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 1.118 by root, Fri Oct 25 20:27:57 2013 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
46 59
47#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) 60#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
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{
112 Zero (json, 1, JSON); 130 static const JSON init = { F_ALLOW_NONREF, 512 };
113 json->max_depth = 512; 131
132 *json = init;
114} 133}
115 134
116///////////////////////////////////////////////////////////////////////////// 135/////////////////////////////////////////////////////////////////////////////
117// utility functions 136// utility functions
118 137
140 SvPV_renew (sv, SvCUR (sv) + 1); 159 SvPV_renew (sv, SvCUR (sv) + 1);
141#endif 160#endif
142 } 161 }
143} 162}
144 163
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
145// 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
146// case of an error. 190// case of an error.
147// we special-case "safe" characters from U+80 .. U+7FF, 191// we special-case "safe" characters from U+80 .. U+7FF,
148// but use the very good perl function to parse anything else. 192// but use the very good perl function to parse anything else.
149// note that we never call this function for a ascii codepoints 193// note that we never call this function for a ascii codepoints
150INLINE UV 194INLINE UV
287 // a recursion depth of ten gives us >>500 bits 331 // a recursion depth of ten gives us >>500 bits
288 json_atof_scan1 (s, &accum, &expo, 0, 10); 332 json_atof_scan1 (s, &accum, &expo, 0, 10);
289 333
290 return neg ? -accum : accum; 334 return neg ? -accum : accum;
291} 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
292///////////////////////////////////////////////////////////////////////////// 379/////////////////////////////////////////////////////////////////////////////
293// encoder 380// encoder
294 381
295// structure used for encoding JSON 382// structure used for encoding JSON
296typedef struct 383typedef struct
304} enc_t; 391} enc_t;
305 392
306INLINE void 393INLINE void
307need (enc_t *enc, STRLEN len) 394need (enc_t *enc, STRLEN len)
308{ 395{
309 if (expect_false (enc->cur + len >= enc->end)) 396 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
310 { 397 {
311 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 398 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
312 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 399 char *buf = json_sv_grow (enc->sv, cur, len);
313 enc->cur = SvPVX (enc->sv) + cur; 400 enc->cur = buf + cur;
314 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 401 enc->end = buf + SvLEN (enc->sv) - 1;
315 } 402 }
316} 403}
317 404
318INLINE void 405INLINE void
319encode_ch (enc_t *enc, char ch) 406encode_ch (enc_t *enc, char ch)
335 422
336 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 423 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
337 { 424 {
338 if (expect_false (ch == '"')) // but with slow exceptions 425 if (expect_false (ch == '"')) // but with slow exceptions
339 { 426 {
340 need (enc, len += 1); 427 need (enc, len + 1);
341 *enc->cur++ = '\\'; 428 *enc->cur++ = '\\';
342 *enc->cur++ = '"'; 429 *enc->cur++ = '"';
343 } 430 }
344 else if (expect_false (ch == '\\')) 431 else if (expect_false (ch == '\\'))
345 { 432 {
346 need (enc, len += 1); 433 need (enc, len + 1);
347 *enc->cur++ = '\\'; 434 *enc->cur++ = '\\';
348 *enc->cur++ = '\\'; 435 *enc->cur++ = '\\';
349 } 436 }
350 else 437 else
351 *enc->cur++ = ch; 438 *enc->cur++ = ch;
354 } 441 }
355 else 442 else
356 { 443 {
357 switch (ch) 444 switch (ch)
358 { 445 {
359 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;
360 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;
361 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;
362 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;
363 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;
364 451
365 default: 452 default:
366 { 453 {
367 STRLEN clen; 454 STRLEN clen;
368 UV uch; 455 UV uch;
384 if (uch >= 0x10000UL) 471 if (uch >= 0x10000UL)
385 { 472 {
386 if (uch >= 0x110000UL) 473 if (uch >= 0x110000UL)
387 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);
388 475
389 need (enc, len += 11); 476 need (enc, len + 11);
390 sprintf (enc->cur, "\\u%04x\\u%04x", 477 sprintf (enc->cur, "\\u%04x\\u%04x",
391 (int)((uch - 0x10000) / 0x400 + 0xD800), 478 (int)((uch - 0x10000) / 0x400 + 0xD800),
392 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 479 (int)((uch - 0x10000) % 0x400 + 0xDC00));
393 enc->cur += 12; 480 enc->cur += 12;
394 } 481 }
395 else 482 else
396 { 483 {
397 need (enc, len += 5); 484 need (enc, len + 5);
398 *enc->cur++ = '\\'; 485 *enc->cur++ = '\\';
399 *enc->cur++ = 'u'; 486 *enc->cur++ = 'u';
400 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 487 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
401 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 488 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
402 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 489 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
410 *enc->cur++ = uch; 497 *enc->cur++ = uch;
411 str += clen; 498 str += clen;
412 } 499 }
413 else if (is_utf8) 500 else if (is_utf8)
414 { 501 {
415 need (enc, len += clen); 502 need (enc, len + clen);
416 do 503 do
417 { 504 {
418 *enc->cur++ = *str++; 505 *enc->cur++ = *str++;
419 } 506 }
420 while (--clen); 507 while (--clen);
421 } 508 }
422 else 509 else
423 { 510 {
424 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
425 enc->cur = encode_utf8 (enc->cur, uch); 512 enc->cur = encode_utf8 (enc->cur, uch);
426 ++str; 513 ++str;
427 } 514 }
428 } 515 }
429 } 516 }
682// encode objects, arrays and special \0=false and \1=true values. 769// encode objects, arrays and special \0=false and \1=true values.
683static void 770static void
684encode_rv (enc_t *enc, SV *sv) 771encode_rv (enc_t *enc, SV *sv)
685{ 772{
686 svtype svt; 773 svtype svt;
774 GV *method;
687 775
688 SvGETMAGIC (sv); 776 SvGETMAGIC (sv);
689 svt = SvTYPE (sv); 777 svt = SvTYPE (sv);
690 778
691 if (expect_false (SvOBJECT (sv))) 779 if (expect_false (SvOBJECT (sv)))
692 { 780 {
693 HV *stash = !JSON_SLOW || json_boolean_stash 781 HV *stash = SvSTASH (sv);
694 ? json_boolean_stash
695 : gv_stashpv ("JSON::XS::Boolean", 1);
696 782
697 if (SvSTASH (sv) == stash) 783 if (stash == bool_stash)
698 { 784 {
699 if (SvIV (sv))
700 encode_str (enc, "true", 4, 0); 785 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
701 else
702 encode_str (enc, "false", 5, 0); 786 else encode_str (enc, "false", 5, 0);
703 } 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);
704 else 860 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", 861 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)))); 862 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
751 }
752 } 863 }
753 else if (svt == SVt_PVHV) 864 else if (svt == SVt_PVHV)
754 encode_hv (enc, (HV *)sv); 865 encode_hv (enc, (HV *)sv);
755 else if (svt == SVt_PVAV) 866 else if (svt == SVt_PVAV)
756 encode_av (enc, (AV *)sv); 867 encode_av (enc, (AV *)sv);
757 else if (svt < SVt_PVAV) 868 else if (svt < SVt_PVAV)
758 { 869 {
759 STRLEN len = 0; 870 int bool_type = ref_bool_type (sv);
760 char *pv = svt ? SvPV (sv, len) : 0;
761 871
762 if (len == 1 && *pv == '1') 872 if (bool_type == 1)
763 encode_str (enc, "true", 4, 0); 873 encode_str (enc, "true", 4, 0);
764 else if (len == 1 && *pv == '0') 874 else if (bool_type == 0)
765 encode_str (enc, "false", 5, 0); 875 encode_str (enc, "false", 5, 0);
766 else if (enc->json.flags & F_ALLOW_UNKNOWN) 876 else if (enc->json.flags & F_ALLOW_UNKNOWN)
767 encode_str (enc, "null", 4, 0); 877 encode_str (enc, "null", 4, 0);
768 else 878 else
769 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",
852static SV * 962static SV *
853encode_json (SV *scalar, JSON *json) 963encode_json (SV *scalar, JSON *json)
854{ 964{
855 enc_t enc; 965 enc_t enc;
856 966
857 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 967 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)"); 968 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
859 969
860 enc.json = *json; 970 enc.json = *json;
861 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 971 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
862 enc.cur = SvPVX (enc.sv); 972 enc.cur = SvPVX (enc.sv);
1069 *cur++ = *dec_cur++; 1179 *cur++ = *dec_cur++;
1070 while (--clen); 1180 while (--clen);
1071 1181
1072 utf8 = 1; 1182 utf8 = 1;
1073 } 1183 }
1184 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1185 *cur++ = ch;
1074 else 1186 else
1075 { 1187 {
1076 --dec_cur; 1188 --dec_cur;
1077 1189
1078 if (!ch) 1190 if (!ch)
1088 1200
1089 if (sv) 1201 if (sv)
1090 { 1202 {
1091 STRLEN cur = SvCUR (sv); 1203 STRLEN cur = SvCUR (sv);
1092 1204
1093 if (SvLEN (sv) <= cur + len) 1205 if (SvLEN (sv) - cur <= len)
1094 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 1206 json_sv_grow (sv, cur, len);
1095 1207
1096 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1208 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1097 SvCUR_set (sv, SvCUR (sv) + len); 1209 SvCUR_set (sv, SvCUR (sv) + len);
1098 } 1210 }
1099 else 1211 else
1400 1512
1401 hv_iterinit (hv); 1513 hv_iterinit (hv);
1402 he = hv_iternext (hv); 1514 he = hv_iternext (hv);
1403 hv_iterinit (hv); 1515 hv_iterinit (hv);
1404 1516
1405 // the next line creates a mortal sv each time its called. 1517 // the next line creates a mortal sv each time it's called.
1406 // might want to optimise this for common cases. 1518 // might want to optimise this for common cases.
1407 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);
1408 1520
1409 if (cb) 1521 if (cb)
1410 { 1522 {
1411 dSP; 1523 dSP;
1412 int count; 1524 int count;
1413 1525
1414 ENTER; SAVETMPS; PUSHMARK (SP); 1526 ENTER; SAVETMPS;
1527 PUSHMARK (SP);
1415 XPUSHs (HeVAL (he)); 1528 XPUSHs (HeVAL (he));
1416 sv_2mortal (sv); 1529 sv_2mortal (sv);
1417 1530
1418 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1531 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1419 1532
1421 { 1534 {
1422 sv = newSVsv (POPs); 1535 sv = newSVsv (POPs);
1423 FREETMPS; LEAVE; 1536 FREETMPS; LEAVE;
1424 return sv; 1537 return sv;
1425 } 1538 }
1539 else if (count)
1540 croak ("filter_json_single_key_object callbacks must not return more than one scalar");
1426 1541
1427 SvREFCNT_inc (sv); 1542 SvREFCNT_inc (sv);
1428 FREETMPS; LEAVE; 1543 FREETMPS; LEAVE;
1429 } 1544 }
1430 } 1545 }
1432 if (dec->json.cb_object) 1547 if (dec->json.cb_object)
1433 { 1548 {
1434 dSP; 1549 dSP;
1435 int count; 1550 int count;
1436 1551
1437 ENTER; SAVETMPS; PUSHMARK (SP); 1552 ENTER; SAVETMPS;
1553 PUSHMARK (SP);
1438 XPUSHs (sv_2mortal (sv)); 1554 XPUSHs (sv_2mortal (sv));
1439 1555
1440 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1556 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1441 1557
1442 if (count == 1) 1558 if (count == 1)
1443 { 1559 {
1444 sv = newSVsv (POPs); 1560 sv = newSVsv (POPs);
1445 FREETMPS; LEAVE; 1561 FREETMPS; LEAVE;
1446 return sv; 1562 return sv;
1447 } 1563 }
1564 else if (count)
1565 croak ("filter_json_object callbacks must not return more than one scalar");
1448 1566
1449 SvREFCNT_inc (sv); 1567 SvREFCNT_inc (sv);
1450 FREETMPS; LEAVE; 1568 FREETMPS; LEAVE;
1451 } 1569 }
1452 } 1570 }
1458 DEC_DEC_DEPTH; 1576 DEC_DEC_DEPTH;
1459 return 0; 1577 return 0;
1460} 1578}
1461 1579
1462static 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 *
1463decode_sv (dec_t *dec) 1664decode_sv (dec_t *dec)
1464{ 1665{
1465 // the beauty of JSON: you need exactly one character lookahead 1666 // the beauty of JSON: you need exactly one character lookahead
1466 // to parse everything. 1667 // to parse everything.
1467 switch (*dec->cur) 1668 switch (*dec->cur)
1468 { 1669 {
1469 case '"': ++dec->cur; return decode_str (dec); 1670 case '"': ++dec->cur; return decode_str (dec);
1470 case '[': ++dec->cur; return decode_av (dec); 1671 case '[': ++dec->cur; return decode_av (dec);
1471 case '{': ++dec->cur; return decode_hv (dec); 1672 case '{': ++dec->cur; return decode_hv (dec);
1673 case '(': return decode_tag (dec);
1472 1674
1473 case '-': 1675 case '-':
1474 case '0': case '1': case '2': case '3': case '4': 1676 case '0': case '1': case '2': case '3': case '4':
1475 case '5': case '6': case '7': case '8': case '9': 1677 case '5': case '6': case '7': case '8': case '9':
1476 return decode_num (dec); 1678 return decode_num (dec);
1477 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
1478 case 't': 1695 case 't':
1479 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1696 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1480 { 1697 {
1481 dec->cur += 4; 1698 dec->cur += 4;
1482#if JSON_SLOW 1699
1483 json_true = get_bool ("JSON::XS::true"); 1700 if (expect_false (!dec->json.v_true))
1484#endif 1701 dec->json.v_true = GET_BOOL (true);
1702
1485 return newSVsv (json_true); 1703 return newSVsv (dec->json.v_true);
1486 } 1704 }
1487 else 1705 else
1488 ERR ("'true' expected"); 1706 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 1707
1504 break; 1708 break;
1505 1709
1506 case 'n': 1710 case 'n':
1507 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1711 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
1513 ERR ("'null' expected"); 1717 ERR ("'null' expected");
1514 1718
1515 break; 1719 break;
1516 1720
1517 default: 1721 default:
1518 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1722 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1519 break; 1723 break;
1520 } 1724 }
1521 1725
1522fail: 1726fail:
1523 return 0; 1727 return 0;
1524} 1728}
1525 1729
1526static SV * 1730static SV *
1527decode_json (SV *string, JSON *json, char **offset_return) 1731decode_json (SV *string, JSON *json, STRLEN *offset_return)
1528{ 1732{
1529 dec_t dec; 1733 dec_t dec;
1530 SV *sv; 1734 SV *sv;
1531 1735
1532 /* work around bugs in 5.10 where manipulating magic values 1736 /* work around bugs in 5.10 where manipulating magic values
1533 * makes perl ignore the magic in subsequent accesses. 1737 * makes perl ignore the magic in subsequent accesses.
1534 * also make a copy of non-PV values, to get them into a clean 1738 * 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). 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.
1536 */ 1743 */
1537 /*SvGETMAGIC (string);*/ 1744 /*SvGETMAGIC (string);*/
1538 if (SvMAGICAL (string) || !SvPOK (string)) 1745 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1539 string = sv_2mortal (newSVsv (string)); 1746 string = sv_2mortal (newSVsv (string));
1540 1747
1541 SvUPGRADE (string, SVt_PV); 1748 SvUPGRADE (string, SVt_PV);
1542 1749
1543 /* 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
1582 1789
1583 decode_ws (&dec); 1790 decode_ws (&dec);
1584 sv = decode_sv (&dec); 1791 sv = decode_sv (&dec);
1585 1792
1586 if (offset_return) 1793 if (offset_return)
1587 *offset_return = dec.cur; 1794 *offset_return = dec.cur - SvPVX (string);
1588 1795 else if (sv)
1589 if (!(offset_return || !sv))
1590 { 1796 {
1591 // check for trailing garbage 1797 // check for trailing garbage
1592 decode_ws (&dec); 1798 decode_ws (&dec);
1593 1799
1594 if (*dec.cur) 1800 if (dec.cur != dec.end)
1595 { 1801 {
1596 dec.err = "garbage after JSON object"; 1802 dec.err = "garbage after JSON object";
1597 SvREFCNT_dec (sv); 1803 SvREFCNT_dec (sv);
1598 sv = 0; 1804 sv = 0;
1599 } 1805 }
1618 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1824 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1619 } 1825 }
1620 1826
1621 sv = sv_2mortal (sv); 1827 sv = sv_2mortal (sv);
1622 1828
1623 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1829 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)"); 1830 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1625 1831
1626 return sv; 1832 return sv;
1627} 1833}
1628 1834
1750 self->incr_mode = INCR_M_STR; 1956 self->incr_mode = INCR_M_STR;
1751 goto incr_m_str; 1957 goto incr_m_str;
1752 1958
1753 case '[': 1959 case '[':
1754 case '{': 1960 case '{':
1961 case '(':
1755 if (++self->incr_nest > self->max_depth) 1962 if (++self->incr_nest > self->max_depth)
1756 croak (ERR_NESTING_EXCEEDED); 1963 croak (ERR_NESTING_EXCEEDED);
1757 break; 1964 break;
1758 1965
1759 case ']': 1966 case ']':
1760 case '}': 1967 case '}':
1761 if (--self->incr_nest <= 0) 1968 if (--self->incr_nest <= 0)
1762 goto interrupt; 1969 goto interrupt;
1970 break;
1971
1972 case ')':
1973 --self->incr_nest;
1763 break; 1974 break;
1764 1975
1765 case '#': 1976 case '#':
1766 self->incr_mode = INCR_M_C1; 1977 self->incr_mode = INCR_M_C1;
1767 goto incr_m_c; 1978 goto incr_m_c;
1793 i >= '0' && i <= '9' ? i - '0' 2004 i >= '0' && i <= '9' ? i - '0'
1794 : i >= 'a' && i <= 'f' ? i - 'a' + 10 2005 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1795 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2006 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1796 : -1; 2007 : -1;
1797 2008
1798 json_stash = gv_stashpv ("JSON::XS" , 1); 2009 json_stash = gv_stashpv ("JSON::XS" , 1);
1799 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 2010 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"); 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);
1803 2016
1804 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 */
1805} 2018}
1806 2019
1807PROTOTYPES: DISABLE 2020PROTOTYPES: DISABLE
1808 2021
1809void CLONE (...) 2022void CLONE (...)
1810 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.
1811 json_stash = 0; 2027 json_stash = 0;
1812 json_boolean_stash = 0; 2028 bool_stash = 0;
2029 bool_false = 0;
2030 bool_true = 0;
1813 2031
1814void new (char *klass) 2032void new (char *klass)
1815 PPCODE: 2033 PPCODE:
1816{ 2034{
1817 SV *pv = NEWSV (0, sizeof (JSON)); 2035 SV *pv = NEWSV (0, sizeof (JSON));
1820 XPUSHs (sv_2mortal (sv_bless ( 2038 XPUSHs (sv_2mortal (sv_bless (
1821 newRV_noinc (pv), 2039 newRV_noinc (pv),
1822 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2040 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1823 ))); 2041 )));
1824} 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 }
1825 2058
1826void ascii (JSON *self, int enable = 1) 2059void ascii (JSON *self, int enable = 1)
1827 ALIAS: 2060 ALIAS:
1828 ascii = F_ASCII 2061 ascii = F_ASCII
1829 latin1 = F_LATIN1 2062 latin1 = F_LATIN1
1837 shrink = F_SHRINK 2070 shrink = F_SHRINK
1838 allow_blessed = F_ALLOW_BLESSED 2071 allow_blessed = F_ALLOW_BLESSED
1839 convert_blessed = F_CONV_BLESSED 2072 convert_blessed = F_CONV_BLESSED
1840 relaxed = F_RELAXED 2073 relaxed = F_RELAXED
1841 allow_unknown = F_ALLOW_UNKNOWN 2074 allow_unknown = F_ALLOW_UNKNOWN
2075 allow_tags = F_ALLOW_TAGS
1842 PPCODE: 2076 PPCODE:
1843{ 2077{
1844 if (enable) 2078 if (enable)
1845 self->flags |= ix; 2079 self->flags |= ix;
1846 else 2080 else
1862 get_shrink = F_SHRINK 2096 get_shrink = F_SHRINK
1863 get_allow_blessed = F_ALLOW_BLESSED 2097 get_allow_blessed = F_ALLOW_BLESSED
1864 get_convert_blessed = F_CONV_BLESSED 2098 get_convert_blessed = F_CONV_BLESSED
1865 get_relaxed = F_RELAXED 2099 get_relaxed = F_RELAXED
1866 get_allow_unknown = F_ALLOW_UNKNOWN 2100 get_allow_unknown = F_ALLOW_UNKNOWN
2101 get_allow_tags = F_ALLOW_TAGS
1867 PPCODE: 2102 PPCODE:
1868 XPUSHs (boolSV (self->flags & ix)); 2103 XPUSHs (boolSV (self->flags & ix));
1869 2104
1870void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 2105void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1871 PPCODE: 2106 PPCODE:
1932 2167
1933void decode_prefix (JSON *self, SV *jsonstr) 2168void decode_prefix (JSON *self, SV *jsonstr)
1934 PPCODE: 2169 PPCODE:
1935{ 2170{
1936 SV *sv; 2171 SV *sv;
1937 char *offset; 2172 STRLEN offset;
1938 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; 2173 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1939 EXTEND (SP, 2); 2174 EXTEND (SP, 2);
1940 PUSHs (sv); 2175 PUSHs (sv);
1941 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2176 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
1942} 2177}
1943 2178
1944void incr_parse (JSON *self, SV *jsonstr = 0) 2179void incr_parse (JSON *self, SV *jsonstr = 0)
1945 PPCODE: 2180 PPCODE:
1946{ 2181{
1980 { 2215 {
1981 STRLEN len; 2216 STRLEN len;
1982 const char *str = SvPV (jsonstr, len); 2217 const char *str = SvPV (jsonstr, len);
1983 STRLEN cur = SvCUR (self->incr_text); 2218 STRLEN cur = SvCUR (self->incr_text);
1984 2219
1985 if (SvLEN (self->incr_text) <= cur + len) 2220 if (SvLEN (self->incr_text) - cur <= len)
1986 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2221 json_sv_grow (self->incr_text, cur, len);
1987 2222
1988 Move (str, SvEND (self->incr_text), len, char); 2223 Move (str, SvEND (self->incr_text), len, char);
1989 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2224 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 2225 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1991 } 2226 }
1993 2228
1994 if (GIMME_V != G_VOID) 2229 if (GIMME_V != G_VOID)
1995 do 2230 do
1996 { 2231 {
1997 SV *sv; 2232 SV *sv;
1998 char *offset; 2233 STRLEN offset;
1999 2234
2000 if (!INCR_DONE (self)) 2235 if (!INCR_DONE (self))
2001 { 2236 {
2002 incr_parse (self); 2237 incr_parse (self);
2003 2238
2019 } 2254 }
2020 2255
2021 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; 2256 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2022 XPUSHs (sv); 2257 XPUSHs (sv);
2023 2258
2024 self->incr_pos -= offset - SvPVX (self->incr_text); 2259 self->incr_pos -= offset;
2025 self->incr_nest = 0; 2260 self->incr_nest = 0;
2026 self->incr_mode = 0; 2261 self->incr_mode = 0;
2027 2262
2028 sv_chop (self->incr_text, offset); 2263 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2029 } 2264 }
2030 while (GIMME_V == G_ARRAY); 2265 while (GIMME_V == G_ARRAY);
2031} 2266}
2032 2267
2033SV *incr_text (JSON *self) 2268SV *incr_text (JSON *self)
2064 self->incr_mode = 0; 2299 self->incr_mode = 0;
2065} 2300}
2066 2301
2067void DESTROY (JSON *self) 2302void DESTROY (JSON *self)
2068 CODE: 2303 CODE:
2304 SvREFCNT_dec (self->v_false);
2305 SvREFCNT_dec (self->v_true);
2069 SvREFCNT_dec (self->cb_sk_object); 2306 SvREFCNT_dec (self->cb_sk_object);
2070 SvREFCNT_dec (self->cb_object); 2307 SvREFCNT_dec (self->cb_object);
2071 SvREFCNT_dec (self->incr_text); 2308 SvREFCNT_dec (self->incr_text);
2072 2309
2073PROTOTYPES: ENABLE 2310PROTOTYPES: ENABLE

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines