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 1.133 by root, Tue Aug 28 16:16:17 2018 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
63 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 78 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
64 79
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
69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 83# define JSON_STASH (expect_true (json_stash) ? json_stash : gv_stashpv ("JSON::XS", 1))
84# define BOOL_STASH (expect_true (bool_stash) ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1))
85# define GET_BOOL(value) (expect_true (bool_ ## value) ? bool_ ## value : get_bool ("Types::Serialiser::" # value))
70#else 86#else
71# define JSON_SLOW 0
72# define JSON_STASH json_stash 87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash
89# define GET_BOOL(value) bool_ ## value
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_false, *bool_true;
97static SV *sv_json;
77 98
78enum { 99enum {
79 INCR_M_WS = 0, // initial whitespace skipping, must be 0 100 INCR_M_WS = 0, // initial whitespace skipping, must be 0
80 INCR_M_STR, // inside string 101 INCR_M_STR, // inside string
81 INCR_M_BS, // inside backslash 102 INCR_M_BS, // inside backslash
97 // for the incremental parser 118 // for the incremental parser
98 SV *incr_text; // the source text so far 119 SV *incr_text; // the source text so far
99 STRLEN incr_pos; // the current offset into the text 120 STRLEN incr_pos; // the current offset into the text
100 int incr_nest; // {[]}-nesting level 121 int incr_nest; // {[]}-nesting level
101 unsigned char incr_mode; 122 unsigned char incr_mode;
123
124 SV *v_false, *v_true;
102} JSON; 125} JSON;
103 126
104INLINE void 127INLINE void
105json_init (JSON *json) 128json_init (JSON *json)
106{ 129{
135 SvPV_renew (sv, SvCUR (sv) + 1); 158 SvPV_renew (sv, SvCUR (sv) + 1);
136#endif 159#endif
137 } 160 }
138} 161}
139 162
163/* adds two STRLENs together, slow, and with paranoia */
164STRLEN
165strlen_sum (STRLEN l1, STRLEN l2)
166{
167 size_t sum = l1 + l2;
168
169 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
170 croak ("JSON::XS: string size overflow");
171
172 return sum;
173}
174
175/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
176static char *
177json_sv_grow (SV *sv, size_t len1, size_t len2)
178{
179 len1 = strlen_sum (len1, len2);
180 len1 = strlen_sum (len1, len1 >> 1);
181
182 if (len1 > 4096 - 24)
183 len1 = (len1 | 4095) - 24;
184
185 return SvGROW (sv, len1);
186}
187
140// decode an utf-8 character and return it, or (UV)-1 in 188// decode a utf-8 character and return it, or (UV)-1 in
141// case of an error. 189// case of an error.
142// we special-case "safe" characters from U+80 .. U+7FF, 190// we special-case "safe" characters from U+80 .. U+7FF,
143// but use the very good perl function to parse anything else. 191// but use the very good perl function to parse anything else.
144// note that we never call this function for a ascii codepoints 192// note that we never call this function for a ascii codepoints
145INLINE UV 193INLINE UV
258 } 306 }
259 307
260 // this relies greatly on the quality of the pow () 308 // this relies greatly on the quality of the pow ()
261 // implementation of the platform, but a good 309 // implementation of the platform, but a good
262 // implementation is hard to beat. 310 // implementation is hard to beat.
311 // (IEEE 754 conformant ones are required to be exact)
263 if (postdp) *expo -= eaccum; 312 if (postdp) *expo -= eaccum;
264 *accum += uaccum * Perl_pow (10., *expo); 313 *accum += uaccum * Perl_pow (10., *expo);
265 *expo += eaccum; 314 *expo += eaccum;
266} 315}
267 316
281 // a recursion depth of ten gives us >>500 bits 330 // a recursion depth of ten gives us >>500 bits
282 json_atof_scan1 (s, &accum, &expo, 0, 10); 331 json_atof_scan1 (s, &accum, &expo, 0, 10);
283 332
284 return neg ? -accum : accum; 333 return neg ? -accum : accum;
285} 334}
335
336// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
337static int
338ref_bool_type (SV *sv)
339{
340 svtype svt = SvTYPE (sv);
341
342 if (svt < SVt_PVAV)
343 {
344 STRLEN len = 0;
345 char *pv = svt ? SvPV (sv, len) : 0;
346
347 if (len == 1)
348 if (*pv == '1')
349 return 1;
350 else if (*pv == '0')
351 return 0;
352 }
353
354 return -1;
355}
356
357// returns whether scalar is not a reference in the sense of allow_nonref
358static int
359json_nonref (SV *scalar)
360{
361 if (!SvROK (scalar))
362 return 1;
363
364 scalar = SvRV (scalar);
365
366 if (SvTYPE (scalar) >= SVt_PVMG)
367 {
368 if (SvSTASH (scalar) == bool_stash)
369 return 1;
370
371 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
372 return 1;
373 }
374
375 return 0;
376}
377
286///////////////////////////////////////////////////////////////////////////// 378/////////////////////////////////////////////////////////////////////////////
287// encoder 379// encoder
288 380
289// structure used for encoding JSON 381// structure used for encoding JSON
290typedef struct 382typedef struct
298} enc_t; 390} enc_t;
299 391
300INLINE void 392INLINE void
301need (enc_t *enc, STRLEN len) 393need (enc_t *enc, STRLEN len)
302{ 394{
303 if (expect_false (enc->cur + len >= enc->end)) 395 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
304 { 396 {
305 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 397 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
306 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 398 char *buf = json_sv_grow (enc->sv, cur, len);
307 enc->cur = SvPVX (enc->sv) + cur; 399 enc->cur = buf + cur;
308 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 400 enc->end = buf + SvLEN (enc->sv) - 1;
309 } 401 }
310} 402}
311 403
312INLINE void 404INLINE void
313encode_ch (enc_t *enc, char ch) 405encode_ch (enc_t *enc, char ch)
329 421
330 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 422 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
331 { 423 {
332 if (expect_false (ch == '"')) // but with slow exceptions 424 if (expect_false (ch == '"')) // but with slow exceptions
333 { 425 {
334 need (enc, len += 1); 426 need (enc, len + 1);
335 *enc->cur++ = '\\'; 427 *enc->cur++ = '\\';
336 *enc->cur++ = '"'; 428 *enc->cur++ = '"';
337 } 429 }
338 else if (expect_false (ch == '\\')) 430 else if (expect_false (ch == '\\'))
339 { 431 {
340 need (enc, len += 1); 432 need (enc, len + 1);
341 *enc->cur++ = '\\'; 433 *enc->cur++ = '\\';
342 *enc->cur++ = '\\'; 434 *enc->cur++ = '\\';
343 } 435 }
344 else 436 else
345 *enc->cur++ = ch; 437 *enc->cur++ = ch;
348 } 440 }
349 else 441 else
350 { 442 {
351 switch (ch) 443 switch (ch)
352 { 444 {
353 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 445 case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
354 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; 446 case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
355 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; 447 case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
356 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; 448 case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
357 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; 449 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
358 450
359 default: 451 default:
360 { 452 {
361 STRLEN clen; 453 STRLEN clen;
362 UV uch; 454 UV uch;
378 if (uch >= 0x10000UL) 470 if (uch >= 0x10000UL)
379 { 471 {
380 if (uch >= 0x110000UL) 472 if (uch >= 0x110000UL)
381 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 473 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
382 474
383 need (enc, len += 11); 475 need (enc, len + 11);
384 sprintf (enc->cur, "\\u%04x\\u%04x", 476 sprintf (enc->cur, "\\u%04x\\u%04x",
385 (int)((uch - 0x10000) / 0x400 + 0xD800), 477 (int)((uch - 0x10000) / 0x400 + 0xD800),
386 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 478 (int)((uch - 0x10000) % 0x400 + 0xDC00));
387 enc->cur += 12; 479 enc->cur += 12;
388 } 480 }
389 else 481 else
390 { 482 {
391 need (enc, len += 5); 483 need (enc, len + 5);
392 *enc->cur++ = '\\'; 484 *enc->cur++ = '\\';
393 *enc->cur++ = 'u'; 485 *enc->cur++ = 'u';
394 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 486 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
395 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 487 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
396 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 488 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
404 *enc->cur++ = uch; 496 *enc->cur++ = uch;
405 str += clen; 497 str += clen;
406 } 498 }
407 else if (is_utf8) 499 else if (is_utf8)
408 { 500 {
409 need (enc, len += clen); 501 need (enc, len + clen);
410 do 502 do
411 { 503 {
412 *enc->cur++ = *str++; 504 *enc->cur++ = *str++;
413 } 505 }
414 while (--clen); 506 while (--clen);
415 } 507 }
416 else 508 else
417 { 509 {
418 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 510 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
419 enc->cur = encode_utf8 (enc->cur, uch); 511 enc->cur = encode_utf8 (enc->cur, uch);
420 ++str; 512 ++str;
421 } 513 }
422 } 514 }
423 } 515 }
477 569
478 if (enc->indent >= enc->json.max_depth) 570 if (enc->indent >= enc->json.max_depth)
479 croak (ERR_NESTING_EXCEEDED); 571 croak (ERR_NESTING_EXCEEDED);
480 572
481 encode_ch (enc, '['); 573 encode_ch (enc, '[');
482 574
483 if (len >= 0) 575 if (len >= 0)
484 { 576 {
485 encode_nl (enc); ++enc->indent; 577 encode_nl (enc); ++enc->indent;
486 578
487 for (i = 0; i <= len; ++i) 579 for (i = 0; i <= len; ++i)
499 encode_comma (enc); 591 encode_comma (enc);
500 } 592 }
501 593
502 encode_nl (enc); --enc->indent; encode_indent (enc); 594 encode_nl (enc); --enc->indent; encode_indent (enc);
503 } 595 }
504 596
505 encode_ch (enc, ']'); 597 encode_ch (enc, ']');
506} 598}
507 599
508static void 600static void
509encode_hk (enc_t *enc, HE *he) 601encode_hk (enc_t *enc, HE *he)
513 if (HeKLEN (he) == HEf_SVKEY) 605 if (HeKLEN (he) == HEf_SVKEY)
514 { 606 {
515 SV *sv = HeSVKEY (he); 607 SV *sv = HeSVKEY (he);
516 STRLEN len; 608 STRLEN len;
517 char *str; 609 char *str;
518 610
519 SvGETMAGIC (sv); 611 SvGETMAGIC (sv);
520 str = SvPV (sv, len); 612 str = SvPV (sv, len);
521 613
522 encode_str (enc, str, len, SvUTF8 (sv)); 614 encode_str (enc, str, len, SvUTF8 (sv));
523 } 615 }
589 } 681 }
590 682
591 if (count) 683 if (count)
592 { 684 {
593 int i, fast = 1; 685 int i, fast = 1;
594#if defined(__BORLANDC__) || defined(_MSC_VER) 686 HE *hes_stack [STACK_HES];
595 HE **hes = _alloca (count * sizeof (HE)); 687 HE **hes = hes_stack;
596#else 688
597 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 689 // allocate larger arrays on the heap
598#endif 690 if (count > STACK_HES)
691 {
692 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
693 hes = (HE **)SvPVX (sv);
694 }
599 695
600 i = 0; 696 i = 0;
601 while ((he = hv_iternext (hv))) 697 while ((he = hv_iternext (hv)))
602 { 698 {
603 hes [i++] = he; 699 hes [i++] = he;
672// encode objects, arrays and special \0=false and \1=true values. 768// encode objects, arrays and special \0=false and \1=true values.
673static void 769static void
674encode_rv (enc_t *enc, SV *sv) 770encode_rv (enc_t *enc, SV *sv)
675{ 771{
676 svtype svt; 772 svtype svt;
773 GV *method;
677 774
678 SvGETMAGIC (sv); 775 SvGETMAGIC (sv);
679 svt = SvTYPE (sv); 776 svt = SvTYPE (sv);
680 777
681 if (expect_false (SvOBJECT (sv))) 778 if (expect_false (SvOBJECT (sv)))
682 { 779 {
683 HV *stash = !JSON_SLOW || json_boolean_stash 780 HV *stash = SvSTASH (sv);
684 ? json_boolean_stash
685 : gv_stashpv ("JSON::XS::Boolean", 1);
686 781
687 if (SvSTASH (sv) == stash) 782 if (stash == bool_stash)
688 { 783 {
689 if (SvIV (sv))
690 encode_str (enc, "true", 4, 0); 784 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
691 else
692 encode_str (enc, "false", 5, 0); 785 else encode_str (enc, "false", 5, 0);
693 } 786 }
787 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
788 {
789 int count;
790 dSP;
791
792 ENTER; SAVETMPS;
793 SAVESTACK_POS ();
794 PUSHMARK (SP);
795 EXTEND (SP, 2);
796 // we re-bless the reference to get overload and other niceties right
797 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
798 PUSHs (sv_json);
799
800 PUTBACK;
801 count = call_sv ((SV *)GvCV (method), G_ARRAY);
802 SPAGAIN;
803
804 // catch this surprisingly common error
805 if (SvROK (TOPs) && SvRV (TOPs) == sv)
806 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
807
808 encode_ch (enc, '(');
809 encode_ch (enc, '"');
810 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
811 encode_ch (enc, '"');
812 encode_ch (enc, ')');
813 encode_ch (enc, '[');
814
815 while (count)
816 {
817 encode_sv (enc, SP[1 - count--]);
818
819 if (count)
820 encode_ch (enc, ',');
821 }
822
823 encode_ch (enc, ']');
824
825 FREETMPS; LEAVE;
826 }
827 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
828 {
829 dSP;
830
831 ENTER; SAVETMPS;
832 PUSHMARK (SP);
833 // we re-bless the reference to get overload and other niceties right
834 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
835
836 // calling with G_SCALAR ensures that we always get a 1 return value
837 PUTBACK;
838 call_sv ((SV *)GvCV (method), G_SCALAR);
839 SPAGAIN;
840
841 // catch this surprisingly common error
842 if (SvROK (TOPs) && SvRV (TOPs) == sv)
843 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
844
845 sv = POPs;
846 PUTBACK;
847
848 encode_sv (enc, sv);
849
850 FREETMPS; LEAVE;
851 }
852 else if (enc->json.flags & F_ALLOW_BLESSED)
853 encode_str (enc, "null", 4, 0);
694 else 854 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", 855 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)))); 856 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
741 }
742 } 857 }
743 else if (svt == SVt_PVHV) 858 else if (svt == SVt_PVHV)
744 encode_hv (enc, (HV *)sv); 859 encode_hv (enc, (HV *)sv);
745 else if (svt == SVt_PVAV) 860 else if (svt == SVt_PVAV)
746 encode_av (enc, (AV *)sv); 861 encode_av (enc, (AV *)sv);
747 else if (svt < SVt_PVAV) 862 else if (svt < SVt_PVAV)
748 { 863 {
749 STRLEN len = 0; 864 int bool_type = ref_bool_type (sv);
750 char *pv = svt ? SvPV (sv, len) : 0;
751 865
752 if (len == 1 && *pv == '1') 866 if (bool_type == 1)
753 encode_str (enc, "true", 4, 0); 867 encode_str (enc, "true", 4, 0);
754 else if (len == 1 && *pv == '0') 868 else if (bool_type == 0)
755 encode_str (enc, "false", 5, 0); 869 encode_str (enc, "false", 5, 0);
756 else if (enc->json.flags & F_ALLOW_UNKNOWN) 870 else if (enc->json.flags & F_ALLOW_UNKNOWN)
757 encode_str (enc, "null", 4, 0); 871 encode_str (enc, "null", 4, 0);
758 else 872 else
759 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 873 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
822 } 936 }
823 else 937 else
824 { 938 {
825 // large integer, use the (rather slow) snprintf way. 939 // large integer, use the (rather slow) snprintf way.
826 need (enc, IVUV_MAXCHARS); 940 need (enc, IVUV_MAXCHARS);
827 enc->cur += 941 enc->cur +=
828 SvIsUV(sv) 942 SvIsUV(sv)
829 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) 943 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
830 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 944 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
831 } 945 }
832 } 946 }
833 else if (SvROK (sv)) 947 else if (SvROK (sv))
834 encode_rv (enc, SvRV (sv)); 948 encode_rv (enc, SvRV (sv));
835 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 949 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
836 encode_str (enc, "null", 4, 0); 950 encode_str (enc, "null", 4, 0);
837 else 951 else
838 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 952 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
839 SvPV_nolen (sv), SvFLAGS (sv)); 953 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
840} 954}
841 955
842static SV * 956static SV *
843encode_json (SV *scalar, JSON *json) 957encode_json (SV *scalar, JSON *json)
844{ 958{
845 enc_t enc; 959 enc_t enc;
846 960
847 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 961 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)"); 962 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
849 963
850 enc.json = *json; 964 enc.json = *json;
851 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 965 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
852 enc.cur = SvPVX (enc.sv); 966 enc.cur = SvPVX (enc.sv);
1059 *cur++ = *dec_cur++; 1173 *cur++ = *dec_cur++;
1060 while (--clen); 1174 while (--clen);
1061 1175
1062 utf8 = 1; 1176 utf8 = 1;
1063 } 1177 }
1178 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1179 *cur++ = ch;
1064 else 1180 else
1065 { 1181 {
1066 --dec_cur; 1182 --dec_cur;
1067 1183
1068 if (!ch) 1184 if (!ch)
1078 1194
1079 if (sv) 1195 if (sv)
1080 { 1196 {
1081 STRLEN cur = SvCUR (sv); 1197 STRLEN cur = SvCUR (sv);
1082 1198
1083 if (SvLEN (sv) <= cur + len) 1199 if (SvLEN (sv) - cur <= len)
1084 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 1200 json_sv_grow (sv, cur, len);
1085 1201
1086 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1202 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1087 SvCUR_set (sv, SvCUR (sv) + len); 1203 SvCUR_set (sv, SvCUR (sv) + len);
1088 } 1204 }
1089 else 1205 else
1256 if (*dec->cur == ']') 1372 if (*dec->cur == ']')
1257 { 1373 {
1258 ++dec->cur; 1374 ++dec->cur;
1259 break; 1375 break;
1260 } 1376 }
1261 1377
1262 if (*dec->cur != ',') 1378 if (*dec->cur != ',')
1263 ERR (", or ] expected while parsing array"); 1379 ERR (", or ] expected while parsing array");
1264 1380
1265 ++dec->cur; 1381 ++dec->cur;
1266 1382
1390 1506
1391 hv_iterinit (hv); 1507 hv_iterinit (hv);
1392 he = hv_iternext (hv); 1508 he = hv_iternext (hv);
1393 hv_iterinit (hv); 1509 hv_iterinit (hv);
1394 1510
1395 // the next line creates a mortal sv each time its called. 1511 // the next line creates a mortal sv each time it's called.
1396 // might want to optimise this for common cases. 1512 // might want to optimise this for common cases.
1397 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1513 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1398 1514
1399 if (cb) 1515 if (cb)
1400 { 1516 {
1401 dSP; 1517 dSP;
1402 int count; 1518 int count;
1403 1519
1404 ENTER; SAVETMPS; PUSHMARK (SP); 1520 ENTER; SAVETMPS;
1521 SAVESTACK_POS ();
1522 PUSHMARK (SP);
1405 XPUSHs (HeVAL (he)); 1523 XPUSHs (HeVAL (he));
1524 sv_2mortal (sv);
1406 1525
1407 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1526 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1408 1527
1409 if (count == 1) 1528 if (count == 1)
1410 { 1529 {
1411 sv = newSVsv (POPs); 1530 sv = newSVsv (POPs);
1412 FREETMPS; LEAVE; 1531 FREETMPS; LEAVE;
1413 return sv; 1532 return sv;
1414 } 1533 }
1415 1534
1535 SvREFCNT_inc (sv);
1416 FREETMPS; LEAVE; 1536 FREETMPS; LEAVE;
1417 } 1537 }
1418 } 1538 }
1419 1539
1420 if (dec->json.cb_object) 1540 if (dec->json.cb_object)
1421 { 1541 {
1422 dSP; 1542 dSP;
1423 int count; 1543 int count;
1424 1544
1425 ENTER; SAVETMPS; PUSHMARK (SP); 1545 ENTER; SAVETMPS;
1546 SAVESTACK_POS ();
1547 PUSHMARK (SP);
1426 XPUSHs (sv_2mortal (sv)); 1548 XPUSHs (sv_2mortal (sv));
1427 1549
1428 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1550 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1429 1551
1430 if (count == 1) 1552 if (count == 1)
1446 DEC_DEC_DEPTH; 1568 DEC_DEC_DEPTH;
1447 return 0; 1569 return 0;
1448} 1570}
1449 1571
1450static SV * 1572static SV *
1573decode_tag (dec_t *dec)
1574{
1575 SV *tag = 0;
1576 SV *val = 0;
1577
1578 if (!(dec->json.flags & F_ALLOW_TAGS))
1579 ERR ("malformed JSON string, neither array, object, number, string or atom");
1580
1581 ++dec->cur;
1582
1583 decode_ws (dec);
1584
1585 tag = decode_sv (dec);
1586 if (!tag)
1587 goto fail;
1588
1589 if (!SvPOK (tag))
1590 ERR ("malformed JSON string, (tag) must be a string");
1591
1592 decode_ws (dec);
1593
1594 if (*dec->cur != ')')
1595 ERR (") expected after tag");
1596
1597 ++dec->cur;
1598
1599 decode_ws (dec);
1600
1601 val = decode_sv (dec);
1602 if (!val)
1603 goto fail;
1604
1605 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1606 ERR ("malformed JSON string, tag value must be an array");
1607
1608 {
1609 AV *av = (AV *)SvRV (val);
1610 int i, len = av_len (av) + 1;
1611 HV *stash = gv_stashsv (tag, 0);
1612 SV *sv;
1613
1614 if (!stash)
1615 ERR ("cannot decode perl-object (package does not exist)");
1616
1617 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1618
1619 if (!method)
1620 ERR ("cannot decode perl-object (package does not have a THAW method)");
1621
1622 dSP;
1623
1624 ENTER; SAVETMPS;
1625 PUSHMARK (SP);
1626 EXTEND (SP, len + 2);
1627 // we re-bless the reference to get overload and other niceties right
1628 PUSHs (tag);
1629 PUSHs (sv_json);
1630
1631 for (i = 0; i < len; ++i)
1632 PUSHs (*av_fetch (av, i, 1));
1633
1634 PUTBACK;
1635 call_sv ((SV *)GvCV (method), G_SCALAR);
1636 SPAGAIN;
1637
1638 SvREFCNT_dec (tag);
1639 SvREFCNT_dec (val);
1640 sv = SvREFCNT_inc (POPs);
1641
1642 PUTBACK;
1643
1644 FREETMPS; LEAVE;
1645
1646 return sv;
1647 }
1648
1649fail:
1650 SvREFCNT_dec (tag);
1651 SvREFCNT_dec (val);
1652 return 0;
1653}
1654
1655static SV *
1451decode_sv (dec_t *dec) 1656decode_sv (dec_t *dec)
1452{ 1657{
1453 // the beauty of JSON: you need exactly one character lookahead 1658 // the beauty of JSON: you need exactly one character lookahead
1454 // to parse everything. 1659 // to parse everything.
1455 switch (*dec->cur) 1660 switch (*dec->cur)
1456 { 1661 {
1457 case '"': ++dec->cur; return decode_str (dec); 1662 case '"': ++dec->cur; return decode_str (dec);
1458 case '[': ++dec->cur; return decode_av (dec); 1663 case '[': ++dec->cur; return decode_av (dec);
1459 case '{': ++dec->cur; return decode_hv (dec); 1664 case '{': ++dec->cur; return decode_hv (dec);
1665 case '(': return decode_tag (dec);
1460 1666
1461 case '-': 1667 case '-':
1462 case '0': case '1': case '2': case '3': case '4': 1668 case '0': case '1': case '2': case '3': case '4':
1463 case '5': case '6': case '7': case '8': case '9': 1669 case '5': case '6': case '7': case '8': case '9':
1464 return decode_num (dec); 1670 return decode_num (dec);
1465 1671
1672 case 'f':
1673 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1674 {
1675 dec->cur += 5;
1676
1677 if (expect_false (!dec->json.v_false))
1678 dec->json.v_false = GET_BOOL (false);
1679
1680 return newSVsv (dec->json.v_false);
1681 }
1682 else
1683 ERR ("'false' expected");
1684
1685 break;
1686
1466 case 't': 1687 case 't':
1467 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1688 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1468 { 1689 {
1469 dec->cur += 4; 1690 dec->cur += 4;
1470#if JSON_SLOW 1691
1471 json_true = get_bool ("JSON::XS::true"); 1692 if (expect_false (!dec->json.v_true))
1472#endif 1693 dec->json.v_true = GET_BOOL (true);
1694
1473 return newSVsv (json_true); 1695 return newSVsv (dec->json.v_true);
1474 } 1696 }
1475 else 1697 else
1476 ERR ("'true' expected"); 1698 ERR ("'true' expected");
1477
1478 break;
1479
1480 case 'f':
1481 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1482 {
1483 dec->cur += 5;
1484#if JSON_SLOW
1485 json_false = get_bool ("JSON::XS::false");
1486#endif
1487 return newSVsv (json_false);
1488 }
1489 else
1490 ERR ("'false' expected");
1491 1699
1492 break; 1700 break;
1493 1701
1494 case 'n': 1702 case 'n':
1495 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1703 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
1501 ERR ("'null' expected"); 1709 ERR ("'null' expected");
1502 1710
1503 break; 1711 break;
1504 1712
1505 default: 1713 default:
1506 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1714 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1507 break; 1715 break;
1508 } 1716 }
1509 1717
1510fail: 1718fail:
1511 return 0; 1719 return 0;
1512} 1720}
1513 1721
1514static SV * 1722static SV *
1515decode_json (SV *string, JSON *json, char **offset_return) 1723decode_json (SV *string, JSON *json, STRLEN *offset_return)
1516{ 1724{
1517 dec_t dec; 1725 dec_t dec;
1518 SV *sv; 1726 SV *sv;
1519 1727
1520 /* work around bugs in 5.10 where manipulating magic values 1728 /* work around bugs in 5.10 where manipulating magic values
1521 * will perl ignore the magic in subsequent accesses 1729 * makes perl ignore the magic in subsequent accesses.
1730 * also make a copy of non-PV values, to get them into a clean
1731 * state (SvPV should do that, but it's buggy, see below).
1732 *
1733 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1734 * as reported by Reini Urban.
1522 */ 1735 */
1523 /*SvGETMAGIC (string);*/ 1736 /*SvGETMAGIC (string);*/
1524 if (SvMAGICAL (string)) 1737 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1525 string = sv_2mortal (newSVsv (string)); 1738 string = sv_2mortal (newSVsv (string));
1526 1739
1527 SvUPGRADE (string, SVt_PV); 1740 SvUPGRADE (string, SVt_PV);
1528 1741
1529 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1742 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1546 if (offset > json->max_size && json->max_size) 1759 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", 1760 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); 1761 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1549 } 1762 }
1550 1763
1551 if (json->flags & F_UTF8) 1764 if (DECODE_WANTS_OCTETS (json))
1552 sv_utf8_downgrade (string, 0); 1765 sv_utf8_downgrade (string, 0);
1553 else 1766 else
1554 sv_utf8_upgrade (string); 1767 sv_utf8_upgrade (string);
1555 1768
1556 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1769 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1568 1781
1569 decode_ws (&dec); 1782 decode_ws (&dec);
1570 sv = decode_sv (&dec); 1783 sv = decode_sv (&dec);
1571 1784
1572 if (offset_return) 1785 if (offset_return)
1573 *offset_return = dec.cur; 1786 *offset_return = dec.cur - SvPVX (string);
1574 1787 else if (sv)
1575 if (!(offset_return || !sv))
1576 { 1788 {
1577 // check for trailing garbage 1789 // check for trailing garbage
1578 decode_ws (&dec); 1790 decode_ws (&dec);
1579 1791
1580 if (*dec.cur) 1792 if (*dec.cur)
1598 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1810 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1599 LEAVE; 1811 LEAVE;
1600 1812
1601 croak ("%s, at character offset %d (before \"%s\")", 1813 croak ("%s, at character offset %d (before \"%s\")",
1602 dec.err, 1814 dec.err,
1603 ptr_to_index (string, dec.cur), 1815 (int)ptr_to_index (string, dec.cur),
1604 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1816 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1605 } 1817 }
1606 1818
1607 sv = sv_2mortal (sv); 1819 sv = sv_2mortal (sv);
1608 1820
1609 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1821 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)"); 1822 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1611 1823
1612 return sv; 1824 return sv;
1613} 1825}
1614 1826
1736 self->incr_mode = INCR_M_STR; 1948 self->incr_mode = INCR_M_STR;
1737 goto incr_m_str; 1949 goto incr_m_str;
1738 1950
1739 case '[': 1951 case '[':
1740 case '{': 1952 case '{':
1953 case '(':
1741 if (++self->incr_nest > self->max_depth) 1954 if (++self->incr_nest > self->max_depth)
1742 croak (ERR_NESTING_EXCEEDED); 1955 croak (ERR_NESTING_EXCEEDED);
1743 break; 1956 break;
1744 1957
1745 case ']': 1958 case ']':
1746 case '}': 1959 case '}':
1747 if (--self->incr_nest <= 0) 1960 if (--self->incr_nest <= 0)
1748 goto interrupt; 1961 goto interrupt;
1962 break;
1963
1964 case ')':
1965 --self->incr_nest;
1749 break; 1966 break;
1750 1967
1751 case '#': 1968 case '#':
1752 self->incr_mode = INCR_M_C1; 1969 self->incr_mode = INCR_M_C1;
1753 goto incr_m_c; 1970 goto incr_m_c;
1779 i >= '0' && i <= '9' ? i - '0' 1996 i >= '0' && i <= '9' ? i - '0'
1780 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1997 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1781 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1998 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1782 : -1; 1999 : -1;
1783 2000
1784 json_stash = gv_stashpv ("JSON::XS" , 1); 2001 json_stash = gv_stashpv ("JSON::XS" , 1);
1785 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 2002 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1786
1787 json_true = get_bool ("JSON::XS::true");
1788 json_false = get_bool ("JSON::XS::false"); 2003 bool_false = get_bool ("Types::Serialiser::false");
2004 bool_true = get_bool ("Types::Serialiser::true");
2005
2006 sv_json = newSVpv ("JSON", 0);
2007 SvREADONLY_on (sv_json);
1789 2008
1790 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ 2009 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1791} 2010}
1792 2011
1793PROTOTYPES: DISABLE 2012PROTOTYPES: DISABLE
1794 2013
1795void CLONE (...) 2014void CLONE (...)
1796 CODE: 2015 CODE:
2016 // as long as these writes are atomic, the race should not matter
2017 // as existing threads either already use 0, or use the old value,
2018 // which is sitll correct for the initial thread.
1797 json_stash = 0; 2019 json_stash = 0;
1798 json_boolean_stash = 0; 2020 bool_stash = 0;
2021 bool_false = 0;
2022 bool_true = 0;
1799 2023
1800void new (char *klass) 2024void new (char *klass)
1801 PPCODE: 2025 PPCODE:
1802{ 2026{
1803 SV *pv = NEWSV (0, sizeof (JSON)); 2027 SV *pv = NEWSV (0, sizeof (JSON));
1804 SvPOK_only (pv); 2028 SvPOK_only (pv);
1805 json_init ((JSON *)SvPVX (pv)); 2029 json_init ((JSON *)SvPVX (pv));
1806 XPUSHs (sv_2mortal (sv_bless ( 2030 XPUSHs (sv_2mortal (sv_bless (
1807 newRV_noinc (pv), 2031 newRV_noinc (pv),
1808 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2032 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1809 ))); 2033 )));
1810} 2034}
2035
2036void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2037 PPCODE:
2038 self->v_false = newSVsv (v_false);
2039 self->v_true = newSVsv (v_true);
2040 XPUSHs (ST (0));
2041
2042void get_boolean_values (JSON *self)
2043 PPCODE:
2044 if (self->v_false && self->v_true)
2045 {
2046 EXTEND (SP, 2);
2047 PUSHs (self->v_false);
2048 PUSHs (self->v_true);
2049 }
1811 2050
1812void ascii (JSON *self, int enable = 1) 2051void ascii (JSON *self, int enable = 1)
1813 ALIAS: 2052 ALIAS:
1814 ascii = F_ASCII 2053 ascii = F_ASCII
1815 latin1 = F_LATIN1 2054 latin1 = F_LATIN1
1823 shrink = F_SHRINK 2062 shrink = F_SHRINK
1824 allow_blessed = F_ALLOW_BLESSED 2063 allow_blessed = F_ALLOW_BLESSED
1825 convert_blessed = F_CONV_BLESSED 2064 convert_blessed = F_CONV_BLESSED
1826 relaxed = F_RELAXED 2065 relaxed = F_RELAXED
1827 allow_unknown = F_ALLOW_UNKNOWN 2066 allow_unknown = F_ALLOW_UNKNOWN
2067 allow_tags = F_ALLOW_TAGS
1828 PPCODE: 2068 PPCODE:
1829{ 2069{
1830 if (enable) 2070 if (enable)
1831 self->flags |= ix; 2071 self->flags |= ix;
1832 else 2072 else
1848 get_shrink = F_SHRINK 2088 get_shrink = F_SHRINK
1849 get_allow_blessed = F_ALLOW_BLESSED 2089 get_allow_blessed = F_ALLOW_BLESSED
1850 get_convert_blessed = F_CONV_BLESSED 2090 get_convert_blessed = F_CONV_BLESSED
1851 get_relaxed = F_RELAXED 2091 get_relaxed = F_RELAXED
1852 get_allow_unknown = F_ALLOW_UNKNOWN 2092 get_allow_unknown = F_ALLOW_UNKNOWN
2093 get_allow_tags = F_ALLOW_TAGS
1853 PPCODE: 2094 PPCODE:
1854 XPUSHs (boolSV (self->flags & ix)); 2095 XPUSHs (boolSV (self->flags & ix));
1855 2096
1856void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 2097void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1857 PPCODE: 2098 PPCODE:
1885} 2126}
1886 2127
1887void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2128void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1888 PPCODE: 2129 PPCODE:
1889{ 2130{
1890 if (!self->cb_sk_object) 2131 if (!self->cb_sk_object)
1891 self->cb_sk_object = newHV (); 2132 self->cb_sk_object = newHV ();
1892 2133
1893 if (SvOK (cb)) 2134 if (SvOK (cb))
1894 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2135 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1895 else 2136 else
1906 XPUSHs (ST (0)); 2147 XPUSHs (ST (0));
1907} 2148}
1908 2149
1909void encode (JSON *self, SV *scalar) 2150void encode (JSON *self, SV *scalar)
1910 PPCODE: 2151 PPCODE:
1911 XPUSHs (encode_json (scalar, self)); 2152 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2153 XPUSHs (scalar);
1912 2154
1913void decode (JSON *self, SV *jsonstr) 2155void decode (JSON *self, SV *jsonstr)
1914 PPCODE: 2156 PPCODE:
1915 XPUSHs (decode_json (jsonstr, self, 0)); 2157 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2158 XPUSHs (jsonstr);
1916 2159
1917void decode_prefix (JSON *self, SV *jsonstr) 2160void decode_prefix (JSON *self, SV *jsonstr)
1918 PPCODE: 2161 PPCODE:
1919{ 2162{
1920 char *offset; 2163 SV *sv;
2164 STRLEN offset;
2165 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1921 EXTEND (SP, 2); 2166 EXTEND (SP, 2);
1922 PUSHs (decode_json (jsonstr, self, &offset)); 2167 PUSHs (sv);
1923 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2168 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
1924} 2169}
1925 2170
1926void incr_parse (JSON *self, SV *jsonstr = 0) 2171void incr_parse (JSON *self, SV *jsonstr = 0)
1927 PPCODE: 2172 PPCODE:
1928{ 2173{
1929 if (!self->incr_text) 2174 if (!self->incr_text)
1930 self->incr_text = newSVpvn ("", 0); 2175 self->incr_text = newSVpvn ("", 0);
2176
2177 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2178 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2179 if (DECODE_WANTS_OCTETS (self))
2180 {
2181 if (self->incr_pos)
2182 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2183 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2184
2185 sv_utf8_downgrade (self->incr_text, 0);
2186 }
2187 else
2188 {
2189 sv_utf8_upgrade (self->incr_text);
2190
2191 if (self->incr_pos)
2192 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2193 - (U8 *)SvPVX (self->incr_text);
2194 }
1931 2195
1932 // append data, if any 2196 // append data, if any
1933 if (jsonstr) 2197 if (jsonstr)
1934 { 2198 {
2199 /* make sure both strings have same encoding */
2200 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1935 if (SvUTF8 (jsonstr)) 2201 if (SvUTF8 (jsonstr))
2202 sv_utf8_downgrade (jsonstr, 0);
1936 { 2203 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); 2204 sv_utf8_upgrade (jsonstr);
1949 2205
2206 /* and then just blindly append */
1950 { 2207 {
1951 STRLEN len; 2208 STRLEN len;
1952 const char *str = SvPV (jsonstr, len); 2209 const char *str = SvPV (jsonstr, len);
1953 STRLEN cur = SvCUR (self->incr_text); 2210 STRLEN cur = SvCUR (self->incr_text);
1954 2211
1955 if (SvLEN (self->incr_text) <= cur + len) 2212 if (SvLEN (self->incr_text) - cur <= len)
1956 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2213 json_sv_grow (self->incr_text, cur, len);
1957 2214
1958 Move (str, SvEND (self->incr_text), len, char); 2215 Move (str, SvEND (self->incr_text), len, char);
1959 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2216 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1960 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 2217 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1961 } 2218 }
1962 } 2219 }
1963 2220
1964 if (GIMME_V != G_VOID) 2221 if (GIMME_V != G_VOID)
1965 do 2222 do
1966 { 2223 {
2224 SV *sv;
1967 char *offset; 2225 STRLEN offset;
1968 2226
1969 if (!INCR_DONE (self)) 2227 if (!INCR_DONE (self))
1970 { 2228 {
1971 incr_parse (self); 2229 incr_parse (self);
1972 2230
1973 if (self->incr_pos > self->max_size && self->max_size) 2231 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", 2232 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); 2233 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1976 2234
1977 if (!INCR_DONE (self)) 2235 if (!INCR_DONE (self))
2236 {
2237 // as an optimisation, do not accumulate white space in the incr buffer
2238 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2239 {
2240 self->incr_pos = 0;
2241 SvCUR_set (self->incr_text, 0);
2242 }
2243
1978 break; 2244 break;
2245 }
1979 } 2246 }
1980 2247
1981 XPUSHs (decode_json (self->incr_text, self, &offset)); 2248 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2249 XPUSHs (sv);
1982 2250
1983 self->incr_pos -= offset - SvPVX (self->incr_text); 2251 self->incr_pos -= offset;
1984 self->incr_nest = 0; 2252 self->incr_nest = 0;
1985 self->incr_mode = 0; 2253 self->incr_mode = 0;
1986 2254
1987 sv_chop (self->incr_text, offset); 2255 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
1988 } 2256 }
1989 while (GIMME_V == G_ARRAY); 2257 while (GIMME_V == G_ARRAY);
1990} 2258}
1991 2259
1992SV *incr_text (JSON *self) 2260SV *incr_text (JSON *self)
2023 self->incr_mode = 0; 2291 self->incr_mode = 0;
2024} 2292}
2025 2293
2026void DESTROY (JSON *self) 2294void DESTROY (JSON *self)
2027 CODE: 2295 CODE:
2296 SvREFCNT_dec (self->v_false);
2297 SvREFCNT_dec (self->v_true);
2028 SvREFCNT_dec (self->cb_sk_object); 2298 SvREFCNT_dec (self->cb_sk_object);
2029 SvREFCNT_dec (self->cb_object); 2299 SvREFCNT_dec (self->cb_object);
2030 SvREFCNT_dec (self->incr_text); 2300 SvREFCNT_dec (self->incr_text);
2031 2301
2032PROTOTYPES: ENABLE 2302PROTOTYPES: ENABLE
2033 2303
2034void encode_json (SV *scalar) 2304void encode_json (SV *scalar)
2035 ALIAS:
2036 to_json_ = 0
2037 encode_json = F_UTF8
2038 PPCODE: 2305 PPCODE:
2039{ 2306{
2040 JSON json; 2307 JSON json;
2041 json_init (&json); 2308 json_init (&json);
2042 json.flags |= ix; 2309 json.flags |= F_UTF8;
2043 XPUSHs (encode_json (scalar, &json)); 2310 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2311 XPUSHs (scalar);
2044} 2312}
2045 2313
2046void decode_json (SV *jsonstr) 2314void decode_json (SV *jsonstr)
2047 ALIAS:
2048 from_json_ = 0
2049 decode_json = F_UTF8
2050 PPCODE: 2315 PPCODE:
2051{ 2316{
2052 JSON json; 2317 JSON json;
2053 json_init (&json); 2318 json_init (&json);
2054 json.flags |= ix; 2319 json.flags |= F_UTF8;
2055 XPUSHs (decode_json (jsonstr, &json, 0)); 2320 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2321 XPUSHs (jsonstr);
2056} 2322}
2057 2323

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines