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 1.102 by root, Sat Oct 10 01:48:50 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// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
20 24
21#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
23#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
24#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
28#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
29#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL 34#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL 35#define F_CONV_BLESSED 0x00000800UL
32#define F_RELAXED 0x00001000UL 36#define F_RELAXED 0x00001000UL
33 37#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 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
39 39
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 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
44#define F_DEFAULT (9UL << S_MAXDEPTH)
45 41
46#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
47#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
48 44
49#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
64 60
65#define IN_RANGE_INC(type,val,beg,end) \ 61#define IN_RANGE_INC(type,val,beg,end) \
66 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 62 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
67 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 63 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
68 64
65#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
66
69#ifdef USE_ITHREADS 67#ifdef USE_ITHREADS
70# define JSON_SLOW 1 68# define JSON_SLOW 1
71# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
72#else 70#else
73# define JSON_SLOW 0 71# define JSON_SLOW 0
75#endif 73#endif
76 74
77static HV *json_stash, *json_boolean_stash; // JSON::XS:: 75static HV *json_stash, *json_boolean_stash; // JSON::XS::
78static SV *json_true, *json_false; 76static SV *json_true, *json_false;
79 77
78enum {
79 INCR_M_WS = 0, // initial whitespace skipping, must be 0
80 INCR_M_STR, // inside string
81 INCR_M_BS, // inside backslash
82 INCR_M_JSON // outside anything, count nesting
83};
84
85#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
86
80typedef struct { 87typedef struct {
81 U32 flags; 88 U32 flags;
89 U32 max_depth;
90 STRLEN max_size;
91
82 SV *cb_object; 92 SV *cb_object;
83 HV *cb_sk_object; 93 HV *cb_sk_object;
94
95 // for the incremental parser
96 SV *incr_text; // the source text so far
97 STRLEN incr_pos; // the current offset into the text
98 int incr_nest; // {[]}-nesting level
99 unsigned char incr_mode;
84} JSON; 100} JSON;
101
102INLINE void
103json_init (JSON *json)
104{
105 Zero (json, 1, JSON);
106 json->max_depth = 512;
107}
85 108
86///////////////////////////////////////////////////////////////////////////// 109/////////////////////////////////////////////////////////////////////////////
87// utility functions 110// utility functions
88 111
112INLINE SV *
113get_bool (const char *name)
114{
115 SV *sv = get_sv (name, 1);
116
117 SvREADONLY_on (sv);
118 SvREADONLY_on (SvRV (sv));
119
120 return sv;
121}
122
89INLINE void 123INLINE void
90shrink (SV *sv) 124shrink (SV *sv)
91{ 125{
92 sv_utf8_downgrade (sv, 1); 126 sv_utf8_downgrade (sv, 1);
127
93 if (SvLEN (sv) > SvCUR (sv) + 1) 128 if (SvLEN (sv) > SvCUR (sv) + 1)
94 { 129 {
95#ifdef SvPV_shrink_to_cur 130#ifdef SvPV_shrink_to_cur
96 SvPV_shrink_to_cur (sv); 131 SvPV_shrink_to_cur (sv);
97#elif defined (SvPV_renew) 132#elif defined (SvPV_renew)
141 *s++ = 0x80 | ( ch & 0x3f); 176 *s++ = 0x80 | ( ch & 0x3f);
142 177
143 return s; 178 return s;
144} 179}
145 180
181// convert offset pointer to character index, sv must be string
182static STRLEN
183ptr_to_index (SV *sv, char *offset)
184{
185 return SvUTF8 (sv)
186 ? utf8_distance (offset, SvPVX (sv))
187 : offset - SvPVX (sv);
188}
189
146///////////////////////////////////////////////////////////////////////////// 190/////////////////////////////////////////////////////////////////////////////
147// encoder 191// encoder
148 192
149// structure used for encoding JSON 193// structure used for encoding JSON
150typedef struct 194typedef struct
152 char *cur; // SvPVX (sv) + current output position 196 char *cur; // SvPVX (sv) + current output position
153 char *end; // SvEND (sv) 197 char *end; // SvEND (sv)
154 SV *sv; // result scalar 198 SV *sv; // result scalar
155 JSON json; 199 JSON json;
156 U32 indent; // indentation level 200 U32 indent; // indentation level
157 U32 maxdepth; // max. indentation/recursion level
158 UV limit; // escape character values >= this value when encoding 201 UV limit; // escape character values >= this value when encoding
159} enc_t; 202} enc_t;
160 203
161INLINE void 204INLINE void
162need (enc_t *enc, STRLEN len) 205need (enc_t *enc, STRLEN len)
163{ 206{
164 if (expect_false (enc->cur + len >= enc->end)) 207 if (expect_false (enc->cur + len >= enc->end))
165 { 208 {
166 STRLEN cur = enc->cur - SvPVX (enc->sv); 209 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
167 SvGROW (enc->sv, cur + len + 1); 210 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
168 enc->cur = SvPVX (enc->sv) + cur; 211 enc->cur = SvPVX (enc->sv) + cur;
169 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 212 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
170 } 213 }
171} 214}
172 215
247 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 290 (int)((uch - 0x10000) % 0x400 + 0xDC00));
248 enc->cur += 12; 291 enc->cur += 12;
249 } 292 }
250 else 293 else
251 { 294 {
252 static char hexdigit [16] = "0123456789abcdef";
253 need (enc, len += 5); 295 need (enc, len += 5);
254 *enc->cur++ = '\\'; 296 *enc->cur++ = '\\';
255 *enc->cur++ = 'u'; 297 *enc->cur++ = 'u';
256 *enc->cur++ = hexdigit [ uch >> 12 ]; 298 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
257 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 299 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
258 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 300 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
259 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 301 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
260 } 302 }
261 303
262 str += clen; 304 str += clen;
263 } 305 }
264 else if (enc->json.flags & F_LATIN1) 306 else if (enc->json.flags & F_LATIN1)
335static void 377static void
336encode_av (enc_t *enc, AV *av) 378encode_av (enc_t *enc, AV *av)
337{ 379{
338 int i, len = av_len (av); 380 int i, len = av_len (av);
339 381
340 if (enc->indent >= enc->maxdepth) 382 if (enc->indent >= enc->json.max_depth)
341 croak ("data structure too deep (hit recursion limit)"); 383 croak (ERR_NESTING_EXCEEDED);
342 384
343 encode_ch (enc, '['); 385 encode_ch (enc, '[');
344 386
345 if (len >= 0) 387 if (len >= 0)
346 { 388 {
420 462
421static void 463static void
422encode_hv (enc_t *enc, HV *hv) 464encode_hv (enc_t *enc, HV *hv)
423{ 465{
424 HE *he; 466 HE *he;
425 int count;
426 467
427 if (enc->indent >= enc->maxdepth) 468 if (enc->indent >= enc->json.max_depth)
428 croak ("data structure too deep (hit recursion limit)"); 469 croak (ERR_NESTING_EXCEEDED);
429 470
430 encode_ch (enc, '{'); 471 encode_ch (enc, '{');
431 472
432 // for canonical output we have to sort by keys first 473 // for canonical output we have to sort by keys first
433 // actually, this is mostly due to the stupid so-called 474 // actually, this is mostly due to the stupid so-called
434 // security workaround added somewhere in 5.8.x. 475 // security workaround added somewhere in 5.8.x
435 // that randomises hash orderings 476 // that randomises hash orderings
436 if (enc->json.flags & F_CANONICAL) 477 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
437 { 478 {
438 int count = hv_iterinit (hv); 479 int count = hv_iterinit (hv);
439 480
440 if (SvMAGICAL (hv)) 481 if (SvMAGICAL (hv))
441 { 482 {
614 655
615 if (len == 1 && *pv == '1') 656 if (len == 1 && *pv == '1')
616 encode_str (enc, "true", 4, 0); 657 encode_str (enc, "true", 4, 0);
617 else if (len == 1 && *pv == '0') 658 else if (len == 1 && *pv == '0')
618 encode_str (enc, "false", 5, 0); 659 encode_str (enc, "false", 5, 0);
660 else if (enc->json.flags & F_ALLOW_UNKNOWN)
661 encode_str (enc, "null", 4, 0);
619 else 662 else
620 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 663 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
621 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 664 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
622 } 665 }
666 else if (enc->json.flags & F_ALLOW_UNKNOWN)
667 encode_str (enc, "null", 4, 0);
623 else 668 else
624 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 669 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
625 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 670 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
626} 671}
627 672
648 else if (SvIOKp (sv)) 693 else if (SvIOKp (sv))
649 { 694 {
650 // we assume we can always read an IV as a UV and vice versa 695 // we assume we can always read an IV as a UV and vice versa
651 // we assume two's complement 696 // we assume two's complement
652 // we assume no aliasing issues in the union 697 // we assume no aliasing issues in the union
653 if (SvIsUV (sv) ? SvUVX (sv) > 59000 698 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
654 : SvIVX (sv) > 59000 || SvIVX (sv) < -59000) 699 : 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 { 700 {
665 // optimise the "small number case" 701 // optimise the "small number case"
666 // code will likely be branchless and use only a single multiplication 702 // code will likely be branchless and use only a single multiplication
667 // works for numbers up to 59074 703 // works for numbers up to 59074
668 I32 i = SvIVX (sv); 704 I32 i = SvIVX (sv);
679 715
680 // now output digit by digit, each time masking out the integer part 716 // 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, 717 // and multiplying by 5 while moving the decimal point one to the right,
682 // resulting in a net multiplication by 10. 718 // resulting in a net multiplication by 10.
683 // we always write the digit to memory but conditionally increment 719 // we always write the digit to memory but conditionally increment
684 // the pointer, to ease the usage of conditional move instructions. 720 // 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; 721 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; 722 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; 723 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; 724 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' 725 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
690 } 726 }
727 else
728 {
729 // large integer, use the (rather slow) snprintf way.
730 need (enc, IVUV_MAXCHARS);
731 enc->cur +=
732 SvIsUV(sv)
733 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
734 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
735 }
691 } 736 }
692 else if (SvROK (sv)) 737 else if (SvROK (sv))
693 encode_rv (enc, SvRV (sv)); 738 encode_rv (enc, SvRV (sv));
694 else if (!SvOK (sv)) 739 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
695 encode_str (enc, "null", 4, 0); 740 encode_str (enc, "null", 4, 0);
696 else 741 else
697 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 742 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
698 SvPV_nolen (sv), SvFLAGS (sv)); 743 SvPV_nolen (sv), SvFLAGS (sv));
699} 744}
709 enc.json = *json; 754 enc.json = *json;
710 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 755 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
711 enc.cur = SvPVX (enc.sv); 756 enc.cur = SvPVX (enc.sv);
712 enc.end = SvEND (enc.sv); 757 enc.end = SvEND (enc.sv);
713 enc.indent = 0; 758 enc.indent = 0;
714 enc.maxdepth = DEC_DEPTH (enc.json.flags);
715 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL 759 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
716 : enc.json.flags & F_LATIN1 ? 0x000100UL 760 : enc.json.flags & F_LATIN1 ? 0x000100UL
717 : 0x110000UL; 761 : 0x110000UL;
718 762
719 SvPOK_only (enc.sv); 763 SvPOK_only (enc.sv);
720 encode_sv (&enc, scalar); 764 encode_sv (&enc, scalar);
765 encode_nl (&enc);
721 766
722 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 767 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
723 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 768 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
724 769
725 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 770 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
786 if (*dec->cur != ch) \ 831 if (*dec->cur != ch) \
787 ERR (# ch " expected"); \ 832 ERR (# ch " expected"); \
788 ++dec->cur; \ 833 ++dec->cur; \
789 SE 834 SE
790 835
791#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 836#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
792#define DEC_DEC_DEPTH --dec->depth 837#define DEC_DEC_DEPTH --dec->depth
793 838
794static SV *decode_sv (dec_t *dec); 839static SV *decode_sv (dec_t *dec);
795 840
796static signed char decode_hexdigit[256]; 841static signed char decode_hexdigit[256];
905 else if (expect_true (ch >= 0x20 && ch < 0x80)) 950 else if (expect_true (ch >= 0x20 && ch < 0x80))
906 *cur++ = ch; 951 *cur++ = ch;
907 else if (ch >= 0x80) 952 else if (ch >= 0x80)
908 { 953 {
909 STRLEN clen; 954 STRLEN clen;
910 UV uch;
911 955
912 --dec_cur; 956 --dec_cur;
913 957
914 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 958 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
915 if (clen == (STRLEN)-1) 959 if (clen == (STRLEN)-1)
916 ERR ("malformed UTF-8 character in JSON string"); 960 ERR ("malformed UTF-8 character in JSON string");
917 961
918 do 962 do
919 *cur++ = *dec_cur++; 963 *cur++ = *dec_cur++;
936 { 980 {
937 STRLEN len = cur - buf; 981 STRLEN len = cur - buf;
938 982
939 if (sv) 983 if (sv)
940 { 984 {
941 SvGROW (sv, SvCUR (sv) + len + 1); 985 STRLEN cur = SvCUR (sv);
986
987 if (SvLEN (sv) <= cur + len)
988 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
989
942 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 990 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
943 SvCUR_set (sv, SvCUR (sv) + len); 991 SvCUR_set (sv, SvCUR (sv) + len);
944 } 992 }
945 else 993 else
946 sv = newSVpvn (buf, len); 994 sv = newSVpvn (buf, len);
1037 1085
1038 // special case the rather common 1..5-digit-int case 1086 // special case the rather common 1..5-digit-int case
1039 if (*start == '-') 1087 if (*start == '-')
1040 switch (len) 1088 switch (len)
1041 { 1089 {
1042 case 2: return newSViv (-( start [1] - '0' * 1)); 1090 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1043 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1091 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)); 1092 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)); 1093 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)); 1094 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1047 } 1095 }
1048 else 1096 else
1049 switch (len) 1097 switch (len)
1050 { 1098 {
1051 case 1: return newSViv ( start [0] - '0' * 1); 1099 case 1: return newSViv ( start [0] - '0' * 1);
1052 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1100 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); 1101 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); 1102 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); 1103 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1056 } 1104 }
1057 1105
1058 { 1106 {
1059 UV uv; 1107 UV uv;
1060 int numtype = grok_number (start, len, &uv); 1108 int numtype = grok_number (start, len, &uv);
1168 char *p = dec->cur; 1216 char *p = dec->cur;
1169 char *e = p + 24; // only try up to 24 bytes 1217 char *e = p + 24; // only try up to 24 bytes
1170 1218
1171 for (;;) 1219 for (;;)
1172 { 1220 {
1173 // the >= 0x80 is true on most architectures 1221 // the >= 0x80 is false on most architectures
1174 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1222 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1175 { 1223 {
1176 // slow path, back up and use decode_str 1224 // slow path, back up and use decode_str
1177 SV *key = decode_str (dec); 1225 SV *key = decode_str (dec);
1178 if (!key) 1226 if (!key)
1326 case 't': 1374 case 't':
1327 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1375 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1328 { 1376 {
1329 dec->cur += 4; 1377 dec->cur += 4;
1330#if JSON_SLOW 1378#if JSON_SLOW
1331 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1379 json_true = get_bool ("JSON::XS::true");
1332#endif 1380#endif
1333 return SvREFCNT_inc (json_true); 1381 return newSVsv (json_true);
1334 } 1382 }
1335 else 1383 else
1336 ERR ("'true' expected"); 1384 ERR ("'true' expected");
1337 1385
1338 break; 1386 break;
1340 case 'f': 1388 case 'f':
1341 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1389 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1342 { 1390 {
1343 dec->cur += 5; 1391 dec->cur += 5;
1344#if JSON_SLOW 1392#if JSON_SLOW
1345 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1393 json_false = get_bool ("JSON::XS::false");
1346#endif 1394#endif
1347 return SvREFCNT_inc (json_false); 1395 return newSVsv (json_false);
1348 } 1396 }
1349 else 1397 else
1350 ERR ("'false' expected"); 1398 ERR ("'false' expected");
1351 1399
1352 break; 1400 break;
1370fail: 1418fail:
1371 return 0; 1419 return 0;
1372} 1420}
1373 1421
1374static SV * 1422static SV *
1375decode_json (SV *string, JSON *json, UV *offset_return) 1423decode_json (SV *string, JSON *json, char **offset_return)
1376{ 1424{
1377 dec_t dec; 1425 dec_t dec;
1378 UV offset;
1379 SV *sv; 1426 SV *sv;
1380 1427
1428 /* work around bugs in 5.10 where manipulating magic values
1429 * will perl ignore the magic in subsequent accesses
1430 */
1381 SvGETMAGIC (string); 1431 /*SvGETMAGIC (string);*/
1432 if (SvMAGICAL (string))
1433 string = sv_2mortal (newSVsv (string));
1434
1382 SvUPGRADE (string, SVt_PV); 1435 SvUPGRADE (string, SVt_PV);
1383 1436
1384 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1437 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1438 * assertion with -DDEBUGGING, although SvCUR is documented to
1439 * return the xpv_cur field which certainly exists after upgrading.
1440 * according to nicholas clark, calling SvPOK fixes this.
1441 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1442 * and hope for the best.
1443 * Damnit, SvPV_nolen still trips over yet another assertion. This
1444 * assertion business is seriously broken, try yet another workaround
1445 * for the broken -DDEBUGGING.
1446 */
1447 {
1448#ifdef DEBUGGING
1449 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1450#else
1451 STRLEN offset = SvCUR (string);
1452#endif
1453
1454 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", 1455 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)); 1456 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1457 }
1387 1458
1388 if (json->flags & F_UTF8) 1459 if (json->flags & F_UTF8)
1389 sv_utf8_downgrade (string, 0); 1460 sv_utf8_downgrade (string, 0);
1390 else 1461 else
1391 sv_utf8_upgrade (string); 1462 sv_utf8_upgrade (string);
1392 1463
1393 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1464 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1394 1465
1395 dec.json = *json; 1466 dec.json = *json;
1396 dec.cur = SvPVX (string); 1467 dec.cur = SvPVX (string);
1397 dec.end = SvEND (string); 1468 dec.end = SvEND (string);
1398 dec.err = 0; 1469 dec.err = 0;
1399 dec.depth = 0; 1470 dec.depth = 0;
1400 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1401 1471
1402 if (dec.json.cb_object || dec.json.cb_sk_object) 1472 if (dec.json.cb_object || dec.json.cb_sk_object)
1403 dec.json.flags |= F_HOOK; 1473 dec.json.flags |= F_HOOK;
1404 1474
1405 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1475 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1406 1476
1407 decode_ws (&dec); 1477 decode_ws (&dec);
1408 sv = decode_sv (&dec); 1478 sv = decode_sv (&dec);
1479
1480 if (offset_return)
1481 *offset_return = dec.cur;
1409 1482
1410 if (!(offset_return || !sv)) 1483 if (!(offset_return || !sv))
1411 { 1484 {
1412 // check for trailing garbage 1485 // check for trailing garbage
1413 decode_ws (&dec); 1486 decode_ws (&dec);
1416 { 1489 {
1417 dec.err = "garbage after JSON object"; 1490 dec.err = "garbage after JSON object";
1418 SvREFCNT_dec (sv); 1491 SvREFCNT_dec (sv);
1419 sv = 0; 1492 sv = 0;
1420 } 1493 }
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 } 1494 }
1432 1495
1433 if (!sv) 1496 if (!sv)
1434 { 1497 {
1435 SV *uni = sv_newmortal (); 1498 SV *uni = sv_newmortal ();
1441 SAVEVPTR (PL_curcop); 1504 SAVEVPTR (PL_curcop);
1442 PL_curcop = &cop; 1505 PL_curcop = &cop;
1443 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1506 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1444 LEAVE; 1507 LEAVE;
1445 1508
1446 croak ("%s, at character offset %d [\"%s\"]", 1509 croak ("%s, at character offset %d (before \"%s\")",
1447 dec.err, 1510 dec.err,
1448 (int)offset, 1511 ptr_to_index (string, dec.cur),
1449 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1512 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1450 } 1513 }
1451 1514
1452 sv = sv_2mortal (sv); 1515 sv = sv_2mortal (sv);
1453 1516
1454 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1517 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)"); 1518 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1456 1519
1457 return sv; 1520 return sv;
1521}
1522
1523/////////////////////////////////////////////////////////////////////////////
1524// incremental parser
1525
1526static void
1527incr_parse (JSON *self)
1528{
1529 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1530
1531 for (;;)
1532 {
1533 //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
1534 switch (self->incr_mode)
1535 {
1536 // only used for intiial whitespace skipping
1537 case INCR_M_WS:
1538 for (;;)
1539 {
1540 if (*p > 0x20)
1541 {
1542 self->incr_mode = INCR_M_JSON;
1543 goto incr_m_json;
1544 }
1545 else if (!*p)
1546 goto interrupt;
1547
1548 ++p;
1549 }
1550
1551 // skip a single char inside a string (for \\-processing)
1552 case INCR_M_BS:
1553 if (!*p)
1554 goto interrupt;
1555
1556 ++p;
1557 self->incr_mode = INCR_M_STR;
1558 goto incr_m_str;
1559
1560 // inside a string
1561 case INCR_M_STR:
1562 incr_m_str:
1563 for (;;)
1564 {
1565 if (*p == '"')
1566 {
1567 ++p;
1568 self->incr_mode = INCR_M_JSON;
1569
1570 if (!self->incr_nest)
1571 goto interrupt;
1572
1573 goto incr_m_json;
1574 }
1575 else if (*p == '\\')
1576 {
1577 ++p; // "virtually" consumes character after \
1578
1579 if (!*p) // if at end of string we have to switch modes
1580 {
1581 self->incr_mode = INCR_M_BS;
1582 goto interrupt;
1583 }
1584 }
1585 else if (!*p)
1586 goto interrupt;
1587
1588 ++p;
1589 }
1590
1591 // after initial ws, outside string
1592 case INCR_M_JSON:
1593 incr_m_json:
1594 for (;;)
1595 {
1596 switch (*p++)
1597 {
1598 case 0:
1599 --p;
1600 goto interrupt;
1601
1602 case 0x09:
1603 case 0x0a:
1604 case 0x0d:
1605 case 0x20:
1606 if (!self->incr_nest)
1607 {
1608 --p; // do not eat the whitespace, let the next round do it
1609 goto interrupt;
1610 }
1611 break;
1612
1613 case '"':
1614 self->incr_mode = INCR_M_STR;
1615 goto incr_m_str;
1616
1617 case '[':
1618 case '{':
1619 if (++self->incr_nest > self->max_depth)
1620 croak (ERR_NESTING_EXCEEDED);
1621 break;
1622
1623 case ']':
1624 case '}':
1625 if (--self->incr_nest <= 0)
1626 goto interrupt;
1627 }
1628 }
1629 }
1630
1631 modechange:
1632 ;
1633 }
1634
1635interrupt:
1636 self->incr_pos = p - SvPVX (self->incr_text);
1637 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1458} 1638}
1459 1639
1460///////////////////////////////////////////////////////////////////////////// 1640/////////////////////////////////////////////////////////////////////////////
1461// XS interface functions 1641// XS interface functions
1462 1642
1474 : -1; 1654 : -1;
1475 1655
1476 json_stash = gv_stashpv ("JSON::XS" , 1); 1656 json_stash = gv_stashpv ("JSON::XS" , 1);
1477 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1657 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1478 1658
1479 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1659 json_true = get_bool ("JSON::XS::true");
1480 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1660 json_false = get_bool ("JSON::XS::false");
1661
1662 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1481} 1663}
1482 1664
1483PROTOTYPES: DISABLE 1665PROTOTYPES: DISABLE
1484 1666
1485void CLONE (...) 1667void CLONE (...)
1490void new (char *klass) 1672void new (char *klass)
1491 PPCODE: 1673 PPCODE:
1492{ 1674{
1493 SV *pv = NEWSV (0, sizeof (JSON)); 1675 SV *pv = NEWSV (0, sizeof (JSON));
1494 SvPOK_only (pv); 1676 SvPOK_only (pv);
1495 Zero (SvPVX (pv), 1, JSON); 1677 json_init ((JSON *)SvPVX (pv));
1496 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1497 XPUSHs (sv_2mortal (sv_bless ( 1678 XPUSHs (sv_2mortal (sv_bless (
1498 newRV_noinc (pv), 1679 newRV_noinc (pv),
1499 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1680 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1500 ))); 1681 )));
1501} 1682}
1513 allow_nonref = F_ALLOW_NONREF 1694 allow_nonref = F_ALLOW_NONREF
1514 shrink = F_SHRINK 1695 shrink = F_SHRINK
1515 allow_blessed = F_ALLOW_BLESSED 1696 allow_blessed = F_ALLOW_BLESSED
1516 convert_blessed = F_CONV_BLESSED 1697 convert_blessed = F_CONV_BLESSED
1517 relaxed = F_RELAXED 1698 relaxed = F_RELAXED
1699 allow_unknown = F_ALLOW_UNKNOWN
1518 PPCODE: 1700 PPCODE:
1519{ 1701{
1520 if (enable) 1702 if (enable)
1521 self->flags |= ix; 1703 self->flags |= ix;
1522 else 1704 else
1537 get_allow_nonref = F_ALLOW_NONREF 1719 get_allow_nonref = F_ALLOW_NONREF
1538 get_shrink = F_SHRINK 1720 get_shrink = F_SHRINK
1539 get_allow_blessed = F_ALLOW_BLESSED 1721 get_allow_blessed = F_ALLOW_BLESSED
1540 get_convert_blessed = F_CONV_BLESSED 1722 get_convert_blessed = F_CONV_BLESSED
1541 get_relaxed = F_RELAXED 1723 get_relaxed = F_RELAXED
1724 get_allow_unknown = F_ALLOW_UNKNOWN
1542 PPCODE: 1725 PPCODE:
1543 XPUSHs (boolSV (self->flags & ix)); 1726 XPUSHs (boolSV (self->flags & ix));
1544 1727
1545void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1728void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1546 PPCODE: 1729 PPCODE:
1547{ 1730 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)); 1731 XPUSHs (ST (0));
1558}
1559 1732
1560U32 get_max_depth (JSON *self) 1733U32 get_max_depth (JSON *self)
1561 CODE: 1734 CODE:
1562 RETVAL = DEC_DEPTH (self->flags); 1735 RETVAL = self->max_depth;
1563 OUTPUT: 1736 OUTPUT:
1564 RETVAL 1737 RETVAL
1565 1738
1566void max_size (JSON *self, UV max_size = 0) 1739void max_size (JSON *self, U32 max_size = 0)
1567 PPCODE: 1740 PPCODE:
1568{ 1741 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)); 1742 XPUSHs (ST (0));
1580}
1581 1743
1582int get_max_size (JSON *self) 1744int get_max_size (JSON *self)
1583 CODE: 1745 CODE:
1584 RETVAL = DEC_SIZE (self->flags); 1746 RETVAL = self->max_size;
1585 OUTPUT: 1747 OUTPUT:
1586 RETVAL 1748 RETVAL
1587 1749
1588void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1750void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1589 PPCODE: 1751 PPCODE:
1625 XPUSHs (decode_json (jsonstr, self, 0)); 1787 XPUSHs (decode_json (jsonstr, self, 0));
1626 1788
1627void decode_prefix (JSON *self, SV *jsonstr) 1789void decode_prefix (JSON *self, SV *jsonstr)
1628 PPCODE: 1790 PPCODE:
1629{ 1791{
1630 UV offset; 1792 char *offset;
1631 EXTEND (SP, 2); 1793 EXTEND (SP, 2);
1632 PUSHs (decode_json (jsonstr, self, &offset)); 1794 PUSHs (decode_json (jsonstr, self, &offset));
1633 PUSHs (sv_2mortal (newSVuv (offset))); 1795 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1796}
1797
1798void incr_parse (JSON *self, SV *jsonstr = 0)
1799 PPCODE:
1800{
1801 if (!self->incr_text)
1802 self->incr_text = newSVpvn ("", 0);
1803
1804 // append data, if any
1805 if (jsonstr)
1806 {
1807 if (SvUTF8 (jsonstr))
1808 {
1809 if (!SvUTF8 (self->incr_text))
1810 {
1811 /* utf-8-ness differs, need to upgrade */
1812 sv_utf8_upgrade (self->incr_text);
1813
1814 if (self->incr_pos)
1815 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1816 - (U8 *)SvPVX (self->incr_text);
1817 }
1818 }
1819 else if (SvUTF8 (self->incr_text))
1820 sv_utf8_upgrade (jsonstr);
1821
1822 {
1823 STRLEN len;
1824 const char *str = SvPV (jsonstr, len);
1825 STRLEN cur = SvCUR (self->incr_text);
1826
1827 if (SvLEN (self->incr_text) <= cur + len)
1828 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1829
1830 Move (str, SvEND (self->incr_text), len, char);
1831 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1832 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1833 }
1834 }
1835
1836 if (GIMME_V != G_VOID)
1837 do
1838 {
1839 char *offset;
1840
1841 if (!INCR_DONE (self))
1842 {
1843 incr_parse (self);
1844
1845 if (self->incr_pos > self->max_size && self->max_size)
1846 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1847 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1848
1849 if (!INCR_DONE (self))
1850 break;
1851 }
1852
1853 XPUSHs (decode_json (self->incr_text, self, &offset));
1854
1855 self->incr_pos -= offset - SvPVX (self->incr_text);
1856 self->incr_nest = 0;
1857 self->incr_mode = 0;
1858
1859 sv_chop (self->incr_text, offset);
1860 }
1861 while (GIMME_V == G_ARRAY);
1862}
1863
1864SV *incr_text (JSON *self)
1865 ATTRS: lvalue
1866 CODE:
1867{
1868 if (self->incr_pos)
1869 croak ("incr_text can not be called when the incremental parser already started parsing");
1870
1871 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1872}
1873 OUTPUT:
1874 RETVAL
1875
1876void incr_skip (JSON *self)
1877 CODE:
1878{
1879 if (self->incr_pos)
1880 {
1881 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1882 self->incr_pos = 0;
1883 self->incr_nest = 0;
1884 self->incr_mode = 0;
1885 }
1886}
1887
1888void incr_reset (JSON *self)
1889 CODE:
1890{
1891 SvREFCNT_dec (self->incr_text);
1892 self->incr_text = 0;
1893 self->incr_pos = 0;
1894 self->incr_nest = 0;
1895 self->incr_mode = 0;
1634} 1896}
1635 1897
1636void DESTROY (JSON *self) 1898void DESTROY (JSON *self)
1637 CODE: 1899 CODE:
1638 SvREFCNT_dec (self->cb_sk_object); 1900 SvREFCNT_dec (self->cb_sk_object);
1639 SvREFCNT_dec (self->cb_object); 1901 SvREFCNT_dec (self->cb_object);
1902 SvREFCNT_dec (self->incr_text);
1640 1903
1641PROTOTYPES: ENABLE 1904PROTOTYPES: ENABLE
1642 1905
1643void encode_json (SV *scalar) 1906void encode_json (SV *scalar)
1907 ALIAS:
1908 to_json_ = 0
1909 encode_json = F_UTF8
1644 PPCODE: 1910 PPCODE:
1645{ 1911{
1646 JSON json = { F_DEFAULT | F_UTF8 }; 1912 JSON json;
1913 json_init (&json);
1914 json.flags |= ix;
1647 XPUSHs (encode_json (scalar, &json)); 1915 XPUSHs (encode_json (scalar, &json));
1648} 1916}
1649 1917
1650void decode_json (SV *jsonstr) 1918void decode_json (SV *jsonstr)
1919 ALIAS:
1920 from_json_ = 0
1921 decode_json = F_UTF8
1651 PPCODE: 1922 PPCODE:
1652{ 1923{
1653 JSON json = { F_DEFAULT | F_UTF8 }; 1924 JSON json;
1925 json_init (&json);
1926 json.flags |= ix;
1654 XPUSHs (decode_json (jsonstr, &json, 0)); 1927 XPUSHs (decode_json (jsonstr, &json, 0));
1655} 1928}
1656 1929
1930

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines