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 1.106 by root, Tue Jan 19 01:36:34 2010 UTC vs.
Revision 1.130 by root, Sat Nov 26 06:09:29 2016 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
12#if defined(__BORLANDC__) || defined(_MSC_VER) 13#if defined(__BORLANDC__) || defined(_MSC_VER)
13# define snprintf _snprintf // C compilers have this in stdio.h 14# define snprintf _snprintf // C compilers have this in stdio.h
14#endif 15#endif
15 16
16// some old perls do not have this, try to make it work, no 17// some old perls do not have this, try to make it work, no
17// guarantees, though. if it breaks, you get to keep the pieces. 18// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES 19#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13 20# define UTF8_MAXBYTES 13
21#endif
22
23// compatibility with perl <5.18
24#ifndef HvNAMELEN_get
25# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
26#endif
27#ifndef HvNAMELEN
28# define HvNAMELEN(hv) HvNAMELEN_get (hv)
29#endif
30#ifndef HvNAMEUTF8
31# define HvNAMEUTF8(hv) 0
20#endif 32#endif
21 33
22// three extra for rounding, sign, and end of string 34// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) 35#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
24 36
33#define F_SHRINK 0x00000200UL 45#define F_SHRINK 0x00000200UL
34#define F_ALLOW_BLESSED 0x00000400UL 46#define F_ALLOW_BLESSED 0x00000400UL
35#define F_CONV_BLESSED 0x00000800UL 47#define F_CONV_BLESSED 0x00000800UL
36#define F_RELAXED 0x00001000UL 48#define F_RELAXED 0x00001000UL
37#define F_ALLOW_UNKNOWN 0x00002000UL 49#define F_ALLOW_UNKNOWN 0x00002000UL
50#define F_ALLOW_TAGS 0x00004000UL
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 51#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
39 52
40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 53#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
41 54
42#define INIT_SIZE 32 // initial scalar size to be allocated 55#define INIT_SIZE 32 // initial scalar size to be allocated
43#define INDENT_STEP 3 // spaces per indentation level 56#define INDENT_STEP 3 // spaces per indentation level
44 57
45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 58#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
59
60#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
46 61
47#define SB do { 62#define SB do {
48#define SE } while (0) 63#define SE } while (0)
49 64
50#if __GNUC__ >= 3 65#if __GNUC__ >= 3
65#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" 80#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
66 81
67#ifdef USE_ITHREADS 82#ifdef USE_ITHREADS
68# define JSON_SLOW 1 83# define JSON_SLOW 1
69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 84# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
85# define BOOL_STASH (bool_stash ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1))
70#else 86#else
71# define JSON_SLOW 0 87# define JSON_SLOW 0
72# define JSON_STASH json_stash 88# define JSON_STASH json_stash
89# define BOOL_STASH bool_stash
73#endif 90#endif
74 91
75static HV *json_stash, *json_boolean_stash; // JSON::XS:: 92// the amount of HEs to allocate on the stack, when sorting keys
76static SV *json_true, *json_false; 93#define STACK_HES 64
94
95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
96static SV *bool_true, *bool_false, *sv_json;
77 97
78enum { 98enum {
79 INCR_M_WS = 0, // initial whitespace skipping, must be 0 99 INCR_M_WS = 0, // initial whitespace skipping, must be 0
80 INCR_M_STR, // inside string 100 INCR_M_STR, // inside string
81 INCR_M_BS, // inside backslash 101 INCR_M_BS, // inside backslash
258 } 278 }
259 279
260 // this relies greatly on the quality of the pow () 280 // this relies greatly on the quality of the pow ()
261 // implementation of the platform, but a good 281 // implementation of the platform, but a good
262 // implementation is hard to beat. 282 // implementation is hard to beat.
283 // (IEEE 754 conformant ones are required to be exact)
263 if (postdp) *expo -= eaccum; 284 if (postdp) *expo -= eaccum;
264 *accum += uaccum * Perl_pow (10., *expo); 285 *accum += uaccum * Perl_pow (10., *expo);
265 *expo += eaccum; 286 *expo += eaccum;
266} 287}
267 288
281 // a recursion depth of ten gives us >>500 bits 302 // a recursion depth of ten gives us >>500 bits
282 json_atof_scan1 (s, &accum, &expo, 0, 10); 303 json_atof_scan1 (s, &accum, &expo, 0, 10);
283 304
284 return neg ? -accum : accum; 305 return neg ? -accum : accum;
285} 306}
307
308// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
309static int
310ref_bool_type (SV *sv)
311{
312 svtype svt = SvTYPE (sv);
313
314 if (svt < SVt_PVAV)
315 {
316 STRLEN len = 0;
317 char *pv = svt ? SvPV (sv, len) : 0;
318
319 if (len == 1)
320 if (*pv == '1')
321 return 1;
322 else if (*pv == '0')
323 return 0;
324 }
325
326 return -1;
327}
328
329// returns whether scalar is not a reference in the sense of allow_nonref
330static int
331json_nonref (SV *scalar)
332{
333 if (!SvROK (scalar))
334 return 1;
335
336 scalar = SvRV (scalar);
337
338 if (SvTYPE (scalar) >= SVt_PVMG)
339 {
340 if (SvSTASH (scalar) == bool_stash)
341 return 1;
342
343 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
344 return 1;
345 }
346
347 return 0;
348}
349
286///////////////////////////////////////////////////////////////////////////// 350/////////////////////////////////////////////////////////////////////////////
287// encoder 351// encoder
288 352
289// structure used for encoding JSON 353// structure used for encoding JSON
290typedef struct 354typedef struct
298} enc_t; 362} enc_t;
299 363
300INLINE void 364INLINE void
301need (enc_t *enc, STRLEN len) 365need (enc_t *enc, STRLEN len)
302{ 366{
303 if (expect_false (enc->cur + len >= enc->end)) 367 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
304 { 368 {
305 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 369 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
306 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 370 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
307 enc->cur = SvPVX (enc->sv) + cur; 371 enc->cur = SvPVX (enc->sv) + cur;
308 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 372 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
477 541
478 if (enc->indent >= enc->json.max_depth) 542 if (enc->indent >= enc->json.max_depth)
479 croak (ERR_NESTING_EXCEEDED); 543 croak (ERR_NESTING_EXCEEDED);
480 544
481 encode_ch (enc, '['); 545 encode_ch (enc, '[');
482 546
483 if (len >= 0) 547 if (len >= 0)
484 { 548 {
485 encode_nl (enc); ++enc->indent; 549 encode_nl (enc); ++enc->indent;
486 550
487 for (i = 0; i <= len; ++i) 551 for (i = 0; i <= len; ++i)
499 encode_comma (enc); 563 encode_comma (enc);
500 } 564 }
501 565
502 encode_nl (enc); --enc->indent; encode_indent (enc); 566 encode_nl (enc); --enc->indent; encode_indent (enc);
503 } 567 }
504 568
505 encode_ch (enc, ']'); 569 encode_ch (enc, ']');
506} 570}
507 571
508static void 572static void
509encode_hk (enc_t *enc, HE *he) 573encode_hk (enc_t *enc, HE *he)
513 if (HeKLEN (he) == HEf_SVKEY) 577 if (HeKLEN (he) == HEf_SVKEY)
514 { 578 {
515 SV *sv = HeSVKEY (he); 579 SV *sv = HeSVKEY (he);
516 STRLEN len; 580 STRLEN len;
517 char *str; 581 char *str;
518 582
519 SvGETMAGIC (sv); 583 SvGETMAGIC (sv);
520 str = SvPV (sv, len); 584 str = SvPV (sv, len);
521 585
522 encode_str (enc, str, len, SvUTF8 (sv)); 586 encode_str (enc, str, len, SvUTF8 (sv));
523 } 587 }
589 } 653 }
590 654
591 if (count) 655 if (count)
592 { 656 {
593 int i, fast = 1; 657 int i, fast = 1;
594#if defined(__BORLANDC__) || defined(_MSC_VER) 658 HE *hes_stack [STACK_HES];
595 HE **hes = _alloca (count * sizeof (HE)); 659 HE **hes = hes_stack;
596#else 660
597 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 661 // allocate larger arrays on the heap
598#endif 662 if (count > STACK_HES)
663 {
664 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
665 hes = (HE **)SvPVX (sv);
666 }
599 667
600 i = 0; 668 i = 0;
601 while ((he = hv_iternext (hv))) 669 while ((he = hv_iternext (hv)))
602 { 670 {
603 hes [i++] = he; 671 hes [i++] = he;
672// encode objects, arrays and special \0=false and \1=true values. 740// encode objects, arrays and special \0=false and \1=true values.
673static void 741static void
674encode_rv (enc_t *enc, SV *sv) 742encode_rv (enc_t *enc, SV *sv)
675{ 743{
676 svtype svt; 744 svtype svt;
745 GV *method;
677 746
678 SvGETMAGIC (sv); 747 SvGETMAGIC (sv);
679 svt = SvTYPE (sv); 748 svt = SvTYPE (sv);
680 749
681 if (expect_false (SvOBJECT (sv))) 750 if (expect_false (SvOBJECT (sv)))
682 { 751 {
683 HV *stash = !JSON_SLOW || json_boolean_stash 752 HV *stash = SvSTASH (sv);
684 ? json_boolean_stash
685 : gv_stashpv ("JSON::XS::Boolean", 1);
686 753
687 if (SvSTASH (sv) == stash) 754 if (stash == bool_stash)
688 { 755 {
689 if (SvIV (sv)) 756 if (SvIV (sv))
690 encode_str (enc, "true", 4, 0); 757 encode_str (enc, "true", 4, 0);
691 else 758 else
692 encode_str (enc, "false", 5, 0); 759 encode_str (enc, "false", 5, 0);
693 } 760 }
761 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
762 {
763 int count;
764 dSP;
765
766 ENTER; SAVETMPS;
767 SAVESTACK_POS ();
768 PUSHMARK (SP);
769 EXTEND (SP, 2);
770 // we re-bless the reference to get overload and other niceties right
771 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
772 PUSHs (sv_json);
773
774 PUTBACK;
775 count = call_sv ((SV *)GvCV (method), G_ARRAY);
776 SPAGAIN;
777
778 // catch this surprisingly common error
779 if (SvROK (TOPs) && SvRV (TOPs) == sv)
780 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
781
782 encode_ch (enc, '(');
783 encode_ch (enc, '"');
784 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
785 encode_ch (enc, '"');
786 encode_ch (enc, ')');
787 encode_ch (enc, '[');
788
789 while (count)
790 {
791 encode_sv (enc, SP[1 - count--]);
792
793 if (count)
794 encode_ch (enc, ',');
795 }
796
797 encode_ch (enc, ']');
798
799 FREETMPS; LEAVE;
800 }
801 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
802 {
803 dSP;
804
805 ENTER; SAVETMPS;
806 PUSHMARK (SP);
807 // we re-bless the reference to get overload and other niceties right
808 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
809
810 // calling with G_SCALAR ensures that we always get a 1 return value
811 PUTBACK;
812 call_sv ((SV *)GvCV (method), G_SCALAR);
813 SPAGAIN;
814
815 // catch this surprisingly common error
816 if (SvROK (TOPs) && SvRV (TOPs) == sv)
817 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
818
819 sv = POPs;
820 PUTBACK;
821
822 encode_sv (enc, sv);
823
824 FREETMPS; LEAVE;
825 }
826 else if (enc->json.flags & F_ALLOW_BLESSED)
827 encode_str (enc, "null", 4, 0);
694 else 828 else
695 {
696#if 0
697 if (0 && sv_derived_from (rv, "JSON::Literal"))
698 {
699 // not yet
700 }
701#endif
702 if (enc->json.flags & F_CONV_BLESSED)
703 {
704 // we re-bless the reference to get overload and other niceties right
705 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
706
707 if (to_json)
708 {
709 dSP;
710
711 ENTER; SAVETMPS; PUSHMARK (SP);
712 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
713
714 // calling with G_SCALAR ensures that we always get a 1 return value
715 PUTBACK;
716 call_sv ((SV *)GvCV (to_json), G_SCALAR);
717 SPAGAIN;
718
719 // catch this surprisingly common error
720 if (SvROK (TOPs) && SvRV (TOPs) == sv)
721 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
722
723 sv = POPs;
724 PUTBACK;
725
726 encode_sv (enc, sv);
727
728 FREETMPS; LEAVE;
729 }
730 else if (enc->json.flags & F_ALLOW_BLESSED)
731 encode_str (enc, "null", 4, 0);
732 else
733 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
734 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
735 }
736 else if (enc->json.flags & F_ALLOW_BLESSED)
737 encode_str (enc, "null", 4, 0);
738 else
739 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 829 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
740 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 830 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
741 }
742 } 831 }
743 else if (svt == SVt_PVHV) 832 else if (svt == SVt_PVHV)
744 encode_hv (enc, (HV *)sv); 833 encode_hv (enc, (HV *)sv);
745 else if (svt == SVt_PVAV) 834 else if (svt == SVt_PVAV)
746 encode_av (enc, (AV *)sv); 835 encode_av (enc, (AV *)sv);
747 else if (svt < SVt_PVAV) 836 else if (svt < SVt_PVAV)
748 { 837 {
749 STRLEN len = 0; 838 int bool_type = ref_bool_type (sv);
750 char *pv = svt ? SvPV (sv, len) : 0;
751 839
752 if (len == 1 && *pv == '1') 840 if (bool_type == 1)
753 encode_str (enc, "true", 4, 0); 841 encode_str (enc, "true", 4, 0);
754 else if (len == 1 && *pv == '0') 842 else if (bool_type == 0)
755 encode_str (enc, "false", 5, 0); 843 encode_str (enc, "false", 5, 0);
756 else if (enc->json.flags & F_ALLOW_UNKNOWN) 844 else if (enc->json.flags & F_ALLOW_UNKNOWN)
757 encode_str (enc, "null", 4, 0); 845 encode_str (enc, "null", 4, 0);
758 else 846 else
759 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 847 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
822 } 910 }
823 else 911 else
824 { 912 {
825 // large integer, use the (rather slow) snprintf way. 913 // large integer, use the (rather slow) snprintf way.
826 need (enc, IVUV_MAXCHARS); 914 need (enc, IVUV_MAXCHARS);
827 enc->cur += 915 enc->cur +=
828 SvIsUV(sv) 916 SvIsUV(sv)
829 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) 917 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
830 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 918 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
831 } 919 }
832 } 920 }
833 else if (SvROK (sv)) 921 else if (SvROK (sv))
834 encode_rv (enc, SvRV (sv)); 922 encode_rv (enc, SvRV (sv));
835 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 923 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
836 encode_str (enc, "null", 4, 0); 924 encode_str (enc, "null", 4, 0);
837 else 925 else
838 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 926 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
839 SvPV_nolen (sv), SvFLAGS (sv)); 927 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
840} 928}
841 929
842static SV * 930static SV *
843encode_json (SV *scalar, JSON *json) 931encode_json (SV *scalar, JSON *json)
844{ 932{
845 enc_t enc; 933 enc_t enc;
846 934
847 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 935 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
848 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 936 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
849 937
850 enc.json = *json; 938 enc.json = *json;
851 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 939 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
852 enc.cur = SvPVX (enc.sv); 940 enc.cur = SvPVX (enc.sv);
1059 *cur++ = *dec_cur++; 1147 *cur++ = *dec_cur++;
1060 while (--clen); 1148 while (--clen);
1061 1149
1062 utf8 = 1; 1150 utf8 = 1;
1063 } 1151 }
1152 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1153 *cur++ = ch;
1064 else 1154 else
1065 { 1155 {
1066 --dec_cur; 1156 --dec_cur;
1067 1157
1068 if (!ch) 1158 if (!ch)
1256 if (*dec->cur == ']') 1346 if (*dec->cur == ']')
1257 { 1347 {
1258 ++dec->cur; 1348 ++dec->cur;
1259 break; 1349 break;
1260 } 1350 }
1261 1351
1262 if (*dec->cur != ',') 1352 if (*dec->cur != ',')
1263 ERR (", or ] expected while parsing array"); 1353 ERR (", or ] expected while parsing array");
1264 1354
1265 ++dec->cur; 1355 ++dec->cur;
1266 1356
1390 1480
1391 hv_iterinit (hv); 1481 hv_iterinit (hv);
1392 he = hv_iternext (hv); 1482 he = hv_iternext (hv);
1393 hv_iterinit (hv); 1483 hv_iterinit (hv);
1394 1484
1395 // the next line creates a mortal sv each time its called. 1485 // the next line creates a mortal sv each time it's called.
1396 // might want to optimise this for common cases. 1486 // might want to optimise this for common cases.
1397 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1487 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1398 1488
1399 if (cb) 1489 if (cb)
1400 { 1490 {
1401 dSP; 1491 dSP;
1402 int count; 1492 int count;
1403 1493
1404 ENTER; SAVETMPS; PUSHMARK (SP); 1494 ENTER; SAVETMPS;
1495 SAVESTACK_POS ();
1496 PUSHMARK (SP);
1405 XPUSHs (HeVAL (he)); 1497 XPUSHs (HeVAL (he));
1498 sv_2mortal (sv);
1406 1499
1407 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1500 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1408 1501
1409 if (count == 1) 1502 if (count == 1)
1410 { 1503 {
1411 sv = newSVsv (POPs); 1504 sv = newSVsv (POPs);
1412 FREETMPS; LEAVE; 1505 FREETMPS; LEAVE;
1413 return sv; 1506 return sv;
1414 } 1507 }
1415 1508
1509 SvREFCNT_inc (sv);
1416 FREETMPS; LEAVE; 1510 FREETMPS; LEAVE;
1417 } 1511 }
1418 } 1512 }
1419 1513
1420 if (dec->json.cb_object) 1514 if (dec->json.cb_object)
1421 { 1515 {
1422 dSP; 1516 dSP;
1423 int count; 1517 int count;
1424 1518
1425 ENTER; SAVETMPS; PUSHMARK (SP); 1519 ENTER; SAVETMPS;
1520 SAVESTACK_POS ();
1521 PUSHMARK (SP);
1426 XPUSHs (sv_2mortal (sv)); 1522 XPUSHs (sv_2mortal (sv));
1427 1523
1428 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1524 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1429 1525
1430 if (count == 1) 1526 if (count == 1)
1446 DEC_DEC_DEPTH; 1542 DEC_DEC_DEPTH;
1447 return 0; 1543 return 0;
1448} 1544}
1449 1545
1450static SV * 1546static SV *
1547decode_tag (dec_t *dec)
1548{
1549 SV *tag = 0;
1550 SV *val = 0;
1551
1552 if (!(dec->json.flags & F_ALLOW_TAGS))
1553 ERR ("malformed JSON string, neither array, object, number, string or atom");
1554
1555 ++dec->cur;
1556
1557 decode_ws (dec);
1558
1559 tag = decode_sv (dec);
1560 if (!tag)
1561 goto fail;
1562
1563 if (!SvPOK (tag))
1564 ERR ("malformed JSON string, (tag) must be a string");
1565
1566 decode_ws (dec);
1567
1568 if (*dec->cur != ')')
1569 ERR (") expected after tag");
1570
1571 ++dec->cur;
1572
1573 decode_ws (dec);
1574
1575 val = decode_sv (dec);
1576 if (!val)
1577 goto fail;
1578
1579 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1580 ERR ("malformed JSON string, tag value must be an array");
1581
1582 {
1583 AV *av = (AV *)SvRV (val);
1584 int i, len = av_len (av) + 1;
1585 HV *stash = gv_stashsv (tag, 0);
1586 SV *sv;
1587
1588 if (!stash)
1589 ERR ("cannot decode perl-object (package does not exist)");
1590
1591 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1592
1593 if (!method)
1594 ERR ("cannot decode perl-object (package does not have a THAW method)");
1595
1596 dSP;
1597
1598 ENTER; SAVETMPS;
1599 PUSHMARK (SP);
1600 EXTEND (SP, len + 2);
1601 // we re-bless the reference to get overload and other niceties right
1602 PUSHs (tag);
1603 PUSHs (sv_json);
1604
1605 for (i = 0; i < len; ++i)
1606 PUSHs (*av_fetch (av, i, 1));
1607
1608 PUTBACK;
1609 call_sv ((SV *)GvCV (method), G_SCALAR);
1610 SPAGAIN;
1611
1612 SvREFCNT_dec (tag);
1613 SvREFCNT_dec (val);
1614 sv = SvREFCNT_inc (POPs);
1615
1616 PUTBACK;
1617
1618 FREETMPS; LEAVE;
1619
1620 return sv;
1621 }
1622
1623fail:
1624 SvREFCNT_dec (tag);
1625 SvREFCNT_dec (val);
1626 return 0;
1627}
1628
1629static SV *
1451decode_sv (dec_t *dec) 1630decode_sv (dec_t *dec)
1452{ 1631{
1453 // the beauty of JSON: you need exactly one character lookahead 1632 // the beauty of JSON: you need exactly one character lookahead
1454 // to parse everything. 1633 // to parse everything.
1455 switch (*dec->cur) 1634 switch (*dec->cur)
1456 { 1635 {
1457 case '"': ++dec->cur; return decode_str (dec); 1636 case '"': ++dec->cur; return decode_str (dec);
1458 case '[': ++dec->cur; return decode_av (dec); 1637 case '[': ++dec->cur; return decode_av (dec);
1459 case '{': ++dec->cur; return decode_hv (dec); 1638 case '{': ++dec->cur; return decode_hv (dec);
1639 case '(': return decode_tag (dec);
1460 1640
1461 case '-': 1641 case '-':
1462 case '0': case '1': case '2': case '3': case '4': 1642 case '0': case '1': case '2': case '3': case '4':
1463 case '5': case '6': case '7': case '8': case '9': 1643 case '5': case '6': case '7': case '8': case '9':
1464 return decode_num (dec); 1644 return decode_num (dec);
1466 case 't': 1646 case 't':
1467 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1647 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1468 { 1648 {
1469 dec->cur += 4; 1649 dec->cur += 4;
1470#if JSON_SLOW 1650#if JSON_SLOW
1471 json_true = get_bool ("JSON::XS::true"); 1651 bool_true = get_bool ("Types::Serialiser::true");
1472#endif 1652#endif
1473 return newSVsv (json_true); 1653 return newSVsv (bool_true);
1474 } 1654 }
1475 else 1655 else
1476 ERR ("'true' expected"); 1656 ERR ("'true' expected");
1477 1657
1478 break; 1658 break;
1480 case 'f': 1660 case 'f':
1481 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1661 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1482 { 1662 {
1483 dec->cur += 5; 1663 dec->cur += 5;
1484#if JSON_SLOW 1664#if JSON_SLOW
1485 json_false = get_bool ("JSON::XS::false"); 1665 bool_false = get_bool ("Types::Serialiser::false");
1486#endif 1666#endif
1487 return newSVsv (json_false); 1667 return newSVsv (bool_false);
1488 } 1668 }
1489 else 1669 else
1490 ERR ("'false' expected"); 1670 ERR ("'false' expected");
1491 1671
1492 break; 1672 break;
1501 ERR ("'null' expected"); 1681 ERR ("'null' expected");
1502 1682
1503 break; 1683 break;
1504 1684
1505 default: 1685 default:
1506 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1686 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1507 break; 1687 break;
1508 } 1688 }
1509 1689
1510fail: 1690fail:
1511 return 0; 1691 return 0;
1512} 1692}
1513 1693
1514static SV * 1694static SV *
1515decode_json (SV *string, JSON *json, char **offset_return) 1695decode_json (SV *string, JSON *json, STRLEN *offset_return)
1516{ 1696{
1517 dec_t dec; 1697 dec_t dec;
1518 SV *sv; 1698 SV *sv;
1519 1699
1520 /* work around bugs in 5.10 where manipulating magic values 1700 /* work around bugs in 5.10 where manipulating magic values
1521 * will perl ignore the magic in subsequent accesses 1701 * makes perl ignore the magic in subsequent accesses.
1702 * also make a copy of non-PV values, to get them into a clean
1703 * state (SvPV should do that, but it's buggy, see below).
1704 *
1705 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1706 * as reported by Reini Urban.
1522 */ 1707 */
1523 /*SvGETMAGIC (string);*/ 1708 /*SvGETMAGIC (string);*/
1524 if (SvMAGICAL (string)) 1709 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1525 string = sv_2mortal (newSVsv (string)); 1710 string = sv_2mortal (newSVsv (string));
1526 1711
1527 SvUPGRADE (string, SVt_PV); 1712 SvUPGRADE (string, SVt_PV);
1528 1713
1529 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1714 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1546 if (offset > json->max_size && json->max_size) 1731 if (offset > json->max_size && json->max_size)
1547 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1732 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1548 (unsigned long)SvCUR (string), (unsigned long)json->max_size); 1733 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1549 } 1734 }
1550 1735
1551 if (json->flags & F_UTF8) 1736 if (DECODE_WANTS_OCTETS (json))
1552 sv_utf8_downgrade (string, 0); 1737 sv_utf8_downgrade (string, 0);
1553 else 1738 else
1554 sv_utf8_upgrade (string); 1739 sv_utf8_upgrade (string);
1555 1740
1556 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1741 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1568 1753
1569 decode_ws (&dec); 1754 decode_ws (&dec);
1570 sv = decode_sv (&dec); 1755 sv = decode_sv (&dec);
1571 1756
1572 if (offset_return) 1757 if (offset_return)
1573 *offset_return = dec.cur; 1758 *offset_return = dec.cur - SvPVX (string);
1574 1759 else if (sv)
1575 if (!(offset_return || !sv))
1576 { 1760 {
1577 // check for trailing garbage 1761 // check for trailing garbage
1578 decode_ws (&dec); 1762 decode_ws (&dec);
1579 1763
1580 if (*dec.cur) 1764 if (*dec.cur)
1598 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1782 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1599 LEAVE; 1783 LEAVE;
1600 1784
1601 croak ("%s, at character offset %d (before \"%s\")", 1785 croak ("%s, at character offset %d (before \"%s\")",
1602 dec.err, 1786 dec.err,
1603 ptr_to_index (string, dec.cur), 1787 (int)ptr_to_index (string, dec.cur),
1604 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1788 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1605 } 1789 }
1606 1790
1607 sv = sv_2mortal (sv); 1791 sv = sv_2mortal (sv);
1608 1792
1609 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1793 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1610 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1794 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1611 1795
1612 return sv; 1796 return sv;
1613} 1797}
1614 1798
1736 self->incr_mode = INCR_M_STR; 1920 self->incr_mode = INCR_M_STR;
1737 goto incr_m_str; 1921 goto incr_m_str;
1738 1922
1739 case '[': 1923 case '[':
1740 case '{': 1924 case '{':
1925 case '(':
1741 if (++self->incr_nest > self->max_depth) 1926 if (++self->incr_nest > self->max_depth)
1742 croak (ERR_NESTING_EXCEEDED); 1927 croak (ERR_NESTING_EXCEEDED);
1743 break; 1928 break;
1744 1929
1745 case ']': 1930 case ']':
1746 case '}': 1931 case '}':
1747 if (--self->incr_nest <= 0) 1932 if (--self->incr_nest <= 0)
1748 goto interrupt; 1933 goto interrupt;
1934 break;
1935
1936 case ')':
1937 --self->incr_nest;
1749 break; 1938 break;
1750 1939
1751 case '#': 1940 case '#':
1752 self->incr_mode = INCR_M_C1; 1941 self->incr_mode = INCR_M_C1;
1753 goto incr_m_c; 1942 goto incr_m_c;
1779 i >= '0' && i <= '9' ? i - '0' 1968 i >= '0' && i <= '9' ? i - '0'
1780 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1969 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1781 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1970 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1782 : -1; 1971 : -1;
1783 1972
1784 json_stash = gv_stashpv ("JSON::XS" , 1); 1973 json_stash = gv_stashpv ("JSON::XS" , 1);
1785 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1974 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1786
1787 json_true = get_bool ("JSON::XS::true"); 1975 bool_true = get_bool ("Types::Serialiser::true");
1788 json_false = get_bool ("JSON::XS::false"); 1976 bool_false = get_bool ("Types::Serialiser::false");
1977
1978 sv_json = newSVpv ("JSON", 0);
1979 SvREADONLY_on (sv_json);
1789 1980
1790 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ 1981 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1791} 1982}
1792 1983
1793PROTOTYPES: DISABLE 1984PROTOTYPES: DISABLE
1794 1985
1795void CLONE (...) 1986void CLONE (...)
1796 CODE: 1987 CODE:
1797 json_stash = 0; 1988 json_stash = 0;
1798 json_boolean_stash = 0; 1989 bool_stash = 0;
1799 1990
1800void new (char *klass) 1991void new (char *klass)
1801 PPCODE: 1992 PPCODE:
1802{ 1993{
1803 SV *pv = NEWSV (0, sizeof (JSON)); 1994 SV *pv = NEWSV (0, sizeof (JSON));
1804 SvPOK_only (pv); 1995 SvPOK_only (pv);
1805 json_init ((JSON *)SvPVX (pv)); 1996 json_init ((JSON *)SvPVX (pv));
1806 XPUSHs (sv_2mortal (sv_bless ( 1997 XPUSHs (sv_2mortal (sv_bless (
1807 newRV_noinc (pv), 1998 newRV_noinc (pv),
1808 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1999 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1823 shrink = F_SHRINK 2014 shrink = F_SHRINK
1824 allow_blessed = F_ALLOW_BLESSED 2015 allow_blessed = F_ALLOW_BLESSED
1825 convert_blessed = F_CONV_BLESSED 2016 convert_blessed = F_CONV_BLESSED
1826 relaxed = F_RELAXED 2017 relaxed = F_RELAXED
1827 allow_unknown = F_ALLOW_UNKNOWN 2018 allow_unknown = F_ALLOW_UNKNOWN
2019 allow_tags = F_ALLOW_TAGS
1828 PPCODE: 2020 PPCODE:
1829{ 2021{
1830 if (enable) 2022 if (enable)
1831 self->flags |= ix; 2023 self->flags |= ix;
1832 else 2024 else
1848 get_shrink = F_SHRINK 2040 get_shrink = F_SHRINK
1849 get_allow_blessed = F_ALLOW_BLESSED 2041 get_allow_blessed = F_ALLOW_BLESSED
1850 get_convert_blessed = F_CONV_BLESSED 2042 get_convert_blessed = F_CONV_BLESSED
1851 get_relaxed = F_RELAXED 2043 get_relaxed = F_RELAXED
1852 get_allow_unknown = F_ALLOW_UNKNOWN 2044 get_allow_unknown = F_ALLOW_UNKNOWN
2045 get_allow_tags = F_ALLOW_TAGS
1853 PPCODE: 2046 PPCODE:
1854 XPUSHs (boolSV (self->flags & ix)); 2047 XPUSHs (boolSV (self->flags & ix));
1855 2048
1856void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 2049void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1857 PPCODE: 2050 PPCODE:
1885} 2078}
1886 2079
1887void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2080void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1888 PPCODE: 2081 PPCODE:
1889{ 2082{
1890 if (!self->cb_sk_object) 2083 if (!self->cb_sk_object)
1891 self->cb_sk_object = newHV (); 2084 self->cb_sk_object = newHV ();
1892 2085
1893 if (SvOK (cb)) 2086 if (SvOK (cb))
1894 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2087 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1895 else 2088 else
1906 XPUSHs (ST (0)); 2099 XPUSHs (ST (0));
1907} 2100}
1908 2101
1909void encode (JSON *self, SV *scalar) 2102void encode (JSON *self, SV *scalar)
1910 PPCODE: 2103 PPCODE:
1911 XPUSHs (encode_json (scalar, self)); 2104 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2105 XPUSHs (scalar);
1912 2106
1913void decode (JSON *self, SV *jsonstr) 2107void decode (JSON *self, SV *jsonstr)
1914 PPCODE: 2108 PPCODE:
1915 XPUSHs (decode_json (jsonstr, self, 0)); 2109 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2110 XPUSHs (jsonstr);
1916 2111
1917void decode_prefix (JSON *self, SV *jsonstr) 2112void decode_prefix (JSON *self, SV *jsonstr)
1918 PPCODE: 2113 PPCODE:
1919{ 2114{
1920 char *offset; 2115 SV *sv;
2116 STRLEN offset;
2117 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1921 EXTEND (SP, 2); 2118 EXTEND (SP, 2);
1922 PUSHs (decode_json (jsonstr, self, &offset)); 2119 PUSHs (sv);
1923 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2120 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
1924} 2121}
1925 2122
1926void incr_parse (JSON *self, SV *jsonstr = 0) 2123void incr_parse (JSON *self, SV *jsonstr = 0)
1927 PPCODE: 2124 PPCODE:
1928{ 2125{
1929 if (!self->incr_text) 2126 if (!self->incr_text)
1930 self->incr_text = newSVpvn ("", 0); 2127 self->incr_text = newSVpvn ("", 0);
2128
2129 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2130 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2131 if (DECODE_WANTS_OCTETS (self))
2132 {
2133 if (self->incr_pos)
2134 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2135 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2136
2137 sv_utf8_downgrade (self->incr_text, 0);
2138 }
2139 else
2140 {
2141 sv_utf8_upgrade (self->incr_text);
2142
2143 if (self->incr_pos)
2144 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2145 - (U8 *)SvPVX (self->incr_text);
2146 }
1931 2147
1932 // append data, if any 2148 // append data, if any
1933 if (jsonstr) 2149 if (jsonstr)
1934 { 2150 {
2151 /* make sure both strings have same encoding */
2152 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1935 if (SvUTF8 (jsonstr)) 2153 if (SvUTF8 (jsonstr))
2154 sv_utf8_downgrade (jsonstr, 0);
1936 { 2155 else
1937 if (!SvUTF8 (self->incr_text))
1938 {
1939 /* utf-8-ness differs, need to upgrade */
1940 sv_utf8_upgrade (self->incr_text);
1941
1942 if (self->incr_pos)
1943 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1944 - (U8 *)SvPVX (self->incr_text);
1945 }
1946 }
1947 else if (SvUTF8 (self->incr_text))
1948 sv_utf8_upgrade (jsonstr); 2156 sv_utf8_upgrade (jsonstr);
1949 2157
2158 /* and then just blindly append */
1950 { 2159 {
1951 STRLEN len; 2160 STRLEN len;
1952 const char *str = SvPV (jsonstr, len); 2161 const char *str = SvPV (jsonstr, len);
1953 STRLEN cur = SvCUR (self->incr_text); 2162 STRLEN cur = SvCUR (self->incr_text);
1954 2163
1962 } 2171 }
1963 2172
1964 if (GIMME_V != G_VOID) 2173 if (GIMME_V != G_VOID)
1965 do 2174 do
1966 { 2175 {
2176 SV *sv;
1967 char *offset; 2177 STRLEN offset;
1968 2178
1969 if (!INCR_DONE (self)) 2179 if (!INCR_DONE (self))
1970 { 2180 {
1971 incr_parse (self); 2181 incr_parse (self);
1972 2182
1973 if (self->incr_pos > self->max_size && self->max_size) 2183 if (self->incr_pos > self->max_size && self->max_size)
1974 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 2184 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1975 (unsigned long)self->incr_pos, (unsigned long)self->max_size); 2185 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1976 2186
1977 if (!INCR_DONE (self)) 2187 if (!INCR_DONE (self))
2188 {
2189 // as an optimisation, do not accumulate white space in the incr buffer
2190 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2191 {
2192 self->incr_pos = 0;
2193 SvCUR_set (self->incr_text, 0);
2194 }
2195
1978 break; 2196 break;
2197 }
1979 } 2198 }
1980 2199
1981 XPUSHs (decode_json (self->incr_text, self, &offset)); 2200 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2201 XPUSHs (sv);
1982 2202
1983 self->incr_pos -= offset - SvPVX (self->incr_text); 2203 self->incr_pos -= offset;
1984 self->incr_nest = 0; 2204 self->incr_nest = 0;
1985 self->incr_mode = 0; 2205 self->incr_mode = 0;
1986 2206
1987 sv_chop (self->incr_text, offset); 2207 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
1988 } 2208 }
1989 while (GIMME_V == G_ARRAY); 2209 while (GIMME_V == G_ARRAY);
1990} 2210}
1991 2211
1992SV *incr_text (JSON *self) 2212SV *incr_text (JSON *self)
2030 SvREFCNT_dec (self->incr_text); 2250 SvREFCNT_dec (self->incr_text);
2031 2251
2032PROTOTYPES: ENABLE 2252PROTOTYPES: ENABLE
2033 2253
2034void encode_json (SV *scalar) 2254void encode_json (SV *scalar)
2035 ALIAS:
2036 to_json_ = 0
2037 encode_json = F_UTF8
2038 PPCODE: 2255 PPCODE:
2039{ 2256{
2040 JSON json; 2257 JSON json;
2041 json_init (&json); 2258 json_init (&json);
2042 json.flags |= ix; 2259 json.flags |= F_UTF8;
2043 XPUSHs (encode_json (scalar, &json)); 2260 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2261 XPUSHs (scalar);
2044} 2262}
2045 2263
2046void decode_json (SV *jsonstr) 2264void decode_json (SV *jsonstr)
2047 ALIAS:
2048 from_json_ = 0
2049 decode_json = F_UTF8
2050 PPCODE: 2265 PPCODE:
2051{ 2266{
2052 JSON json; 2267 JSON json;
2053 json_init (&json); 2268 json_init (&json);
2054 json.flags |= ix; 2269 json.flags |= F_UTF8;
2055 XPUSHs (decode_json (jsonstr, &json, 0)); 2270 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2271 XPUSHs (jsonstr);
2056} 2272}
2057 2273

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines