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 1.120 by root, Mon Oct 28 23:23:10 2013 UTC vs.
Revision 1.139 by root, Wed Mar 6 07:32:06 2019 UTC

… …
6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
12#if defined(__BORLANDC__) || defined(_MSC_VER) 13#if defined(__BORLANDC__) || defined(_MSC_VER)
13# define snprintf _snprintf // C compilers have this in stdio.h 14# define snprintf _snprintf // C compilers have this in stdio.h
14#endif 15#endif
15 16
16// some old perls do not have this, try to make it work, no 17// some old perls do not have this, try to make it work, no
17// guarantees, though. if it breaks, you get to keep the pieces. 18// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES 19#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13 20# define UTF8_MAXBYTES 13
21#endif
22
23// compatibility with perl <5.18
24#ifndef HvNAMELEN_get
25# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
26#endif
27#ifndef HvNAMELEN
28# define HvNAMELEN(hv) HvNAMELEN_get (hv)
29#endif
30#ifndef HvNAMEUTF8
31# define HvNAMEUTF8(hv) 0
20#endif 32#endif
21 33
22// three extra for rounding, sign, and end of string 34// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) 35#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
24 36
… …
38#define F_ALLOW_TAGS 0x00004000UL 50#define F_ALLOW_TAGS 0x00004000UL
39#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 51#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
40 52
41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 53#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
42 54
43#define INIT_SIZE 32 // initial scalar size to be allocated 55#define INIT_SIZE 64 // initial scalar size to be allocated
44#define INDENT_STEP 3 // spaces per indentation level 56#define INDENT_STEP 3 // spaces per indentation level
45 57
46#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
47 59
48#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) 60#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
… …
66 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 78 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
67 79
68#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?)"
69 81
70#ifdef USE_ITHREADS 82#ifdef USE_ITHREADS
71# define JSON_SLOW 1
72# 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))
73#else 86#else
74# define JSON_SLOW 0
75# define JSON_STASH json_stash 87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash
89# define GET_BOOL(value) bool_ ## value
76#endif 90#endif
77 91
78// 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
79#define STACK_HES 64 93#define STACK_HES 64
80 94
81static HV *json_stash, *types_boolean_stash; // JSON::XS:: 95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
82static SV *types_true, *types_false, *sv_json; 96static SV *bool_false, *bool_true;
97static SV *sv_json;
83 98
84enum { 99enum {
85 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
86 INCR_M_STR, // inside string 103 INCR_M_STR, // inside string
87 INCR_M_BS, // inside backslash 104 INCR_M_BS, // inside backslash
88 INCR_M_C0, // inside comment in initial whitespace sequence 105 INCR_M_C0, // inside comment in initial whitespace sequence
89 INCR_M_C1, // inside comment in other places 106 INCR_M_C1, // inside comment in other places
90 INCR_M_JSON // outside anything, count nesting 107 INCR_M_JSON // outside anything, count nesting
… …
103 // for the incremental parser 120 // for the incremental parser
104 SV *incr_text; // the source text so far 121 SV *incr_text; // the source text so far
105 STRLEN incr_pos; // the current offset into the text 122 STRLEN incr_pos; // the current offset into the text
106 int incr_nest; // {[]}-nesting level 123 int incr_nest; // {[]}-nesting level
107 unsigned char incr_mode; 124 unsigned char incr_mode;
125
126 SV *v_false, *v_true;
108} JSON; 127} JSON;
109 128
110INLINE void 129INLINE void
111json_init (JSON *json) 130json_init (JSON *json)
112{ 131{
113 Zero (json, 1, JSON); 132 static const JSON init = { F_ALLOW_NONREF, 512 };
114 json->max_depth = 512; 133
134 *json = init;
115} 135}
116 136
117///////////////////////////////////////////////////////////////////////////// 137/////////////////////////////////////////////////////////////////////////////
118// utility functions 138// utility functions
119 139
… …
141 SvPV_renew (sv, SvCUR (sv) + 1); 161 SvPV_renew (sv, SvCUR (sv) + 1);
142#endif 162#endif
143 } 163 }
144} 164}
145 165
166/* adds two STRLENs together, slow, and with paranoia */
167STRLEN
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
146// 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
147// case of an error. 192// case of an error.
148// we special-case "safe" characters from U+80 .. U+7FF, 193// we special-case "safe" characters from U+80 .. U+7FF,
149// but use the very good perl function to parse anything else. 194// but use the very good perl function to parse anything else.
150// note that we never call this function for a ascii codepoints 195// note that we never call this function for a ascii codepoints
151INLINE UV 196INLINE UV
… …
288 // a recursion depth of ten gives us >>500 bits 333 // a recursion depth of ten gives us >>500 bits
289 json_atof_scan1 (s, &accum, &expo, 0, 10); 334 json_atof_scan1 (s, &accum, &expo, 0, 10);
290 335
291 return neg ? -accum : accum; 336 return neg ? -accum : accum;
292} 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
293///////////////////////////////////////////////////////////////////////////// 381/////////////////////////////////////////////////////////////////////////////
294// encoder 382// encoder
295 383
296// structure used for encoding JSON 384// structure used for encoding JSON
297typedef struct 385typedef struct
… …
305} enc_t; 393} enc_t;
306 394
307INLINE void 395INLINE void
308need (enc_t *enc, STRLEN len) 396need (enc_t *enc, STRLEN len)
309{ 397{
310 if (expect_false (enc->cur + len >= enc->end)) 398 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
311 { 399 {
312 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 400 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
313 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 401 char *buf = json_sv_grow (enc->sv, cur, len);
314 enc->cur = SvPVX (enc->sv) + cur; 402 enc->cur = buf + cur;
315 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 403 enc->end = buf + SvLEN (enc->sv) - 1;
316 } 404 }
317} 405}
318 406
319INLINE void 407INLINE void
320encode_ch (enc_t *enc, char ch) 408encode_ch (enc_t *enc, char ch)
… …
336 424
337 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 425 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
338 { 426 {
339 if (expect_false (ch == '"')) // but with slow exceptions 427 if (expect_false (ch == '"')) // but with slow exceptions
340 { 428 {
341 need (enc, len += 1); 429 need (enc, len + 1);
342 *enc->cur++ = '\\'; 430 *enc->cur++ = '\\';
343 *enc->cur++ = '"'; 431 *enc->cur++ = '"';
344 } 432 }
345 else if (expect_false (ch == '\\')) 433 else if (expect_false (ch == '\\'))
346 { 434 {
347 need (enc, len += 1); 435 need (enc, len + 1);
348 *enc->cur++ = '\\'; 436 *enc->cur++ = '\\';
349 *enc->cur++ = '\\'; 437 *enc->cur++ = '\\';
350 } 438 }
351 else 439 else
352 *enc->cur++ = ch; 440 *enc->cur++ = ch;
… …
355 } 443 }
356 else 444 else
357 { 445 {
358 switch (ch) 446 switch (ch)
359 { 447 {
360 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;
361 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;
362 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;
363 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;
364 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;
365 453
366 default: 454 default:
367 { 455 {
368 STRLEN clen; 456 STRLEN clen;
369 UV uch; 457 UV uch;
… …
385 if (uch >= 0x10000UL) 473 if (uch >= 0x10000UL)
386 { 474 {
387 if (uch >= 0x110000UL) 475 if (uch >= 0x110000UL)
388 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);
389 477
390 need (enc, len += 11); 478 need (enc, len + 11);
391 sprintf (enc->cur, "\\u%04x\\u%04x", 479 sprintf (enc->cur, "\\u%04x\\u%04x",
392 (int)((uch - 0x10000) / 0x400 + 0xD800), 480 (int)((uch - 0x10000) / 0x400 + 0xD800),
393 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 481 (int)((uch - 0x10000) % 0x400 + 0xDC00));
394 enc->cur += 12; 482 enc->cur += 12;
395 } 483 }
396 else 484 else
397 { 485 {
398 need (enc, len += 5); 486 need (enc, len + 5);
399 *enc->cur++ = '\\'; 487 *enc->cur++ = '\\';
400 *enc->cur++ = 'u'; 488 *enc->cur++ = 'u';
401 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 489 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
402 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 490 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
403 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 491 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
… …
411 *enc->cur++ = uch; 499 *enc->cur++ = uch;
412 str += clen; 500 str += clen;
413 } 501 }
414 else if (is_utf8) 502 else if (is_utf8)
415 { 503 {
416 need (enc, len += clen); 504 need (enc, len + clen);
417 do 505 do
418 { 506 {
419 *enc->cur++ = *str++; 507 *enc->cur++ = *str++;
420 } 508 }
421 while (--clen); 509 while (--clen);
422 } 510 }
423 else 511 else
424 { 512 {
425 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
426 enc->cur = encode_utf8 (enc->cur, uch); 514 enc->cur = encode_utf8 (enc->cur, uch);
427 ++str; 515 ++str;
428 } 516 }
429 } 517 }
430 } 518 }
… …
690 SvGETMAGIC (sv); 778 SvGETMAGIC (sv);
691 svt = SvTYPE (sv); 779 svt = SvTYPE (sv);
692 780
693 if (expect_false (SvOBJECT (sv))) 781 if (expect_false (SvOBJECT (sv)))
694 { 782 {
695 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
696 ? types_boolean_stash
697 : gv_stashpv ("Types::Serialiser::Boolean", 1);
698 HV *stash = SvSTASH (sv); 783 HV *stash = SvSTASH (sv);
699 784
700 if (stash == boolean_stash) 785 if (stash == bool_stash)
701 { 786 {
702 if (SvIV (sv))
703 encode_str (enc, "true", 4, 0); 787 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
704 else
705 encode_str (enc, "false", 5, 0); 788 else encode_str (enc, "false", 5, 0);
706 } 789 }
707 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)))
708 { 791 {
709 int count; 792 int count;
710 dSP; 793 dSP;
711 794
712 ENTER; SAVETMPS; PUSHMARK (SP); 795 ENTER; SAVETMPS;
796 PUSHMARK (SP);
713 EXTEND (SP, 2); 797 EXTEND (SP, 2);
714 // 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
715 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 799 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
716 PUSHs (sv_json); 800 PUSHs (sv_json);
717 801
… …
719 count = call_sv ((SV *)GvCV (method), G_ARRAY); 803 count = call_sv ((SV *)GvCV (method), G_ARRAY);
720 SPAGAIN; 804 SPAGAIN;
721 805
722 // catch this surprisingly common error 806 // catch this surprisingly common error
723 if (SvROK (TOPs) && SvRV (TOPs) == sv) 807 if (SvROK (TOPs) && SvRV (TOPs) == sv)
724 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)));
725 809
726 encode_ch (enc, '('); 810 encode_ch (enc, '(');
727 encode_ch (enc, '"'); 811 encode_ch (enc, '"');
728 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); 812 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
729 encode_ch (enc, '"'); 813 encode_ch (enc, '"');
730 encode_ch (enc, ')'); 814 encode_ch (enc, ')');
731 encode_ch (enc, '['); 815 encode_ch (enc, '[');
732 816
733 while (count) 817 if (count)
734 { 818 {
819 int i;
820
821 for (i = 0; i < count - 1; ++i)
822 {
735 encode_sv (enc, SP[1 - count--]); 823 encode_sv (enc, SP[i + 1 - count]);
736
737 if (count)
738 encode_ch (enc, ','); 824 encode_ch (enc, ',');
825 }
826
827 encode_sv (enc, TOPs);
828 SP -= count;
739 } 829 }
740 830
831 PUTBACK;
832
741 encode_ch (enc, ']'); 833 encode_ch (enc, ']');
742
743 PUTBACK;
744 834
745 FREETMPS; LEAVE; 835 FREETMPS; LEAVE;
746 } 836 }
747 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)))
748 { 838 {
749 dSP; 839 dSP;
750 840
751 ENTER; SAVETMPS; PUSHMARK (SP); 841 ENTER; SAVETMPS;
842 PUSHMARK (SP);
752 // 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
753 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 844 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
754 845
755 // 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
756 PUTBACK; 847 PUTBACK;
… …
778 encode_hv (enc, (HV *)sv); 869 encode_hv (enc, (HV *)sv);
779 else if (svt == SVt_PVAV) 870 else if (svt == SVt_PVAV)
780 encode_av (enc, (AV *)sv); 871 encode_av (enc, (AV *)sv);
781 else if (svt < SVt_PVAV) 872 else if (svt < SVt_PVAV)
782 { 873 {
783 STRLEN len = 0; 874 int bool_type = ref_bool_type (sv);
784 char *pv = svt ? SvPV (sv, len) : 0;
785 875
786 if (len == 1 && *pv == '1') 876 if (bool_type == 1)
787 encode_str (enc, "true", 4, 0); 877 encode_str (enc, "true", 4, 0);
788 else if (len == 1 && *pv == '0') 878 else if (bool_type == 0)
789 encode_str (enc, "false", 5, 0); 879 encode_str (enc, "false", 5, 0);
790 else if (enc->json.flags & F_ALLOW_UNKNOWN) 880 else if (enc->json.flags & F_ALLOW_UNKNOWN)
791 encode_str (enc, "null", 4, 0); 881 encode_str (enc, "null", 4, 0);
792 else 882 else
793 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",
… …
876static SV * 966static SV *
877encode_json (SV *scalar, JSON *json) 967encode_json (SV *scalar, JSON *json)
878{ 968{
879 enc_t enc; 969 enc_t enc;
880 970
881 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 971 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
882 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)");
883 973
884 enc.json = *json; 974 enc.json = *json;
885 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 975 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
886 enc.cur = SvPVX (enc.sv); 976 enc.cur = SvPVX (enc.sv);
… …
1093 *cur++ = *dec_cur++; 1183 *cur++ = *dec_cur++;
1094 while (--clen); 1184 while (--clen);
1095 1185
1096 utf8 = 1; 1186 utf8 = 1;
1097 } 1187 }
1188 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1189 *cur++ = ch;
1098 else 1190 else
1099 { 1191 {
1100 --dec_cur; 1192 --dec_cur;
1101 1193
1102 if (!ch) 1194 if (!ch)
… …
1112 1204
1113 if (sv) 1205 if (sv)
1114 { 1206 {
1115 STRLEN cur = SvCUR (sv); 1207 STRLEN cur = SvCUR (sv);
1116 1208
1117 if (SvLEN (sv) <= cur + len) 1209 if (SvLEN (sv) - cur <= len)
1118 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 1210 json_sv_grow (sv, cur, len);
1119 1211
1120 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1212 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1121 SvCUR_set (sv, SvCUR (sv) + len); 1213 SvCUR_set (sv, SvCUR (sv) + len);
1122 } 1214 }
1123 else 1215 else
… …
1414 1506
1415 DEC_DEC_DEPTH; 1507 DEC_DEC_DEPTH;
1416 sv = newRV_noinc ((SV *)hv); 1508 sv = newRV_noinc ((SV *)hv);
1417 1509
1418 // check filter callbacks 1510 // check filter callbacks
1419 if (dec->json.flags & F_HOOK) 1511 if (expect_false (dec->json.flags & F_HOOK))
1420 { 1512 {
1421 if (dec->json.cb_sk_object && HvKEYS (hv) == 1) 1513 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1422 { 1514 {
1423 HE *cb, *he; 1515 HE *cb, *he;
1424 1516
1425 hv_iterinit (hv); 1517 hv_iterinit (hv);
1426 he = hv_iternext (hv); 1518 he = hv_iternext (hv);
1427 hv_iterinit (hv); 1519 hv_iterinit (hv);
1428 1520
1429 // the next line creates a mortal sv each time its called. 1521 // the next line creates a mortal sv each time it's called.
1430 // might want to optimise this for common cases. 1522 // might want to optimise this for common cases.
1431 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);
1432 1524
1433 if (cb) 1525 if (cb)
1434 { 1526 {
1435 dSP; 1527 dSP;
1436 int count; 1528 int count;
1437 1529
1438 ENTER; SAVETMPS; PUSHMARK (SP); 1530 ENTER; SAVETMPS;
1531 PUSHMARK (SP);
1439 XPUSHs (HeVAL (he)); 1532 XPUSHs (HeVAL (he));
1440 sv_2mortal (sv); 1533 sv_2mortal (sv);
1441 1534
1442 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1535 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1443 1536
1444 if (count == 1) 1537 if (count == 1)
1445 { 1538 {
1446 sv = newSVsv (POPs); 1539 sv = newSVsv (POPs);
1540 PUTBACK;
1447 FREETMPS; LEAVE; 1541 FREETMPS; LEAVE;
1448 return sv; 1542 return sv;
1449 } 1543 }
1544 else if (count)
1545 croak ("filter_json_single_key_object callbacks must not return more than one scalar");
1546
1547 PUTBACK;
1450 1548
1451 SvREFCNT_inc (sv); 1549 SvREFCNT_inc (sv);
1550
1452 FREETMPS; LEAVE; 1551 FREETMPS; LEAVE;
1453 } 1552 }
1454 } 1553 }
1455 1554
1456 if (dec->json.cb_object) 1555 if (dec->json.cb_object)
1457 { 1556 {
1458 dSP; 1557 dSP;
1459 int count; 1558 int count;
1460 1559
1461 ENTER; SAVETMPS; PUSHMARK (SP); 1560 ENTER; SAVETMPS;
1561 PUSHMARK (SP);
1462 XPUSHs (sv_2mortal (sv)); 1562 XPUSHs (sv_2mortal (sv));
1463 1563
1464 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1564 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1465 1565
1466 if (count == 1) 1566 if (count == 1)
1467 {
1468 sv = newSVsv (POPs); 1567 sv = newSVsv (POPs);
1469 FREETMPS; LEAVE; 1568 else if (count == 0)
1470 return sv;
1471 }
1472
1473 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
1474 FREETMPS; LEAVE; 1575 FREETMPS; LEAVE;
1475 } 1576 }
1476 } 1577 }
1477 1578
1478 return sv; 1579 return sv;
… …
1492 if (!(dec->json.flags & F_ALLOW_TAGS)) 1593 if (!(dec->json.flags & F_ALLOW_TAGS))
1493 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1594 ERR ("malformed JSON string, neither array, object, number, string or atom");
1494 1595
1495 ++dec->cur; 1596 ++dec->cur;
1496 1597
1598 decode_ws (dec);
1599
1497 tag = decode_sv (dec); 1600 tag = decode_sv (dec);
1498 if (!tag) 1601 if (!tag)
1499 goto fail; 1602 goto fail;
1500 1603
1501 if (!SvPOK (tag)) 1604 if (!SvPOK (tag))
1502 ERR ("malformed JSON string, (tag) must be a string"); 1605 ERR ("malformed JSON string, (tag) must be a string");
1503 1606
1607 decode_ws (dec);
1608
1504 if (*dec->cur != ')') 1609 if (*dec->cur != ')')
1505 ERR (") expected after tag"); 1610 ERR (") expected after tag");
1506 1611
1507 ++dec->cur; 1612 ++dec->cur;
1613
1614 decode_ws (dec);
1508 1615
1509 val = decode_sv (dec); 1616 val = decode_sv (dec);
1510 if (!val) 1617 if (!val)
1511 goto fail; 1618 goto fail;
1512 1619
… …
1527 if (!method) 1634 if (!method)
1528 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)");
1529 1636
1530 dSP; 1637 dSP;
1531 1638
1532 ENTER; SAVETMPS; PUSHMARK (SP); 1639 ENTER; SAVETMPS;
1640 PUSHMARK (SP);
1533 EXTEND (SP, len + 2); 1641 EXTEND (SP, len + 2);
1534 // 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
1535 PUSHs (tag); 1643 PUSHs (tag);
1536 PUSHs (sv_json); 1644 PUSHs (sv_json);
1537 1645
… …
1574 case '-': 1682 case '-':
1575 case '0': case '1': case '2': case '3': case '4': 1683 case '0': case '1': case '2': case '3': case '4':
1576 case '5': case '6': case '7': case '8': case '9': 1684 case '5': case '6': case '7': case '8': case '9':
1577 return decode_num (dec); 1685 return decode_num (dec);
1578 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
1579 case 't': 1702 case 't':
1580 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1703 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1581 { 1704 {
1582 dec->cur += 4; 1705 dec->cur += 4;
1583#if JSON_SLOW 1706
1584 types_true = get_bool ("Types::Serialiser::true"); 1707 if (expect_false (!dec->json.v_true))
1585#endif 1708 dec->json.v_true = GET_BOOL (true);
1709
1586 return newSVsv (types_true); 1710 return newSVsv (dec->json.v_true);
1587 } 1711 }
1588 else 1712 else
1589 ERR ("'true' expected"); 1713 ERR ("'true' expected");
1590
1591 break;
1592
1593 case 'f':
1594 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1595 {
1596 dec->cur += 5;
1597#if JSON_SLOW
1598 types_false = get_bool ("Types::Serialiser::false");
1599#endif
1600 return newSVsv (types_false);
1601 }
1602 else
1603 ERR ("'false' expected");
1604 1714
1605 break; 1715 break;
1606 1716
1607 case 'n': 1717 case 'n':
1608 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1718 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
… …
1623fail: 1733fail:
1624 return 0; 1734 return 0;
1625} 1735}
1626 1736
1627static SV * 1737static SV *
1628decode_json (SV *string, JSON *json, char **offset_return) 1738decode_json (SV *string, JSON *json, STRLEN *offset_return)
1629{ 1739{
1630 dec_t dec; 1740 dec_t dec;
1631 SV *sv; 1741 SV *sv;
1632 1742
1633 /* work around bugs in 5.10 where manipulating magic values 1743 /* work around bugs in 5.10 where manipulating magic values
1634 * makes perl ignore the magic in subsequent accesses. 1744 * makes perl ignore the magic in subsequent accesses.
1635 * 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
1636 * 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.
1637 */ 1750 */
1638 /*SvGETMAGIC (string);*/ 1751 /*SvGETMAGIC (string);*/
1639 if (SvMAGICAL (string) || !SvPOK (string)) 1752 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1640 string = sv_2mortal (newSVsv (string)); 1753 string = sv_2mortal (newSVsv (string));
1641 1754
1642 SvUPGRADE (string, SVt_PV); 1755 SvUPGRADE (string, SVt_PV);
1643 1756
1644 /* 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
… …
1683 1796
1684 decode_ws (&dec); 1797 decode_ws (&dec);
1685 sv = decode_sv (&dec); 1798 sv = decode_sv (&dec);
1686 1799
1687 if (offset_return) 1800 if (offset_return)
1688 *offset_return = dec.cur; 1801 *offset_return = dec.cur - SvPVX (string);
1689 1802 else if (sv)
1690 if (!(offset_return || !sv))
1691 { 1803 {
1692 // check for trailing garbage 1804 // check for trailing garbage
1693 decode_ws (&dec); 1805 decode_ws (&dec);
1694 1806
1695 if (*dec.cur) 1807 if (dec.cur != dec.end)
1696 { 1808 {
1697 dec.err = "garbage after JSON object"; 1809 dec.err = "garbage after JSON object";
1698 SvREFCNT_dec (sv); 1810 SvREFCNT_dec (sv);
1699 sv = 0; 1811 sv = 0;
1700 } 1812 }
… …
1719 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1831 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1720 } 1832 }
1721 1833
1722 sv = sv_2mortal (sv); 1834 sv = sv_2mortal (sv);
1723 1835
1724 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1836 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1725 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)");
1726 1838
1727 return sv; 1839 return sv;
1728} 1840}
1729 1841
… …
1738 // 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
1739 // but this would make it slower, so... 1851 // but this would make it slower, so...
1740 1852
1741 for (;;) 1853 for (;;)
1742 { 1854 {
1743 //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
1744 switch (self->incr_mode) 1855 switch (self->incr_mode)
1745 { 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
1746 // only used for initial whitespace skipping 1866 // only used for initial whitespace skipping
1747 case INCR_M_WS: 1867 case INCR_M_WS:
1748 for (;;) 1868 for (;;)
1749 { 1869 {
1750 if (*p > 0x20) 1870 if (*p > 0x20)
… …
1792 ++p; 1912 ++p;
1793 } 1913 }
1794 1914
1795 break; 1915 break;
1796 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
1797 // inside a string 1951 // inside a string
1798 case INCR_M_STR: 1952 case INCR_M_STR:
1799 incr_m_str: 1953 incr_m_str:
1800 for (;;) 1954 for (;;)
1801 { 1955 {
1802 if (*p == '"') 1956 if (*p == '"')
1803 { 1957 {
1804 ++p; 1958 ++p;
1805 self->incr_mode = INCR_M_JSON;
1806
1807 if (!self->incr_nest)
1808 goto interrupt;
1809
1810 goto incr_m_json; 1959 goto end_of_scalar;
1811 } 1960 }
1812 else if (*p == '\\') 1961 else if (*p == '\\')
1813 { 1962 {
1814 ++p; // "virtually" consumes character after \ 1963 ++p; // "virtually" consumes character after \
1815 1964
… …
1845 --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
1846 goto interrupt; 1995 goto interrupt;
1847 } 1996 }
1848 break; 1997 break;
1849 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
1850 case '"': 2014 case '"':
1851 self->incr_mode = INCR_M_STR; 2015 self->incr_mode = INCR_M_STR;
1852 goto incr_m_str; 2016 goto incr_m_str;
1853 2017
1854 case '[': 2018 case '[':
… …
1899 i >= '0' && i <= '9' ? i - '0' 2063 i >= '0' && i <= '9' ? i - '0'
1900 : i >= 'a' && i <= 'f' ? i - 'a' + 10 2064 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1901 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2065 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1902 : -1; 2066 : -1;
1903 2067
1904 json_stash = gv_stashpv ("JSON::XS" , 1); 2068 json_stash = gv_stashpv ("JSON::XS" , 1);
1905 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); 2069 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1906
1907 types_true = get_bool ("Types::Serialiser::true");
1908 types_false = get_bool ("Types::Serialiser::false"); 2070 bool_false = get_bool ("Types::Serialiser::false");
2071 bool_true = get_bool ("Types::Serialiser::true");
1909 2072
1910 sv_json = newSVpv ("JSON", 0); 2073 sv_json = newSVpv ("JSON", 0);
1911 SvREADONLY_on (sv_json); 2074 SvREADONLY_on (sv_json);
1912 2075
1913 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 */
… …
1915 2078
1916PROTOTYPES: DISABLE 2079PROTOTYPES: DISABLE
1917 2080
1918void CLONE (...) 2081void CLONE (...)
1919 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.
1920 json_stash = 0; 2086 json_stash = 0;
1921 types_boolean_stash = 0; 2087 bool_stash = 0;
2088 bool_false = 0;
2089 bool_true = 0;
1922 2090
1923void new (char *klass) 2091void new (char *klass)
1924 PPCODE: 2092 PPCODE:
1925{ 2093{
1926 SV *pv = NEWSV (0, sizeof (JSON)); 2094 SV *pv = NEWSV (0, sizeof (JSON));
… …
1929 XPUSHs (sv_2mortal (sv_bless ( 2097 XPUSHs (sv_2mortal (sv_bless (
1930 newRV_noinc (pv), 2098 newRV_noinc (pv),
1931 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2099 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1932 ))); 2100 )));
1933} 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 }
1934 2117
1935void ascii (JSON *self, int enable = 1) 2118void ascii (JSON *self, int enable = 1)
1936 ALIAS: 2119 ALIAS:
1937 ascii = F_ASCII 2120 ascii = F_ASCII
1938 latin1 = F_LATIN1 2121 latin1 = F_LATIN1
… …
2043 2226
2044void decode_prefix (JSON *self, SV *jsonstr) 2227void decode_prefix (JSON *self, SV *jsonstr)
2045 PPCODE: 2228 PPCODE:
2046{ 2229{
2047 SV *sv; 2230 SV *sv;
2048 char *offset; 2231 STRLEN offset;
2049 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; 2232 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2050 EXTEND (SP, 2); 2233 EXTEND (SP, 2);
2051 PUSHs (sv); 2234 PUSHs (sv);
2052 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2235 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
2053} 2236}
2054 2237
2055void incr_parse (JSON *self, SV *jsonstr = 0) 2238void incr_parse (JSON *self, SV *jsonstr = 0)
2056 PPCODE: 2239 PPCODE:
2057{ 2240{
… …
2091 { 2274 {
2092 STRLEN len; 2275 STRLEN len;
2093 const char *str = SvPV (jsonstr, len); 2276 const char *str = SvPV (jsonstr, len);
2094 STRLEN cur = SvCUR (self->incr_text); 2277 STRLEN cur = SvCUR (self->incr_text);
2095 2278
2096 if (SvLEN (self->incr_text) <= cur + len) 2279 if (SvLEN (self->incr_text) - cur <= len)
2097 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2280 json_sv_grow (self->incr_text, cur, len);
2098 2281
2099 Move (str, SvEND (self->incr_text), len, char); 2282 Move (str, SvEND (self->incr_text), len, char);
2100 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2283 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2101 *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
2102 } 2285 }
… …
2104 2287
2105 if (GIMME_V != G_VOID) 2288 if (GIMME_V != G_VOID)
2106 do 2289 do
2107 { 2290 {
2108 SV *sv; 2291 SV *sv;
2109 char *offset; 2292 STRLEN offset;
2110 2293
2111 if (!INCR_DONE (self)) 2294 if (!INCR_DONE (self))
2112 { 2295 {
2113 incr_parse (self); 2296 incr_parse (self);
2114 2297
… …
2130 } 2313 }
2131 2314
2132 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; 2315 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2133 XPUSHs (sv); 2316 XPUSHs (sv);
2134 2317
2135 self->incr_pos -= offset - SvPVX (self->incr_text); 2318 self->incr_pos -= offset;
2136 self->incr_nest = 0; 2319 self->incr_nest = 0;
2137 self->incr_mode = 0; 2320 self->incr_mode = 0;
2138 2321
2139 sv_chop (self->incr_text, offset); 2322 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2140 } 2323 }
2141 while (GIMME_V == G_ARRAY); 2324 while (GIMME_V == G_ARRAY);
2142} 2325}
2143 2326
2144SV *incr_text (JSON *self) 2327SV *incr_text (JSON *self)
… …
2175 self->incr_mode = 0; 2358 self->incr_mode = 0;
2176} 2359}
2177 2360
2178void DESTROY (JSON *self) 2361void DESTROY (JSON *self)
2179 CODE: 2362 CODE:
2363 SvREFCNT_dec (self->v_false);
2364 SvREFCNT_dec (self->v_true);
2180 SvREFCNT_dec (self->cb_sk_object); 2365 SvREFCNT_dec (self->cb_sk_object);
2181 SvREFCNT_dec (self->cb_object); 2366 SvREFCNT_dec (self->cb_object);
2182 SvREFCNT_dec (self->incr_text); 2367 SvREFCNT_dec (self->incr_text);
2183 2368
2184PROTOTYPES: ENABLE 2369PROTOTYPES: ENABLE

Diff Legend

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