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 1.107 by root, Wed Mar 17 01:45:43 2010 UTC vs.
Revision 1.136 by root, Thu Nov 15 22:35:35 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
101 INCR_M_TFN, // inside true/false/null
102 INCR_M_NUM, // inside number
80 INCR_M_STR, // inside string 103 INCR_M_STR, // inside string
81 INCR_M_BS, // inside backslash 104 INCR_M_BS, // inside backslash
82 INCR_M_C0, // inside comment in initial whitespace sequence 105 INCR_M_C0, // inside comment in initial whitespace sequence
83 INCR_M_C1, // inside comment in other places 106 INCR_M_C1, // inside comment in other places
84 INCR_M_JSON // outside anything, count nesting 107 INCR_M_JSON // outside anything, count nesting
97 // for the incremental parser 120 // for the incremental parser
98 SV *incr_text; // the source text so far 121 SV *incr_text; // the source text so far
99 STRLEN incr_pos; // the current offset into the text 122 STRLEN incr_pos; // the current offset into the text
100 int incr_nest; // {[]}-nesting level 123 int incr_nest; // {[]}-nesting level
101 unsigned char incr_mode; 124 unsigned char incr_mode;
125
126 SV *v_false, *v_true;
102} JSON; 127} JSON;
103 128
104INLINE void 129INLINE void
105json_init (JSON *json) 130json_init (JSON *json)
106{ 131{
107 Zero (json, 1, JSON); 132 static const JSON init = { F_ALLOW_NONREF, 512 };
108 json->max_depth = 512; 133
134 *json = init;
109} 135}
110 136
111///////////////////////////////////////////////////////////////////////////// 137/////////////////////////////////////////////////////////////////////////////
112// utility functions 138// utility functions
113 139
135 SvPV_renew (sv, SvCUR (sv) + 1); 161 SvPV_renew (sv, SvCUR (sv) + 1);
136#endif 162#endif
137 } 163 }
138} 164}
139 165
166/* adds two STRLENs together, slow, and with paranoia */
167STRLEN
168strlen_sum (STRLEN l1, STRLEN l2)
169{
170 size_t sum = l1 + l2;
171
172 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
173 croak ("JSON::XS: string size overflow");
174
175 return sum;
176}
177
178/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
179static char *
180json_sv_grow (SV *sv, size_t len1, size_t len2)
181{
182 len1 = strlen_sum (len1, len2);
183 len1 = strlen_sum (len1, len1 >> 1);
184
185 if (len1 > 4096 - 24)
186 len1 = (len1 | 4095) - 24;
187
188 return SvGROW (sv, len1);
189}
190
140// decode an utf-8 character and return it, or (UV)-1 in 191// decode a utf-8 character and return it, or (UV)-1 in
141// case of an error. 192// case of an error.
142// we special-case "safe" characters from U+80 .. U+7FF, 193// we special-case "safe" characters from U+80 .. U+7FF,
143// but use the very good perl function to parse anything else. 194// but use the very good perl function to parse anything else.
144// note that we never call this function for a ascii codepoints 195// note that we never call this function for a ascii codepoints
145INLINE UV 196INLINE UV
258 } 309 }
259 310
260 // this relies greatly on the quality of the pow () 311 // this relies greatly on the quality of the pow ()
261 // implementation of the platform, but a good 312 // implementation of the platform, but a good
262 // implementation is hard to beat. 313 // implementation is hard to beat.
314 // (IEEE 754 conformant ones are required to be exact)
263 if (postdp) *expo -= eaccum; 315 if (postdp) *expo -= eaccum;
264 *accum += uaccum * Perl_pow (10., *expo); 316 *accum += uaccum * Perl_pow (10., *expo);
265 *expo += eaccum; 317 *expo += eaccum;
266} 318}
267 319
281 // a recursion depth of ten gives us >>500 bits 333 // a recursion depth of ten gives us >>500 bits
282 json_atof_scan1 (s, &accum, &expo, 0, 10); 334 json_atof_scan1 (s, &accum, &expo, 0, 10);
283 335
284 return neg ? -accum : accum; 336 return neg ? -accum : accum;
285} 337}
338
339// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
340static int
341ref_bool_type (SV *sv)
342{
343 svtype svt = SvTYPE (sv);
344
345 if (svt < SVt_PVAV)
346 {
347 STRLEN len = 0;
348 char *pv = svt ? SvPV (sv, len) : 0;
349
350 if (len == 1)
351 if (*pv == '1')
352 return 1;
353 else if (*pv == '0')
354 return 0;
355 }
356
357 return -1;
358}
359
360// returns whether scalar is not a reference in the sense of allow_nonref
361static int
362json_nonref (SV *scalar)
363{
364 if (!SvROK (scalar))
365 return 1;
366
367 scalar = SvRV (scalar);
368
369 if (SvTYPE (scalar) >= SVt_PVMG)
370 {
371 if (SvSTASH (scalar) == bool_stash)
372 return 1;
373
374 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
375 return 1;
376 }
377
378 return 0;
379}
380
286///////////////////////////////////////////////////////////////////////////// 381/////////////////////////////////////////////////////////////////////////////
287// encoder 382// encoder
288 383
289// structure used for encoding JSON 384// structure used for encoding JSON
290typedef struct 385typedef struct
298} enc_t; 393} enc_t;
299 394
300INLINE void 395INLINE void
301need (enc_t *enc, STRLEN len) 396need (enc_t *enc, STRLEN len)
302{ 397{
303 if (expect_false (enc->cur + len >= enc->end)) 398 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
304 { 399 {
305 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 400 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
306 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 401 char *buf = json_sv_grow (enc->sv, cur, len);
307 enc->cur = SvPVX (enc->sv) + cur; 402 enc->cur = buf + cur;
308 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 403 enc->end = buf + SvLEN (enc->sv) - 1;
309 } 404 }
310} 405}
311 406
312INLINE void 407INLINE void
313encode_ch (enc_t *enc, char ch) 408encode_ch (enc_t *enc, char ch)
329 424
330 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 425 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
331 { 426 {
332 if (expect_false (ch == '"')) // but with slow exceptions 427 if (expect_false (ch == '"')) // but with slow exceptions
333 { 428 {
334 need (enc, len += 1); 429 need (enc, len + 1);
335 *enc->cur++ = '\\'; 430 *enc->cur++ = '\\';
336 *enc->cur++ = '"'; 431 *enc->cur++ = '"';
337 } 432 }
338 else if (expect_false (ch == '\\')) 433 else if (expect_false (ch == '\\'))
339 { 434 {
340 need (enc, len += 1); 435 need (enc, len + 1);
341 *enc->cur++ = '\\'; 436 *enc->cur++ = '\\';
342 *enc->cur++ = '\\'; 437 *enc->cur++ = '\\';
343 } 438 }
344 else 439 else
345 *enc->cur++ = ch; 440 *enc->cur++ = ch;
348 } 443 }
349 else 444 else
350 { 445 {
351 switch (ch) 446 switch (ch)
352 { 447 {
353 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 448 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; 449 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; 450 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; 451 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; 452 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
358 453
359 default: 454 default:
360 { 455 {
361 STRLEN clen; 456 STRLEN clen;
362 UV uch; 457 UV uch;
378 if (uch >= 0x10000UL) 473 if (uch >= 0x10000UL)
379 { 474 {
380 if (uch >= 0x110000UL) 475 if (uch >= 0x110000UL)
381 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 476 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
382 477
383 need (enc, len += 11); 478 need (enc, len + 11);
384 sprintf (enc->cur, "\\u%04x\\u%04x", 479 sprintf (enc->cur, "\\u%04x\\u%04x",
385 (int)((uch - 0x10000) / 0x400 + 0xD800), 480 (int)((uch - 0x10000) / 0x400 + 0xD800),
386 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 481 (int)((uch - 0x10000) % 0x400 + 0xDC00));
387 enc->cur += 12; 482 enc->cur += 12;
388 } 483 }
389 else 484 else
390 { 485 {
391 need (enc, len += 5); 486 need (enc, len + 5);
392 *enc->cur++ = '\\'; 487 *enc->cur++ = '\\';
393 *enc->cur++ = 'u'; 488 *enc->cur++ = 'u';
394 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 489 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
395 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 490 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
396 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 491 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
404 *enc->cur++ = uch; 499 *enc->cur++ = uch;
405 str += clen; 500 str += clen;
406 } 501 }
407 else if (is_utf8) 502 else if (is_utf8)
408 { 503 {
409 need (enc, len += clen); 504 need (enc, len + clen);
410 do 505 do
411 { 506 {
412 *enc->cur++ = *str++; 507 *enc->cur++ = *str++;
413 } 508 }
414 while (--clen); 509 while (--clen);
415 } 510 }
416 else 511 else
417 { 512 {
418 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 513 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
419 enc->cur = encode_utf8 (enc->cur, uch); 514 enc->cur = encode_utf8 (enc->cur, uch);
420 ++str; 515 ++str;
421 } 516 }
422 } 517 }
423 } 518 }
477 572
478 if (enc->indent >= enc->json.max_depth) 573 if (enc->indent >= enc->json.max_depth)
479 croak (ERR_NESTING_EXCEEDED); 574 croak (ERR_NESTING_EXCEEDED);
480 575
481 encode_ch (enc, '['); 576 encode_ch (enc, '[');
482 577
483 if (len >= 0) 578 if (len >= 0)
484 { 579 {
485 encode_nl (enc); ++enc->indent; 580 encode_nl (enc); ++enc->indent;
486 581
487 for (i = 0; i <= len; ++i) 582 for (i = 0; i <= len; ++i)
499 encode_comma (enc); 594 encode_comma (enc);
500 } 595 }
501 596
502 encode_nl (enc); --enc->indent; encode_indent (enc); 597 encode_nl (enc); --enc->indent; encode_indent (enc);
503 } 598 }
504 599
505 encode_ch (enc, ']'); 600 encode_ch (enc, ']');
506} 601}
507 602
508static void 603static void
509encode_hk (enc_t *enc, HE *he) 604encode_hk (enc_t *enc, HE *he)
513 if (HeKLEN (he) == HEf_SVKEY) 608 if (HeKLEN (he) == HEf_SVKEY)
514 { 609 {
515 SV *sv = HeSVKEY (he); 610 SV *sv = HeSVKEY (he);
516 STRLEN len; 611 STRLEN len;
517 char *str; 612 char *str;
518 613
519 SvGETMAGIC (sv); 614 SvGETMAGIC (sv);
520 str = SvPV (sv, len); 615 str = SvPV (sv, len);
521 616
522 encode_str (enc, str, len, SvUTF8 (sv)); 617 encode_str (enc, str, len, SvUTF8 (sv));
523 } 618 }
589 } 684 }
590 685
591 if (count) 686 if (count)
592 { 687 {
593 int i, fast = 1; 688 int i, fast = 1;
594#if defined(__BORLANDC__) || defined(_MSC_VER) 689 HE *hes_stack [STACK_HES];
595 HE **hes = _alloca (count * sizeof (HE)); 690 HE **hes = hes_stack;
596#else 691
597 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 692 // allocate larger arrays on the heap
598#endif 693 if (count > STACK_HES)
694 {
695 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
696 hes = (HE **)SvPVX (sv);
697 }
599 698
600 i = 0; 699 i = 0;
601 while ((he = hv_iternext (hv))) 700 while ((he = hv_iternext (hv)))
602 { 701 {
603 hes [i++] = he; 702 hes [i++] = he;
672// encode objects, arrays and special \0=false and \1=true values. 771// encode objects, arrays and special \0=false and \1=true values.
673static void 772static void
674encode_rv (enc_t *enc, SV *sv) 773encode_rv (enc_t *enc, SV *sv)
675{ 774{
676 svtype svt; 775 svtype svt;
776 GV *method;
677 777
678 SvGETMAGIC (sv); 778 SvGETMAGIC (sv);
679 svt = SvTYPE (sv); 779 svt = SvTYPE (sv);
680 780
681 if (expect_false (SvOBJECT (sv))) 781 if (expect_false (SvOBJECT (sv)))
682 { 782 {
683 HV *stash = !JSON_SLOW || json_boolean_stash 783 HV *stash = SvSTASH (sv);
684 ? json_boolean_stash
685 : gv_stashpv ("JSON::XS::Boolean", 1);
686 784
687 if (SvSTASH (sv) == stash) 785 if (stash == bool_stash)
688 { 786 {
689 if (SvIV (sv))
690 encode_str (enc, "true", 4, 0); 787 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
691 else
692 encode_str (enc, "false", 5, 0); 788 else encode_str (enc, "false", 5, 0);
693 } 789 }
790 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
791 {
792 int count;
793 dSP;
794
795 ENTER; SAVETMPS;
796 PUSHMARK (SP);
797 EXTEND (SP, 2);
798 // we re-bless the reference to get overload and other niceties right
799 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
800 PUSHs (sv_json);
801
802 PUTBACK;
803 count = call_sv ((SV *)GvCV (method), G_ARRAY);
804 SPAGAIN;
805
806 // catch this surprisingly common error
807 if (SvROK (TOPs) && SvRV (TOPs) == sv)
808 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
809
810 encode_ch (enc, '(');
811 encode_ch (enc, '"');
812 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
813 encode_ch (enc, '"');
814 encode_ch (enc, ')');
815 encode_ch (enc, '[');
816
817 if (count)
818 {
819 int i;
820
821 for (i = 0; i < count - 1; ++i)
822 {
823 encode_sv (enc, SP[i + 1 - count]);
824 encode_ch (enc, ',');
825 }
826
827 encode_sv (enc, TOPs);
828 SP -= count;
829 }
830
831 encode_ch (enc, ']');
832
833 FREETMPS; LEAVE;
834 }
835 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
836 {
837 dSP;
838
839 ENTER; SAVETMPS;
840 PUSHMARK (SP);
841 // we re-bless the reference to get overload and other niceties right
842 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
843
844 // calling with G_SCALAR ensures that we always get a 1 return value
845 PUTBACK;
846 call_sv ((SV *)GvCV (method), G_SCALAR);
847 SPAGAIN;
848
849 // catch this surprisingly common error
850 if (SvROK (TOPs) && SvRV (TOPs) == sv)
851 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
852
853 sv = POPs;
854 PUTBACK;
855
856 encode_sv (enc, sv);
857
858 FREETMPS; LEAVE;
859 }
860 else if (enc->json.flags & F_ALLOW_BLESSED)
861 encode_str (enc, "null", 4, 0);
694 else 862 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", 863 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)))); 864 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
741 }
742 } 865 }
743 else if (svt == SVt_PVHV) 866 else if (svt == SVt_PVHV)
744 encode_hv (enc, (HV *)sv); 867 encode_hv (enc, (HV *)sv);
745 else if (svt == SVt_PVAV) 868 else if (svt == SVt_PVAV)
746 encode_av (enc, (AV *)sv); 869 encode_av (enc, (AV *)sv);
747 else if (svt < SVt_PVAV) 870 else if (svt < SVt_PVAV)
748 { 871 {
749 STRLEN len = 0; 872 int bool_type = ref_bool_type (sv);
750 char *pv = svt ? SvPV (sv, len) : 0;
751 873
752 if (len == 1 && *pv == '1') 874 if (bool_type == 1)
753 encode_str (enc, "true", 4, 0); 875 encode_str (enc, "true", 4, 0);
754 else if (len == 1 && *pv == '0') 876 else if (bool_type == 0)
755 encode_str (enc, "false", 5, 0); 877 encode_str (enc, "false", 5, 0);
756 else if (enc->json.flags & F_ALLOW_UNKNOWN) 878 else if (enc->json.flags & F_ALLOW_UNKNOWN)
757 encode_str (enc, "null", 4, 0); 879 encode_str (enc, "null", 4, 0);
758 else 880 else
759 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 881 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
822 } 944 }
823 else 945 else
824 { 946 {
825 // large integer, use the (rather slow) snprintf way. 947 // large integer, use the (rather slow) snprintf way.
826 need (enc, IVUV_MAXCHARS); 948 need (enc, IVUV_MAXCHARS);
827 enc->cur += 949 enc->cur +=
828 SvIsUV(sv) 950 SvIsUV(sv)
829 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) 951 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
830 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 952 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
831 } 953 }
832 } 954 }
833 else if (SvROK (sv)) 955 else if (SvROK (sv))
834 encode_rv (enc, SvRV (sv)); 956 encode_rv (enc, SvRV (sv));
835 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 957 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
836 encode_str (enc, "null", 4, 0); 958 encode_str (enc, "null", 4, 0);
837 else 959 else
838 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 960 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
839 SvPV_nolen (sv), SvFLAGS (sv)); 961 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
840} 962}
841 963
842static SV * 964static SV *
843encode_json (SV *scalar, JSON *json) 965encode_json (SV *scalar, JSON *json)
844{ 966{
845 enc_t enc; 967 enc_t enc;
846 968
847 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 969 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)"); 970 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
849 971
850 enc.json = *json; 972 enc.json = *json;
851 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 973 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
852 enc.cur = SvPVX (enc.sv); 974 enc.cur = SvPVX (enc.sv);
1059 *cur++ = *dec_cur++; 1181 *cur++ = *dec_cur++;
1060 while (--clen); 1182 while (--clen);
1061 1183
1062 utf8 = 1; 1184 utf8 = 1;
1063 } 1185 }
1186 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1187 *cur++ = ch;
1064 else 1188 else
1065 { 1189 {
1066 --dec_cur; 1190 --dec_cur;
1067 1191
1068 if (!ch) 1192 if (!ch)
1078 1202
1079 if (sv) 1203 if (sv)
1080 { 1204 {
1081 STRLEN cur = SvCUR (sv); 1205 STRLEN cur = SvCUR (sv);
1082 1206
1083 if (SvLEN (sv) <= cur + len) 1207 if (SvLEN (sv) - cur <= len)
1084 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 1208 json_sv_grow (sv, cur, len);
1085 1209
1086 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1210 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1087 SvCUR_set (sv, SvCUR (sv) + len); 1211 SvCUR_set (sv, SvCUR (sv) + len);
1088 } 1212 }
1089 else 1213 else
1256 if (*dec->cur == ']') 1380 if (*dec->cur == ']')
1257 { 1381 {
1258 ++dec->cur; 1382 ++dec->cur;
1259 break; 1383 break;
1260 } 1384 }
1261 1385
1262 if (*dec->cur != ',') 1386 if (*dec->cur != ',')
1263 ERR (", or ] expected while parsing array"); 1387 ERR (", or ] expected while parsing array");
1264 1388
1265 ++dec->cur; 1389 ++dec->cur;
1266 1390
1390 1514
1391 hv_iterinit (hv); 1515 hv_iterinit (hv);
1392 he = hv_iternext (hv); 1516 he = hv_iternext (hv);
1393 hv_iterinit (hv); 1517 hv_iterinit (hv);
1394 1518
1395 // the next line creates a mortal sv each time its called. 1519 // the next line creates a mortal sv each time it's called.
1396 // might want to optimise this for common cases. 1520 // might want to optimise this for common cases.
1397 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1521 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1398 1522
1399 if (cb) 1523 if (cb)
1400 { 1524 {
1401 dSP; 1525 dSP;
1402 int count; 1526 int count;
1403 1527
1404 ENTER; SAVETMPS; PUSHMARK (SP); 1528 ENTER; SAVETMPS;
1529 PUSHMARK (SP);
1405 XPUSHs (HeVAL (he)); 1530 XPUSHs (HeVAL (he));
1406 sv_2mortal (sv); 1531 sv_2mortal (sv);
1407 1532
1408 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1533 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1409 1534
1411 { 1536 {
1412 sv = newSVsv (POPs); 1537 sv = newSVsv (POPs);
1413 FREETMPS; LEAVE; 1538 FREETMPS; LEAVE;
1414 return sv; 1539 return sv;
1415 } 1540 }
1541 else if (count)
1542 croak ("filter_json_single_key_object callbacks must not return more than one scalar");
1416 1543
1417 SvREFCNT_inc (sv); 1544 SvREFCNT_inc (sv);
1418 FREETMPS; LEAVE; 1545 FREETMPS; LEAVE;
1419 } 1546 }
1420 } 1547 }
1422 if (dec->json.cb_object) 1549 if (dec->json.cb_object)
1423 { 1550 {
1424 dSP; 1551 dSP;
1425 int count; 1552 int count;
1426 1553
1427 ENTER; SAVETMPS; PUSHMARK (SP); 1554 ENTER; SAVETMPS;
1555 PUSHMARK (SP);
1428 XPUSHs (sv_2mortal (sv)); 1556 XPUSHs (sv_2mortal (sv));
1429 1557
1430 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1558 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1431 1559
1432 if (count == 1) 1560 if (count == 1)
1433 { 1561 {
1434 sv = newSVsv (POPs); 1562 sv = newSVsv (POPs);
1435 FREETMPS; LEAVE; 1563 FREETMPS; LEAVE;
1436 return sv; 1564 return sv;
1437 } 1565 }
1566 else if (count)
1567 croak ("filter_json_object callbacks must not return more than one scalar");
1438 1568
1439 SvREFCNT_inc (sv); 1569 SvREFCNT_inc (sv);
1440 FREETMPS; LEAVE; 1570 FREETMPS; LEAVE;
1441 } 1571 }
1442 } 1572 }
1448 DEC_DEC_DEPTH; 1578 DEC_DEC_DEPTH;
1449 return 0; 1579 return 0;
1450} 1580}
1451 1581
1452static SV * 1582static SV *
1583decode_tag (dec_t *dec)
1584{
1585 SV *tag = 0;
1586 SV *val = 0;
1587
1588 if (!(dec->json.flags & F_ALLOW_TAGS))
1589 ERR ("malformed JSON string, neither array, object, number, string or atom");
1590
1591 ++dec->cur;
1592
1593 decode_ws (dec);
1594
1595 tag = decode_sv (dec);
1596 if (!tag)
1597 goto fail;
1598
1599 if (!SvPOK (tag))
1600 ERR ("malformed JSON string, (tag) must be a string");
1601
1602 decode_ws (dec);
1603
1604 if (*dec->cur != ')')
1605 ERR (") expected after tag");
1606
1607 ++dec->cur;
1608
1609 decode_ws (dec);
1610
1611 val = decode_sv (dec);
1612 if (!val)
1613 goto fail;
1614
1615 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1616 ERR ("malformed JSON string, tag value must be an array");
1617
1618 {
1619 AV *av = (AV *)SvRV (val);
1620 int i, len = av_len (av) + 1;
1621 HV *stash = gv_stashsv (tag, 0);
1622 SV *sv;
1623
1624 if (!stash)
1625 ERR ("cannot decode perl-object (package does not exist)");
1626
1627 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1628
1629 if (!method)
1630 ERR ("cannot decode perl-object (package does not have a THAW method)");
1631
1632 dSP;
1633
1634 ENTER; SAVETMPS;
1635 PUSHMARK (SP);
1636 EXTEND (SP, len + 2);
1637 // we re-bless the reference to get overload and other niceties right
1638 PUSHs (tag);
1639 PUSHs (sv_json);
1640
1641 for (i = 0; i < len; ++i)
1642 PUSHs (*av_fetch (av, i, 1));
1643
1644 PUTBACK;
1645 call_sv ((SV *)GvCV (method), G_SCALAR);
1646 SPAGAIN;
1647
1648 SvREFCNT_dec (tag);
1649 SvREFCNT_dec (val);
1650 sv = SvREFCNT_inc (POPs);
1651
1652 PUTBACK;
1653
1654 FREETMPS; LEAVE;
1655
1656 return sv;
1657 }
1658
1659fail:
1660 SvREFCNT_dec (tag);
1661 SvREFCNT_dec (val);
1662 return 0;
1663}
1664
1665static SV *
1453decode_sv (dec_t *dec) 1666decode_sv (dec_t *dec)
1454{ 1667{
1455 // the beauty of JSON: you need exactly one character lookahead 1668 // the beauty of JSON: you need exactly one character lookahead
1456 // to parse everything. 1669 // to parse everything.
1457 switch (*dec->cur) 1670 switch (*dec->cur)
1458 { 1671 {
1459 case '"': ++dec->cur; return decode_str (dec); 1672 case '"': ++dec->cur; return decode_str (dec);
1460 case '[': ++dec->cur; return decode_av (dec); 1673 case '[': ++dec->cur; return decode_av (dec);
1461 case '{': ++dec->cur; return decode_hv (dec); 1674 case '{': ++dec->cur; return decode_hv (dec);
1675 case '(': return decode_tag (dec);
1462 1676
1463 case '-': 1677 case '-':
1464 case '0': case '1': case '2': case '3': case '4': 1678 case '0': case '1': case '2': case '3': case '4':
1465 case '5': case '6': case '7': case '8': case '9': 1679 case '5': case '6': case '7': case '8': case '9':
1466 return decode_num (dec); 1680 return decode_num (dec);
1467 1681
1682 case 'f':
1683 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1684 {
1685 dec->cur += 5;
1686
1687 if (expect_false (!dec->json.v_false))
1688 dec->json.v_false = GET_BOOL (false);
1689
1690 return newSVsv (dec->json.v_false);
1691 }
1692 else
1693 ERR ("'false' expected");
1694
1695 break;
1696
1468 case 't': 1697 case 't':
1469 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1698 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1470 { 1699 {
1471 dec->cur += 4; 1700 dec->cur += 4;
1472#if JSON_SLOW 1701
1473 json_true = get_bool ("JSON::XS::true"); 1702 if (expect_false (!dec->json.v_true))
1474#endif 1703 dec->json.v_true = GET_BOOL (true);
1704
1475 return newSVsv (json_true); 1705 return newSVsv (dec->json.v_true);
1476 } 1706 }
1477 else 1707 else
1478 ERR ("'true' expected"); 1708 ERR ("'true' expected");
1479
1480 break;
1481
1482 case 'f':
1483 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1484 {
1485 dec->cur += 5;
1486#if JSON_SLOW
1487 json_false = get_bool ("JSON::XS::false");
1488#endif
1489 return newSVsv (json_false);
1490 }
1491 else
1492 ERR ("'false' expected");
1493 1709
1494 break; 1710 break;
1495 1711
1496 case 'n': 1712 case 'n':
1497 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1713 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
1503 ERR ("'null' expected"); 1719 ERR ("'null' expected");
1504 1720
1505 break; 1721 break;
1506 1722
1507 default: 1723 default:
1508 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1724 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1509 break; 1725 break;
1510 } 1726 }
1511 1727
1512fail: 1728fail:
1513 return 0; 1729 return 0;
1514} 1730}
1515 1731
1516static SV * 1732static SV *
1517decode_json (SV *string, JSON *json, char **offset_return) 1733decode_json (SV *string, JSON *json, STRLEN *offset_return)
1518{ 1734{
1519 dec_t dec; 1735 dec_t dec;
1520 SV *sv; 1736 SV *sv;
1521 1737
1522 /* work around bugs in 5.10 where manipulating magic values 1738 /* work around bugs in 5.10 where manipulating magic values
1523 * will perl ignore the magic in subsequent accesses 1739 * makes perl ignore the magic in subsequent accesses.
1740 * also make a copy of non-PV values, to get them into a clean
1741 * state (SvPV should do that, but it's buggy, see below).
1742 *
1743 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1744 * as reported by Reini Urban.
1524 */ 1745 */
1525 /*SvGETMAGIC (string);*/ 1746 /*SvGETMAGIC (string);*/
1526 if (SvMAGICAL (string)) 1747 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1527 string = sv_2mortal (newSVsv (string)); 1748 string = sv_2mortal (newSVsv (string));
1528 1749
1529 SvUPGRADE (string, SVt_PV); 1750 SvUPGRADE (string, SVt_PV);
1530 1751
1531 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1752 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1548 if (offset > json->max_size && json->max_size) 1769 if (offset > json->max_size && json->max_size)
1549 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1770 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1550 (unsigned long)SvCUR (string), (unsigned long)json->max_size); 1771 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1551 } 1772 }
1552 1773
1553 if (json->flags & F_UTF8) 1774 if (DECODE_WANTS_OCTETS (json))
1554 sv_utf8_downgrade (string, 0); 1775 sv_utf8_downgrade (string, 0);
1555 else 1776 else
1556 sv_utf8_upgrade (string); 1777 sv_utf8_upgrade (string);
1557 1778
1558 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1779 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1570 1791
1571 decode_ws (&dec); 1792 decode_ws (&dec);
1572 sv = decode_sv (&dec); 1793 sv = decode_sv (&dec);
1573 1794
1574 if (offset_return) 1795 if (offset_return)
1575 *offset_return = dec.cur; 1796 *offset_return = dec.cur - SvPVX (string);
1576 1797 else if (sv)
1577 if (!(offset_return || !sv))
1578 { 1798 {
1579 // check for trailing garbage 1799 // check for trailing garbage
1580 decode_ws (&dec); 1800 decode_ws (&dec);
1581 1801
1582 if (*dec.cur) 1802 if (dec.cur != dec.end)
1583 { 1803 {
1584 dec.err = "garbage after JSON object"; 1804 dec.err = "garbage after JSON object";
1585 SvREFCNT_dec (sv); 1805 SvREFCNT_dec (sv);
1586 sv = 0; 1806 sv = 0;
1587 } 1807 }
1600 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1820 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1601 LEAVE; 1821 LEAVE;
1602 1822
1603 croak ("%s, at character offset %d (before \"%s\")", 1823 croak ("%s, at character offset %d (before \"%s\")",
1604 dec.err, 1824 dec.err,
1605 ptr_to_index (string, dec.cur), 1825 (int)ptr_to_index (string, dec.cur),
1606 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1826 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1607 } 1827 }
1608 1828
1609 sv = sv_2mortal (sv); 1829 sv = sv_2mortal (sv);
1610 1830
1611 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1831 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1612 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1832 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1613 1833
1614 return sv; 1834 return sv;
1615} 1835}
1616 1836
1625 // the state machine here is a bit convoluted and could be simplified a lot 1845 // the state machine here is a bit convoluted and could be simplified a lot
1626 // but this would make it slower, so... 1846 // but this would make it slower, so...
1627 1847
1628 for (;;) 1848 for (;;)
1629 { 1849 {
1630 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1631 switch (self->incr_mode) 1850 switch (self->incr_mode)
1632 { 1851 {
1852 // reached end of a scalar, see if we are inside a nested structure or not
1853 end_of_scalar:
1854 self->incr_mode = INCR_M_JSON;
1855
1856 if (self->incr_nest) // end of a scalar inside array, object or tag
1857 goto incr_m_json;
1858 else // end of scalar outside structure, json text ends here
1859 goto interrupt;
1860
1633 // only used for initial whitespace skipping 1861 // only used for initial whitespace skipping
1634 case INCR_M_WS: 1862 case INCR_M_WS:
1635 for (;;) 1863 for (;;)
1636 { 1864 {
1637 if (*p > 0x20) 1865 if (*p > 0x20)
1679 ++p; 1907 ++p;
1680 } 1908 }
1681 1909
1682 break; 1910 break;
1683 1911
1912 // inside true/false/null
1913 case INCR_M_TFN:
1914 incr_m_tfn:
1915 for (;;)
1916 switch (*p++)
1917 {
1918 case 'r': case 'u': case 'e': // tRUE, falsE, nUll
1919 case 'a': case 'l': case 's': // fALSe, nuLL
1920 // allowed
1921 break;
1922
1923 default:
1924 --p;
1925 goto end_of_scalar;
1926 }
1927
1928 // inside a number
1929 case INCR_M_NUM:
1930 incr_m_num:
1931 for (;;)
1932 switch (*p++)
1933 {
1934 case 'e': case 'E': case '.': case '+':
1935 case '-':
1936 case '0': case '1': case '2': case '3': case '4':
1937 case '5': case '6': case '7': case '8': case '9':
1938 // allowed
1939 break;
1940
1941 default:
1942 --p;
1943 goto end_of_scalar;
1944 }
1945
1684 // inside a string 1946 // inside a string
1685 case INCR_M_STR: 1947 case INCR_M_STR:
1686 incr_m_str: 1948 incr_m_str:
1687 for (;;) 1949 for (;;)
1688 { 1950 {
1689 if (*p == '"') 1951 if (*p == '"')
1690 { 1952 {
1691 ++p; 1953 ++p;
1692 self->incr_mode = INCR_M_JSON;
1693
1694 if (!self->incr_nest)
1695 goto interrupt;
1696
1697 goto incr_m_json; 1954 goto end_of_scalar;
1698 } 1955 }
1699 else if (*p == '\\') 1956 else if (*p == '\\')
1700 { 1957 {
1701 ++p; // "virtually" consumes character after \ 1958 ++p; // "virtually" consumes character after \
1702 1959
1732 --p; // do not eat the whitespace, let the next round do it 1989 --p; // do not eat the whitespace, let the next round do it
1733 goto interrupt; 1990 goto interrupt;
1734 } 1991 }
1735 break; 1992 break;
1736 1993
1994 // the following three blocks handle scalars. this makes the parser
1995 // more strict than required inside arrays or objects, and could
1996 // be moved to a special case on the toplevel (except strings)
1997 case 't':
1998 case 'f':
1999 case 'n':
2000 self->incr_mode = INCR_M_TFN;
2001 goto incr_m_tfn;
2002
2003 case '-':
2004 case '0': case '1': case '2': case '3': case '4':
2005 case '5': case '6': case '7': case '8': case '9':
2006 self->incr_mode = INCR_M_NUM;
2007 goto incr_m_num;
2008
1737 case '"': 2009 case '"':
1738 self->incr_mode = INCR_M_STR; 2010 self->incr_mode = INCR_M_STR;
1739 goto incr_m_str; 2011 goto incr_m_str;
1740 2012
1741 case '[': 2013 case '[':
1742 case '{': 2014 case '{':
2015 case '(':
1743 if (++self->incr_nest > self->max_depth) 2016 if (++self->incr_nest > self->max_depth)
1744 croak (ERR_NESTING_EXCEEDED); 2017 croak (ERR_NESTING_EXCEEDED);
1745 break; 2018 break;
1746 2019
1747 case ']': 2020 case ']':
1748 case '}': 2021 case '}':
1749 if (--self->incr_nest <= 0) 2022 if (--self->incr_nest <= 0)
1750 goto interrupt; 2023 goto interrupt;
2024 break;
2025
2026 case ')':
2027 --self->incr_nest;
1751 break; 2028 break;
1752 2029
1753 case '#': 2030 case '#':
1754 self->incr_mode = INCR_M_C1; 2031 self->incr_mode = INCR_M_C1;
1755 goto incr_m_c; 2032 goto incr_m_c;
1781 i >= '0' && i <= '9' ? i - '0' 2058 i >= '0' && i <= '9' ? i - '0'
1782 : i >= 'a' && i <= 'f' ? i - 'a' + 10 2059 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1783 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2060 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1784 : -1; 2061 : -1;
1785 2062
1786 json_stash = gv_stashpv ("JSON::XS" , 1); 2063 json_stash = gv_stashpv ("JSON::XS" , 1);
1787 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 2064 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1788
1789 json_true = get_bool ("JSON::XS::true");
1790 json_false = get_bool ("JSON::XS::false"); 2065 bool_false = get_bool ("Types::Serialiser::false");
2066 bool_true = get_bool ("Types::Serialiser::true");
2067
2068 sv_json = newSVpv ("JSON", 0);
2069 SvREADONLY_on (sv_json);
1791 2070
1792 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ 2071 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1793} 2072}
1794 2073
1795PROTOTYPES: DISABLE 2074PROTOTYPES: DISABLE
1796 2075
1797void CLONE (...) 2076void CLONE (...)
1798 CODE: 2077 CODE:
2078 // as long as these writes are atomic, the race should not matter
2079 // as existing threads either already use 0, or use the old value,
2080 // which is sitll correct for the initial thread.
1799 json_stash = 0; 2081 json_stash = 0;
1800 json_boolean_stash = 0; 2082 bool_stash = 0;
2083 bool_false = 0;
2084 bool_true = 0;
1801 2085
1802void new (char *klass) 2086void new (char *klass)
1803 PPCODE: 2087 PPCODE:
1804{ 2088{
1805 SV *pv = NEWSV (0, sizeof (JSON)); 2089 SV *pv = NEWSV (0, sizeof (JSON));
1806 SvPOK_only (pv); 2090 SvPOK_only (pv);
1807 json_init ((JSON *)SvPVX (pv)); 2091 json_init ((JSON *)SvPVX (pv));
1808 XPUSHs (sv_2mortal (sv_bless ( 2092 XPUSHs (sv_2mortal (sv_bless (
1809 newRV_noinc (pv), 2093 newRV_noinc (pv),
1810 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2094 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1811 ))); 2095 )));
1812} 2096}
2097
2098void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2099 PPCODE:
2100 self->v_false = newSVsv (v_false);
2101 self->v_true = newSVsv (v_true);
2102 XPUSHs (ST (0));
2103
2104void get_boolean_values (JSON *self)
2105 PPCODE:
2106 if (self->v_false && self->v_true)
2107 {
2108 EXTEND (SP, 2);
2109 PUSHs (self->v_false);
2110 PUSHs (self->v_true);
2111 }
1813 2112
1814void ascii (JSON *self, int enable = 1) 2113void ascii (JSON *self, int enable = 1)
1815 ALIAS: 2114 ALIAS:
1816 ascii = F_ASCII 2115 ascii = F_ASCII
1817 latin1 = F_LATIN1 2116 latin1 = F_LATIN1
1825 shrink = F_SHRINK 2124 shrink = F_SHRINK
1826 allow_blessed = F_ALLOW_BLESSED 2125 allow_blessed = F_ALLOW_BLESSED
1827 convert_blessed = F_CONV_BLESSED 2126 convert_blessed = F_CONV_BLESSED
1828 relaxed = F_RELAXED 2127 relaxed = F_RELAXED
1829 allow_unknown = F_ALLOW_UNKNOWN 2128 allow_unknown = F_ALLOW_UNKNOWN
2129 allow_tags = F_ALLOW_TAGS
1830 PPCODE: 2130 PPCODE:
1831{ 2131{
1832 if (enable) 2132 if (enable)
1833 self->flags |= ix; 2133 self->flags |= ix;
1834 else 2134 else
1850 get_shrink = F_SHRINK 2150 get_shrink = F_SHRINK
1851 get_allow_blessed = F_ALLOW_BLESSED 2151 get_allow_blessed = F_ALLOW_BLESSED
1852 get_convert_blessed = F_CONV_BLESSED 2152 get_convert_blessed = F_CONV_BLESSED
1853 get_relaxed = F_RELAXED 2153 get_relaxed = F_RELAXED
1854 get_allow_unknown = F_ALLOW_UNKNOWN 2154 get_allow_unknown = F_ALLOW_UNKNOWN
2155 get_allow_tags = F_ALLOW_TAGS
1855 PPCODE: 2156 PPCODE:
1856 XPUSHs (boolSV (self->flags & ix)); 2157 XPUSHs (boolSV (self->flags & ix));
1857 2158
1858void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 2159void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1859 PPCODE: 2160 PPCODE:
1887} 2188}
1888 2189
1889void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2190void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1890 PPCODE: 2191 PPCODE:
1891{ 2192{
1892 if (!self->cb_sk_object) 2193 if (!self->cb_sk_object)
1893 self->cb_sk_object = newHV (); 2194 self->cb_sk_object = newHV ();
1894 2195
1895 if (SvOK (cb)) 2196 if (SvOK (cb))
1896 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2197 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1897 else 2198 else
1908 XPUSHs (ST (0)); 2209 XPUSHs (ST (0));
1909} 2210}
1910 2211
1911void encode (JSON *self, SV *scalar) 2212void encode (JSON *self, SV *scalar)
1912 PPCODE: 2213 PPCODE:
1913 XPUSHs (encode_json (scalar, self)); 2214 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2215 XPUSHs (scalar);
1914 2216
1915void decode (JSON *self, SV *jsonstr) 2217void decode (JSON *self, SV *jsonstr)
1916 PPCODE: 2218 PPCODE:
1917 XPUSHs (decode_json (jsonstr, self, 0)); 2219 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2220 XPUSHs (jsonstr);
1918 2221
1919void decode_prefix (JSON *self, SV *jsonstr) 2222void decode_prefix (JSON *self, SV *jsonstr)
1920 PPCODE: 2223 PPCODE:
1921{ 2224{
1922 char *offset; 2225 SV *sv;
2226 STRLEN offset;
2227 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1923 EXTEND (SP, 2); 2228 EXTEND (SP, 2);
1924 PUSHs (decode_json (jsonstr, self, &offset)); 2229 PUSHs (sv);
1925 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2230 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
1926} 2231}
1927 2232
1928void incr_parse (JSON *self, SV *jsonstr = 0) 2233void incr_parse (JSON *self, SV *jsonstr = 0)
1929 PPCODE: 2234 PPCODE:
1930{ 2235{
1931 if (!self->incr_text) 2236 if (!self->incr_text)
1932 self->incr_text = newSVpvn ("", 0); 2237 self->incr_text = newSVpvn ("", 0);
2238
2239 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2240 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2241 if (DECODE_WANTS_OCTETS (self))
2242 {
2243 if (self->incr_pos)
2244 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2245 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2246
2247 sv_utf8_downgrade (self->incr_text, 0);
2248 }
2249 else
2250 {
2251 sv_utf8_upgrade (self->incr_text);
2252
2253 if (self->incr_pos)
2254 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2255 - (U8 *)SvPVX (self->incr_text);
2256 }
1933 2257
1934 // append data, if any 2258 // append data, if any
1935 if (jsonstr) 2259 if (jsonstr)
1936 { 2260 {
2261 /* make sure both strings have same encoding */
2262 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1937 if (SvUTF8 (jsonstr)) 2263 if (SvUTF8 (jsonstr))
2264 sv_utf8_downgrade (jsonstr, 0);
1938 { 2265 else
1939 if (!SvUTF8 (self->incr_text))
1940 {
1941 /* utf-8-ness differs, need to upgrade */
1942 sv_utf8_upgrade (self->incr_text);
1943
1944 if (self->incr_pos)
1945 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1946 - (U8 *)SvPVX (self->incr_text);
1947 }
1948 }
1949 else if (SvUTF8 (self->incr_text))
1950 sv_utf8_upgrade (jsonstr); 2266 sv_utf8_upgrade (jsonstr);
1951 2267
2268 /* and then just blindly append */
1952 { 2269 {
1953 STRLEN len; 2270 STRLEN len;
1954 const char *str = SvPV (jsonstr, len); 2271 const char *str = SvPV (jsonstr, len);
1955 STRLEN cur = SvCUR (self->incr_text); 2272 STRLEN cur = SvCUR (self->incr_text);
1956 2273
1957 if (SvLEN (self->incr_text) <= cur + len) 2274 if (SvLEN (self->incr_text) - cur <= len)
1958 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2275 json_sv_grow (self->incr_text, cur, len);
1959 2276
1960 Move (str, SvEND (self->incr_text), len, char); 2277 Move (str, SvEND (self->incr_text), len, char);
1961 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2278 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1962 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 2279 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1963 } 2280 }
1964 } 2281 }
1965 2282
1966 if (GIMME_V != G_VOID) 2283 if (GIMME_V != G_VOID)
1967 do 2284 do
1968 { 2285 {
2286 SV *sv;
1969 char *offset; 2287 STRLEN offset;
1970 2288
1971 if (!INCR_DONE (self)) 2289 if (!INCR_DONE (self))
1972 { 2290 {
1973 incr_parse (self); 2291 incr_parse (self);
1974 2292
1975 if (self->incr_pos > self->max_size && self->max_size) 2293 if (self->incr_pos > self->max_size && self->max_size)
1976 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 2294 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1977 (unsigned long)self->incr_pos, (unsigned long)self->max_size); 2295 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1978 2296
1979 if (!INCR_DONE (self)) 2297 if (!INCR_DONE (self))
2298 {
2299 // as an optimisation, do not accumulate white space in the incr buffer
2300 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2301 {
2302 self->incr_pos = 0;
2303 SvCUR_set (self->incr_text, 0);
2304 }
2305
1980 break; 2306 break;
2307 }
1981 } 2308 }
1982 2309
1983 XPUSHs (decode_json (self->incr_text, self, &offset)); 2310 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2311 XPUSHs (sv);
1984 2312
1985 self->incr_pos -= offset - SvPVX (self->incr_text); 2313 self->incr_pos -= offset;
1986 self->incr_nest = 0; 2314 self->incr_nest = 0;
1987 self->incr_mode = 0; 2315 self->incr_mode = 0;
1988 2316
1989 sv_chop (self->incr_text, offset); 2317 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
1990 } 2318 }
1991 while (GIMME_V == G_ARRAY); 2319 while (GIMME_V == G_ARRAY);
1992} 2320}
1993 2321
1994SV *incr_text (JSON *self) 2322SV *incr_text (JSON *self)
2025 self->incr_mode = 0; 2353 self->incr_mode = 0;
2026} 2354}
2027 2355
2028void DESTROY (JSON *self) 2356void DESTROY (JSON *self)
2029 CODE: 2357 CODE:
2358 SvREFCNT_dec (self->v_false);
2359 SvREFCNT_dec (self->v_true);
2030 SvREFCNT_dec (self->cb_sk_object); 2360 SvREFCNT_dec (self->cb_sk_object);
2031 SvREFCNT_dec (self->cb_object); 2361 SvREFCNT_dec (self->cb_object);
2032 SvREFCNT_dec (self->incr_text); 2362 SvREFCNT_dec (self->incr_text);
2033 2363
2034PROTOTYPES: ENABLE 2364PROTOTYPES: ENABLE
2035 2365
2036void encode_json (SV *scalar) 2366void encode_json (SV *scalar)
2037 ALIAS:
2038 to_json_ = 0
2039 encode_json = F_UTF8
2040 PPCODE: 2367 PPCODE:
2041{ 2368{
2042 JSON json; 2369 JSON json;
2043 json_init (&json); 2370 json_init (&json);
2044 json.flags |= ix; 2371 json.flags |= F_UTF8;
2045 XPUSHs (encode_json (scalar, &json)); 2372 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2373 XPUSHs (scalar);
2046} 2374}
2047 2375
2048void decode_json (SV *jsonstr) 2376void decode_json (SV *jsonstr)
2049 ALIAS:
2050 from_json_ = 0
2051 decode_json = F_UTF8
2052 PPCODE: 2377 PPCODE:
2053{ 2378{
2054 JSON json; 2379 JSON json;
2055 json_init (&json); 2380 json_init (&json);
2056 json.flags |= ix; 2381 json.flags |= F_UTF8;
2057 XPUSHs (decode_json (jsonstr, &json, 0)); 2382 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2383 XPUSHs (jsonstr);
2058} 2384}
2059 2385

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines