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 1.117 by root, Fri Oct 25 20:19:57 2013 UTC vs.
Revision 1.133 by root, Tue Aug 28 16:16:17 2018 UTC

… …
6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
12#if defined(__BORLANDC__) || defined(_MSC_VER) 13#if defined(__BORLANDC__) || defined(_MSC_VER)
13# define snprintf _snprintf // C compilers have this in stdio.h 14# define snprintf _snprintf // C compilers have this in stdio.h
14#endif 15#endif
15 16
16// some old perls do not have this, try to make it work, no 17// some old perls do not have this, try to make it work, no
17// guarantees, though. if it breaks, you get to keep the pieces. 18// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES 19#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13 20# define UTF8_MAXBYTES 13
21#endif
22
23// compatibility with perl <5.18
24#ifndef HvNAMELEN_get
25# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
26#endif
27#ifndef HvNAMELEN
28# define HvNAMELEN(hv) HvNAMELEN_get (hv)
29#endif
30#ifndef HvNAMEUTF8
31# define HvNAMEUTF8(hv) 0
20#endif 32#endif
21 33
22// three extra for rounding, sign, and end of string 34// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) 35#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
24 36
… …
33#define F_SHRINK 0x00000200UL 45#define F_SHRINK 0x00000200UL
34#define F_ALLOW_BLESSED 0x00000400UL 46#define F_ALLOW_BLESSED 0x00000400UL
35#define F_CONV_BLESSED 0x00000800UL 47#define F_CONV_BLESSED 0x00000800UL
36#define F_RELAXED 0x00001000UL 48#define F_RELAXED 0x00001000UL
37#define F_ALLOW_UNKNOWN 0x00002000UL 49#define F_ALLOW_UNKNOWN 0x00002000UL
50#define F_ALLOW_TAGS 0x00004000UL
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 51#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
39 52
40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 53#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
41 54
42#define INIT_SIZE 32 // initial scalar size to be allocated 55#define INIT_SIZE 32 // initial scalar size to be allocated
… …
65 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 78 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
66 79
67#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" 80#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
68 81
69#ifdef USE_ITHREADS 82#ifdef USE_ITHREADS
70# define JSON_SLOW 1
71# 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))
72#else 86#else
73# define JSON_SLOW 0
74# define JSON_STASH json_stash 87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash
89# define GET_BOOL(value) bool_ ## value
75#endif 90#endif
76 91
77// the amount of HEs to allocate on the stack, when sorting keys 92// the amount of HEs to allocate on the stack, when sorting keys
78#define STACK_HES 64 93#define STACK_HES 64
79 94
80static HV *json_stash, *json_boolean_stash; // JSON::XS:: 95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
81static SV *json_true, *json_false; 96static SV *bool_false, *bool_true;
97static SV *sv_json;
82 98
83enum { 99enum {
84 INCR_M_WS = 0, // initial whitespace skipping, must be 0 100 INCR_M_WS = 0, // initial whitespace skipping, must be 0
85 INCR_M_STR, // inside string 101 INCR_M_STR, // inside string
86 INCR_M_BS, // inside backslash 102 INCR_M_BS, // inside backslash
… …
102 // for the incremental parser 118 // for the incremental parser
103 SV *incr_text; // the source text so far 119 SV *incr_text; // the source text so far
104 STRLEN incr_pos; // the current offset into the text 120 STRLEN incr_pos; // the current offset into the text
105 int incr_nest; // {[]}-nesting level 121 int incr_nest; // {[]}-nesting level
106 unsigned char incr_mode; 122 unsigned char incr_mode;
123
124 SV *v_false, *v_true;
107} JSON; 125} JSON;
108 126
109INLINE void 127INLINE void
110json_init (JSON *json) 128json_init (JSON *json)
111{ 129{
… …
140 SvPV_renew (sv, SvCUR (sv) + 1); 158 SvPV_renew (sv, SvCUR (sv) + 1);
141#endif 159#endif
142 } 160 }
143} 161}
144 162
163/* adds two STRLENs together, slow, and with paranoia */
164STRLEN
165strlen_sum (STRLEN l1, STRLEN l2)
166{
167 size_t sum = l1 + l2;
168
169 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
170 croak ("JSON::XS: string size overflow");
171
172 return sum;
173}
174
175/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
176static char *
177json_sv_grow (SV *sv, size_t len1, size_t len2)
178{
179 len1 = strlen_sum (len1, len2);
180 len1 = strlen_sum (len1, len1 >> 1);
181
182 if (len1 > 4096 - 24)
183 len1 = (len1 | 4095) - 24;
184
185 return SvGROW (sv, len1);
186}
187
145// decode an utf-8 character and return it, or (UV)-1 in 188// decode a utf-8 character and return it, or (UV)-1 in
146// case of an error. 189// case of an error.
147// we special-case "safe" characters from U+80 .. U+7FF, 190// we special-case "safe" characters from U+80 .. U+7FF,
148// but use the very good perl function to parse anything else. 191// but use the very good perl function to parse anything else.
149// note that we never call this function for a ascii codepoints 192// note that we never call this function for a ascii codepoints
150INLINE UV 193INLINE UV
… …
287 // a recursion depth of ten gives us >>500 bits 330 // a recursion depth of ten gives us >>500 bits
288 json_atof_scan1 (s, &accum, &expo, 0, 10); 331 json_atof_scan1 (s, &accum, &expo, 0, 10);
289 332
290 return neg ? -accum : accum; 333 return neg ? -accum : accum;
291} 334}
335
336// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
337static int
338ref_bool_type (SV *sv)
339{
340 svtype svt = SvTYPE (sv);
341
342 if (svt < SVt_PVAV)
343 {
344 STRLEN len = 0;
345 char *pv = svt ? SvPV (sv, len) : 0;
346
347 if (len == 1)
348 if (*pv == '1')
349 return 1;
350 else if (*pv == '0')
351 return 0;
352 }
353
354 return -1;
355}
356
357// returns whether scalar is not a reference in the sense of allow_nonref
358static int
359json_nonref (SV *scalar)
360{
361 if (!SvROK (scalar))
362 return 1;
363
364 scalar = SvRV (scalar);
365
366 if (SvTYPE (scalar) >= SVt_PVMG)
367 {
368 if (SvSTASH (scalar) == bool_stash)
369 return 1;
370
371 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
372 return 1;
373 }
374
375 return 0;
376}
377
292///////////////////////////////////////////////////////////////////////////// 378/////////////////////////////////////////////////////////////////////////////
293// encoder 379// encoder
294 380
295// structure used for encoding JSON 381// structure used for encoding JSON
296typedef struct 382typedef struct
… …
304} enc_t; 390} enc_t;
305 391
306INLINE void 392INLINE void
307need (enc_t *enc, STRLEN len) 393need (enc_t *enc, STRLEN len)
308{ 394{
309 if (expect_false (enc->cur + len >= enc->end)) 395 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
310 { 396 {
311 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 397 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
312 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 398 char *buf = json_sv_grow (enc->sv, cur, len);
313 enc->cur = SvPVX (enc->sv) + cur; 399 enc->cur = buf + cur;
314 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 400 enc->end = buf + SvLEN (enc->sv) - 1;
315 } 401 }
316} 402}
317 403
318INLINE void 404INLINE void
319encode_ch (enc_t *enc, char ch) 405encode_ch (enc_t *enc, char ch)
… …
335 421
336 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 422 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
337 { 423 {
338 if (expect_false (ch == '"')) // but with slow exceptions 424 if (expect_false (ch == '"')) // but with slow exceptions
339 { 425 {
340 need (enc, len += 1); 426 need (enc, len + 1);
341 *enc->cur++ = '\\'; 427 *enc->cur++ = '\\';
342 *enc->cur++ = '"'; 428 *enc->cur++ = '"';
343 } 429 }
344 else if (expect_false (ch == '\\')) 430 else if (expect_false (ch == '\\'))
345 { 431 {
346 need (enc, len += 1); 432 need (enc, len + 1);
347 *enc->cur++ = '\\'; 433 *enc->cur++ = '\\';
348 *enc->cur++ = '\\'; 434 *enc->cur++ = '\\';
349 } 435 }
350 else 436 else
351 *enc->cur++ = ch; 437 *enc->cur++ = ch;
… …
354 } 440 }
355 else 441 else
356 { 442 {
357 switch (ch) 443 switch (ch)
358 { 444 {
359 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 445 case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
360 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; 446 case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
361 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; 447 case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
362 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; 448 case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
363 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; 449 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
364 450
365 default: 451 default:
366 { 452 {
367 STRLEN clen; 453 STRLEN clen;
368 UV uch; 454 UV uch;
… …
384 if (uch >= 0x10000UL) 470 if (uch >= 0x10000UL)
385 { 471 {
386 if (uch >= 0x110000UL) 472 if (uch >= 0x110000UL)
387 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 473 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
388 474
389 need (enc, len += 11); 475 need (enc, len + 11);
390 sprintf (enc->cur, "\\u%04x\\u%04x", 476 sprintf (enc->cur, "\\u%04x\\u%04x",
391 (int)((uch - 0x10000) / 0x400 + 0xD800), 477 (int)((uch - 0x10000) / 0x400 + 0xD800),
392 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 478 (int)((uch - 0x10000) % 0x400 + 0xDC00));
393 enc->cur += 12; 479 enc->cur += 12;
394 } 480 }
395 else 481 else
396 { 482 {
397 need (enc, len += 5); 483 need (enc, len + 5);
398 *enc->cur++ = '\\'; 484 *enc->cur++ = '\\';
399 *enc->cur++ = 'u'; 485 *enc->cur++ = 'u';
400 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 486 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
401 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 487 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
402 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 488 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
… …
410 *enc->cur++ = uch; 496 *enc->cur++ = uch;
411 str += clen; 497 str += clen;
412 } 498 }
413 else if (is_utf8) 499 else if (is_utf8)
414 { 500 {
415 need (enc, len += clen); 501 need (enc, len + clen);
416 do 502 do
417 { 503 {
418 *enc->cur++ = *str++; 504 *enc->cur++ = *str++;
419 } 505 }
420 while (--clen); 506 while (--clen);
421 } 507 }
422 else 508 else
423 { 509 {
424 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 510 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
425 enc->cur = encode_utf8 (enc->cur, uch); 511 enc->cur = encode_utf8 (enc->cur, uch);
426 ++str; 512 ++str;
427 } 513 }
428 } 514 }
429 } 515 }
… …
682// encode objects, arrays and special \0=false and \1=true values. 768// encode objects, arrays and special \0=false and \1=true values.
683static void 769static void
684encode_rv (enc_t *enc, SV *sv) 770encode_rv (enc_t *enc, SV *sv)
685{ 771{
686 svtype svt; 772 svtype svt;
773 GV *method;
687 774
688 SvGETMAGIC (sv); 775 SvGETMAGIC (sv);
689 svt = SvTYPE (sv); 776 svt = SvTYPE (sv);
690 777
691 if (expect_false (SvOBJECT (sv))) 778 if (expect_false (SvOBJECT (sv)))
692 { 779 {
693 HV *stash = !JSON_SLOW || json_boolean_stash 780 HV *stash = SvSTASH (sv);
694 ? json_boolean_stash
695 : gv_stashpv ("JSON::XS::Boolean", 1);
696 781
697 if (SvSTASH (sv) == stash) 782 if (stash == bool_stash)
698 { 783 {
699 if (SvIV (sv))
700 encode_str (enc, "true", 4, 0); 784 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
701 else
702 encode_str (enc, "false", 5, 0); 785 else encode_str (enc, "false", 5, 0);
703 } 786 }
787 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
788 {
789 int count;
790 dSP;
791
792 ENTER; SAVETMPS;
793 SAVESTACK_POS ();
794 PUSHMARK (SP);
795 EXTEND (SP, 2);
796 // we re-bless the reference to get overload and other niceties right
797 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
798 PUSHs (sv_json);
799
800 PUTBACK;
801 count = call_sv ((SV *)GvCV (method), G_ARRAY);
802 SPAGAIN;
803
804 // catch this surprisingly common error
805 if (SvROK (TOPs) && SvRV (TOPs) == sv)
806 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
807
808 encode_ch (enc, '(');
809 encode_ch (enc, '"');
810 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
811 encode_ch (enc, '"');
812 encode_ch (enc, ')');
813 encode_ch (enc, '[');
814
815 while (count)
816 {
817 encode_sv (enc, SP[1 - count--]);
818
819 if (count)
820 encode_ch (enc, ',');
821 }
822
823 encode_ch (enc, ']');
824
825 FREETMPS; LEAVE;
826 }
827 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
828 {
829 dSP;
830
831 ENTER; SAVETMPS;
832 PUSHMARK (SP);
833 // we re-bless the reference to get overload and other niceties right
834 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
835
836 // calling with G_SCALAR ensures that we always get a 1 return value
837 PUTBACK;
838 call_sv ((SV *)GvCV (method), G_SCALAR);
839 SPAGAIN;
840
841 // catch this surprisingly common error
842 if (SvROK (TOPs) && SvRV (TOPs) == sv)
843 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
844
845 sv = POPs;
846 PUTBACK;
847
848 encode_sv (enc, sv);
849
850 FREETMPS; LEAVE;
851 }
852 else if (enc->json.flags & F_ALLOW_BLESSED)
853 encode_str (enc, "null", 4, 0);
704 else 854 else
705 {
706#if 0
707 if (0 && sv_derived_from (rv, "JSON::Literal"))
708 {
709 // not yet
710 }
711#endif
712 if (enc->json.flags & F_CONV_BLESSED)
713 {
714 // we re-bless the reference to get overload and other niceties right
715 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
716
717 if (to_json)
718 {
719 dSP;
720
721 ENTER; SAVETMPS; PUSHMARK (SP);
722 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
723
724 // calling with G_SCALAR ensures that we always get a 1 return value
725 PUTBACK;
726 call_sv ((SV *)GvCV (to_json), G_SCALAR);
727 SPAGAIN;
728
729 // catch this surprisingly common error
730 if (SvROK (TOPs) && SvRV (TOPs) == sv)
731 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
732
733 sv = POPs;
734 PUTBACK;
735
736 encode_sv (enc, sv);
737
738 FREETMPS; LEAVE;
739 }
740 else if (enc->json.flags & F_ALLOW_BLESSED)
741 encode_str (enc, "null", 4, 0);
742 else
743 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
744 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
745 }
746 else if (enc->json.flags & F_ALLOW_BLESSED)
747 encode_str (enc, "null", 4, 0);
748 else
749 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 855 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
750 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 856 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
751 }
752 } 857 }
753 else if (svt == SVt_PVHV) 858 else if (svt == SVt_PVHV)
754 encode_hv (enc, (HV *)sv); 859 encode_hv (enc, (HV *)sv);
755 else if (svt == SVt_PVAV) 860 else if (svt == SVt_PVAV)
756 encode_av (enc, (AV *)sv); 861 encode_av (enc, (AV *)sv);
757 else if (svt < SVt_PVAV) 862 else if (svt < SVt_PVAV)
758 { 863 {
759 STRLEN len = 0; 864 int bool_type = ref_bool_type (sv);
760 char *pv = svt ? SvPV (sv, len) : 0;
761 865
762 if (len == 1 && *pv == '1') 866 if (bool_type == 1)
763 encode_str (enc, "true", 4, 0); 867 encode_str (enc, "true", 4, 0);
764 else if (len == 1 && *pv == '0') 868 else if (bool_type == 0)
765 encode_str (enc, "false", 5, 0); 869 encode_str (enc, "false", 5, 0);
766 else if (enc->json.flags & F_ALLOW_UNKNOWN) 870 else if (enc->json.flags & F_ALLOW_UNKNOWN)
767 encode_str (enc, "null", 4, 0); 871 encode_str (enc, "null", 4, 0);
768 else 872 else
769 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 873 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
… …
852static SV * 956static SV *
853encode_json (SV *scalar, JSON *json) 957encode_json (SV *scalar, JSON *json)
854{ 958{
855 enc_t enc; 959 enc_t enc;
856 960
857 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 961 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
858 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 962 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
859 963
860 enc.json = *json; 964 enc.json = *json;
861 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 965 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
862 enc.cur = SvPVX (enc.sv); 966 enc.cur = SvPVX (enc.sv);
… …
1069 *cur++ = *dec_cur++; 1173 *cur++ = *dec_cur++;
1070 while (--clen); 1174 while (--clen);
1071 1175
1072 utf8 = 1; 1176 utf8 = 1;
1073 } 1177 }
1178 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1179 *cur++ = ch;
1074 else 1180 else
1075 { 1181 {
1076 --dec_cur; 1182 --dec_cur;
1077 1183
1078 if (!ch) 1184 if (!ch)
… …
1088 1194
1089 if (sv) 1195 if (sv)
1090 { 1196 {
1091 STRLEN cur = SvCUR (sv); 1197 STRLEN cur = SvCUR (sv);
1092 1198
1093 if (SvLEN (sv) <= cur + len) 1199 if (SvLEN (sv) - cur <= len)
1094 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 1200 json_sv_grow (sv, cur, len);
1095 1201
1096 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1202 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1097 SvCUR_set (sv, SvCUR (sv) + len); 1203 SvCUR_set (sv, SvCUR (sv) + len);
1098 } 1204 }
1099 else 1205 else
… …
1400 1506
1401 hv_iterinit (hv); 1507 hv_iterinit (hv);
1402 he = hv_iternext (hv); 1508 he = hv_iternext (hv);
1403 hv_iterinit (hv); 1509 hv_iterinit (hv);
1404 1510
1405 // the next line creates a mortal sv each time its called. 1511 // the next line creates a mortal sv each time it's called.
1406 // might want to optimise this for common cases. 1512 // might want to optimise this for common cases.
1407 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1513 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1408 1514
1409 if (cb) 1515 if (cb)
1410 { 1516 {
1411 dSP; 1517 dSP;
1412 int count; 1518 int count;
1413 1519
1414 ENTER; SAVETMPS; PUSHMARK (SP); 1520 ENTER; SAVETMPS;
1521 SAVESTACK_POS ();
1522 PUSHMARK (SP);
1415 XPUSHs (HeVAL (he)); 1523 XPUSHs (HeVAL (he));
1416 sv_2mortal (sv); 1524 sv_2mortal (sv);
1417 1525
1418 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1526 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1419 1527
… …
1432 if (dec->json.cb_object) 1540 if (dec->json.cb_object)
1433 { 1541 {
1434 dSP; 1542 dSP;
1435 int count; 1543 int count;
1436 1544
1437 ENTER; SAVETMPS; PUSHMARK (SP); 1545 ENTER; SAVETMPS;
1546 SAVESTACK_POS ();
1547 PUSHMARK (SP);
1438 XPUSHs (sv_2mortal (sv)); 1548 XPUSHs (sv_2mortal (sv));
1439 1549
1440 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1550 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1441 1551
1442 if (count == 1) 1552 if (count == 1)
… …
1458 DEC_DEC_DEPTH; 1568 DEC_DEC_DEPTH;
1459 return 0; 1569 return 0;
1460} 1570}
1461 1571
1462static SV * 1572static SV *
1573decode_tag (dec_t *dec)
1574{
1575 SV *tag = 0;
1576 SV *val = 0;
1577
1578 if (!(dec->json.flags & F_ALLOW_TAGS))
1579 ERR ("malformed JSON string, neither array, object, number, string or atom");
1580
1581 ++dec->cur;
1582
1583 decode_ws (dec);
1584
1585 tag = decode_sv (dec);
1586 if (!tag)
1587 goto fail;
1588
1589 if (!SvPOK (tag))
1590 ERR ("malformed JSON string, (tag) must be a string");
1591
1592 decode_ws (dec);
1593
1594 if (*dec->cur != ')')
1595 ERR (") expected after tag");
1596
1597 ++dec->cur;
1598
1599 decode_ws (dec);
1600
1601 val = decode_sv (dec);
1602 if (!val)
1603 goto fail;
1604
1605 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1606 ERR ("malformed JSON string, tag value must be an array");
1607
1608 {
1609 AV *av = (AV *)SvRV (val);
1610 int i, len = av_len (av) + 1;
1611 HV *stash = gv_stashsv (tag, 0);
1612 SV *sv;
1613
1614 if (!stash)
1615 ERR ("cannot decode perl-object (package does not exist)");
1616
1617 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1618
1619 if (!method)
1620 ERR ("cannot decode perl-object (package does not have a THAW method)");
1621
1622 dSP;
1623
1624 ENTER; SAVETMPS;
1625 PUSHMARK (SP);
1626 EXTEND (SP, len + 2);
1627 // we re-bless the reference to get overload and other niceties right
1628 PUSHs (tag);
1629 PUSHs (sv_json);
1630
1631 for (i = 0; i < len; ++i)
1632 PUSHs (*av_fetch (av, i, 1));
1633
1634 PUTBACK;
1635 call_sv ((SV *)GvCV (method), G_SCALAR);
1636 SPAGAIN;
1637
1638 SvREFCNT_dec (tag);
1639 SvREFCNT_dec (val);
1640 sv = SvREFCNT_inc (POPs);
1641
1642 PUTBACK;
1643
1644 FREETMPS; LEAVE;
1645
1646 return sv;
1647 }
1648
1649fail:
1650 SvREFCNT_dec (tag);
1651 SvREFCNT_dec (val);
1652 return 0;
1653}
1654
1655static SV *
1463decode_sv (dec_t *dec) 1656decode_sv (dec_t *dec)
1464{ 1657{
1465 // the beauty of JSON: you need exactly one character lookahead 1658 // the beauty of JSON: you need exactly one character lookahead
1466 // to parse everything. 1659 // to parse everything.
1467 switch (*dec->cur) 1660 switch (*dec->cur)
1468 { 1661 {
1469 case '"': ++dec->cur; return decode_str (dec); 1662 case '"': ++dec->cur; return decode_str (dec);
1470 case '[': ++dec->cur; return decode_av (dec); 1663 case '[': ++dec->cur; return decode_av (dec);
1471 case '{': ++dec->cur; return decode_hv (dec); 1664 case '{': ++dec->cur; return decode_hv (dec);
1665 case '(': return decode_tag (dec);
1472 1666
1473 case '-': 1667 case '-':
1474 case '0': case '1': case '2': case '3': case '4': 1668 case '0': case '1': case '2': case '3': case '4':
1475 case '5': case '6': case '7': case '8': case '9': 1669 case '5': case '6': case '7': case '8': case '9':
1476 return decode_num (dec); 1670 return decode_num (dec);
1477 1671
1672 case 'f':
1673 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1674 {
1675 dec->cur += 5;
1676
1677 if (expect_false (!dec->json.v_false))
1678 dec->json.v_false = GET_BOOL (false);
1679
1680 return newSVsv (dec->json.v_false);
1681 }
1682 else
1683 ERR ("'false' expected");
1684
1685 break;
1686
1478 case 't': 1687 case 't':
1479 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1688 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1480 { 1689 {
1481 dec->cur += 4; 1690 dec->cur += 4;
1482#if JSON_SLOW 1691
1483 json_true = get_bool ("JSON::XS::true"); 1692 if (expect_false (!dec->json.v_true))
1484#endif 1693 dec->json.v_true = GET_BOOL (true);
1694
1485 return newSVsv (json_true); 1695 return newSVsv (dec->json.v_true);
1486 } 1696 }
1487 else 1697 else
1488 ERR ("'true' expected"); 1698 ERR ("'true' expected");
1489
1490 break;
1491
1492 case 'f':
1493 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1494 {
1495 dec->cur += 5;
1496#if JSON_SLOW
1497 json_false = get_bool ("JSON::XS::false");
1498#endif
1499 return newSVsv (json_false);
1500 }
1501 else
1502 ERR ("'false' expected");
1503 1699
1504 break; 1700 break;
1505 1701
1506 case 'n': 1702 case 'n':
1507 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1703 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
… …
1513 ERR ("'null' expected"); 1709 ERR ("'null' expected");
1514 1710
1515 break; 1711 break;
1516 1712
1517 default: 1713 default:
1518 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1714 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1519 break; 1715 break;
1520 } 1716 }
1521 1717
1522fail: 1718fail:
1523 return 0; 1719 return 0;
1524} 1720}
1525 1721
1526static SV * 1722static SV *
1527decode_json (SV *string, JSON *json, char **offset_return) 1723decode_json (SV *string, JSON *json, STRLEN *offset_return)
1528{ 1724{
1529 dec_t dec; 1725 dec_t dec;
1530 SV *sv; 1726 SV *sv;
1531 1727
1532 /* work around bugs in 5.10 where manipulating magic values 1728 /* work around bugs in 5.10 where manipulating magic values
1533 * makes perl ignore the magic in subsequent accesses. 1729 * makes perl ignore the magic in subsequent accesses.
1534 * also make a copy of non-PV values, to get them into a clean 1730 * also make a copy of non-PV values, to get them into a clean
1535 * state (SvPV should do that, but it's buggy, see below). 1731 * state (SvPV should do that, but it's buggy, see below).
1732 *
1733 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1734 * as reported by Reini Urban.
1536 */ 1735 */
1537 /*SvGETMAGIC (string);*/ 1736 /*SvGETMAGIC (string);*/
1538 if (SvMAGICAL (string) || !SvPOK (string)) 1737 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1539 string = sv_2mortal (newSVsv (string)); 1738 string = sv_2mortal (newSVsv (string));
1540 1739
1541 SvUPGRADE (string, SVt_PV); 1740 SvUPGRADE (string, SVt_PV);
1542 1741
1543 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1742 /* work around a bug in perl 5.10, which causes SvCUR to fail an
… …
1582 1781
1583 decode_ws (&dec); 1782 decode_ws (&dec);
1584 sv = decode_sv (&dec); 1783 sv = decode_sv (&dec);
1585 1784
1586 if (offset_return) 1785 if (offset_return)
1587 *offset_return = dec.cur; 1786 *offset_return = dec.cur - SvPVX (string);
1588 1787 else if (sv)
1589 if (!(offset_return || !sv))
1590 { 1788 {
1591 // check for trailing garbage 1789 // check for trailing garbage
1592 decode_ws (&dec); 1790 decode_ws (&dec);
1593 1791
1594 if (*dec.cur) 1792 if (*dec.cur)
… …
1618 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1816 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1619 } 1817 }
1620 1818
1621 sv = sv_2mortal (sv); 1819 sv = sv_2mortal (sv);
1622 1820
1623 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1821 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1624 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1822 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1625 1823
1626 return sv; 1824 return sv;
1627} 1825}
1628 1826
… …
1750 self->incr_mode = INCR_M_STR; 1948 self->incr_mode = INCR_M_STR;
1751 goto incr_m_str; 1949 goto incr_m_str;
1752 1950
1753 case '[': 1951 case '[':
1754 case '{': 1952 case '{':
1953 case '(':
1755 if (++self->incr_nest > self->max_depth) 1954 if (++self->incr_nest > self->max_depth)
1756 croak (ERR_NESTING_EXCEEDED); 1955 croak (ERR_NESTING_EXCEEDED);
1757 break; 1956 break;
1758 1957
1759 case ']': 1958 case ']':
1760 case '}': 1959 case '}':
1761 if (--self->incr_nest <= 0) 1960 if (--self->incr_nest <= 0)
1762 goto interrupt; 1961 goto interrupt;
1962 break;
1963
1964 case ')':
1965 --self->incr_nest;
1763 break; 1966 break;
1764 1967
1765 case '#': 1968 case '#':
1766 self->incr_mode = INCR_M_C1; 1969 self->incr_mode = INCR_M_C1;
1767 goto incr_m_c; 1970 goto incr_m_c;
… …
1793 i >= '0' && i <= '9' ? i - '0' 1996 i >= '0' && i <= '9' ? i - '0'
1794 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1997 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1795 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1998 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1796 : -1; 1999 : -1;
1797 2000
1798 json_stash = gv_stashpv ("JSON::XS" , 1); 2001 json_stash = gv_stashpv ("JSON::XS" , 1);
1799 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 2002 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1800
1801 json_true = get_bool ("JSON::XS::true");
1802 json_false = get_bool ("JSON::XS::false"); 2003 bool_false = get_bool ("Types::Serialiser::false");
2004 bool_true = get_bool ("Types::Serialiser::true");
2005
2006 sv_json = newSVpv ("JSON", 0);
2007 SvREADONLY_on (sv_json);
1803 2008
1804 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ 2009 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1805} 2010}
1806 2011
1807PROTOTYPES: DISABLE 2012PROTOTYPES: DISABLE
1808 2013
1809void CLONE (...) 2014void CLONE (...)
1810 CODE: 2015 CODE:
2016 // as long as these writes are atomic, the race should not matter
2017 // as existing threads either already use 0, or use the old value,
2018 // which is sitll correct for the initial thread.
1811 json_stash = 0; 2019 json_stash = 0;
1812 json_boolean_stash = 0; 2020 bool_stash = 0;
2021 bool_false = 0;
2022 bool_true = 0;
1813 2023
1814void new (char *klass) 2024void new (char *klass)
1815 PPCODE: 2025 PPCODE:
1816{ 2026{
1817 SV *pv = NEWSV (0, sizeof (JSON)); 2027 SV *pv = NEWSV (0, sizeof (JSON));
… …
1820 XPUSHs (sv_2mortal (sv_bless ( 2030 XPUSHs (sv_2mortal (sv_bless (
1821 newRV_noinc (pv), 2031 newRV_noinc (pv),
1822 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2032 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1823 ))); 2033 )));
1824} 2034}
2035
2036void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2037 PPCODE:
2038 self->v_false = newSVsv (v_false);
2039 self->v_true = newSVsv (v_true);
2040 XPUSHs (ST (0));
2041
2042void get_boolean_values (JSON *self)
2043 PPCODE:
2044 if (self->v_false && self->v_true)
2045 {
2046 EXTEND (SP, 2);
2047 PUSHs (self->v_false);
2048 PUSHs (self->v_true);
2049 }
1825 2050
1826void ascii (JSON *self, int enable = 1) 2051void ascii (JSON *self, int enable = 1)
1827 ALIAS: 2052 ALIAS:
1828 ascii = F_ASCII 2053 ascii = F_ASCII
1829 latin1 = F_LATIN1 2054 latin1 = F_LATIN1
… …
1837 shrink = F_SHRINK 2062 shrink = F_SHRINK
1838 allow_blessed = F_ALLOW_BLESSED 2063 allow_blessed = F_ALLOW_BLESSED
1839 convert_blessed = F_CONV_BLESSED 2064 convert_blessed = F_CONV_BLESSED
1840 relaxed = F_RELAXED 2065 relaxed = F_RELAXED
1841 allow_unknown = F_ALLOW_UNKNOWN 2066 allow_unknown = F_ALLOW_UNKNOWN
2067 allow_tags = F_ALLOW_TAGS
1842 PPCODE: 2068 PPCODE:
1843{ 2069{
1844 if (enable) 2070 if (enable)
1845 self->flags |= ix; 2071 self->flags |= ix;
1846 else 2072 else
… …
1862 get_shrink = F_SHRINK 2088 get_shrink = F_SHRINK
1863 get_allow_blessed = F_ALLOW_BLESSED 2089 get_allow_blessed = F_ALLOW_BLESSED
1864 get_convert_blessed = F_CONV_BLESSED 2090 get_convert_blessed = F_CONV_BLESSED
1865 get_relaxed = F_RELAXED 2091 get_relaxed = F_RELAXED
1866 get_allow_unknown = F_ALLOW_UNKNOWN 2092 get_allow_unknown = F_ALLOW_UNKNOWN
2093 get_allow_tags = F_ALLOW_TAGS
1867 PPCODE: 2094 PPCODE:
1868 XPUSHs (boolSV (self->flags & ix)); 2095 XPUSHs (boolSV (self->flags & ix));
1869 2096
1870void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 2097void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1871 PPCODE: 2098 PPCODE:
… …
1932 2159
1933void decode_prefix (JSON *self, SV *jsonstr) 2160void decode_prefix (JSON *self, SV *jsonstr)
1934 PPCODE: 2161 PPCODE:
1935{ 2162{
1936 SV *sv; 2163 SV *sv;
1937 char *offset; 2164 STRLEN offset;
1938 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; 2165 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1939 EXTEND (SP, 2); 2166 EXTEND (SP, 2);
1940 PUSHs (sv); 2167 PUSHs (sv);
1941 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2168 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
1942} 2169}
1943 2170
1944void incr_parse (JSON *self, SV *jsonstr = 0) 2171void incr_parse (JSON *self, SV *jsonstr = 0)
1945 PPCODE: 2172 PPCODE:
1946{ 2173{
… …
1980 { 2207 {
1981 STRLEN len; 2208 STRLEN len;
1982 const char *str = SvPV (jsonstr, len); 2209 const char *str = SvPV (jsonstr, len);
1983 STRLEN cur = SvCUR (self->incr_text); 2210 STRLEN cur = SvCUR (self->incr_text);
1984 2211
1985 if (SvLEN (self->incr_text) <= cur + len) 2212 if (SvLEN (self->incr_text) - cur <= len)
1986 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2213 json_sv_grow (self->incr_text, cur, len);
1987 2214
1988 Move (str, SvEND (self->incr_text), len, char); 2215 Move (str, SvEND (self->incr_text), len, char);
1989 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2216 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1990 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 2217 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1991 } 2218 }
… …
1993 2220
1994 if (GIMME_V != G_VOID) 2221 if (GIMME_V != G_VOID)
1995 do 2222 do
1996 { 2223 {
1997 SV *sv; 2224 SV *sv;
1998 char *offset; 2225 STRLEN offset;
1999 2226
2000 if (!INCR_DONE (self)) 2227 if (!INCR_DONE (self))
2001 { 2228 {
2002 incr_parse (self); 2229 incr_parse (self);
2003 2230
… …
2019 } 2246 }
2020 2247
2021 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; 2248 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2022 XPUSHs (sv); 2249 XPUSHs (sv);
2023 2250
2024 self->incr_pos -= offset - SvPVX (self->incr_text); 2251 self->incr_pos -= offset;
2025 self->incr_nest = 0; 2252 self->incr_nest = 0;
2026 self->incr_mode = 0; 2253 self->incr_mode = 0;
2027 2254
2028 sv_chop (self->incr_text, offset); 2255 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2029 } 2256 }
2030 while (GIMME_V == G_ARRAY); 2257 while (GIMME_V == G_ARRAY);
2031} 2258}
2032 2259
2033SV *incr_text (JSON *self) 2260SV *incr_text (JSON *self)
… …
2064 self->incr_mode = 0; 2291 self->incr_mode = 0;
2065} 2292}
2066 2293
2067void DESTROY (JSON *self) 2294void DESTROY (JSON *self)
2068 CODE: 2295 CODE:
2296 SvREFCNT_dec (self->v_false);
2297 SvREFCNT_dec (self->v_true);
2069 SvREFCNT_dec (self->cb_sk_object); 2298 SvREFCNT_dec (self->cb_sk_object);
2070 SvREFCNT_dec (self->cb_object); 2299 SvREFCNT_dec (self->cb_object);
2071 SvREFCNT_dec (self->incr_text); 2300 SvREFCNT_dec (self->incr_text);
2072 2301
2073PROTOTYPES: ENABLE 2302PROTOTYPES: ENABLE
2074 2303
2075void encode_json (SV *scalar) 2304void encode_json (SV *scalar)
2076 ALIAS:
2077 to_json_ = 0
2078 encode_json = F_UTF8
2079 PPCODE: 2305 PPCODE:
2080{ 2306{
2081 JSON json; 2307 JSON json;
2082 json_init (&json); 2308 json_init (&json);
2083 json.flags |= ix; 2309 json.flags |= F_UTF8;
2084 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN; 2310 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2085 XPUSHs (scalar); 2311 XPUSHs (scalar);
2086} 2312}
2087 2313
2088void decode_json (SV *jsonstr) 2314void decode_json (SV *jsonstr)
2089 ALIAS:
2090 from_json_ = 0
2091 decode_json = F_UTF8
2092 PPCODE: 2315 PPCODE:
2093{ 2316{
2094 JSON json; 2317 JSON json;
2095 json_init (&json); 2318 json_init (&json);
2096 json.flags |= ix; 2319 json.flags |= F_UTF8;
2097 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN; 2320 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2098 XPUSHs (jsonstr); 2321 XPUSHs (jsonstr);
2099} 2322}
2100 2323

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines