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 1.74 by root, Wed Mar 19 15:17:53 2008 UTC vs.
Revision 1.101 by root, Tue Sep 8 18:00:03 2009 UTC

4 4
5#include <assert.h> 5#include <assert.h>
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 <float.h> 10#include <float.h>
10 11
11#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
13#endif 14#endif
14 15
15// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
19#endif 20#endif
21
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
20 23
21#define F_ASCII 0x00000001UL 24#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL 25#define F_LATIN1 0x00000002UL
23#define F_UTF8 0x00000004UL 26#define F_UTF8 0x00000004UL
24#define F_INDENT 0x00000008UL 27#define F_INDENT 0x00000008UL
28#define F_ALLOW_NONREF 0x00000100UL 31#define F_ALLOW_NONREF 0x00000100UL
29#define F_SHRINK 0x00000200UL 32#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL 33#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL 34#define F_CONV_BLESSED 0x00000800UL
32#define F_RELAXED 0x00001000UL 35#define F_RELAXED 0x00001000UL
33 36#define F_ALLOW_UNKNOWN 0x00002000UL
34#define F_MAXDEPTH 0xf8000000UL
35#define S_MAXDEPTH 27
36#define F_MAXSIZE 0x01f00000UL
37#define S_MAXSIZE 20
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
39 38
40#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
41#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
42
43#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
44#define F_DEFAULT (9UL << S_MAXDEPTH)
45 40
46#define INIT_SIZE 32 // initial scalar size to be allocated 41#define INIT_SIZE 32 // initial scalar size to be allocated
47#define INDENT_STEP 3 // spaces per indentation level 42#define INDENT_STEP 3 // spaces per indentation level
48 43
49#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
64 59
65#define IN_RANGE_INC(type,val,beg,end) \ 60#define IN_RANGE_INC(type,val,beg,end) \
66 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 61 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
67 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 62 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
68 63
64#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
65
69#ifdef USE_ITHREADS 66#ifdef USE_ITHREADS
70# define JSON_SLOW 1 67# define JSON_SLOW 1
71# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 68# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
72#else 69#else
73# define JSON_SLOW 0 70# define JSON_SLOW 0
75#endif 72#endif
76 73
77static HV *json_stash, *json_boolean_stash; // JSON::XS:: 74static HV *json_stash, *json_boolean_stash; // JSON::XS::
78static SV *json_true, *json_false; 75static SV *json_true, *json_false;
79 76
77enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0
79 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash
81 INCR_M_JSON // outside anything, count nesting
82};
83
84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85
80typedef struct { 86typedef struct {
81 U32 flags; 87 U32 flags;
88 U32 max_depth;
89 STRLEN max_size;
90
82 SV *cb_object; 91 SV *cb_object;
83 HV *cb_sk_object; 92 HV *cb_sk_object;
93
94 // for the incremental parser
95 SV *incr_text; // the source text so far
96 STRLEN incr_pos; // the current offset into the text
97 int incr_nest; // {[]}-nesting level
98 unsigned char incr_mode;
84} JSON; 99} JSON;
100
101INLINE void
102json_init (JSON *json)
103{
104 Zero (json, 1, JSON);
105 json->max_depth = 512;
106}
85 107
86///////////////////////////////////////////////////////////////////////////// 108/////////////////////////////////////////////////////////////////////////////
87// utility functions 109// utility functions
88 110
111INLINE SV *
112get_bool (const char *name)
113{
114 SV *sv = get_sv (name, 1);
115
116 SvREADONLY_on (sv);
117 SvREADONLY_on (SvRV (sv));
118
119 return sv;
120}
121
89INLINE void 122INLINE void
90shrink (SV *sv) 123shrink (SV *sv)
91{ 124{
92 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
126
93 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
94 { 128 {
95#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
96 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
97#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
141 *s++ = 0x80 | ( ch & 0x3f); 175 *s++ = 0x80 | ( ch & 0x3f);
142 176
143 return s; 177 return s;
144} 178}
145 179
180// convert offset pointer to character index, sv must be string
181static STRLEN
182ptr_to_index (SV *sv, char *offset)
183{
184 return SvUTF8 (sv)
185 ? utf8_distance (offset, SvPVX (sv))
186 : offset - SvPVX (sv);
187}
188
146///////////////////////////////////////////////////////////////////////////// 189/////////////////////////////////////////////////////////////////////////////
147// encoder 190// encoder
148 191
149// structure used for encoding JSON 192// structure used for encoding JSON
150typedef struct 193typedef struct
152 char *cur; // SvPVX (sv) + current output position 195 char *cur; // SvPVX (sv) + current output position
153 char *end; // SvEND (sv) 196 char *end; // SvEND (sv)
154 SV *sv; // result scalar 197 SV *sv; // result scalar
155 JSON json; 198 JSON json;
156 U32 indent; // indentation level 199 U32 indent; // indentation level
157 U32 maxdepth; // max. indentation/recursion level
158 UV limit; // escape character values >= this value when encoding 200 UV limit; // escape character values >= this value when encoding
159} enc_t; 201} enc_t;
160 202
161INLINE void 203INLINE void
162need (enc_t *enc, STRLEN len) 204need (enc_t *enc, STRLEN len)
163{ 205{
164 if (expect_false (enc->cur + len >= enc->end)) 206 if (expect_false (enc->cur + len >= enc->end))
165 { 207 {
166 STRLEN cur = enc->cur - SvPVX (enc->sv); 208 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
167 SvGROW (enc->sv, cur + len + 1); 209 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
168 enc->cur = SvPVX (enc->sv) + cur; 210 enc->cur = SvPVX (enc->sv) + cur;
169 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 211 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
170 } 212 }
171} 213}
172 214
247 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 289 (int)((uch - 0x10000) % 0x400 + 0xDC00));
248 enc->cur += 12; 290 enc->cur += 12;
249 } 291 }
250 else 292 else
251 { 293 {
252 static char hexdigit [16] = "0123456789abcdef";
253 need (enc, len += 5); 294 need (enc, len += 5);
254 *enc->cur++ = '\\'; 295 *enc->cur++ = '\\';
255 *enc->cur++ = 'u'; 296 *enc->cur++ = 'u';
256 *enc->cur++ = hexdigit [ uch >> 12 ]; 297 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
257 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 298 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
258 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 299 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
259 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 300 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
260 } 301 }
261 302
262 str += clen; 303 str += clen;
263 } 304 }
264 else if (enc->json.flags & F_LATIN1) 305 else if (enc->json.flags & F_LATIN1)
335static void 376static void
336encode_av (enc_t *enc, AV *av) 377encode_av (enc_t *enc, AV *av)
337{ 378{
338 int i, len = av_len (av); 379 int i, len = av_len (av);
339 380
340 if (enc->indent >= enc->maxdepth) 381 if (enc->indent >= enc->json.max_depth)
341 croak ("data structure too deep (hit recursion limit)"); 382 croak (ERR_NESTING_EXCEEDED);
342 383
343 encode_ch (enc, '['); 384 encode_ch (enc, '[');
344 385
345 if (len >= 0) 386 if (len >= 0)
346 { 387 {
420 461
421static void 462static void
422encode_hv (enc_t *enc, HV *hv) 463encode_hv (enc_t *enc, HV *hv)
423{ 464{
424 HE *he; 465 HE *he;
425 int count;
426 466
427 if (enc->indent >= enc->maxdepth) 467 if (enc->indent >= enc->json.max_depth)
428 croak ("data structure too deep (hit recursion limit)"); 468 croak (ERR_NESTING_EXCEEDED);
429 469
430 encode_ch (enc, '{'); 470 encode_ch (enc, '{');
431 471
432 // for canonical output we have to sort by keys first 472 // for canonical output we have to sort by keys first
433 // actually, this is mostly due to the stupid so-called 473 // actually, this is mostly due to the stupid so-called
434 // security workaround added somewhere in 5.8.x. 474 // security workaround added somewhere in 5.8.x
435 // that randomises hash orderings 475 // that randomises hash orderings
436 if (enc->json.flags & F_CANONICAL) 476 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
437 { 477 {
438 int count = hv_iterinit (hv); 478 int count = hv_iterinit (hv);
439 479
440 if (SvMAGICAL (hv)) 480 if (SvMAGICAL (hv))
441 { 481 {
614 654
615 if (len == 1 && *pv == '1') 655 if (len == 1 && *pv == '1')
616 encode_str (enc, "true", 4, 0); 656 encode_str (enc, "true", 4, 0);
617 else if (len == 1 && *pv == '0') 657 else if (len == 1 && *pv == '0')
618 encode_str (enc, "false", 5, 0); 658 encode_str (enc, "false", 5, 0);
659 else if (enc->json.flags & F_ALLOW_UNKNOWN)
660 encode_str (enc, "null", 4, 0);
619 else 661 else
620 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 662 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
621 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 663 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
622 } 664 }
665 else if (enc->json.flags & F_ALLOW_UNKNOWN)
666 encode_str (enc, "null", 4, 0);
623 else 667 else
624 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 668 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
625 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 669 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
626} 670}
627 671
648 else if (SvIOKp (sv)) 692 else if (SvIOKp (sv))
649 { 693 {
650 // we assume we can always read an IV as a UV and vice versa 694 // we assume we can always read an IV as a UV and vice versa
651 // we assume two's complement 695 // we assume two's complement
652 // we assume no aliasing issues in the union 696 // we assume no aliasing issues in the union
653 if (SvIsUV (sv) ? SvUVX (sv) > 59000 697 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
654 : SvIVX (sv) > 59000 || SvIVX (sv) < -59000) 698 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
655 {
656 // large integer, use the (rather slow) snprintf way.
657 need (enc, sizeof (UV) * 5 / 2 + 1); // CHAR_BIT is at least 8
658 enc->cur +=
659 SvIsUV(sv)
660 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
661 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
662 }
663 else
664 { 699 {
665 // optimise the "small number case" 700 // optimise the "small number case"
666 // code will likely be branchless and use only a single multiplication 701 // code will likely be branchless and use only a single multiplication
667 // works for numbers up to 59074 702 // works for numbers up to 59074
668 I32 i = SvIVX (sv); 703 I32 i = SvIVX (sv);
679 714
680 // now output digit by digit, each time masking out the integer part 715 // now output digit by digit, each time masking out the integer part
681 // and multiplying by 5 while moving the decimal point one to the right, 716 // and multiplying by 5 while moving the decimal point one to the right,
682 // resulting in a net multiplication by 10. 717 // resulting in a net multiplication by 10.
683 // we always write the digit to memory but conditionally increment 718 // we always write the digit to memory but conditionally increment
684 // the pointer, to ease the usage of conditional move instructions. 719 // the pointer, to enable the use of conditional move instructions.
685 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 720 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
686 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 721 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
687 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 722 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
688 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 723 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
689 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 724 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
690 } 725 }
726 else
727 {
728 // large integer, use the (rather slow) snprintf way.
729 need (enc, IVUV_MAXCHARS);
730 enc->cur +=
731 SvIsUV(sv)
732 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
733 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
734 }
691 } 735 }
692 else if (SvROK (sv)) 736 else if (SvROK (sv))
693 encode_rv (enc, SvRV (sv)); 737 encode_rv (enc, SvRV (sv));
694 else if (!SvOK (sv)) 738 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
695 encode_str (enc, "null", 4, 0); 739 encode_str (enc, "null", 4, 0);
696 else 740 else
697 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 741 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
698 SvPV_nolen (sv), SvFLAGS (sv)); 742 SvPV_nolen (sv), SvFLAGS (sv));
699} 743}
709 enc.json = *json; 753 enc.json = *json;
710 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 754 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
711 enc.cur = SvPVX (enc.sv); 755 enc.cur = SvPVX (enc.sv);
712 enc.end = SvEND (enc.sv); 756 enc.end = SvEND (enc.sv);
713 enc.indent = 0; 757 enc.indent = 0;
714 enc.maxdepth = DEC_DEPTH (enc.json.flags);
715 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL 758 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
716 : enc.json.flags & F_LATIN1 ? 0x000100UL 759 : enc.json.flags & F_LATIN1 ? 0x000100UL
717 : 0x110000UL; 760 : 0x110000UL;
718 761
719 SvPOK_only (enc.sv); 762 SvPOK_only (enc.sv);
720 encode_sv (&enc, scalar); 763 encode_sv (&enc, scalar);
764 encode_nl (&enc);
721 765
722 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 766 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
723 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 767 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
724 768
725 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 769 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
786 if (*dec->cur != ch) \ 830 if (*dec->cur != ch) \
787 ERR (# ch " expected"); \ 831 ERR (# ch " expected"); \
788 ++dec->cur; \ 832 ++dec->cur; \
789 SE 833 SE
790 834
791#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 835#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
792#define DEC_DEC_DEPTH --dec->depth 836#define DEC_DEC_DEPTH --dec->depth
793 837
794static SV *decode_sv (dec_t *dec); 838static SV *decode_sv (dec_t *dec);
795 839
796static signed char decode_hexdigit[256]; 840static signed char decode_hexdigit[256];
905 else if (expect_true (ch >= 0x20 && ch < 0x80)) 949 else if (expect_true (ch >= 0x20 && ch < 0x80))
906 *cur++ = ch; 950 *cur++ = ch;
907 else if (ch >= 0x80) 951 else if (ch >= 0x80)
908 { 952 {
909 STRLEN clen; 953 STRLEN clen;
910 UV uch;
911 954
912 --dec_cur; 955 --dec_cur;
913 956
914 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 957 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
915 if (clen == (STRLEN)-1) 958 if (clen == (STRLEN)-1)
916 ERR ("malformed UTF-8 character in JSON string"); 959 ERR ("malformed UTF-8 character in JSON string");
917 960
918 do 961 do
919 *cur++ = *dec_cur++; 962 *cur++ = *dec_cur++;
936 { 979 {
937 STRLEN len = cur - buf; 980 STRLEN len = cur - buf;
938 981
939 if (sv) 982 if (sv)
940 { 983 {
941 SvGROW (sv, SvCUR (sv) + len + 1); 984 STRLEN cur = SvCUR (sv);
985
986 if (SvLEN (sv) <= cur + len)
987 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
988
942 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 989 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
943 SvCUR_set (sv, SvCUR (sv) + len); 990 SvCUR_set (sv, SvCUR (sv) + len);
944 } 991 }
945 else 992 else
946 sv = newSVpvn (buf, len); 993 sv = newSVpvn (buf, len);
1037 1084
1038 // special case the rather common 1..5-digit-int case 1085 // special case the rather common 1..5-digit-int case
1039 if (*start == '-') 1086 if (*start == '-')
1040 switch (len) 1087 switch (len)
1041 { 1088 {
1042 case 2: return newSViv (-( start [1] - '0' * 1)); 1089 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1043 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1090 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1044 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1091 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1045 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1092 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1046 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); 1093 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1047 } 1094 }
1048 else 1095 else
1049 switch (len) 1096 switch (len)
1050 { 1097 {
1051 case 1: return newSViv ( start [0] - '0' * 1); 1098 case 1: return newSViv ( start [0] - '0' * 1);
1052 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1099 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1053 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1100 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1054 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1101 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1055 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); 1102 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1056 } 1103 }
1057 1104
1058 { 1105 {
1059 UV uv; 1106 UV uv;
1060 int numtype = grok_number (start, len, &uv); 1107 int numtype = grok_number (start, len, &uv);
1168 char *p = dec->cur; 1215 char *p = dec->cur;
1169 char *e = p + 24; // only try up to 24 bytes 1216 char *e = p + 24; // only try up to 24 bytes
1170 1217
1171 for (;;) 1218 for (;;)
1172 { 1219 {
1173 // the >= 0x80 is true on most architectures 1220 // the >= 0x80 is false on most architectures
1174 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1221 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1175 { 1222 {
1176 // slow path, back up and use decode_str 1223 // slow path, back up and use decode_str
1177 SV *key = decode_str (dec); 1224 SV *key = decode_str (dec);
1178 if (!key) 1225 if (!key)
1326 case 't': 1373 case 't':
1327 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1374 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1328 { 1375 {
1329 dec->cur += 4; 1376 dec->cur += 4;
1330#if JSON_SLOW 1377#if JSON_SLOW
1331 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1378 json_true = get_bool ("JSON::XS::true");
1332#endif 1379#endif
1333 return SvREFCNT_inc (json_true); 1380 return newSVsv (json_true);
1334 } 1381 }
1335 else 1382 else
1336 ERR ("'true' expected"); 1383 ERR ("'true' expected");
1337 1384
1338 break; 1385 break;
1340 case 'f': 1387 case 'f':
1341 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1388 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1342 { 1389 {
1343 dec->cur += 5; 1390 dec->cur += 5;
1344#if JSON_SLOW 1391#if JSON_SLOW
1345 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1392 json_false = get_bool ("JSON::XS::false");
1346#endif 1393#endif
1347 return SvREFCNT_inc (json_false); 1394 return newSVsv (json_false);
1348 } 1395 }
1349 else 1396 else
1350 ERR ("'false' expected"); 1397 ERR ("'false' expected");
1351 1398
1352 break; 1399 break;
1370fail: 1417fail:
1371 return 0; 1418 return 0;
1372} 1419}
1373 1420
1374static SV * 1421static SV *
1375decode_json (SV *string, JSON *json, UV *offset_return) 1422decode_json (SV *string, JSON *json, char **offset_return)
1376{ 1423{
1377 dec_t dec; 1424 dec_t dec;
1378 UV offset;
1379 SV *sv; 1425 SV *sv;
1380 1426
1427 /* work around bugs in 5.10 where manipulating magic values
1428 * will perl ignore the magic in subsequent accesses
1429 */
1381 SvGETMAGIC (string); 1430 /*SvGETMAGIC (string);*/
1431 if (SvMAGICAL (string))
1432 string = sv_2mortal (newSVsv (string));
1433
1382 SvUPGRADE (string, SVt_PV); 1434 SvUPGRADE (string, SVt_PV);
1383 1435
1384 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1436 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1437 * assertion with -DDEBUGGING, although SvCUR is documented to
1438 * return the xpv_cur field which certainly exists after upgrading.
1439 * according to nicholas clark, calling SvPOK fixes this.
1440 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1441 * and hope for the best.
1442 * Damnit, SvPV_nolen still trips over yet another assertion. This
1443 * assertion business is seriously broken, try yet another workaround
1444 * for the broken -DDEBUGGING.
1445 */
1446 {
1447#ifdef DEBUGGING
1448 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1449#else
1450 STRLEN offset = SvCUR (string);
1451#endif
1452
1453 if (offset > json->max_size && json->max_size)
1385 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1454 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1386 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1455 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1456 }
1387 1457
1388 if (json->flags & F_UTF8) 1458 if (json->flags & F_UTF8)
1389 sv_utf8_downgrade (string, 0); 1459 sv_utf8_downgrade (string, 0);
1390 else 1460 else
1391 sv_utf8_upgrade (string); 1461 sv_utf8_upgrade (string);
1392 1462
1393 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1463 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1394 1464
1395 dec.json = *json; 1465 dec.json = *json;
1396 dec.cur = SvPVX (string); 1466 dec.cur = SvPVX (string);
1397 dec.end = SvEND (string); 1467 dec.end = SvEND (string);
1398 dec.err = 0; 1468 dec.err = 0;
1399 dec.depth = 0; 1469 dec.depth = 0;
1400 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1401 1470
1402 if (dec.json.cb_object || dec.json.cb_sk_object) 1471 if (dec.json.cb_object || dec.json.cb_sk_object)
1403 dec.json.flags |= F_HOOK; 1472 dec.json.flags |= F_HOOK;
1404 1473
1405 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1474 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1406 1475
1407 decode_ws (&dec); 1476 decode_ws (&dec);
1408 sv = decode_sv (&dec); 1477 sv = decode_sv (&dec);
1478
1479 if (offset_return)
1480 *offset_return = dec.cur;
1409 1481
1410 if (!(offset_return || !sv)) 1482 if (!(offset_return || !sv))
1411 { 1483 {
1412 // check for trailing garbage 1484 // check for trailing garbage
1413 decode_ws (&dec); 1485 decode_ws (&dec);
1416 { 1488 {
1417 dec.err = "garbage after JSON object"; 1489 dec.err = "garbage after JSON object";
1418 SvREFCNT_dec (sv); 1490 SvREFCNT_dec (sv);
1419 sv = 0; 1491 sv = 0;
1420 } 1492 }
1421 }
1422
1423 if (offset_return || !sv)
1424 {
1425 offset = dec.json.flags & F_UTF8
1426 ? dec.cur - SvPVX (string)
1427 : utf8_distance (dec.cur, SvPVX (string));
1428
1429 if (offset_return)
1430 *offset_return = offset;
1431 } 1493 }
1432 1494
1433 if (!sv) 1495 if (!sv)
1434 { 1496 {
1435 SV *uni = sv_newmortal (); 1497 SV *uni = sv_newmortal ();
1441 SAVEVPTR (PL_curcop); 1503 SAVEVPTR (PL_curcop);
1442 PL_curcop = &cop; 1504 PL_curcop = &cop;
1443 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1505 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1444 LEAVE; 1506 LEAVE;
1445 1507
1446 croak ("%s, at character offset %d [\"%s\"]", 1508 croak ("%s, at character offset %d (before \"%s\")",
1447 dec.err, 1509 dec.err,
1448 (int)offset, 1510 ptr_to_index (string, dec.cur),
1449 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1511 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1450 } 1512 }
1451 1513
1452 sv = sv_2mortal (sv); 1514 sv = sv_2mortal (sv);
1453 1515
1454 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1516 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1455 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1517 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1456 1518
1457 return sv; 1519 return sv;
1520}
1521
1522/////////////////////////////////////////////////////////////////////////////
1523// incremental parser
1524
1525static void
1526incr_parse (JSON *self)
1527{
1528 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1529
1530 for (;;)
1531 {
1532 //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
1533 switch (self->incr_mode)
1534 {
1535 // only used for intiial whitespace skipping
1536 case INCR_M_WS:
1537 for (;;)
1538 {
1539 if (*p > 0x20)
1540 {
1541 self->incr_mode = INCR_M_JSON;
1542 goto incr_m_json;
1543 }
1544 else if (!*p)
1545 goto interrupt;
1546
1547 ++p;
1548 }
1549
1550 // skip a single char inside a string (for \\-processing)
1551 case INCR_M_BS:
1552 if (!*p)
1553 goto interrupt;
1554
1555 ++p;
1556 self->incr_mode = INCR_M_STR;
1557 goto incr_m_str;
1558
1559 // inside a string
1560 case INCR_M_STR:
1561 incr_m_str:
1562 for (;;)
1563 {
1564 if (*p == '"')
1565 {
1566 ++p;
1567 self->incr_mode = INCR_M_JSON;
1568
1569 if (!self->incr_nest)
1570 goto interrupt;
1571
1572 goto incr_m_json;
1573 }
1574 else if (*p == '\\')
1575 {
1576 ++p; // "virtually" consumes character after \
1577
1578 if (!*p) // if at end of string we have to switch modes
1579 {
1580 self->incr_mode = INCR_M_BS;
1581 goto interrupt;
1582 }
1583 }
1584 else if (!*p)
1585 goto interrupt;
1586
1587 ++p;
1588 }
1589
1590 // after initial ws, outside string
1591 case INCR_M_JSON:
1592 incr_m_json:
1593 for (;;)
1594 {
1595 switch (*p++)
1596 {
1597 case 0:
1598 --p;
1599 goto interrupt;
1600
1601 case 0x09:
1602 case 0x0a:
1603 case 0x0d:
1604 case 0x20:
1605 if (!self->incr_nest)
1606 {
1607 --p; // do not eat the whitespace, let the next round do it
1608 goto interrupt;
1609 }
1610 break;
1611
1612 case '"':
1613 self->incr_mode = INCR_M_STR;
1614 goto incr_m_str;
1615
1616 case '[':
1617 case '{':
1618 if (++self->incr_nest > self->max_depth)
1619 croak (ERR_NESTING_EXCEEDED);
1620 break;
1621
1622 case ']':
1623 case '}':
1624 if (--self->incr_nest <= 0)
1625 goto interrupt;
1626 }
1627 }
1628 }
1629
1630 modechange:
1631 ;
1632 }
1633
1634interrupt:
1635 self->incr_pos = p - SvPVX (self->incr_text);
1636 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1458} 1637}
1459 1638
1460///////////////////////////////////////////////////////////////////////////// 1639/////////////////////////////////////////////////////////////////////////////
1461// XS interface functions 1640// XS interface functions
1462 1641
1474 : -1; 1653 : -1;
1475 1654
1476 json_stash = gv_stashpv ("JSON::XS" , 1); 1655 json_stash = gv_stashpv ("JSON::XS" , 1);
1477 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1656 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1478 1657
1479 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1658 json_true = get_bool ("JSON::XS::true");
1480 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1659 json_false = get_bool ("JSON::XS::false");
1660
1661 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1481} 1662}
1482 1663
1483PROTOTYPES: DISABLE 1664PROTOTYPES: DISABLE
1484 1665
1485void CLONE (...) 1666void CLONE (...)
1490void new (char *klass) 1671void new (char *klass)
1491 PPCODE: 1672 PPCODE:
1492{ 1673{
1493 SV *pv = NEWSV (0, sizeof (JSON)); 1674 SV *pv = NEWSV (0, sizeof (JSON));
1494 SvPOK_only (pv); 1675 SvPOK_only (pv);
1495 Zero (SvPVX (pv), 1, JSON); 1676 json_init ((JSON *)SvPVX (pv));
1496 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1497 XPUSHs (sv_2mortal (sv_bless ( 1677 XPUSHs (sv_2mortal (sv_bless (
1498 newRV_noinc (pv), 1678 newRV_noinc (pv),
1499 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1679 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1500 ))); 1680 )));
1501} 1681}
1513 allow_nonref = F_ALLOW_NONREF 1693 allow_nonref = F_ALLOW_NONREF
1514 shrink = F_SHRINK 1694 shrink = F_SHRINK
1515 allow_blessed = F_ALLOW_BLESSED 1695 allow_blessed = F_ALLOW_BLESSED
1516 convert_blessed = F_CONV_BLESSED 1696 convert_blessed = F_CONV_BLESSED
1517 relaxed = F_RELAXED 1697 relaxed = F_RELAXED
1698 allow_unknown = F_ALLOW_UNKNOWN
1518 PPCODE: 1699 PPCODE:
1519{ 1700{
1520 if (enable) 1701 if (enable)
1521 self->flags |= ix; 1702 self->flags |= ix;
1522 else 1703 else
1537 get_allow_nonref = F_ALLOW_NONREF 1718 get_allow_nonref = F_ALLOW_NONREF
1538 get_shrink = F_SHRINK 1719 get_shrink = F_SHRINK
1539 get_allow_blessed = F_ALLOW_BLESSED 1720 get_allow_blessed = F_ALLOW_BLESSED
1540 get_convert_blessed = F_CONV_BLESSED 1721 get_convert_blessed = F_CONV_BLESSED
1541 get_relaxed = F_RELAXED 1722 get_relaxed = F_RELAXED
1723 get_allow_unknown = F_ALLOW_UNKNOWN
1542 PPCODE: 1724 PPCODE:
1543 XPUSHs (boolSV (self->flags & ix)); 1725 XPUSHs (boolSV (self->flags & ix));
1544 1726
1545void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1727void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1546 PPCODE: 1728 PPCODE:
1547{ 1729 self->max_depth = max_depth;
1548 UV log2 = 0;
1549
1550 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1551
1552 while ((1UL << log2) < max_depth)
1553 ++log2;
1554
1555 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1556
1557 XPUSHs (ST (0)); 1730 XPUSHs (ST (0));
1558}
1559 1731
1560U32 get_max_depth (JSON *self) 1732U32 get_max_depth (JSON *self)
1561 CODE: 1733 CODE:
1562 RETVAL = DEC_DEPTH (self->flags); 1734 RETVAL = self->max_depth;
1563 OUTPUT: 1735 OUTPUT:
1564 RETVAL 1736 RETVAL
1565 1737
1566void max_size (JSON *self, UV max_size = 0) 1738void max_size (JSON *self, U32 max_size = 0)
1567 PPCODE: 1739 PPCODE:
1568{ 1740 self->max_size = max_size;
1569 UV log2 = 0;
1570
1571 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1572 if (max_size == 1) max_size = 2;
1573
1574 while ((1UL << log2) < max_size)
1575 ++log2;
1576
1577 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1578
1579 XPUSHs (ST (0)); 1741 XPUSHs (ST (0));
1580}
1581 1742
1582int get_max_size (JSON *self) 1743int get_max_size (JSON *self)
1583 CODE: 1744 CODE:
1584 RETVAL = DEC_SIZE (self->flags); 1745 RETVAL = self->max_size;
1585 OUTPUT: 1746 OUTPUT:
1586 RETVAL 1747 RETVAL
1587 1748
1588void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1749void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1589 PPCODE: 1750 PPCODE:
1625 XPUSHs (decode_json (jsonstr, self, 0)); 1786 XPUSHs (decode_json (jsonstr, self, 0));
1626 1787
1627void decode_prefix (JSON *self, SV *jsonstr) 1788void decode_prefix (JSON *self, SV *jsonstr)
1628 PPCODE: 1789 PPCODE:
1629{ 1790{
1630 UV offset; 1791 char *offset;
1631 EXTEND (SP, 2); 1792 EXTEND (SP, 2);
1632 PUSHs (decode_json (jsonstr, self, &offset)); 1793 PUSHs (decode_json (jsonstr, self, &offset));
1633 PUSHs (sv_2mortal (newSVuv (offset))); 1794 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1795}
1796
1797void incr_parse (JSON *self, SV *jsonstr = 0)
1798 PPCODE:
1799{
1800 if (!self->incr_text)
1801 self->incr_text = newSVpvn ("", 0);
1802
1803 // append data, if any
1804 if (jsonstr)
1805 {
1806 if (SvUTF8 (jsonstr))
1807 {
1808 if (!SvUTF8 (self->incr_text))
1809 {
1810 /* utf-8-ness differs, need to upgrade */
1811 sv_utf8_upgrade (self->incr_text);
1812
1813 if (self->incr_pos)
1814 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1815 - (U8 *)SvPVX (self->incr_text);
1816 }
1817 }
1818 else if (SvUTF8 (self->incr_text))
1819 sv_utf8_upgrade (jsonstr);
1820
1821 {
1822 STRLEN len;
1823 const char *str = SvPV (jsonstr, len);
1824 STRLEN cur = SvCUR (self->incr_text);
1825
1826 if (SvLEN (self->incr_text) <= cur + len)
1827 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1828
1829 Move (str, SvEND (self->incr_text), len, char);
1830 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1831 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1832 }
1833 }
1834
1835 if (GIMME_V != G_VOID)
1836 do
1837 {
1838 char *offset;
1839
1840 if (!INCR_DONE (self))
1841 {
1842 incr_parse (self);
1843
1844 if (self->incr_pos > self->max_size && self->max_size)
1845 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1846 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1847
1848 if (!INCR_DONE (self))
1849 break;
1850 }
1851
1852 XPUSHs (decode_json (self->incr_text, self, &offset));
1853
1854 self->incr_pos -= offset - SvPVX (self->incr_text);
1855 self->incr_nest = 0;
1856 self->incr_mode = 0;
1857
1858 sv_chop (self->incr_text, offset);
1859 }
1860 while (GIMME_V == G_ARRAY);
1861}
1862
1863SV *incr_text (JSON *self)
1864 ATTRS: lvalue
1865 CODE:
1866{
1867 if (self->incr_pos)
1868 croak ("incr_text can not be called when the incremental parser already started parsing");
1869
1870 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1871}
1872 OUTPUT:
1873 RETVAL
1874
1875void incr_skip (JSON *self)
1876 CODE:
1877{
1878 if (self->incr_pos)
1879 {
1880 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1881 self->incr_pos = 0;
1882 self->incr_nest = 0;
1883 self->incr_mode = 0;
1884 }
1885}
1886
1887void incr_reset (JSON *self)
1888 CODE:
1889{
1890 SvREFCNT_dec (self->incr_text);
1891 self->incr_text = 0;
1892 self->incr_pos = 0;
1893 self->incr_nest = 0;
1894 self->incr_mode = 0;
1634} 1895}
1635 1896
1636void DESTROY (JSON *self) 1897void DESTROY (JSON *self)
1637 CODE: 1898 CODE:
1638 SvREFCNT_dec (self->cb_sk_object); 1899 SvREFCNT_dec (self->cb_sk_object);
1639 SvREFCNT_dec (self->cb_object); 1900 SvREFCNT_dec (self->cb_object);
1901 SvREFCNT_dec (self->incr_text);
1640 1902
1641PROTOTYPES: ENABLE 1903PROTOTYPES: ENABLE
1642 1904
1643void encode_json (SV *scalar) 1905void encode_json (SV *scalar)
1906 ALIAS:
1907 to_json_ = 0
1908 encode_json = F_UTF8
1644 PPCODE: 1909 PPCODE:
1645{ 1910{
1646 JSON json = { F_DEFAULT | F_UTF8 }; 1911 JSON json;
1912 json_init (&json);
1913 json.flags |= ix;
1647 XPUSHs (encode_json (scalar, &json)); 1914 XPUSHs (encode_json (scalar, &json));
1648} 1915}
1649 1916
1650void decode_json (SV *jsonstr) 1917void decode_json (SV *jsonstr)
1918 ALIAS:
1919 from_json_ = 0
1920 decode_json = F_UTF8
1651 PPCODE: 1921 PPCODE:
1652{ 1922{
1653 JSON json = { F_DEFAULT | F_UTF8 }; 1923 JSON json;
1924 json_init (&json);
1925 json.flags |= ix;
1654 XPUSHs (decode_json (jsonstr, &json, 0)); 1926 XPUSHs (decode_json (jsonstr, &json, 0));
1655} 1927}
1656 1928
1929

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines