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 1.135 by root, Thu Nov 15 20:49:12 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 64 // 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{
107 Zero (json, 1, JSON); 130 static const JSON init = { F_ALLOW_NONREF, 512 };
108 json->max_depth = 512; 131
132 *json = init;
109} 133}
110 134
111///////////////////////////////////////////////////////////////////////////// 135/////////////////////////////////////////////////////////////////////////////
112// utility functions 136// utility functions
113 137
135 SvPV_renew (sv, SvCUR (sv) + 1); 159 SvPV_renew (sv, SvCUR (sv) + 1);
136#endif 160#endif
137 } 161 }
138} 162}
139 163
164/* adds two STRLENs together, slow, and with paranoia */
165STRLEN
166strlen_sum (STRLEN l1, STRLEN l2)
167{
168 size_t sum = l1 + l2;
169
170 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
171 croak ("JSON::XS: string size overflow");
172
173 return sum;
174}
175
176/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
177static char *
178json_sv_grow (SV *sv, size_t len1, size_t len2)
179{
180 len1 = strlen_sum (len1, len2);
181 len1 = strlen_sum (len1, len1 >> 1);
182
183 if (len1 > 4096 - 24)
184 len1 = (len1 | 4095) - 24;
185
186 return SvGROW (sv, len1);
187}
188
140// decode an utf-8 character and return it, or (UV)-1 in 189// decode a utf-8 character and return it, or (UV)-1 in
141// case of an error. 190// case of an error.
142// we special-case "safe" characters from U+80 .. U+7FF, 191// we special-case "safe" characters from U+80 .. U+7FF,
143// but use the very good perl function to parse anything else. 192// but use the very good perl function to parse anything else.
144// note that we never call this function for a ascii codepoints 193// note that we never call this function for a ascii codepoints
145INLINE UV 194INLINE UV
258 } 307 }
259 308
260 // this relies greatly on the quality of the pow () 309 // this relies greatly on the quality of the pow ()
261 // implementation of the platform, but a good 310 // implementation of the platform, but a good
262 // implementation is hard to beat. 311 // implementation is hard to beat.
312 // (IEEE 754 conformant ones are required to be exact)
263 if (postdp) *expo -= eaccum; 313 if (postdp) *expo -= eaccum;
264 *accum += uaccum * Perl_pow (10., *expo); 314 *accum += uaccum * Perl_pow (10., *expo);
265 *expo += eaccum; 315 *expo += eaccum;
266} 316}
267 317
281 // a recursion depth of ten gives us >>500 bits 331 // a recursion depth of ten gives us >>500 bits
282 json_atof_scan1 (s, &accum, &expo, 0, 10); 332 json_atof_scan1 (s, &accum, &expo, 0, 10);
283 333
284 return neg ? -accum : accum; 334 return neg ? -accum : accum;
285} 335}
336
337// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
338static int
339ref_bool_type (SV *sv)
340{
341 svtype svt = SvTYPE (sv);
342
343 if (svt < SVt_PVAV)
344 {
345 STRLEN len = 0;
346 char *pv = svt ? SvPV (sv, len) : 0;
347
348 if (len == 1)
349 if (*pv == '1')
350 return 1;
351 else if (*pv == '0')
352 return 0;
353 }
354
355 return -1;
356}
357
358// returns whether scalar is not a reference in the sense of allow_nonref
359static int
360json_nonref (SV *scalar)
361{
362 if (!SvROK (scalar))
363 return 1;
364
365 scalar = SvRV (scalar);
366
367 if (SvTYPE (scalar) >= SVt_PVMG)
368 {
369 if (SvSTASH (scalar) == bool_stash)
370 return 1;
371
372 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
373 return 1;
374 }
375
376 return 0;
377}
378
286///////////////////////////////////////////////////////////////////////////// 379/////////////////////////////////////////////////////////////////////////////
287// encoder 380// encoder
288 381
289// structure used for encoding JSON 382// structure used for encoding JSON
290typedef struct 383typedef struct
298} enc_t; 391} enc_t;
299 392
300INLINE void 393INLINE void
301need (enc_t *enc, STRLEN len) 394need (enc_t *enc, STRLEN len)
302{ 395{
303 if (expect_false (enc->cur + len >= enc->end)) 396 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
304 { 397 {
305 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 398 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
306 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 399 char *buf = json_sv_grow (enc->sv, cur, len);
307 enc->cur = SvPVX (enc->sv) + cur; 400 enc->cur = buf + cur;
308 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 401 enc->end = buf + SvLEN (enc->sv) - 1;
309 } 402 }
310} 403}
311 404
312INLINE void 405INLINE void
313encode_ch (enc_t *enc, char ch) 406encode_ch (enc_t *enc, char ch)
329 422
330 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 423 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
331 { 424 {
332 if (expect_false (ch == '"')) // but with slow exceptions 425 if (expect_false (ch == '"')) // but with slow exceptions
333 { 426 {
334 need (enc, len += 1); 427 need (enc, len + 1);
335 *enc->cur++ = '\\'; 428 *enc->cur++ = '\\';
336 *enc->cur++ = '"'; 429 *enc->cur++ = '"';
337 } 430 }
338 else if (expect_false (ch == '\\')) 431 else if (expect_false (ch == '\\'))
339 { 432 {
340 need (enc, len += 1); 433 need (enc, len + 1);
341 *enc->cur++ = '\\'; 434 *enc->cur++ = '\\';
342 *enc->cur++ = '\\'; 435 *enc->cur++ = '\\';
343 } 436 }
344 else 437 else
345 *enc->cur++ = ch; 438 *enc->cur++ = ch;
348 } 441 }
349 else 442 else
350 { 443 {
351 switch (ch) 444 switch (ch)
352 { 445 {
353 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 446 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; 447 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; 448 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; 449 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; 450 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
358 451
359 default: 452 default:
360 { 453 {
361 STRLEN clen; 454 STRLEN clen;
362 UV uch; 455 UV uch;
378 if (uch >= 0x10000UL) 471 if (uch >= 0x10000UL)
379 { 472 {
380 if (uch >= 0x110000UL) 473 if (uch >= 0x110000UL)
381 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 474 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
382 475
383 need (enc, len += 11); 476 need (enc, len + 11);
384 sprintf (enc->cur, "\\u%04x\\u%04x", 477 sprintf (enc->cur, "\\u%04x\\u%04x",
385 (int)((uch - 0x10000) / 0x400 + 0xD800), 478 (int)((uch - 0x10000) / 0x400 + 0xD800),
386 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 479 (int)((uch - 0x10000) % 0x400 + 0xDC00));
387 enc->cur += 12; 480 enc->cur += 12;
388 } 481 }
389 else 482 else
390 { 483 {
391 need (enc, len += 5); 484 need (enc, len + 5);
392 *enc->cur++ = '\\'; 485 *enc->cur++ = '\\';
393 *enc->cur++ = 'u'; 486 *enc->cur++ = 'u';
394 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 487 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
395 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 488 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
396 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 489 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
404 *enc->cur++ = uch; 497 *enc->cur++ = uch;
405 str += clen; 498 str += clen;
406 } 499 }
407 else if (is_utf8) 500 else if (is_utf8)
408 { 501 {
409 need (enc, len += clen); 502 need (enc, len + clen);
410 do 503 do
411 { 504 {
412 *enc->cur++ = *str++; 505 *enc->cur++ = *str++;
413 } 506 }
414 while (--clen); 507 while (--clen);
415 } 508 }
416 else 509 else
417 { 510 {
418 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 511 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
419 enc->cur = encode_utf8 (enc->cur, uch); 512 enc->cur = encode_utf8 (enc->cur, uch);
420 ++str; 513 ++str;
421 } 514 }
422 } 515 }
423 } 516 }
477 570
478 if (enc->indent >= enc->json.max_depth) 571 if (enc->indent >= enc->json.max_depth)
479 croak (ERR_NESTING_EXCEEDED); 572 croak (ERR_NESTING_EXCEEDED);
480 573
481 encode_ch (enc, '['); 574 encode_ch (enc, '[');
482 575
483 if (len >= 0) 576 if (len >= 0)
484 { 577 {
485 encode_nl (enc); ++enc->indent; 578 encode_nl (enc); ++enc->indent;
486 579
487 for (i = 0; i <= len; ++i) 580 for (i = 0; i <= len; ++i)
499 encode_comma (enc); 592 encode_comma (enc);
500 } 593 }
501 594
502 encode_nl (enc); --enc->indent; encode_indent (enc); 595 encode_nl (enc); --enc->indent; encode_indent (enc);
503 } 596 }
504 597
505 encode_ch (enc, ']'); 598 encode_ch (enc, ']');
506} 599}
507 600
508static void 601static void
509encode_hk (enc_t *enc, HE *he) 602encode_hk (enc_t *enc, HE *he)
513 if (HeKLEN (he) == HEf_SVKEY) 606 if (HeKLEN (he) == HEf_SVKEY)
514 { 607 {
515 SV *sv = HeSVKEY (he); 608 SV *sv = HeSVKEY (he);
516 STRLEN len; 609 STRLEN len;
517 char *str; 610 char *str;
518 611
519 SvGETMAGIC (sv); 612 SvGETMAGIC (sv);
520 str = SvPV (sv, len); 613 str = SvPV (sv, len);
521 614
522 encode_str (enc, str, len, SvUTF8 (sv)); 615 encode_str (enc, str, len, SvUTF8 (sv));
523 } 616 }
589 } 682 }
590 683
591 if (count) 684 if (count)
592 { 685 {
593 int i, fast = 1; 686 int i, fast = 1;
594#if defined(__BORLANDC__) || defined(_MSC_VER) 687 HE *hes_stack [STACK_HES];
595 HE **hes = _alloca (count * sizeof (HE)); 688 HE **hes = hes_stack;
596#else 689
597 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 690 // allocate larger arrays on the heap
598#endif 691 if (count > STACK_HES)
692 {
693 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
694 hes = (HE **)SvPVX (sv);
695 }
599 696
600 i = 0; 697 i = 0;
601 while ((he = hv_iternext (hv))) 698 while ((he = hv_iternext (hv)))
602 { 699 {
603 hes [i++] = he; 700 hes [i++] = he;
672// encode objects, arrays and special \0=false and \1=true values. 769// encode objects, arrays and special \0=false and \1=true values.
673static void 770static void
674encode_rv (enc_t *enc, SV *sv) 771encode_rv (enc_t *enc, SV *sv)
675{ 772{
676 svtype svt; 773 svtype svt;
774 GV *method;
677 775
678 SvGETMAGIC (sv); 776 SvGETMAGIC (sv);
679 svt = SvTYPE (sv); 777 svt = SvTYPE (sv);
680 778
681 if (expect_false (SvOBJECT (sv))) 779 if (expect_false (SvOBJECT (sv)))
682 { 780 {
683 HV *stash = !JSON_SLOW || json_boolean_stash 781 HV *stash = SvSTASH (sv);
684 ? json_boolean_stash
685 : gv_stashpv ("JSON::XS::Boolean", 1);
686 782
687 if (SvSTASH (sv) == stash) 783 if (stash == bool_stash)
688 { 784 {
689 if (SvIV (sv))
690 encode_str (enc, "true", 4, 0); 785 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
691 else
692 encode_str (enc, "false", 5, 0); 786 else encode_str (enc, "false", 5, 0);
693 } 787 }
788 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
789 {
790 int count;
791 dSP;
792
793 ENTER; SAVETMPS;
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 if (count)
816 {
817 int i;
818
819 for (i = 0; i < count - 1; ++i)
820 {
821 encode_sv (enc, SP[i + 1 - count]);
822 encode_ch (enc, ',');
823 }
824
825 encode_sv (enc, TOPs);
826 SP -= count;
827 }
828
829 encode_ch (enc, ']');
830
831 FREETMPS; LEAVE;
832 }
833 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
834 {
835 dSP;
836
837 ENTER; SAVETMPS;
838 PUSHMARK (SP);
839 // we re-bless the reference to get overload and other niceties right
840 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
841
842 // calling with G_SCALAR ensures that we always get a 1 return value
843 PUTBACK;
844 call_sv ((SV *)GvCV (method), G_SCALAR);
845 SPAGAIN;
846
847 // catch this surprisingly common error
848 if (SvROK (TOPs) && SvRV (TOPs) == sv)
849 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
850
851 sv = POPs;
852 PUTBACK;
853
854 encode_sv (enc, sv);
855
856 FREETMPS; LEAVE;
857 }
858 else if (enc->json.flags & F_ALLOW_BLESSED)
859 encode_str (enc, "null", 4, 0);
694 else 860 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", 861 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)))); 862 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
741 }
742 } 863 }
743 else if (svt == SVt_PVHV) 864 else if (svt == SVt_PVHV)
744 encode_hv (enc, (HV *)sv); 865 encode_hv (enc, (HV *)sv);
745 else if (svt == SVt_PVAV) 866 else if (svt == SVt_PVAV)
746 encode_av (enc, (AV *)sv); 867 encode_av (enc, (AV *)sv);
747 else if (svt < SVt_PVAV) 868 else if (svt < SVt_PVAV)
748 { 869 {
749 STRLEN len = 0; 870 int bool_type = ref_bool_type (sv);
750 char *pv = svt ? SvPV (sv, len) : 0;
751 871
752 if (len == 1 && *pv == '1') 872 if (bool_type == 1)
753 encode_str (enc, "true", 4, 0); 873 encode_str (enc, "true", 4, 0);
754 else if (len == 1 && *pv == '0') 874 else if (bool_type == 0)
755 encode_str (enc, "false", 5, 0); 875 encode_str (enc, "false", 5, 0);
756 else if (enc->json.flags & F_ALLOW_UNKNOWN) 876 else if (enc->json.flags & F_ALLOW_UNKNOWN)
757 encode_str (enc, "null", 4, 0); 877 encode_str (enc, "null", 4, 0);
758 else 878 else
759 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 879 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
822 } 942 }
823 else 943 else
824 { 944 {
825 // large integer, use the (rather slow) snprintf way. 945 // large integer, use the (rather slow) snprintf way.
826 need (enc, IVUV_MAXCHARS); 946 need (enc, IVUV_MAXCHARS);
827 enc->cur += 947 enc->cur +=
828 SvIsUV(sv) 948 SvIsUV(sv)
829 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) 949 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
830 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 950 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
831 } 951 }
832 } 952 }
833 else if (SvROK (sv)) 953 else if (SvROK (sv))
834 encode_rv (enc, SvRV (sv)); 954 encode_rv (enc, SvRV (sv));
835 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 955 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
836 encode_str (enc, "null", 4, 0); 956 encode_str (enc, "null", 4, 0);
837 else 957 else
838 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 958 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
839 SvPV_nolen (sv), SvFLAGS (sv)); 959 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
840} 960}
841 961
842static SV * 962static SV *
843encode_json (SV *scalar, JSON *json) 963encode_json (SV *scalar, JSON *json)
844{ 964{
845 enc_t enc; 965 enc_t enc;
846 966
847 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 967 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)"); 968 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
849 969
850 enc.json = *json; 970 enc.json = *json;
851 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 971 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
852 enc.cur = SvPVX (enc.sv); 972 enc.cur = SvPVX (enc.sv);
1059 *cur++ = *dec_cur++; 1179 *cur++ = *dec_cur++;
1060 while (--clen); 1180 while (--clen);
1061 1181
1062 utf8 = 1; 1182 utf8 = 1;
1063 } 1183 }
1184 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1185 *cur++ = ch;
1064 else 1186 else
1065 { 1187 {
1066 --dec_cur; 1188 --dec_cur;
1067 1189
1068 if (!ch) 1190 if (!ch)
1078 1200
1079 if (sv) 1201 if (sv)
1080 { 1202 {
1081 STRLEN cur = SvCUR (sv); 1203 STRLEN cur = SvCUR (sv);
1082 1204
1083 if (SvLEN (sv) <= cur + len) 1205 if (SvLEN (sv) - cur <= len)
1084 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 1206 json_sv_grow (sv, cur, len);
1085 1207
1086 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1208 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1087 SvCUR_set (sv, SvCUR (sv) + len); 1209 SvCUR_set (sv, SvCUR (sv) + len);
1088 } 1210 }
1089 else 1211 else
1256 if (*dec->cur == ']') 1378 if (*dec->cur == ']')
1257 { 1379 {
1258 ++dec->cur; 1380 ++dec->cur;
1259 break; 1381 break;
1260 } 1382 }
1261 1383
1262 if (*dec->cur != ',') 1384 if (*dec->cur != ',')
1263 ERR (", or ] expected while parsing array"); 1385 ERR (", or ] expected while parsing array");
1264 1386
1265 ++dec->cur; 1387 ++dec->cur;
1266 1388
1390 1512
1391 hv_iterinit (hv); 1513 hv_iterinit (hv);
1392 he = hv_iternext (hv); 1514 he = hv_iternext (hv);
1393 hv_iterinit (hv); 1515 hv_iterinit (hv);
1394 1516
1395 // the next line creates a mortal sv each time its called. 1517 // the next line creates a mortal sv each time it's called.
1396 // might want to optimise this for common cases. 1518 // might want to optimise this for common cases.
1397 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1519 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1398 1520
1399 if (cb) 1521 if (cb)
1400 { 1522 {
1401 dSP; 1523 dSP;
1402 int count; 1524 int count;
1403 1525
1404 ENTER; SAVETMPS; PUSHMARK (SP); 1526 ENTER; SAVETMPS;
1527 PUSHMARK (SP);
1405 XPUSHs (HeVAL (he)); 1528 XPUSHs (HeVAL (he));
1529 sv_2mortal (sv);
1406 1530
1407 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1531 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1408 1532
1409 if (count == 1) 1533 if (count == 1)
1410 { 1534 {
1411 sv = newSVsv (POPs); 1535 sv = newSVsv (POPs);
1412 FREETMPS; LEAVE; 1536 FREETMPS; LEAVE;
1413 return sv; 1537 return sv;
1414 } 1538 }
1539 else if (count)
1540 croak ("filter_json_single_key_object callbacks must not return more than one scalar");
1415 1541
1542 SvREFCNT_inc (sv);
1416 FREETMPS; LEAVE; 1543 FREETMPS; LEAVE;
1417 } 1544 }
1418 } 1545 }
1419 1546
1420 if (dec->json.cb_object) 1547 if (dec->json.cb_object)
1421 { 1548 {
1422 dSP; 1549 dSP;
1423 int count; 1550 int count;
1424 1551
1425 ENTER; SAVETMPS; PUSHMARK (SP); 1552 ENTER; SAVETMPS;
1553 PUSHMARK (SP);
1426 XPUSHs (sv_2mortal (sv)); 1554 XPUSHs (sv_2mortal (sv));
1427 1555
1428 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1556 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1429 1557
1430 if (count == 1) 1558 if (count == 1)
1431 { 1559 {
1432 sv = newSVsv (POPs); 1560 sv = newSVsv (POPs);
1433 FREETMPS; LEAVE; 1561 FREETMPS; LEAVE;
1434 return sv; 1562 return sv;
1435 } 1563 }
1564 else if (count)
1565 croak ("filter_json_object callbacks must not return more than one scalar");
1436 1566
1437 SvREFCNT_inc (sv); 1567 SvREFCNT_inc (sv);
1438 FREETMPS; LEAVE; 1568 FREETMPS; LEAVE;
1439 } 1569 }
1440 } 1570 }
1446 DEC_DEC_DEPTH; 1576 DEC_DEC_DEPTH;
1447 return 0; 1577 return 0;
1448} 1578}
1449 1579
1450static SV * 1580static SV *
1581decode_tag (dec_t *dec)
1582{
1583 SV *tag = 0;
1584 SV *val = 0;
1585
1586 if (!(dec->json.flags & F_ALLOW_TAGS))
1587 ERR ("malformed JSON string, neither array, object, number, string or atom");
1588
1589 ++dec->cur;
1590
1591 decode_ws (dec);
1592
1593 tag = decode_sv (dec);
1594 if (!tag)
1595 goto fail;
1596
1597 if (!SvPOK (tag))
1598 ERR ("malformed JSON string, (tag) must be a string");
1599
1600 decode_ws (dec);
1601
1602 if (*dec->cur != ')')
1603 ERR (") expected after tag");
1604
1605 ++dec->cur;
1606
1607 decode_ws (dec);
1608
1609 val = decode_sv (dec);
1610 if (!val)
1611 goto fail;
1612
1613 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1614 ERR ("malformed JSON string, tag value must be an array");
1615
1616 {
1617 AV *av = (AV *)SvRV (val);
1618 int i, len = av_len (av) + 1;
1619 HV *stash = gv_stashsv (tag, 0);
1620 SV *sv;
1621
1622 if (!stash)
1623 ERR ("cannot decode perl-object (package does not exist)");
1624
1625 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1626
1627 if (!method)
1628 ERR ("cannot decode perl-object (package does not have a THAW method)");
1629
1630 dSP;
1631
1632 ENTER; SAVETMPS;
1633 PUSHMARK (SP);
1634 EXTEND (SP, len + 2);
1635 // we re-bless the reference to get overload and other niceties right
1636 PUSHs (tag);
1637 PUSHs (sv_json);
1638
1639 for (i = 0; i < len; ++i)
1640 PUSHs (*av_fetch (av, i, 1));
1641
1642 PUTBACK;
1643 call_sv ((SV *)GvCV (method), G_SCALAR);
1644 SPAGAIN;
1645
1646 SvREFCNT_dec (tag);
1647 SvREFCNT_dec (val);
1648 sv = SvREFCNT_inc (POPs);
1649
1650 PUTBACK;
1651
1652 FREETMPS; LEAVE;
1653
1654 return sv;
1655 }
1656
1657fail:
1658 SvREFCNT_dec (tag);
1659 SvREFCNT_dec (val);
1660 return 0;
1661}
1662
1663static SV *
1451decode_sv (dec_t *dec) 1664decode_sv (dec_t *dec)
1452{ 1665{
1453 // the beauty of JSON: you need exactly one character lookahead 1666 // the beauty of JSON: you need exactly one character lookahead
1454 // to parse everything. 1667 // to parse everything.
1455 switch (*dec->cur) 1668 switch (*dec->cur)
1456 { 1669 {
1457 case '"': ++dec->cur; return decode_str (dec); 1670 case '"': ++dec->cur; return decode_str (dec);
1458 case '[': ++dec->cur; return decode_av (dec); 1671 case '[': ++dec->cur; return decode_av (dec);
1459 case '{': ++dec->cur; return decode_hv (dec); 1672 case '{': ++dec->cur; return decode_hv (dec);
1673 case '(': return decode_tag (dec);
1460 1674
1461 case '-': 1675 case '-':
1462 case '0': case '1': case '2': case '3': case '4': 1676 case '0': case '1': case '2': case '3': case '4':
1463 case '5': case '6': case '7': case '8': case '9': 1677 case '5': case '6': case '7': case '8': case '9':
1464 return decode_num (dec); 1678 return decode_num (dec);
1465 1679
1680 case 'f':
1681 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1682 {
1683 dec->cur += 5;
1684
1685 if (expect_false (!dec->json.v_false))
1686 dec->json.v_false = GET_BOOL (false);
1687
1688 return newSVsv (dec->json.v_false);
1689 }
1690 else
1691 ERR ("'false' expected");
1692
1693 break;
1694
1466 case 't': 1695 case 't':
1467 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1696 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1468 { 1697 {
1469 dec->cur += 4; 1698 dec->cur += 4;
1470#if JSON_SLOW 1699
1471 json_true = get_bool ("JSON::XS::true"); 1700 if (expect_false (!dec->json.v_true))
1472#endif 1701 dec->json.v_true = GET_BOOL (true);
1702
1473 return newSVsv (json_true); 1703 return newSVsv (dec->json.v_true);
1474 } 1704 }
1475 else 1705 else
1476 ERR ("'true' expected"); 1706 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 1707
1492 break; 1708 break;
1493 1709
1494 case 'n': 1710 case 'n':
1495 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1711 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
1501 ERR ("'null' expected"); 1717 ERR ("'null' expected");
1502 1718
1503 break; 1719 break;
1504 1720
1505 default: 1721 default:
1506 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1722 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1507 break; 1723 break;
1508 } 1724 }
1509 1725
1510fail: 1726fail:
1511 return 0; 1727 return 0;
1512} 1728}
1513 1729
1514static SV * 1730static SV *
1515decode_json (SV *string, JSON *json, char **offset_return) 1731decode_json (SV *string, JSON *json, STRLEN *offset_return)
1516{ 1732{
1517 dec_t dec; 1733 dec_t dec;
1518 SV *sv; 1734 SV *sv;
1519 1735
1520 /* work around bugs in 5.10 where manipulating magic values 1736 /* work around bugs in 5.10 where manipulating magic values
1521 * will perl ignore the magic in subsequent accesses 1737 * makes perl ignore the magic in subsequent accesses.
1738 * also make a copy of non-PV values, to get them into a clean
1739 * state (SvPV should do that, but it's buggy, see below).
1740 *
1741 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1742 * as reported by Reini Urban.
1522 */ 1743 */
1523 /*SvGETMAGIC (string);*/ 1744 /*SvGETMAGIC (string);*/
1524 if (SvMAGICAL (string)) 1745 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1525 string = sv_2mortal (newSVsv (string)); 1746 string = sv_2mortal (newSVsv (string));
1526 1747
1527 SvUPGRADE (string, SVt_PV); 1748 SvUPGRADE (string, SVt_PV);
1528 1749
1529 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1750 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1546 if (offset > json->max_size && json->max_size) 1767 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", 1768 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); 1769 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1549 } 1770 }
1550 1771
1551 if (json->flags & F_UTF8) 1772 if (DECODE_WANTS_OCTETS (json))
1552 sv_utf8_downgrade (string, 0); 1773 sv_utf8_downgrade (string, 0);
1553 else 1774 else
1554 sv_utf8_upgrade (string); 1775 sv_utf8_upgrade (string);
1555 1776
1556 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1777 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1568 1789
1569 decode_ws (&dec); 1790 decode_ws (&dec);
1570 sv = decode_sv (&dec); 1791 sv = decode_sv (&dec);
1571 1792
1572 if (offset_return) 1793 if (offset_return)
1573 *offset_return = dec.cur; 1794 *offset_return = dec.cur - SvPVX (string);
1574 1795 else if (sv)
1575 if (!(offset_return || !sv))
1576 { 1796 {
1577 // check for trailing garbage 1797 // check for trailing garbage
1578 decode_ws (&dec); 1798 decode_ws (&dec);
1579 1799
1580 if (*dec.cur) 1800 if (dec.cur != dec.end)
1581 { 1801 {
1582 dec.err = "garbage after JSON object"; 1802 dec.err = "garbage after JSON object";
1583 SvREFCNT_dec (sv); 1803 SvREFCNT_dec (sv);
1584 sv = 0; 1804 sv = 0;
1585 } 1805 }
1598 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1818 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1599 LEAVE; 1819 LEAVE;
1600 1820
1601 croak ("%s, at character offset %d (before \"%s\")", 1821 croak ("%s, at character offset %d (before \"%s\")",
1602 dec.err, 1822 dec.err,
1603 ptr_to_index (string, dec.cur), 1823 (int)ptr_to_index (string, dec.cur),
1604 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1824 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1605 } 1825 }
1606 1826
1607 sv = sv_2mortal (sv); 1827 sv = sv_2mortal (sv);
1608 1828
1609 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1829 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)"); 1830 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1611 1831
1612 return sv; 1832 return sv;
1613} 1833}
1614 1834
1736 self->incr_mode = INCR_M_STR; 1956 self->incr_mode = INCR_M_STR;
1737 goto incr_m_str; 1957 goto incr_m_str;
1738 1958
1739 case '[': 1959 case '[':
1740 case '{': 1960 case '{':
1961 case '(':
1741 if (++self->incr_nest > self->max_depth) 1962 if (++self->incr_nest > self->max_depth)
1742 croak (ERR_NESTING_EXCEEDED); 1963 croak (ERR_NESTING_EXCEEDED);
1743 break; 1964 break;
1744 1965
1745 case ']': 1966 case ']':
1746 case '}': 1967 case '}':
1747 if (--self->incr_nest <= 0) 1968 if (--self->incr_nest <= 0)
1748 goto interrupt; 1969 goto interrupt;
1970 break;
1971
1972 case ')':
1973 --self->incr_nest;
1749 break; 1974 break;
1750 1975
1751 case '#': 1976 case '#':
1752 self->incr_mode = INCR_M_C1; 1977 self->incr_mode = INCR_M_C1;
1753 goto incr_m_c; 1978 goto incr_m_c;
1779 i >= '0' && i <= '9' ? i - '0' 2004 i >= '0' && i <= '9' ? i - '0'
1780 : i >= 'a' && i <= 'f' ? i - 'a' + 10 2005 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1781 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2006 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1782 : -1; 2007 : -1;
1783 2008
1784 json_stash = gv_stashpv ("JSON::XS" , 1); 2009 json_stash = gv_stashpv ("JSON::XS" , 1);
1785 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 2010 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"); 2011 bool_false = get_bool ("Types::Serialiser::false");
2012 bool_true = get_bool ("Types::Serialiser::true");
2013
2014 sv_json = newSVpv ("JSON", 0);
2015 SvREADONLY_on (sv_json);
1789 2016
1790 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ 2017 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1791} 2018}
1792 2019
1793PROTOTYPES: DISABLE 2020PROTOTYPES: DISABLE
1794 2021
1795void CLONE (...) 2022void CLONE (...)
1796 CODE: 2023 CODE:
2024 // as long as these writes are atomic, the race should not matter
2025 // as existing threads either already use 0, or use the old value,
2026 // which is sitll correct for the initial thread.
1797 json_stash = 0; 2027 json_stash = 0;
1798 json_boolean_stash = 0; 2028 bool_stash = 0;
2029 bool_false = 0;
2030 bool_true = 0;
1799 2031
1800void new (char *klass) 2032void new (char *klass)
1801 PPCODE: 2033 PPCODE:
1802{ 2034{
1803 SV *pv = NEWSV (0, sizeof (JSON)); 2035 SV *pv = NEWSV (0, sizeof (JSON));
1804 SvPOK_only (pv); 2036 SvPOK_only (pv);
1805 json_init ((JSON *)SvPVX (pv)); 2037 json_init ((JSON *)SvPVX (pv));
1806 XPUSHs (sv_2mortal (sv_bless ( 2038 XPUSHs (sv_2mortal (sv_bless (
1807 newRV_noinc (pv), 2039 newRV_noinc (pv),
1808 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2040 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1809 ))); 2041 )));
1810} 2042}
2043
2044void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2045 PPCODE:
2046 self->v_false = newSVsv (v_false);
2047 self->v_true = newSVsv (v_true);
2048 XPUSHs (ST (0));
2049
2050void get_boolean_values (JSON *self)
2051 PPCODE:
2052 if (self->v_false && self->v_true)
2053 {
2054 EXTEND (SP, 2);
2055 PUSHs (self->v_false);
2056 PUSHs (self->v_true);
2057 }
1811 2058
1812void ascii (JSON *self, int enable = 1) 2059void ascii (JSON *self, int enable = 1)
1813 ALIAS: 2060 ALIAS:
1814 ascii = F_ASCII 2061 ascii = F_ASCII
1815 latin1 = F_LATIN1 2062 latin1 = F_LATIN1
1823 shrink = F_SHRINK 2070 shrink = F_SHRINK
1824 allow_blessed = F_ALLOW_BLESSED 2071 allow_blessed = F_ALLOW_BLESSED
1825 convert_blessed = F_CONV_BLESSED 2072 convert_blessed = F_CONV_BLESSED
1826 relaxed = F_RELAXED 2073 relaxed = F_RELAXED
1827 allow_unknown = F_ALLOW_UNKNOWN 2074 allow_unknown = F_ALLOW_UNKNOWN
2075 allow_tags = F_ALLOW_TAGS
1828 PPCODE: 2076 PPCODE:
1829{ 2077{
1830 if (enable) 2078 if (enable)
1831 self->flags |= ix; 2079 self->flags |= ix;
1832 else 2080 else
1848 get_shrink = F_SHRINK 2096 get_shrink = F_SHRINK
1849 get_allow_blessed = F_ALLOW_BLESSED 2097 get_allow_blessed = F_ALLOW_BLESSED
1850 get_convert_blessed = F_CONV_BLESSED 2098 get_convert_blessed = F_CONV_BLESSED
1851 get_relaxed = F_RELAXED 2099 get_relaxed = F_RELAXED
1852 get_allow_unknown = F_ALLOW_UNKNOWN 2100 get_allow_unknown = F_ALLOW_UNKNOWN
2101 get_allow_tags = F_ALLOW_TAGS
1853 PPCODE: 2102 PPCODE:
1854 XPUSHs (boolSV (self->flags & ix)); 2103 XPUSHs (boolSV (self->flags & ix));
1855 2104
1856void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 2105void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1857 PPCODE: 2106 PPCODE:
1885} 2134}
1886 2135
1887void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2136void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1888 PPCODE: 2137 PPCODE:
1889{ 2138{
1890 if (!self->cb_sk_object) 2139 if (!self->cb_sk_object)
1891 self->cb_sk_object = newHV (); 2140 self->cb_sk_object = newHV ();
1892 2141
1893 if (SvOK (cb)) 2142 if (SvOK (cb))
1894 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2143 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1895 else 2144 else
1906 XPUSHs (ST (0)); 2155 XPUSHs (ST (0));
1907} 2156}
1908 2157
1909void encode (JSON *self, SV *scalar) 2158void encode (JSON *self, SV *scalar)
1910 PPCODE: 2159 PPCODE:
1911 XPUSHs (encode_json (scalar, self)); 2160 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2161 XPUSHs (scalar);
1912 2162
1913void decode (JSON *self, SV *jsonstr) 2163void decode (JSON *self, SV *jsonstr)
1914 PPCODE: 2164 PPCODE:
1915 XPUSHs (decode_json (jsonstr, self, 0)); 2165 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2166 XPUSHs (jsonstr);
1916 2167
1917void decode_prefix (JSON *self, SV *jsonstr) 2168void decode_prefix (JSON *self, SV *jsonstr)
1918 PPCODE: 2169 PPCODE:
1919{ 2170{
1920 char *offset; 2171 SV *sv;
2172 STRLEN offset;
2173 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1921 EXTEND (SP, 2); 2174 EXTEND (SP, 2);
1922 PUSHs (decode_json (jsonstr, self, &offset)); 2175 PUSHs (sv);
1923 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2176 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
1924} 2177}
1925 2178
1926void incr_parse (JSON *self, SV *jsonstr = 0) 2179void incr_parse (JSON *self, SV *jsonstr = 0)
1927 PPCODE: 2180 PPCODE:
1928{ 2181{
1929 if (!self->incr_text) 2182 if (!self->incr_text)
1930 self->incr_text = newSVpvn ("", 0); 2183 self->incr_text = newSVpvn ("", 0);
2184
2185 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2186 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2187 if (DECODE_WANTS_OCTETS (self))
2188 {
2189 if (self->incr_pos)
2190 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2191 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2192
2193 sv_utf8_downgrade (self->incr_text, 0);
2194 }
2195 else
2196 {
2197 sv_utf8_upgrade (self->incr_text);
2198
2199 if (self->incr_pos)
2200 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2201 - (U8 *)SvPVX (self->incr_text);
2202 }
1931 2203
1932 // append data, if any 2204 // append data, if any
1933 if (jsonstr) 2205 if (jsonstr)
1934 { 2206 {
2207 /* make sure both strings have same encoding */
2208 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1935 if (SvUTF8 (jsonstr)) 2209 if (SvUTF8 (jsonstr))
2210 sv_utf8_downgrade (jsonstr, 0);
1936 { 2211 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); 2212 sv_utf8_upgrade (jsonstr);
1949 2213
2214 /* and then just blindly append */
1950 { 2215 {
1951 STRLEN len; 2216 STRLEN len;
1952 const char *str = SvPV (jsonstr, len); 2217 const char *str = SvPV (jsonstr, len);
1953 STRLEN cur = SvCUR (self->incr_text); 2218 STRLEN cur = SvCUR (self->incr_text);
1954 2219
1955 if (SvLEN (self->incr_text) <= cur + len) 2220 if (SvLEN (self->incr_text) - cur <= len)
1956 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2221 json_sv_grow (self->incr_text, cur, len);
1957 2222
1958 Move (str, SvEND (self->incr_text), len, char); 2223 Move (str, SvEND (self->incr_text), len, char);
1959 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2224 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 2225 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1961 } 2226 }
1962 } 2227 }
1963 2228
1964 if (GIMME_V != G_VOID) 2229 if (GIMME_V != G_VOID)
1965 do 2230 do
1966 { 2231 {
2232 SV *sv;
1967 char *offset; 2233 STRLEN offset;
1968 2234
1969 if (!INCR_DONE (self)) 2235 if (!INCR_DONE (self))
1970 { 2236 {
1971 incr_parse (self); 2237 incr_parse (self);
1972 2238
1973 if (self->incr_pos > self->max_size && self->max_size) 2239 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", 2240 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); 2241 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1976 2242
1977 if (!INCR_DONE (self)) 2243 if (!INCR_DONE (self))
2244 {
2245 // as an optimisation, do not accumulate white space in the incr buffer
2246 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2247 {
2248 self->incr_pos = 0;
2249 SvCUR_set (self->incr_text, 0);
2250 }
2251
1978 break; 2252 break;
2253 }
1979 } 2254 }
1980 2255
1981 XPUSHs (decode_json (self->incr_text, self, &offset)); 2256 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2257 XPUSHs (sv);
1982 2258
1983 self->incr_pos -= offset - SvPVX (self->incr_text); 2259 self->incr_pos -= offset;
1984 self->incr_nest = 0; 2260 self->incr_nest = 0;
1985 self->incr_mode = 0; 2261 self->incr_mode = 0;
1986 2262
1987 sv_chop (self->incr_text, offset); 2263 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
1988 } 2264 }
1989 while (GIMME_V == G_ARRAY); 2265 while (GIMME_V == G_ARRAY);
1990} 2266}
1991 2267
1992SV *incr_text (JSON *self) 2268SV *incr_text (JSON *self)
2023 self->incr_mode = 0; 2299 self->incr_mode = 0;
2024} 2300}
2025 2301
2026void DESTROY (JSON *self) 2302void DESTROY (JSON *self)
2027 CODE: 2303 CODE:
2304 SvREFCNT_dec (self->v_false);
2305 SvREFCNT_dec (self->v_true);
2028 SvREFCNT_dec (self->cb_sk_object); 2306 SvREFCNT_dec (self->cb_sk_object);
2029 SvREFCNT_dec (self->cb_object); 2307 SvREFCNT_dec (self->cb_object);
2030 SvREFCNT_dec (self->incr_text); 2308 SvREFCNT_dec (self->incr_text);
2031 2309
2032PROTOTYPES: ENABLE 2310PROTOTYPES: ENABLE
2033 2311
2034void encode_json (SV *scalar) 2312void encode_json (SV *scalar)
2035 ALIAS:
2036 to_json_ = 0
2037 encode_json = F_UTF8
2038 PPCODE: 2313 PPCODE:
2039{ 2314{
2040 JSON json; 2315 JSON json;
2041 json_init (&json); 2316 json_init (&json);
2042 json.flags |= ix; 2317 json.flags |= F_UTF8;
2043 XPUSHs (encode_json (scalar, &json)); 2318 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2319 XPUSHs (scalar);
2044} 2320}
2045 2321
2046void decode_json (SV *jsonstr) 2322void decode_json (SV *jsonstr)
2047 ALIAS:
2048 from_json_ = 0
2049 decode_json = F_UTF8
2050 PPCODE: 2323 PPCODE:
2051{ 2324{
2052 JSON json; 2325 JSON json;
2053 json_init (&json); 2326 json_init (&json);
2054 json.flags |= ix; 2327 json.flags |= F_UTF8;
2055 XPUSHs (decode_json (jsonstr, &json, 0)); 2328 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2329 XPUSHs (jsonstr);
2056} 2330}
2057 2331

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines