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 1.122 by root, Tue Oct 29 15:55:49 2013 UTC vs.
Revision 1.142 by root, Mon Jun 21 00:12:57 2021 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
… …
49#define F_ALLOW_TAGS 0x00004000UL 50#define F_ALLOW_TAGS 0x00004000UL
50#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 51#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
51 52
52#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 53#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
53 54
54#define INIT_SIZE 32 // initial scalar size to be allocated 55#define INIT_SIZE 64 // initial scalar size to be allocated
55#define INDENT_STEP 3 // spaces per indentation level 56#define INDENT_STEP 3 // spaces per indentation level
56 57
57#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
58 59
59#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) 60#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
… …
77 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 78 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
78 79
79#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?)"
80 81
81#ifdef USE_ITHREADS 82#ifdef USE_ITHREADS
82# define JSON_SLOW 1
83# 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))
84#else 86#else
85# define JSON_SLOW 0
86# define JSON_STASH json_stash 87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash
89# define GET_BOOL(value) bool_ ## value
87#endif 90#endif
88 91
89// 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
90#define STACK_HES 64 93#define STACK_HES 64
91 94
92static HV *json_stash, *types_boolean_stash; // JSON::XS:: 95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
93static SV *types_true, *types_false, *sv_json; 96static SV *bool_false, *bool_true;
97static SV *sv_json;
94 98
95enum { 99enum {
96 INCR_M_WS = 0, // initial whitespace skipping, must be 0 100 INCR_M_WS = 0, // initial whitespace skipping, must be 0
101 INCR_M_TFN, // inside true/false/null
102 INCR_M_NUM, // inside number
97 INCR_M_STR, // inside string 103 INCR_M_STR, // inside string
98 INCR_M_BS, // inside backslash 104 INCR_M_BS, // inside backslash
99 INCR_M_C0, // inside comment in initial whitespace sequence 105 INCR_M_C0, // inside comment in initial whitespace sequence
100 INCR_M_C1, // inside comment in other places 106 INCR_M_C1, // inside comment in other places
101 INCR_M_JSON // outside anything, count nesting 107 INCR_M_JSON // outside anything, count nesting
… …
114 // for the incremental parser 120 // for the incremental parser
115 SV *incr_text; // the source text so far 121 SV *incr_text; // the source text so far
116 STRLEN incr_pos; // the current offset into the text 122 STRLEN incr_pos; // the current offset into the text
117 int incr_nest; // {[]}-nesting level 123 int incr_nest; // {[]}-nesting level
118 unsigned char incr_mode; 124 unsigned char incr_mode;
125
126 SV *v_false, *v_true;
119} JSON; 127} JSON;
120 128
121INLINE void 129INLINE void
122json_init (JSON *json) 130json_init (JSON *json)
123{ 131{
124 Zero (json, 1, JSON); 132 static const JSON init = { F_ALLOW_NONREF, 512 };
125 json->max_depth = 512; 133
134 *json = init;
126} 135}
127 136
128///////////////////////////////////////////////////////////////////////////// 137/////////////////////////////////////////////////////////////////////////////
129// utility functions 138// utility functions
130 139
… …
152 SvPV_renew (sv, SvCUR (sv) + 1); 161 SvPV_renew (sv, SvCUR (sv) + 1);
153#endif 162#endif
154 } 163 }
155} 164}
156 165
166/* adds two STRLENs together, slow, and with paranoia */
167static STRLEN
168strlen_sum (STRLEN l1, STRLEN l2)
169{
170 size_t sum = l1 + l2;
171
172 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
173 croak ("JSON::XS: string size overflow");
174
175 return sum;
176}
177
178/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
179static char *
180json_sv_grow (SV *sv, size_t len1, size_t len2)
181{
182 len1 = strlen_sum (len1, len2);
183 len1 = strlen_sum (len1, len1 >> 1);
184
185 if (len1 > 4096 - 24)
186 len1 = (len1 | 4095) - 24;
187
188 return SvGROW (sv, len1);
189}
190
157// decode an utf-8 character and return it, or (UV)-1 in 191// decode a utf-8 character and return it, or (UV)-1 in
158// case of an error. 192// case of an error.
159// we special-case "safe" characters from U+80 .. U+7FF, 193// we special-case "safe" characters from U+80 .. U+7FF,
160// but use the very good perl function to parse anything else. 194// but use the very good perl function to parse anything else.
161// note that we never call this function for a ascii codepoints 195// note that we never call this function for a ascii codepoints
162INLINE UV 196INLINE UV
… …
299 // a recursion depth of ten gives us >>500 bits 333 // a recursion depth of ten gives us >>500 bits
300 json_atof_scan1 (s, &accum, &expo, 0, 10); 334 json_atof_scan1 (s, &accum, &expo, 0, 10);
301 335
302 return neg ? -accum : accum; 336 return neg ? -accum : accum;
303} 337}
338
339// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
340static int
341ref_bool_type (SV *sv)
342{
343 svtype svt = SvTYPE (sv);
344
345 if (svt < SVt_PVAV)
346 {
347 STRLEN len = 0;
348 char *pv = svt ? SvPV (sv, len) : 0;
349
350 if (len == 1)
351 if (*pv == '1')
352 return 1;
353 else if (*pv == '0')
354 return 0;
355 }
356
357 return -1;
358}
359
360// returns whether scalar is not a reference in the sense of allow_nonref
361static int
362json_nonref (SV *scalar)
363{
364 if (!SvROK (scalar))
365 return 1;
366
367 scalar = SvRV (scalar);
368
369 if (SvTYPE (scalar) >= SVt_PVMG)
370 {
371 if (SvSTASH (scalar) == bool_stash)
372 return 1;
373
374 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
375 return 1;
376 }
377
378 return 0;
379}
380
304///////////////////////////////////////////////////////////////////////////// 381/////////////////////////////////////////////////////////////////////////////
305// encoder 382// encoder
306 383
307// structure used for encoding JSON 384// structure used for encoding JSON
308typedef struct 385typedef struct
… …
316} enc_t; 393} enc_t;
317 394
318INLINE void 395INLINE void
319need (enc_t *enc, STRLEN len) 396need (enc_t *enc, STRLEN len)
320{ 397{
321 if (expect_false (enc->cur + len >= enc->end)) 398 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
322 { 399 {
323 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 400 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
324 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 401 char *buf = json_sv_grow (enc->sv, cur, len);
325 enc->cur = SvPVX (enc->sv) + cur; 402 enc->cur = buf + cur;
326 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 403 enc->end = buf + SvLEN (enc->sv) - 1;
327 } 404 }
328} 405}
329 406
330INLINE void 407INLINE void
331encode_ch (enc_t *enc, char ch) 408encode_ch (enc_t *enc, char ch)
… …
347 424
348 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 425 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
349 { 426 {
350 if (expect_false (ch == '"')) // but with slow exceptions 427 if (expect_false (ch == '"')) // but with slow exceptions
351 { 428 {
352 need (enc, len += 1); 429 need (enc, len + 1);
353 *enc->cur++ = '\\'; 430 *enc->cur++ = '\\';
354 *enc->cur++ = '"'; 431 *enc->cur++ = '"';
355 } 432 }
356 else if (expect_false (ch == '\\')) 433 else if (expect_false (ch == '\\'))
357 { 434 {
358 need (enc, len += 1); 435 need (enc, len + 1);
359 *enc->cur++ = '\\'; 436 *enc->cur++ = '\\';
360 *enc->cur++ = '\\'; 437 *enc->cur++ = '\\';
361 } 438 }
362 else 439 else
363 *enc->cur++ = ch; 440 *enc->cur++ = ch;
… …
366 } 443 }
367 else 444 else
368 { 445 {
369 switch (ch) 446 switch (ch)
370 { 447 {
371 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 448 case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
372 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; 449 case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
373 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; 450 case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
374 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; 451 case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
375 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; 452 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
376 453
377 default: 454 default:
378 { 455 {
379 STRLEN clen; 456 STRLEN clen;
380 UV uch; 457 UV uch;
… …
396 if (uch >= 0x10000UL) 473 if (uch >= 0x10000UL)
397 { 474 {
398 if (uch >= 0x110000UL) 475 if (uch >= 0x110000UL)
399 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 476 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
400 477
401 need (enc, len += 11); 478 need (enc, len + 11);
402 sprintf (enc->cur, "\\u%04x\\u%04x", 479 sprintf (enc->cur, "\\u%04x\\u%04x",
403 (int)((uch - 0x10000) / 0x400 + 0xD800), 480 (int)((uch - 0x10000) / 0x400 + 0xD800),
404 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 481 (int)((uch - 0x10000) % 0x400 + 0xDC00));
405 enc->cur += 12; 482 enc->cur += 12;
406 } 483 }
407 else 484 else
408 { 485 {
409 need (enc, len += 5); 486 need (enc, len + 5);
410 *enc->cur++ = '\\'; 487 *enc->cur++ = '\\';
411 *enc->cur++ = 'u'; 488 *enc->cur++ = 'u';
412 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 489 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
413 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 490 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
414 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 491 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
… …
422 *enc->cur++ = uch; 499 *enc->cur++ = uch;
423 str += clen; 500 str += clen;
424 } 501 }
425 else if (is_utf8) 502 else if (is_utf8)
426 { 503 {
427 need (enc, len += clen); 504 need (enc, len + clen);
428 do 505 do
429 { 506 {
430 *enc->cur++ = *str++; 507 *enc->cur++ = *str++;
431 } 508 }
432 while (--clen); 509 while (--clen);
433 } 510 }
434 else 511 else
435 { 512 {
436 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 513 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
437 enc->cur = encode_utf8 (enc->cur, uch); 514 enc->cur = encode_utf8 (enc->cur, uch);
438 ++str; 515 ++str;
439 } 516 }
440 } 517 }
441 } 518 }
… …
701 SvGETMAGIC (sv); 778 SvGETMAGIC (sv);
702 svt = SvTYPE (sv); 779 svt = SvTYPE (sv);
703 780
704 if (expect_false (SvOBJECT (sv))) 781 if (expect_false (SvOBJECT (sv)))
705 { 782 {
706 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
707 ? types_boolean_stash
708 : gv_stashpv ("Types::Serialiser::Boolean", 1);
709 HV *stash = SvSTASH (sv); 783 HV *stash = SvSTASH (sv);
710 784
711 if (stash == boolean_stash) 785 if (stash == bool_stash)
712 { 786 {
713 if (SvIV (sv))
714 encode_str (enc, "true", 4, 0); 787 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
715 else
716 encode_str (enc, "false", 5, 0); 788 else encode_str (enc, "false", 5, 0);
717 } 789 }
718 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0))) 790 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
719 { 791 {
720 int count; 792 int count;
721 dSP; 793 dSP;
722 794
723 ENTER; SAVETMPS; PUSHMARK (SP); 795 ENTER; SAVETMPS;
796 PUSHMARK (SP);
724 EXTEND (SP, 2); 797 EXTEND (SP, 2);
725 // we re-bless the reference to get overload and other niceties right 798 // we re-bless the reference to get overload and other niceties right
726 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 799 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
727 PUSHs (sv_json); 800 PUSHs (sv_json);
728 801
… …
730 count = call_sv ((SV *)GvCV (method), G_ARRAY); 803 count = call_sv ((SV *)GvCV (method), G_ARRAY);
731 SPAGAIN; 804 SPAGAIN;
732 805
733 // catch this surprisingly common error 806 // catch this surprisingly common error
734 if (SvROK (TOPs) && SvRV (TOPs) == sv) 807 if (SvROK (TOPs) && SvRV (TOPs) == sv)
735 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 808 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
736 809
737 encode_ch (enc, '('); 810 encode_ch (enc, '(');
738 encode_ch (enc, '"'); 811 encode_ch (enc, '"');
739 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); 812 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
740 encode_ch (enc, '"'); 813 encode_ch (enc, '"');
741 encode_ch (enc, ')'); 814 encode_ch (enc, ')');
742 encode_ch (enc, '['); 815 encode_ch (enc, '[');
743 816
744 while (count) 817 if (count)
745 { 818 {
819 int i;
820
821 for (i = 0; i < count - 1; ++i)
822 {
746 encode_sv (enc, SP[1 - count--]); 823 encode_sv (enc, SP[i + 1 - count]);
747
748 if (count)
749 encode_ch (enc, ','); 824 encode_ch (enc, ',');
825 }
826
827 encode_sv (enc, TOPs);
828 SP -= count;
750 } 829 }
751 830
831 PUTBACK;
832
752 encode_ch (enc, ']'); 833 encode_ch (enc, ']');
753
754 PUTBACK;
755 834
756 FREETMPS; LEAVE; 835 FREETMPS; LEAVE;
757 } 836 }
758 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0))) 837 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
759 { 838 {
760 dSP; 839 dSP;
761 840
762 ENTER; SAVETMPS; PUSHMARK (SP); 841 ENTER; SAVETMPS;
842 PUSHMARK (SP);
763 // we re-bless the reference to get overload and other niceties right 843 // we re-bless the reference to get overload and other niceties right
764 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 844 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
765 845
766 // calling with G_SCALAR ensures that we always get a 1 return value 846 // calling with G_SCALAR ensures that we always get a 1 return value
767 PUTBACK; 847 PUTBACK;
… …
789 encode_hv (enc, (HV *)sv); 869 encode_hv (enc, (HV *)sv);
790 else if (svt == SVt_PVAV) 870 else if (svt == SVt_PVAV)
791 encode_av (enc, (AV *)sv); 871 encode_av (enc, (AV *)sv);
792 else if (svt < SVt_PVAV) 872 else if (svt < SVt_PVAV)
793 { 873 {
794 STRLEN len = 0; 874 int bool_type = ref_bool_type (sv);
795 char *pv = svt ? SvPV (sv, len) : 0;
796 875
797 if (len == 1 && *pv == '1') 876 if (bool_type == 1)
798 encode_str (enc, "true", 4, 0); 877 encode_str (enc, "true", 4, 0);
799 else if (len == 1 && *pv == '0') 878 else if (bool_type == 0)
800 encode_str (enc, "false", 5, 0); 879 encode_str (enc, "false", 5, 0);
801 else if (enc->json.flags & F_ALLOW_UNKNOWN) 880 else if (enc->json.flags & F_ALLOW_UNKNOWN)
802 encode_str (enc, "null", 4, 0); 881 encode_str (enc, "null", 4, 0);
803 else 882 else
804 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 883 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
… …
839 if (SvIsUV (sv) ? SvUVX (sv) <= 59000 918 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
840 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) 919 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
841 { 920 {
842 // optimise the "small number case" 921 // optimise the "small number case"
843 // code will likely be branchless and use only a single multiplication 922 // code will likely be branchless and use only a single multiplication
844 // works for numbers up to 59074 923 // 4.28 works for numbers up to 59074, 4.29 works up to 59076
845 I32 i = SvIVX (sv); 924 I32 i = SvIVX (sv);
846 U32 u; 925 U32 u;
847 char digit, nz = 0; 926 char digit, nz = 0;
848 927
849 need (enc, 6); 928 need (enc, 6);
… …
887static SV * 966static SV *
888encode_json (SV *scalar, JSON *json) 967encode_json (SV *scalar, JSON *json)
889{ 968{
890 enc_t enc; 969 enc_t enc;
891 970
892 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 971 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
893 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 972 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
894 973
895 enc.json = *json; 974 enc.json = *json;
896 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 975 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
897 enc.cur = SvPVX (enc.sv); 976 enc.cur = SvPVX (enc.sv);
… …
959 else 1038 else
960 break; 1039 break;
961 } 1040 }
962 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) 1041 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
963 break; // parse error, but let higher level handle it, gives better error messages 1042 break; // parse error, but let higher level handle it, gives better error messages
964 1043 else
965 ++dec->cur; 1044 ++dec->cur;
966 } 1045 }
967} 1046}
968 1047
969#define ERR(reason) SB dec->err = reason; goto fail; SE 1048#define ERR(reason) SB dec->err = reason; goto fail; SE
970 1049
… …
1104 *cur++ = *dec_cur++; 1183 *cur++ = *dec_cur++;
1105 while (--clen); 1184 while (--clen);
1106 1185
1107 utf8 = 1; 1186 utf8 = 1;
1108 } 1187 }
1188 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1189 *cur++ = ch;
1109 else 1190 else
1110 { 1191 {
1111 --dec_cur; 1192 --dec_cur;
1112 1193
1113 if (!ch) 1194 if (!ch)
… …
1123 1204
1124 if (sv) 1205 if (sv)
1125 { 1206 {
1126 STRLEN cur = SvCUR (sv); 1207 STRLEN cur = SvCUR (sv);
1127 1208
1128 if (SvLEN (sv) <= cur + len) 1209 if (SvLEN (sv) - cur <= len)
1129 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 1210 json_sv_grow (sv, cur, len);
1130 1211
1131 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1212 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1132 SvCUR_set (sv, SvCUR (sv) + len); 1213 SvCUR_set (sv, SvCUR (sv) + len);
1133 } 1214 }
1134 else 1215 else
… …
1425 1506
1426 DEC_DEC_DEPTH; 1507 DEC_DEC_DEPTH;
1427 sv = newRV_noinc ((SV *)hv); 1508 sv = newRV_noinc ((SV *)hv);
1428 1509
1429 // check filter callbacks 1510 // check filter callbacks
1430 if (dec->json.flags & F_HOOK) 1511 if (expect_false (dec->json.flags & F_HOOK))
1431 { 1512 {
1432 if (dec->json.cb_sk_object && HvKEYS (hv) == 1) 1513 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1433 { 1514 {
1434 HE *cb, *he; 1515 HE *cb, *he;
1435 1516
1436 hv_iterinit (hv); 1517 hv_iterinit (hv);
1437 he = hv_iternext (hv); 1518 he = hv_iternext (hv);
1438 hv_iterinit (hv); 1519 hv_iterinit (hv);
1439 1520
1440 // the next line creates a mortal sv each time its called. 1521 // the next line creates a mortal sv each time it's called.
1441 // might want to optimise this for common cases. 1522 // might want to optimise this for common cases.
1442 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1523 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1443 1524
1444 if (cb) 1525 if (cb)
1445 { 1526 {
1446 dSP; 1527 dSP;
1447 int count; 1528 int count;
1448 1529
1449 ENTER; SAVETMPS; PUSHMARK (SP); 1530 ENTER; SAVETMPS;
1531 PUSHMARK (SP);
1450 XPUSHs (HeVAL (he)); 1532 XPUSHs (HeVAL (he));
1451 sv_2mortal (sv); 1533 sv_2mortal (sv);
1452 1534
1453 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1535 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1454 1536
1455 if (count == 1) 1537 if (count == 1)
1456 { 1538 {
1457 sv = newSVsv (POPs); 1539 sv = newSVsv (POPs);
1540 PUTBACK;
1458 FREETMPS; LEAVE; 1541 FREETMPS; LEAVE;
1459 return sv; 1542 return sv;
1460 } 1543 }
1544 else if (count)
1545 croak ("filter_json_single_key_object callbacks must not return more than one scalar");
1546
1547 PUTBACK;
1461 1548
1462 SvREFCNT_inc (sv); 1549 SvREFCNT_inc (sv);
1550
1463 FREETMPS; LEAVE; 1551 FREETMPS; LEAVE;
1464 } 1552 }
1465 } 1553 }
1466 1554
1467 if (dec->json.cb_object) 1555 if (dec->json.cb_object)
1468 { 1556 {
1469 dSP; 1557 dSP;
1470 int count; 1558 int count;
1471 1559
1472 ENTER; SAVETMPS; PUSHMARK (SP); 1560 ENTER; SAVETMPS;
1561 PUSHMARK (SP);
1473 XPUSHs (sv_2mortal (sv)); 1562 XPUSHs (sv_2mortal (sv));
1474 1563
1475 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1564 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1476 1565
1477 if (count == 1) 1566 if (count == 1)
1478 {
1479 sv = newSVsv (POPs); 1567 sv = newSVsv (POPs);
1480 FREETMPS; LEAVE; 1568 else if (count == 0)
1481 return sv;
1482 }
1483
1484 SvREFCNT_inc (sv); 1569 SvREFCNT_inc (sv);
1570 else
1571 croak ("filter_json_object callbacks must not return more than one scalar");
1572
1573 PUTBACK;
1574
1485 FREETMPS; LEAVE; 1575 FREETMPS; LEAVE;
1486 } 1576 }
1487 } 1577 }
1488 1578
1489 return sv; 1579 return sv;
… …
1544 if (!method) 1634 if (!method)
1545 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1635 ERR ("cannot decode perl-object (package does not have a THAW method)");
1546 1636
1547 dSP; 1637 dSP;
1548 1638
1549 ENTER; SAVETMPS; PUSHMARK (SP); 1639 ENTER; SAVETMPS;
1640 PUSHMARK (SP);
1550 EXTEND (SP, len + 2); 1641 EXTEND (SP, len + 2);
1551 // we re-bless the reference to get overload and other niceties right 1642 // we re-bless the reference to get overload and other niceties right
1552 PUSHs (tag); 1643 PUSHs (tag);
1553 PUSHs (sv_json); 1644 PUSHs (sv_json);
1554 1645
… …
1591 case '-': 1682 case '-':
1592 case '0': case '1': case '2': case '3': case '4': 1683 case '0': case '1': case '2': case '3': case '4':
1593 case '5': case '6': case '7': case '8': case '9': 1684 case '5': case '6': case '7': case '8': case '9':
1594 return decode_num (dec); 1685 return decode_num (dec);
1595 1686
1687 case 'f':
1688 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1689 {
1690 dec->cur += 5;
1691
1692 if (expect_false (!dec->json.v_false))
1693 dec->json.v_false = GET_BOOL (false);
1694
1695 return newSVsv (dec->json.v_false);
1696 }
1697 else
1698 ERR ("'false' expected");
1699
1700 break;
1701
1596 case 't': 1702 case 't':
1597 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1703 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1598 { 1704 {
1599 dec->cur += 4; 1705 dec->cur += 4;
1600#if JSON_SLOW 1706
1601 types_true = get_bool ("Types::Serialiser::true"); 1707 if (expect_false (!dec->json.v_true))
1602#endif 1708 dec->json.v_true = GET_BOOL (true);
1709
1603 return newSVsv (types_true); 1710 return newSVsv (dec->json.v_true);
1604 } 1711 }
1605 else 1712 else
1606 ERR ("'true' expected"); 1713 ERR ("'true' expected");
1607
1608 break;
1609
1610 case 'f':
1611 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1612 {
1613 dec->cur += 5;
1614#if JSON_SLOW
1615 types_false = get_bool ("Types::Serialiser::false");
1616#endif
1617 return newSVsv (types_false);
1618 }
1619 else
1620 ERR ("'false' expected");
1621 1714
1622 break; 1715 break;
1623 1716
1624 case 'n': 1717 case 'n':
1625 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1718 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
… …
1640fail: 1733fail:
1641 return 0; 1734 return 0;
1642} 1735}
1643 1736
1644static SV * 1737static SV *
1645decode_json (SV *string, JSON *json, char **offset_return) 1738decode_json (SV *string, JSON *json, STRLEN *offset_return)
1646{ 1739{
1647 dec_t dec; 1740 dec_t dec;
1648 SV *sv; 1741 SV *sv;
1649 1742
1650 /* work around bugs in 5.10 where manipulating magic values 1743 /* work around bugs in 5.10 where manipulating magic values
1651 * makes perl ignore the magic in subsequent accesses. 1744 * makes perl ignore the magic in subsequent accesses.
1652 * also make a copy of non-PV values, to get them into a clean 1745 * also make a copy of non-PV values, to get them into a clean
1653 * state (SvPV should do that, but it's buggy, see below). 1746 * state (SvPV should do that, but it's buggy, see below).
1747 *
1748 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1749 * as reported by Reini Urban.
1654 */ 1750 */
1655 /*SvGETMAGIC (string);*/ 1751 /*SvGETMAGIC (string);*/
1656 if (SvMAGICAL (string) || !SvPOK (string)) 1752 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1657 string = sv_2mortal (newSVsv (string)); 1753 string = sv_2mortal (newSVsv (string));
1658 1754
1659 SvUPGRADE (string, SVt_PV); 1755 SvUPGRADE (string, SVt_PV);
1660 1756
1661 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1757 /* work around a bug in perl 5.10, which causes SvCUR to fail an
… …
1700 1796
1701 decode_ws (&dec); 1797 decode_ws (&dec);
1702 sv = decode_sv (&dec); 1798 sv = decode_sv (&dec);
1703 1799
1704 if (offset_return) 1800 if (offset_return)
1705 *offset_return = dec.cur; 1801 *offset_return = dec.cur - SvPVX (string);
1706 1802 else if (sv)
1707 if (!(offset_return || !sv))
1708 { 1803 {
1709 // check for trailing garbage 1804 // check for trailing garbage
1710 decode_ws (&dec); 1805 decode_ws (&dec);
1711 1806
1712 if (*dec.cur) 1807 if (dec.cur != dec.end)
1713 { 1808 {
1714 dec.err = "garbage after JSON object"; 1809 dec.err = "garbage after JSON object";
1715 SvREFCNT_dec (sv); 1810 SvREFCNT_dec (sv);
1716 sv = 0; 1811 sv = 0;
1717 } 1812 }
… …
1736 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1831 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1737 } 1832 }
1738 1833
1739 sv = sv_2mortal (sv); 1834 sv = sv_2mortal (sv);
1740 1835
1741 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1836 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1742 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1837 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1743 1838
1744 return sv; 1839 return sv;
1745} 1840}
1746 1841
… …
1755 // the state machine here is a bit convoluted and could be simplified a lot 1850 // the state machine here is a bit convoluted and could be simplified a lot
1756 // but this would make it slower, so... 1851 // but this would make it slower, so...
1757 1852
1758 for (;;) 1853 for (;;)
1759 { 1854 {
1760 //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
1761 switch (self->incr_mode) 1855 switch (self->incr_mode)
1762 { 1856 {
1857 // reached end of a scalar, see if we are inside a nested structure or not
1858 end_of_scalar:
1859 self->incr_mode = INCR_M_JSON;
1860
1861 if (self->incr_nest) // end of a scalar inside array, object or tag
1862 goto incr_m_json;
1863 else // end of scalar outside structure, json text ends here
1864 goto interrupt;
1865
1763 // only used for initial whitespace skipping 1866 // only used for initial whitespace skipping
1764 case INCR_M_WS: 1867 case INCR_M_WS:
1765 for (;;) 1868 for (;;)
1766 { 1869 {
1767 if (*p > 0x20) 1870 if (*p > 0x20)
… …
1809 ++p; 1912 ++p;
1810 } 1913 }
1811 1914
1812 break; 1915 break;
1813 1916
1917 // inside true/false/null
1918 case INCR_M_TFN:
1919 incr_m_tfn:
1920 for (;;)
1921 switch (*p++)
1922 {
1923 case 'r': case 'u': case 'e': // tRUE, falsE, nUll
1924 case 'a': case 'l': case 's': // fALSe, nuLL
1925 // allowed
1926 break;
1927
1928 default:
1929 --p;
1930 goto end_of_scalar;
1931 }
1932
1933 // inside a number
1934 case INCR_M_NUM:
1935 incr_m_num:
1936 for (;;)
1937 switch (*p++)
1938 {
1939 case 'e': case 'E': case '.': case '+':
1940 case '-':
1941 case '0': case '1': case '2': case '3': case '4':
1942 case '5': case '6': case '7': case '8': case '9':
1943 // allowed
1944 break;
1945
1946 default:
1947 --p;
1948 goto end_of_scalar;
1949 }
1950
1814 // inside a string 1951 // inside a string
1815 case INCR_M_STR: 1952 case INCR_M_STR:
1816 incr_m_str: 1953 incr_m_str:
1817 for (;;) 1954 for (;;)
1818 { 1955 {
1819 if (*p == '"') 1956 if (*p == '"')
1820 { 1957 {
1821 ++p; 1958 ++p;
1822 self->incr_mode = INCR_M_JSON;
1823
1824 if (!self->incr_nest)
1825 goto interrupt;
1826
1827 goto incr_m_json; 1959 goto end_of_scalar;
1828 } 1960 }
1829 else if (*p == '\\') 1961 else if (*p == '\\')
1830 { 1962 {
1831 ++p; // "virtually" consumes character after \ 1963 ++p; // "virtually" consumes character after \
1832 1964
… …
1862 --p; // do not eat the whitespace, let the next round do it 1994 --p; // do not eat the whitespace, let the next round do it
1863 goto interrupt; 1995 goto interrupt;
1864 } 1996 }
1865 break; 1997 break;
1866 1998
1999 // the following three blocks handle scalars. this makes the parser
2000 // more strict than required inside arrays or objects, and could
2001 // be moved to a special case on the toplevel (except strings)
2002 case 't':
2003 case 'f':
2004 case 'n':
2005 self->incr_mode = INCR_M_TFN;
2006 goto incr_m_tfn;
2007
2008 case '-':
2009 case '0': case '1': case '2': case '3': case '4':
2010 case '5': case '6': case '7': case '8': case '9':
2011 self->incr_mode = INCR_M_NUM;
2012 goto incr_m_num;
2013
1867 case '"': 2014 case '"':
1868 self->incr_mode = INCR_M_STR; 2015 self->incr_mode = INCR_M_STR;
1869 goto incr_m_str; 2016 goto incr_m_str;
1870 2017
1871 case '[': 2018 case '[':
… …
1916 i >= '0' && i <= '9' ? i - '0' 2063 i >= '0' && i <= '9' ? i - '0'
1917 : i >= 'a' && i <= 'f' ? i - 'a' + 10 2064 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1918 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2065 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1919 : -1; 2066 : -1;
1920 2067
1921 json_stash = gv_stashpv ("JSON::XS" , 1); 2068 json_stash = gv_stashpv ("JSON::XS" , 1);
1922 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); 2069 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1923
1924 types_true = get_bool ("Types::Serialiser::true");
1925 types_false = get_bool ("Types::Serialiser::false"); 2070 bool_false = get_bool ("Types::Serialiser::false");
2071 bool_true = get_bool ("Types::Serialiser::true");
1926 2072
1927 sv_json = newSVpv ("JSON", 0); 2073 sv_json = newSVpv ("JSON", 0);
1928 SvREADONLY_on (sv_json); 2074 SvREADONLY_on (sv_json);
1929 2075
1930 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ 2076 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
… …
1932 2078
1933PROTOTYPES: DISABLE 2079PROTOTYPES: DISABLE
1934 2080
1935void CLONE (...) 2081void CLONE (...)
1936 CODE: 2082 CODE:
2083 // as long as these writes are atomic, the race should not matter
2084 // as existing threads either already use 0, or use the old value,
2085 // which is sitll correct for the initial thread.
1937 json_stash = 0; 2086 json_stash = 0;
1938 types_boolean_stash = 0; 2087 bool_stash = 0;
2088 bool_false = 0;
2089 bool_true = 0;
1939 2090
1940void new (char *klass) 2091void new (char *klass)
1941 PPCODE: 2092 PPCODE:
1942{ 2093{
1943 SV *pv = NEWSV (0, sizeof (JSON)); 2094 SV *pv = NEWSV (0, sizeof (JSON));
… …
1946 XPUSHs (sv_2mortal (sv_bless ( 2097 XPUSHs (sv_2mortal (sv_bless (
1947 newRV_noinc (pv), 2098 newRV_noinc (pv),
1948 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2099 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1949 ))); 2100 )));
1950} 2101}
2102
2103void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2104 PPCODE:
2105 self->v_false = newSVsv (v_false);
2106 self->v_true = newSVsv (v_true);
2107 XPUSHs (ST (0));
2108
2109void get_boolean_values (JSON *self)
2110 PPCODE:
2111 if (self->v_false && self->v_true)
2112 {
2113 EXTEND (SP, 2);
2114 PUSHs (self->v_false);
2115 PUSHs (self->v_true);
2116 }
1951 2117
1952void ascii (JSON *self, int enable = 1) 2118void ascii (JSON *self, int enable = 1)
1953 ALIAS: 2119 ALIAS:
1954 ascii = F_ASCII 2120 ascii = F_ASCII
1955 latin1 = F_LATIN1 2121 latin1 = F_LATIN1
… …
2060 2226
2061void decode_prefix (JSON *self, SV *jsonstr) 2227void decode_prefix (JSON *self, SV *jsonstr)
2062 PPCODE: 2228 PPCODE:
2063{ 2229{
2064 SV *sv; 2230 SV *sv;
2065 char *offset; 2231 STRLEN offset;
2066 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; 2232 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2067 EXTEND (SP, 2); 2233 EXTEND (SP, 2);
2068 PUSHs (sv); 2234 PUSHs (sv);
2069 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2235 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
2070} 2236}
2071 2237
2072void incr_parse (JSON *self, SV *jsonstr = 0) 2238void incr_parse (JSON *self, SV *jsonstr = 0)
2073 PPCODE: 2239 PPCODE:
2074{ 2240{
… …
2108 { 2274 {
2109 STRLEN len; 2275 STRLEN len;
2110 const char *str = SvPV (jsonstr, len); 2276 const char *str = SvPV (jsonstr, len);
2111 STRLEN cur = SvCUR (self->incr_text); 2277 STRLEN cur = SvCUR (self->incr_text);
2112 2278
2113 if (SvLEN (self->incr_text) <= cur + len) 2279 if (SvLEN (self->incr_text) - cur <= len)
2114 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2280 json_sv_grow (self->incr_text, cur, len);
2115 2281
2116 Move (str, SvEND (self->incr_text), len, char); 2282 Move (str, SvEND (self->incr_text), len, char);
2117 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2283 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2118 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 2284 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2119 } 2285 }
… …
2121 2287
2122 if (GIMME_V != G_VOID) 2288 if (GIMME_V != G_VOID)
2123 do 2289 do
2124 { 2290 {
2125 SV *sv; 2291 SV *sv;
2126 char *offset; 2292 STRLEN offset;
2127 2293
2128 if (!INCR_DONE (self)) 2294 if (!INCR_DONE (self))
2129 { 2295 {
2130 incr_parse (self); 2296 incr_parse (self);
2131 2297
… …
2147 } 2313 }
2148 2314
2149 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; 2315 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2150 XPUSHs (sv); 2316 XPUSHs (sv);
2151 2317
2152 self->incr_pos -= offset - SvPVX (self->incr_text); 2318 self->incr_pos -= offset;
2153 self->incr_nest = 0; 2319 self->incr_nest = 0;
2154 self->incr_mode = 0; 2320 self->incr_mode = 0;
2155 2321
2156 sv_chop (self->incr_text, offset); 2322 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2157 } 2323 }
2158 while (GIMME_V == G_ARRAY); 2324 while (GIMME_V == G_ARRAY);
2159} 2325}
2160 2326
2161SV *incr_text (JSON *self) 2327SV *incr_text (JSON *self)
… …
2192 self->incr_mode = 0; 2358 self->incr_mode = 0;
2193} 2359}
2194 2360
2195void DESTROY (JSON *self) 2361void DESTROY (JSON *self)
2196 CODE: 2362 CODE:
2363 SvREFCNT_dec (self->v_false);
2364 SvREFCNT_dec (self->v_true);
2197 SvREFCNT_dec (self->cb_sk_object); 2365 SvREFCNT_dec (self->cb_sk_object);
2198 SvREFCNT_dec (self->cb_object); 2366 SvREFCNT_dec (self->cb_object);
2199 SvREFCNT_dec (self->incr_text); 2367 SvREFCNT_dec (self->incr_text);
2200 2368
2201PROTOTYPES: ENABLE 2369PROTOTYPES: ENABLE

Diff Legend

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