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 1.116 by root, Thu May 23 09:31:32 2013 UTC vs.
Revision 1.132 by root, Tue Sep 5 13:07:09 2017 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
67#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?)"
68 81
69#ifdef USE_ITHREADS 82#ifdef USE_ITHREADS
70# define JSON_SLOW 1 83# define JSON_SLOW 1
71# 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))
72#else 86#else
73# define JSON_SLOW 0 87# define JSON_SLOW 0
74# define JSON_STASH json_stash 88# define JSON_STASH json_stash
89# define BOOL_STASH bool_stash
75#endif 90#endif
76 91
77// the amount of HEs to allocate on the stack, when sorting keys 92// the amount of HEs to allocate on the stack, when sorting keys
78#define STACK_HES 64 93#define STACK_HES 64
79 94
80static HV *json_stash, *json_boolean_stash; // JSON::XS:: 95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
81static SV *json_true, *json_false; 96static SV *bool_true, *bool_false, *sv_json;
82 97
83enum { 98enum {
84 INCR_M_WS = 0, // initial whitespace skipping, must be 0 99 INCR_M_WS = 0, // initial whitespace skipping, must be 0
85 INCR_M_STR, // inside string 100 INCR_M_STR, // inside string
86 INCR_M_BS, // inside backslash 101 INCR_M_BS, // inside backslash
140 SvPV_renew (sv, SvCUR (sv) + 1); 155 SvPV_renew (sv, SvCUR (sv) + 1);
141#endif 156#endif
142 } 157 }
143} 158}
144 159
160/* adds two STRLENs together, slow, and with paranoia */
161STRLEN
162strlen_sum (STRLEN l1, STRLEN l2)
163{
164 size_t sum = l1 + l2;
165
166 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
167 croak ("JSON::XS: string size overflow");
168
169 return sum;
170}
171
172/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
173static char *
174json_sv_grow (SV *sv, size_t len1, size_t len2)
175{
176 len1 = strlen_sum (len1, len2);
177 len1 = strlen_sum (len1, len1 >> 1);
178
179 if (len1 > 4096 - 24)
180 len1 = (len1 | 4095) - 24;
181
182 return SvGROW (sv, len1);
183}
184
145// decode an utf-8 character and return it, or (UV)-1 in 185// decode a utf-8 character and return it, or (UV)-1 in
146// case of an error. 186// case of an error.
147// we special-case "safe" characters from U+80 .. U+7FF, 187// we special-case "safe" characters from U+80 .. U+7FF,
148// but use the very good perl function to parse anything else. 188// but use the very good perl function to parse anything else.
149// note that we never call this function for a ascii codepoints 189// note that we never call this function for a ascii codepoints
150INLINE UV 190INLINE UV
287 // a recursion depth of ten gives us >>500 bits 327 // a recursion depth of ten gives us >>500 bits
288 json_atof_scan1 (s, &accum, &expo, 0, 10); 328 json_atof_scan1 (s, &accum, &expo, 0, 10);
289 329
290 return neg ? -accum : accum; 330 return neg ? -accum : accum;
291} 331}
332
333// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
334static int
335ref_bool_type (SV *sv)
336{
337 svtype svt = SvTYPE (sv);
338
339 if (svt < SVt_PVAV)
340 {
341 STRLEN len = 0;
342 char *pv = svt ? SvPV (sv, len) : 0;
343
344 if (len == 1)
345 if (*pv == '1')
346 return 1;
347 else if (*pv == '0')
348 return 0;
349 }
350
351 return -1;
352}
353
354// returns whether scalar is not a reference in the sense of allow_nonref
355static int
356json_nonref (SV *scalar)
357{
358 if (!SvROK (scalar))
359 return 1;
360
361 scalar = SvRV (scalar);
362
363 if (SvTYPE (scalar) >= SVt_PVMG)
364 {
365 if (SvSTASH (scalar) == bool_stash)
366 return 1;
367
368 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
369 return 1;
370 }
371
372 return 0;
373}
374
292///////////////////////////////////////////////////////////////////////////// 375/////////////////////////////////////////////////////////////////////////////
293// encoder 376// encoder
294 377
295// structure used for encoding JSON 378// structure used for encoding JSON
296typedef struct 379typedef struct
304} enc_t; 387} enc_t;
305 388
306INLINE void 389INLINE void
307need (enc_t *enc, STRLEN len) 390need (enc_t *enc, STRLEN len)
308{ 391{
309 if (expect_false (enc->cur + len >= enc->end)) 392 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
310 { 393 {
311 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 394 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
312 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 395 char *buf = json_sv_grow (enc->sv, cur, len);
313 enc->cur = SvPVX (enc->sv) + cur; 396 enc->cur = buf + cur;
314 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 397 enc->end = buf + SvLEN (enc->sv) - 1;
315 } 398 }
316} 399}
317 400
318INLINE void 401INLINE void
319encode_ch (enc_t *enc, char ch) 402encode_ch (enc_t *enc, char ch)
335 418
336 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 419 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
337 { 420 {
338 if (expect_false (ch == '"')) // but with slow exceptions 421 if (expect_false (ch == '"')) // but with slow exceptions
339 { 422 {
340 need (enc, len += 1); 423 need (enc, len + 1);
341 *enc->cur++ = '\\'; 424 *enc->cur++ = '\\';
342 *enc->cur++ = '"'; 425 *enc->cur++ = '"';
343 } 426 }
344 else if (expect_false (ch == '\\')) 427 else if (expect_false (ch == '\\'))
345 { 428 {
346 need (enc, len += 1); 429 need (enc, len + 1);
347 *enc->cur++ = '\\'; 430 *enc->cur++ = '\\';
348 *enc->cur++ = '\\'; 431 *enc->cur++ = '\\';
349 } 432 }
350 else 433 else
351 *enc->cur++ = ch; 434 *enc->cur++ = ch;
354 } 437 }
355 else 438 else
356 { 439 {
357 switch (ch) 440 switch (ch)
358 { 441 {
359 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 442 case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
360 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; 443 case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
361 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; 444 case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
362 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; 445 case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
363 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; 446 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
364 447
365 default: 448 default:
366 { 449 {
367 STRLEN clen; 450 STRLEN clen;
368 UV uch; 451 UV uch;
384 if (uch >= 0x10000UL) 467 if (uch >= 0x10000UL)
385 { 468 {
386 if (uch >= 0x110000UL) 469 if (uch >= 0x110000UL)
387 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 470 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
388 471
389 need (enc, len += 11); 472 need (enc, len + 11);
390 sprintf (enc->cur, "\\u%04x\\u%04x", 473 sprintf (enc->cur, "\\u%04x\\u%04x",
391 (int)((uch - 0x10000) / 0x400 + 0xD800), 474 (int)((uch - 0x10000) / 0x400 + 0xD800),
392 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 475 (int)((uch - 0x10000) % 0x400 + 0xDC00));
393 enc->cur += 12; 476 enc->cur += 12;
394 } 477 }
395 else 478 else
396 { 479 {
397 need (enc, len += 5); 480 need (enc, len + 5);
398 *enc->cur++ = '\\'; 481 *enc->cur++ = '\\';
399 *enc->cur++ = 'u'; 482 *enc->cur++ = 'u';
400 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 483 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
401 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 484 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
402 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 485 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
410 *enc->cur++ = uch; 493 *enc->cur++ = uch;
411 str += clen; 494 str += clen;
412 } 495 }
413 else if (is_utf8) 496 else if (is_utf8)
414 { 497 {
415 need (enc, len += clen); 498 need (enc, len + clen);
416 do 499 do
417 { 500 {
418 *enc->cur++ = *str++; 501 *enc->cur++ = *str++;
419 } 502 }
420 while (--clen); 503 while (--clen);
421 } 504 }
422 else 505 else
423 { 506 {
424 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 507 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
425 enc->cur = encode_utf8 (enc->cur, uch); 508 enc->cur = encode_utf8 (enc->cur, uch);
426 ++str; 509 ++str;
427 } 510 }
428 } 511 }
429 } 512 }
483 566
484 if (enc->indent >= enc->json.max_depth) 567 if (enc->indent >= enc->json.max_depth)
485 croak (ERR_NESTING_EXCEEDED); 568 croak (ERR_NESTING_EXCEEDED);
486 569
487 encode_ch (enc, '['); 570 encode_ch (enc, '[');
488 571
489 if (len >= 0) 572 if (len >= 0)
490 { 573 {
491 encode_nl (enc); ++enc->indent; 574 encode_nl (enc); ++enc->indent;
492 575
493 for (i = 0; i <= len; ++i) 576 for (i = 0; i <= len; ++i)
505 encode_comma (enc); 588 encode_comma (enc);
506 } 589 }
507 590
508 encode_nl (enc); --enc->indent; encode_indent (enc); 591 encode_nl (enc); --enc->indent; encode_indent (enc);
509 } 592 }
510 593
511 encode_ch (enc, ']'); 594 encode_ch (enc, ']');
512} 595}
513 596
514static void 597static void
515encode_hk (enc_t *enc, HE *he) 598encode_hk (enc_t *enc, HE *he)
519 if (HeKLEN (he) == HEf_SVKEY) 602 if (HeKLEN (he) == HEf_SVKEY)
520 { 603 {
521 SV *sv = HeSVKEY (he); 604 SV *sv = HeSVKEY (he);
522 STRLEN len; 605 STRLEN len;
523 char *str; 606 char *str;
524 607
525 SvGETMAGIC (sv); 608 SvGETMAGIC (sv);
526 str = SvPV (sv, len); 609 str = SvPV (sv, len);
527 610
528 encode_str (enc, str, len, SvUTF8 (sv)); 611 encode_str (enc, str, len, SvUTF8 (sv));
529 } 612 }
597 if (count) 680 if (count)
598 { 681 {
599 int i, fast = 1; 682 int i, fast = 1;
600 HE *hes_stack [STACK_HES]; 683 HE *hes_stack [STACK_HES];
601 HE **hes = hes_stack; 684 HE **hes = hes_stack;
602 685
603 // allocate larger arrays on the heap 686 // allocate larger arrays on the heap
604 if (count > STACK_HES) 687 if (count > STACK_HES)
605 { 688 {
606 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes))); 689 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
607 hes = (HE **)SvPVX (sv); 690 hes = (HE **)SvPVX (sv);
682// encode objects, arrays and special \0=false and \1=true values. 765// encode objects, arrays and special \0=false and \1=true values.
683static void 766static void
684encode_rv (enc_t *enc, SV *sv) 767encode_rv (enc_t *enc, SV *sv)
685{ 768{
686 svtype svt; 769 svtype svt;
770 GV *method;
687 771
688 SvGETMAGIC (sv); 772 SvGETMAGIC (sv);
689 svt = SvTYPE (sv); 773 svt = SvTYPE (sv);
690 774
691 if (expect_false (SvOBJECT (sv))) 775 if (expect_false (SvOBJECT (sv)))
692 { 776 {
693 HV *stash = !JSON_SLOW || json_boolean_stash 777 HV *stash = SvSTASH (sv);
694 ? json_boolean_stash
695 : gv_stashpv ("JSON::XS::Boolean", 1);
696 778
697 if (SvSTASH (sv) == stash) 779 if (stash == bool_stash)
698 { 780 {
699 if (SvIV (sv)) 781 if (SvIV (sv))
700 encode_str (enc, "true", 4, 0); 782 encode_str (enc, "true", 4, 0);
701 else 783 else
702 encode_str (enc, "false", 5, 0); 784 encode_str (enc, "false", 5, 0);
703 } 785 }
786 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
787 {
788 int count;
789 dSP;
790
791 ENTER; SAVETMPS;
792 SAVESTACK_POS ();
793 PUSHMARK (SP);
794 EXTEND (SP, 2);
795 // we re-bless the reference to get overload and other niceties right
796 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
797 PUSHs (sv_json);
798
799 PUTBACK;
800 count = call_sv ((SV *)GvCV (method), G_ARRAY);
801 SPAGAIN;
802
803 // catch this surprisingly common error
804 if (SvROK (TOPs) && SvRV (TOPs) == sv)
805 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
806
807 encode_ch (enc, '(');
808 encode_ch (enc, '"');
809 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
810 encode_ch (enc, '"');
811 encode_ch (enc, ')');
812 encode_ch (enc, '[');
813
814 while (count)
815 {
816 encode_sv (enc, SP[1 - count--]);
817
818 if (count)
819 encode_ch (enc, ',');
820 }
821
822 encode_ch (enc, ']');
823
824 FREETMPS; LEAVE;
825 }
826 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
827 {
828 dSP;
829
830 ENTER; SAVETMPS;
831 PUSHMARK (SP);
832 // we re-bless the reference to get overload and other niceties right
833 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
834
835 // calling with G_SCALAR ensures that we always get a 1 return value
836 PUTBACK;
837 call_sv ((SV *)GvCV (method), G_SCALAR);
838 SPAGAIN;
839
840 // catch this surprisingly common error
841 if (SvROK (TOPs) && SvRV (TOPs) == sv)
842 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
843
844 sv = POPs;
845 PUTBACK;
846
847 encode_sv (enc, sv);
848
849 FREETMPS; LEAVE;
850 }
851 else if (enc->json.flags & F_ALLOW_BLESSED)
852 encode_str (enc, "null", 4, 0);
704 else 853 else
705 {
706#if 0
707 if (0 && sv_derived_from (rv, "JSON::Literal"))
708 {
709 // not yet
710 }
711#endif
712 if (enc->json.flags & F_CONV_BLESSED)
713 {
714 // we re-bless the reference to get overload and other niceties right
715 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
716
717 if (to_json)
718 {
719 dSP;
720
721 ENTER; SAVETMPS; PUSHMARK (SP);
722 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
723
724 // calling with G_SCALAR ensures that we always get a 1 return value
725 PUTBACK;
726 call_sv ((SV *)GvCV (to_json), G_SCALAR);
727 SPAGAIN;
728
729 // catch this surprisingly common error
730 if (SvROK (TOPs) && SvRV (TOPs) == sv)
731 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
732
733 sv = POPs;
734 PUTBACK;
735
736 encode_sv (enc, sv);
737
738 FREETMPS; LEAVE;
739 }
740 else if (enc->json.flags & F_ALLOW_BLESSED)
741 encode_str (enc, "null", 4, 0);
742 else
743 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
744 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
745 }
746 else if (enc->json.flags & F_ALLOW_BLESSED)
747 encode_str (enc, "null", 4, 0);
748 else
749 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 854 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
750 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 855 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
751 }
752 } 856 }
753 else if (svt == SVt_PVHV) 857 else if (svt == SVt_PVHV)
754 encode_hv (enc, (HV *)sv); 858 encode_hv (enc, (HV *)sv);
755 else if (svt == SVt_PVAV) 859 else if (svt == SVt_PVAV)
756 encode_av (enc, (AV *)sv); 860 encode_av (enc, (AV *)sv);
757 else if (svt < SVt_PVAV) 861 else if (svt < SVt_PVAV)
758 { 862 {
759 STRLEN len = 0; 863 int bool_type = ref_bool_type (sv);
760 char *pv = svt ? SvPV (sv, len) : 0;
761 864
762 if (len == 1 && *pv == '1') 865 if (bool_type == 1)
763 encode_str (enc, "true", 4, 0); 866 encode_str (enc, "true", 4, 0);
764 else if (len == 1 && *pv == '0') 867 else if (bool_type == 0)
765 encode_str (enc, "false", 5, 0); 868 encode_str (enc, "false", 5, 0);
766 else if (enc->json.flags & F_ALLOW_UNKNOWN) 869 else if (enc->json.flags & F_ALLOW_UNKNOWN)
767 encode_str (enc, "null", 4, 0); 870 encode_str (enc, "null", 4, 0);
768 else 871 else
769 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 872 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
843 else if (SvROK (sv)) 946 else if (SvROK (sv))
844 encode_rv (enc, SvRV (sv)); 947 encode_rv (enc, SvRV (sv));
845 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 948 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
846 encode_str (enc, "null", 4, 0); 949 encode_str (enc, "null", 4, 0);
847 else 950 else
848 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 951 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
849 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 952 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
850} 953}
851 954
852static SV * 955static SV *
853encode_json (SV *scalar, JSON *json) 956encode_json (SV *scalar, JSON *json)
854{ 957{
855 enc_t enc; 958 enc_t enc;
856 959
857 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 960 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
858 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 961 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
859 962
860 enc.json = *json; 963 enc.json = *json;
861 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 964 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
862 enc.cur = SvPVX (enc.sv); 965 enc.cur = SvPVX (enc.sv);
1069 *cur++ = *dec_cur++; 1172 *cur++ = *dec_cur++;
1070 while (--clen); 1173 while (--clen);
1071 1174
1072 utf8 = 1; 1175 utf8 = 1;
1073 } 1176 }
1177 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1178 *cur++ = ch;
1074 else 1179 else
1075 { 1180 {
1076 --dec_cur; 1181 --dec_cur;
1077 1182
1078 if (!ch) 1183 if (!ch)
1088 1193
1089 if (sv) 1194 if (sv)
1090 { 1195 {
1091 STRLEN cur = SvCUR (sv); 1196 STRLEN cur = SvCUR (sv);
1092 1197
1093 if (SvLEN (sv) <= cur + len) 1198 if (SvLEN (sv) - cur <= len)
1094 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 1199 json_sv_grow (sv, cur, len);
1095 1200
1096 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1201 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1097 SvCUR_set (sv, SvCUR (sv) + len); 1202 SvCUR_set (sv, SvCUR (sv) + len);
1098 } 1203 }
1099 else 1204 else
1266 if (*dec->cur == ']') 1371 if (*dec->cur == ']')
1267 { 1372 {
1268 ++dec->cur; 1373 ++dec->cur;
1269 break; 1374 break;
1270 } 1375 }
1271 1376
1272 if (*dec->cur != ',') 1377 if (*dec->cur != ',')
1273 ERR (", or ] expected while parsing array"); 1378 ERR (", or ] expected while parsing array");
1274 1379
1275 ++dec->cur; 1380 ++dec->cur;
1276 1381
1400 1505
1401 hv_iterinit (hv); 1506 hv_iterinit (hv);
1402 he = hv_iternext (hv); 1507 he = hv_iternext (hv);
1403 hv_iterinit (hv); 1508 hv_iterinit (hv);
1404 1509
1405 // the next line creates a mortal sv each time its called. 1510 // the next line creates a mortal sv each time it's called.
1406 // might want to optimise this for common cases. 1511 // might want to optimise this for common cases.
1407 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1512 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1408 1513
1409 if (cb) 1514 if (cb)
1410 { 1515 {
1411 dSP; 1516 dSP;
1412 int count; 1517 int count;
1413 1518
1414 ENTER; SAVETMPS; PUSHMARK (SP); 1519 ENTER; SAVETMPS;
1520 SAVESTACK_POS ();
1521 PUSHMARK (SP);
1415 XPUSHs (HeVAL (he)); 1522 XPUSHs (HeVAL (he));
1416 sv_2mortal (sv); 1523 sv_2mortal (sv);
1417 1524
1418 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1525 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1419 1526
1432 if (dec->json.cb_object) 1539 if (dec->json.cb_object)
1433 { 1540 {
1434 dSP; 1541 dSP;
1435 int count; 1542 int count;
1436 1543
1437 ENTER; SAVETMPS; PUSHMARK (SP); 1544 ENTER; SAVETMPS;
1545 SAVESTACK_POS ();
1546 PUSHMARK (SP);
1438 XPUSHs (sv_2mortal (sv)); 1547 XPUSHs (sv_2mortal (sv));
1439 1548
1440 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1549 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1441 1550
1442 if (count == 1) 1551 if (count == 1)
1458 DEC_DEC_DEPTH; 1567 DEC_DEC_DEPTH;
1459 return 0; 1568 return 0;
1460} 1569}
1461 1570
1462static SV * 1571static SV *
1572decode_tag (dec_t *dec)
1573{
1574 SV *tag = 0;
1575 SV *val = 0;
1576
1577 if (!(dec->json.flags & F_ALLOW_TAGS))
1578 ERR ("malformed JSON string, neither array, object, number, string or atom");
1579
1580 ++dec->cur;
1581
1582 decode_ws (dec);
1583
1584 tag = decode_sv (dec);
1585 if (!tag)
1586 goto fail;
1587
1588 if (!SvPOK (tag))
1589 ERR ("malformed JSON string, (tag) must be a string");
1590
1591 decode_ws (dec);
1592
1593 if (*dec->cur != ')')
1594 ERR (") expected after tag");
1595
1596 ++dec->cur;
1597
1598 decode_ws (dec);
1599
1600 val = decode_sv (dec);
1601 if (!val)
1602 goto fail;
1603
1604 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1605 ERR ("malformed JSON string, tag value must be an array");
1606
1607 {
1608 AV *av = (AV *)SvRV (val);
1609 int i, len = av_len (av) + 1;
1610 HV *stash = gv_stashsv (tag, 0);
1611 SV *sv;
1612
1613 if (!stash)
1614 ERR ("cannot decode perl-object (package does not exist)");
1615
1616 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1617
1618 if (!method)
1619 ERR ("cannot decode perl-object (package does not have a THAW method)");
1620
1621 dSP;
1622
1623 ENTER; SAVETMPS;
1624 PUSHMARK (SP);
1625 EXTEND (SP, len + 2);
1626 // we re-bless the reference to get overload and other niceties right
1627 PUSHs (tag);
1628 PUSHs (sv_json);
1629
1630 for (i = 0; i < len; ++i)
1631 PUSHs (*av_fetch (av, i, 1));
1632
1633 PUTBACK;
1634 call_sv ((SV *)GvCV (method), G_SCALAR);
1635 SPAGAIN;
1636
1637 SvREFCNT_dec (tag);
1638 SvREFCNT_dec (val);
1639 sv = SvREFCNT_inc (POPs);
1640
1641 PUTBACK;
1642
1643 FREETMPS; LEAVE;
1644
1645 return sv;
1646 }
1647
1648fail:
1649 SvREFCNT_dec (tag);
1650 SvREFCNT_dec (val);
1651 return 0;
1652}
1653
1654static SV *
1463decode_sv (dec_t *dec) 1655decode_sv (dec_t *dec)
1464{ 1656{
1465 // the beauty of JSON: you need exactly one character lookahead 1657 // the beauty of JSON: you need exactly one character lookahead
1466 // to parse everything. 1658 // to parse everything.
1467 switch (*dec->cur) 1659 switch (*dec->cur)
1468 { 1660 {
1469 case '"': ++dec->cur; return decode_str (dec); 1661 case '"': ++dec->cur; return decode_str (dec);
1470 case '[': ++dec->cur; return decode_av (dec); 1662 case '[': ++dec->cur; return decode_av (dec);
1471 case '{': ++dec->cur; return decode_hv (dec); 1663 case '{': ++dec->cur; return decode_hv (dec);
1664 case '(': return decode_tag (dec);
1472 1665
1473 case '-': 1666 case '-':
1474 case '0': case '1': case '2': case '3': case '4': 1667 case '0': case '1': case '2': case '3': case '4':
1475 case '5': case '6': case '7': case '8': case '9': 1668 case '5': case '6': case '7': case '8': case '9':
1476 return decode_num (dec); 1669 return decode_num (dec);
1478 case 't': 1671 case 't':
1479 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1672 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1480 { 1673 {
1481 dec->cur += 4; 1674 dec->cur += 4;
1482#if JSON_SLOW 1675#if JSON_SLOW
1483 json_true = get_bool ("JSON::XS::true"); 1676 bool_true = get_bool ("Types::Serialiser::true");
1484#endif 1677#endif
1485 return newSVsv (json_true); 1678 return newSVsv (bool_true);
1486 } 1679 }
1487 else 1680 else
1488 ERR ("'true' expected"); 1681 ERR ("'true' expected");
1489 1682
1490 break; 1683 break;
1492 case 'f': 1685 case 'f':
1493 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1686 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1494 { 1687 {
1495 dec->cur += 5; 1688 dec->cur += 5;
1496#if JSON_SLOW 1689#if JSON_SLOW
1497 json_false = get_bool ("JSON::XS::false"); 1690 bool_false = get_bool ("Types::Serialiser::false");
1498#endif 1691#endif
1499 return newSVsv (json_false); 1692 return newSVsv (bool_false);
1500 } 1693 }
1501 else 1694 else
1502 ERR ("'false' expected"); 1695 ERR ("'false' expected");
1503 1696
1504 break; 1697 break;
1513 ERR ("'null' expected"); 1706 ERR ("'null' expected");
1514 1707
1515 break; 1708 break;
1516 1709
1517 default: 1710 default:
1518 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1711 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1519 break; 1712 break;
1520 } 1713 }
1521 1714
1522fail: 1715fail:
1523 return 0; 1716 return 0;
1524} 1717}
1525 1718
1526static SV * 1719static SV *
1527decode_json (SV *string, JSON *json, char **offset_return) 1720decode_json (SV *string, JSON *json, STRLEN *offset_return)
1528{ 1721{
1529 dec_t dec; 1722 dec_t dec;
1530 SV *sv; 1723 SV *sv;
1531 1724
1532 /* work around bugs in 5.10 where manipulating magic values 1725 /* work around bugs in 5.10 where manipulating magic values
1533 * makes perl ignore the magic in subsequent accesses. 1726 * makes perl ignore the magic in subsequent accesses.
1534 * also make a copy of non-PV values, to get them into a clean 1727 * also make a copy of non-PV values, to get them into a clean
1535 * state (SvPV should do that, but it's buggy, see below). 1728 * state (SvPV should do that, but it's buggy, see below).
1729 *
1730 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1731 * as reported by Reini Urban.
1536 */ 1732 */
1537 /*SvGETMAGIC (string);*/ 1733 /*SvGETMAGIC (string);*/
1538 if (SvMAGICAL (string) || !SvPOK (string)) 1734 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1539 string = sv_2mortal (newSVsv (string)); 1735 string = sv_2mortal (newSVsv (string));
1540 1736
1541 SvUPGRADE (string, SVt_PV); 1737 SvUPGRADE (string, SVt_PV);
1542 1738
1543 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1739 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1582 1778
1583 decode_ws (&dec); 1779 decode_ws (&dec);
1584 sv = decode_sv (&dec); 1780 sv = decode_sv (&dec);
1585 1781
1586 if (offset_return) 1782 if (offset_return)
1587 *offset_return = dec.cur; 1783 *offset_return = dec.cur - SvPVX (string);
1588 1784 else if (sv)
1589 if (!(offset_return || !sv))
1590 { 1785 {
1591 // check for trailing garbage 1786 // check for trailing garbage
1592 decode_ws (&dec); 1787 decode_ws (&dec);
1593 1788
1594 if (*dec.cur) 1789 if (*dec.cur)
1618 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1813 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1619 } 1814 }
1620 1815
1621 sv = sv_2mortal (sv); 1816 sv = sv_2mortal (sv);
1622 1817
1623 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1818 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1624 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1819 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1625 1820
1626 return sv; 1821 return sv;
1627} 1822}
1628 1823
1750 self->incr_mode = INCR_M_STR; 1945 self->incr_mode = INCR_M_STR;
1751 goto incr_m_str; 1946 goto incr_m_str;
1752 1947
1753 case '[': 1948 case '[':
1754 case '{': 1949 case '{':
1950 case '(':
1755 if (++self->incr_nest > self->max_depth) 1951 if (++self->incr_nest > self->max_depth)
1756 croak (ERR_NESTING_EXCEEDED); 1952 croak (ERR_NESTING_EXCEEDED);
1757 break; 1953 break;
1758 1954
1759 case ']': 1955 case ']':
1760 case '}': 1956 case '}':
1761 if (--self->incr_nest <= 0) 1957 if (--self->incr_nest <= 0)
1762 goto interrupt; 1958 goto interrupt;
1959 break;
1960
1961 case ')':
1962 --self->incr_nest;
1763 break; 1963 break;
1764 1964
1765 case '#': 1965 case '#':
1766 self->incr_mode = INCR_M_C1; 1966 self->incr_mode = INCR_M_C1;
1767 goto incr_m_c; 1967 goto incr_m_c;
1793 i >= '0' && i <= '9' ? i - '0' 1993 i >= '0' && i <= '9' ? i - '0'
1794 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1994 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1795 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1995 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1796 : -1; 1996 : -1;
1797 1997
1798 json_stash = gv_stashpv ("JSON::XS" , 1); 1998 json_stash = gv_stashpv ("JSON::XS" , 1);
1799 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1999 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1800
1801 json_true = get_bool ("JSON::XS::true"); 2000 bool_true = get_bool ("Types::Serialiser::true");
1802 json_false = get_bool ("JSON::XS::false"); 2001 bool_false = get_bool ("Types::Serialiser::false");
2002
2003 sv_json = newSVpv ("JSON", 0);
2004 SvREADONLY_on (sv_json);
1803 2005
1804 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ 2006 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1805} 2007}
1806 2008
1807PROTOTYPES: DISABLE 2009PROTOTYPES: DISABLE
1808 2010
1809void CLONE (...) 2011void CLONE (...)
1810 CODE: 2012 CODE:
1811 json_stash = 0; 2013 json_stash = 0;
1812 json_boolean_stash = 0; 2014 bool_stash = 0;
1813 2015
1814void new (char *klass) 2016void new (char *klass)
1815 PPCODE: 2017 PPCODE:
1816{ 2018{
1817 SV *pv = NEWSV (0, sizeof (JSON)); 2019 SV *pv = NEWSV (0, sizeof (JSON));
1837 shrink = F_SHRINK 2039 shrink = F_SHRINK
1838 allow_blessed = F_ALLOW_BLESSED 2040 allow_blessed = F_ALLOW_BLESSED
1839 convert_blessed = F_CONV_BLESSED 2041 convert_blessed = F_CONV_BLESSED
1840 relaxed = F_RELAXED 2042 relaxed = F_RELAXED
1841 allow_unknown = F_ALLOW_UNKNOWN 2043 allow_unknown = F_ALLOW_UNKNOWN
2044 allow_tags = F_ALLOW_TAGS
1842 PPCODE: 2045 PPCODE:
1843{ 2046{
1844 if (enable) 2047 if (enable)
1845 self->flags |= ix; 2048 self->flags |= ix;
1846 else 2049 else
1862 get_shrink = F_SHRINK 2065 get_shrink = F_SHRINK
1863 get_allow_blessed = F_ALLOW_BLESSED 2066 get_allow_blessed = F_ALLOW_BLESSED
1864 get_convert_blessed = F_CONV_BLESSED 2067 get_convert_blessed = F_CONV_BLESSED
1865 get_relaxed = F_RELAXED 2068 get_relaxed = F_RELAXED
1866 get_allow_unknown = F_ALLOW_UNKNOWN 2069 get_allow_unknown = F_ALLOW_UNKNOWN
2070 get_allow_tags = F_ALLOW_TAGS
1867 PPCODE: 2071 PPCODE:
1868 XPUSHs (boolSV (self->flags & ix)); 2072 XPUSHs (boolSV (self->flags & ix));
1869 2073
1870void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 2074void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1871 PPCODE: 2075 PPCODE:
1932 2136
1933void decode_prefix (JSON *self, SV *jsonstr) 2137void decode_prefix (JSON *self, SV *jsonstr)
1934 PPCODE: 2138 PPCODE:
1935{ 2139{
1936 SV *sv; 2140 SV *sv;
1937 char *offset; 2141 STRLEN offset;
1938 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; 2142 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1939 EXTEND (SP, 2); 2143 EXTEND (SP, 2);
1940 PUSHs (sv); 2144 PUSHs (sv);
1941 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2145 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
1942} 2146}
1943 2147
1944void incr_parse (JSON *self, SV *jsonstr = 0) 2148void incr_parse (JSON *self, SV *jsonstr = 0)
1945 PPCODE: 2149 PPCODE:
1946{ 2150{
1980 { 2184 {
1981 STRLEN len; 2185 STRLEN len;
1982 const char *str = SvPV (jsonstr, len); 2186 const char *str = SvPV (jsonstr, len);
1983 STRLEN cur = SvCUR (self->incr_text); 2187 STRLEN cur = SvCUR (self->incr_text);
1984 2188
1985 if (SvLEN (self->incr_text) <= cur + len) 2189 if (SvLEN (self->incr_text) - cur <= len)
1986 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2190 json_sv_grow (self->incr_text, cur, len);
1987 2191
1988 Move (str, SvEND (self->incr_text), len, char); 2192 Move (str, SvEND (self->incr_text), len, char);
1989 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2193 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1990 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 2194 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1991 } 2195 }
1993 2197
1994 if (GIMME_V != G_VOID) 2198 if (GIMME_V != G_VOID)
1995 do 2199 do
1996 { 2200 {
1997 SV *sv; 2201 SV *sv;
1998 char *offset; 2202 STRLEN offset;
1999 2203
2000 if (!INCR_DONE (self)) 2204 if (!INCR_DONE (self))
2001 { 2205 {
2002 incr_parse (self); 2206 incr_parse (self);
2003 2207
2019 } 2223 }
2020 2224
2021 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; 2225 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2022 XPUSHs (sv); 2226 XPUSHs (sv);
2023 2227
2024 self->incr_pos -= offset - SvPVX (self->incr_text); 2228 self->incr_pos -= offset;
2025 self->incr_nest = 0; 2229 self->incr_nest = 0;
2026 self->incr_mode = 0; 2230 self->incr_mode = 0;
2027 2231
2028 sv_chop (self->incr_text, offset); 2232 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2029 } 2233 }
2030 while (GIMME_V == G_ARRAY); 2234 while (GIMME_V == G_ARRAY);
2031} 2235}
2032 2236
2033SV *incr_text (JSON *self) 2237SV *incr_text (JSON *self)
2071 SvREFCNT_dec (self->incr_text); 2275 SvREFCNT_dec (self->incr_text);
2072 2276
2073PROTOTYPES: ENABLE 2277PROTOTYPES: ENABLE
2074 2278
2075void encode_json (SV *scalar) 2279void encode_json (SV *scalar)
2076 ALIAS:
2077 to_json_ = 0
2078 encode_json = F_UTF8
2079 PPCODE: 2280 PPCODE:
2080{ 2281{
2081 JSON json; 2282 JSON json;
2082 json_init (&json); 2283 json_init (&json);
2083 json.flags |= ix; 2284 json.flags |= F_UTF8;
2084 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN; 2285 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2085 XPUSHs (scalar); 2286 XPUSHs (scalar);
2086} 2287}
2087 2288
2088void decode_json (SV *jsonstr) 2289void decode_json (SV *jsonstr)
2089 ALIAS:
2090 from_json_ = 0
2091 decode_json = F_UTF8
2092 PPCODE: 2290 PPCODE:
2093{ 2291{
2094 JSON json; 2292 JSON json;
2095 json_init (&json); 2293 json_init (&json);
2096 json.flags |= ix; 2294 json.flags |= F_UTF8;
2097 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN; 2295 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2098 XPUSHs (jsonstr); 2296 XPUSHs (jsonstr);
2099} 2297}
2100 2298

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines