;
1502 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1719 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1503 LEAVE; 1720 LEAVE;
1504 1721
1505 croak ("%s, at character offset %d [\"%s\"]", 1722 croak ("%s, at character offset %d (before \"%s\")",
1506 dec.err, 1723 dec.err,
1507 (int)offset, 1724 (int)ptr_to_index (string, dec.cur),
1508 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1725 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1509 } 1726 }
1510 1727
1511 sv = sv_2mortal (sv); 1728 sv = sv_2mortal (sv);
1512 1729
… …
1521 1738
1522static void 1739static void
1523incr_parse (JSON *self) 1740incr_parse (JSON *self)
1524{ 1741{
1525 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1742 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1743
1744 // the state machine here is a bit convoluted and could be simplified a lot
1745 // but this would make it slower, so...
1526 1746
1527 for (;;) 1747 for (;;)
1528 { 1748 {
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 1749 //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) 1750 switch (self->incr_mode)
1531 { 1751 {
1532 // only used for intiial whitespace skipping 1752 // only used for initial whitespace skipping
1533 case INCR_M_WS: 1753 case INCR_M_WS:
1534 for (;;) 1754 for (;;)
1535 { 1755 {
1536 if (*p > 0x20) 1756 if (*p > 0x20)
1537 { 1757 {
1758 if (*p == '#')
1759 {
1760 self->incr_mode = INCR_M_C0;
1761 goto incr_m_c;
1762 }
1763 else
1764 {
1538 self->incr_mode = INCR_M_JSON; 1765 self->incr_mode = INCR_M_JSON;
1539 goto incr_m_json; 1766 goto incr_m_json;
1767 }
1540 } 1768 }
1541 else if (!*p) 1769 else if (!*p)
1542 goto interrupt; 1770 goto interrupt;
1543 1771
1544 ++p; 1772 ++p;
… …
1550 goto interrupt; 1778 goto interrupt;
1551 1779
1552 ++p; 1780 ++p;
1553 self->incr_mode = INCR_M_STR; 1781 self->incr_mode = INCR_M_STR;
1554 goto incr_m_str; 1782 goto incr_m_str;
1783
1784 // inside #-style comments
1785 case INCR_M_C0:
1786 case INCR_M_C1:
1787 incr_m_c:
1788 for (;;)
1789 {
1790 if (*p == '\n')
1791 {
1792 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1793 break;
1794 }
1795 else if (!*p)
1796 goto interrupt;
1797
1798 ++p;
1799 }
1800
1801 break;
1555 1802
1556 // inside a string 1803 // inside a string
1557 case INCR_M_STR: 1804 case INCR_M_STR:
1558 incr_m_str: 1805 incr_m_str:
1559 for (;;) 1806 for (;;)
… …
1610 self->incr_mode = INCR_M_STR; 1857 self->incr_mode = INCR_M_STR;
1611 goto incr_m_str; 1858 goto incr_m_str;
1612 1859
1613 case '[': 1860 case '[':
1614 case '{': 1861 case '{':
1862 case '(':
1615 if (++self->incr_nest > self->max_depth) 1863 if (++self->incr_nest > self->max_depth)
1616 croak (ERR_NESTING_EXCEEDED); 1864 croak (ERR_NESTING_EXCEEDED);
1617 break; 1865 break;
1618 1866
1619 case ']': 1867 case ']':
1620 case '}': 1868 case '}':
1621 if (--self->incr_nest <= 0) 1869 if (--self->incr_nest <= 0)
1622 goto interrupt; 1870 goto interrupt;
1871 break;
1872
1873 case ')':
1874 --self->incr_nest;
1875 break;
1876
1877 case '#':
1878 self->incr_mode = INCR_M_C1;
1879 goto incr_m_c;
1623 } 1880 }
1624 } 1881 }
1625 } 1882 }
1626 1883
1627 modechange: 1884 modechange:
1628 ; 1885 ;
1629 } 1886 }
1630 1887
1631interrupt: 1888interrupt:
1632 self->incr_pos = p - SvPVX (self->incr_text); 1889 self->incr_pos = p - SvPVX (self->incr_text);
1890 //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 1891 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1634} 1892}
1635 1893
1636///////////////////////////////////////////////////////////////////////////// 1894/////////////////////////////////////////////////////////////////////////////
1637// XS interface functions 1895// XS interface functions
… …
1647 i >= '0' && i <= '9' ? i - '0' 1905 i >= '0' && i <= '9' ? i - '0'
1648 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1906 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1649 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1907 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1650 : -1; 1908 : -1;
1651 1909
1652 json_stash = gv_stashpv ("JSON::XS" , 1); 1910 json_stash = gv_stashpv ("JSON::XS" , 1);
1653 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1911 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1654 1912
1655 json_true = get_bool ("JSON::XS::true"); 1913 types_true = get_bool ("Types::Serialiser::true");
1656 json_false = get_bool ("JSON::XS::false"); 1914 types_false = get_bool ("Types::Serialiser::false");
1915
1916 sv_json = newSVpv ("JSON", 0);
1917 SvREADONLY_on (sv_json);
1918
1919 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1657} 1920}
1658 1921
1659PROTOTYPES: DISABLE 1922PROTOTYPES: DISABLE
1660 1923
1661void CLONE (...) 1924void CLONE (...)
1662 CODE: 1925 CODE:
1663 json_stash = 0; 1926 json_stash = 0;
1664 json_boolean_stash = 0; 1927 types_boolean_stash = 0;
1665 1928
1666void new (char *klass) 1929void new (char *klass)
1667 PPCODE: 1930 PPCODE:
1668{ 1931{
1669 SV *pv = NEWSV (0, sizeof (JSON)); 1932 SV *pv = NEWSV (0, sizeof (JSON));
1670 SvPOK_only (pv); 1933 SvPOK_only (pv);
1671 json_init ((JSON *)SvPVX (pv)); 1934 json_init ((JSON *)SvPVX (pv));
1672 XPUSHs (sv_2mortal (sv_bless ( 1935 XPUSHs (sv_2mortal (sv_bless (
1673 newRV_noinc (pv), 1936 newRV_noinc (pv),
1674 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1937 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
… …
1689 shrink = F_SHRINK 1952 shrink = F_SHRINK
1690 allow_blessed = F_ALLOW_BLESSED 1953 allow_blessed = F_ALLOW_BLESSED
1691 convert_blessed = F_CONV_BLESSED 1954 convert_blessed = F_CONV_BLESSED
1692 relaxed = F_RELAXED 1955 relaxed = F_RELAXED
1693 allow_unknown = F_ALLOW_UNKNOWN 1956 allow_unknown = F_ALLOW_UNKNOWN
1957 allow_tags = F_ALLOW_TAGS
1694 PPCODE: 1958 PPCODE:
1695{ 1959{
1696 if (enable) 1960 if (enable)
1697 self->flags |= ix; 1961 self->flags |= ix;
1698 else 1962 else
… …
1714 get_shrink = F_SHRINK 1978 get_shrink = F_SHRINK
1715 get_allow_blessed = F_ALLOW_BLESSED 1979 get_allow_blessed = F_ALLOW_BLESSED
1716 get_convert_blessed = F_CONV_BLESSED 1980 get_convert_blessed = F_CONV_BLESSED
1717 get_relaxed = F_RELAXED 1981 get_relaxed = F_RELAXED
1718 get_allow_unknown = F_ALLOW_UNKNOWN 1982 get_allow_unknown = F_ALLOW_UNKNOWN
1983 get_allow_tags = F_ALLOW_TAGS
1719 PPCODE: 1984 PPCODE:
1720 XPUSHs (boolSV (self->flags & ix)); 1985 XPUSHs (boolSV (self->flags & ix));
1721 1986
1722void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 1987void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1723 PPCODE: 1988 PPCODE:
… …
1751} 2016}
1752 2017
1753void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2018void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1754 PPCODE: 2019 PPCODE:
1755{ 2020{
1756 if (!self->cb_sk_object) 2021 if (!self->cb_sk_object)
1757 self->cb_sk_object = newHV (); 2022 self->cb_sk_object = newHV ();
1758 2023
1759 if (SvOK (cb)) 2024 if (SvOK (cb))
1760 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2025 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1761 else 2026 else
… …
1772 XPUSHs (ST (0)); 2037 XPUSHs (ST (0));
1773} 2038}
1774 2039
1775void encode (JSON *self, SV *scalar) 2040void encode (JSON *self, SV *scalar)
1776 PPCODE: 2041 PPCODE:
1777 XPUSHs (encode_json (scalar, self)); 2042 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2043 XPUSHs (scalar);
1778 2044
1779void decode (JSON *self, SV *jsonstr) 2045void decode (JSON *self, SV *jsonstr)
1780 PPCODE: 2046 PPCODE:
1781 XPUSHs (decode_json (jsonstr, self, 0)); 2047 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2048 XPUSHs (jsonstr);
1782 2049
1783void decode_prefix (JSON *self, SV *jsonstr) 2050void decode_prefix (JSON *self, SV *jsonstr)
1784 PPCODE: 2051 PPCODE:
1785{ 2052{
1786 STRLEN offset; 2053 SV *sv;
2054 char *offset;
2055 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1787 EXTEND (SP, 2); 2056 EXTEND (SP, 2);
1788 PUSHs (decode_json (jsonstr, self, &offset)); 2057 PUSHs (sv);
1789 PUSHs (sv_2mortal (newSVuv (offset))); 2058 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1790} 2059}
1791 2060
1792void incr_parse (JSON *self, SV *jsonstr = 0) 2061void incr_parse (JSON *self, SV *jsonstr = 0)
1793 PPCODE: 2062 PPCODE:
1794{ 2063{
1795 if (!self->incr_text) 2064 if (!self->incr_text)
1796 self->incr_text = newSVpvn ("", 0); 2065 self->incr_text = newSVpvn ("", 0);
2066
2067 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2068 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2069 if (DECODE_WANTS_OCTETS (self))
2070 {
2071 if (self->incr_pos)
2072 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2073 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2074
2075 sv_utf8_downgrade (self->incr_text, 0);
2076 }
2077 else
2078 {
2079 sv_utf8_upgrade (self->incr_text);
2080
2081 if (self->incr_pos)
2082 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2083 - (U8 *)SvPVX (self->incr_text);
2084 }
1797 2085
1798 // append data, if any 2086 // append data, if any
1799 if (jsonstr) 2087 if (jsonstr)
1800 { 2088 {
2089 /* make sure both strings have same encoding */
1801 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 2090 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2091 if (SvUTF8 (jsonstr))
2092 sv_utf8_downgrade (jsonstr, 0);
1802 { 2093 else
1803 /* utf-8-ness differs, need to upgrade */
1804 sv_utf8_upgrade (self->incr_text); 2094 sv_utf8_upgrade (jsonstr);
1805 2095
1806 if (self->incr_pos) 2096 /* 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 { 2097 {
1812 STRLEN len; 2098 STRLEN len;
1813 const char *str = SvPV (jsonstr, len); 2099 const char *str = SvPV (jsonstr, len);
1814 STRLEN cur = SvCUR (self->incr_text); 2100 STRLEN cur = SvCUR (self->incr_text);
1815 2101
… …
1823 } 2109 }
1824 2110
1825 if (GIMME_V != G_VOID) 2111 if (GIMME_V != G_VOID)
1826 do 2112 do
1827 { 2113 {
2114 SV *sv;
1828 STRLEN offset; 2115 char *offset;
1829 2116
1830 if (!INCR_DONE (self)) 2117 if (!INCR_DONE (self))
1831 { 2118 {
1832 incr_parse (self); 2119 incr_parse (self);
1833 2120
1834 if (self->incr_pos > self->max_size && self->max_size) 2121 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", 2122 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); 2123 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1837 2124
1838 if (!INCR_DONE (self)) 2125 if (!INCR_DONE (self))
2126 {
2127 // as an optimisation, do not accumulate white space in the incr buffer
2128 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2129 {
2130 self->incr_pos = 0;
2131 SvCUR_set (self->incr_text, 0);
2132 }
2133
1839 break; 2134 break;
2135 }
1840 } 2136 }
1841 2137
1842 XPUSHs (decode_json (self->incr_text, self, &offset)); 2138 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2139 XPUSHs (sv);
1843 2140
1844 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1845 self->incr_pos -= offset; 2141 self->incr_pos -= offset - SvPVX (self->incr_text);
1846 self->incr_nest = 0; 2142 self->incr_nest = 0;
1847 self->incr_mode = 0; 2143 self->incr_mode = 0;
2144
2145 sv_chop (self->incr_text, offset);
1848 } 2146 }
1849 while (GIMME_V == G_ARRAY); 2147 while (GIMME_V == G_ARRAY);
1850} 2148}
1851 2149
1852SV *incr_text (JSON *self) 2150SV *incr_text (JSON *self)
… …
1890 SvREFCNT_dec (self->incr_text); 2188 SvREFCNT_dec (self->incr_text);
1891 2189
1892PROTOTYPES: ENABLE 2190PROTOTYPES: ENABLE
1893 2191
1894void encode_json (SV *scalar) 2192void encode_json (SV *scalar)
1895 ALIAS:
1896 to_json_ = 0
1897 encode_json = F_UTF8
1898 PPCODE: 2193 PPCODE:
1899{ 2194{
1900 JSON json; 2195 JSON json;
1901 json_init (&json); 2196 json_init (&json);
1902 json.flags |= ix; 2197 json.flags |= F_UTF8;
1903 XPUSHs (encode_json (scalar, &json)); 2198 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2199 XPUSHs (scalar);
1904} 2200}
1905 2201
1906void decode_json (SV *jsonstr) 2202void decode_json (SV *jsonstr)
1907 ALIAS:
1908 from_json_ = 0
1909 decode_json = F_UTF8
1910 PPCODE: 2203 PPCODE:
1911{ 2204{
1912 JSON json; 2205 JSON json;
1913 json_init (&json); 2206 json_init (&json);
1914 json.flags |= ix; 2207 json.flags |= F_UTF8;
1915 XPUSHs (decode_json (jsonstr, &json, 0)); 2208 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2209 XPUSHs (jsonstr);
1916} 2210}
1917 2211
1918

Diff Legend

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