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 1.90 by root, Sun Jul 20 17:55:19 2008 UTC vs.
Revision 1.124 by root, Mon Nov 24 18:42:51 2014 UTC

… …
12#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
13# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
14#endif 14#endif
15 15
16// 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
17// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
20#endif 20#endif
21 21
22// compatibility with perl <5.18
23#ifndef HvNAMELEN_get
24# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
25#endif
26#ifndef HvNAMELEN
27# define HvNAMELEN(hv) HvNAMELEN_get (hv)
28#endif
29#ifndef HvNAMEUTF8
30# define HvNAMEUTF8(hv) 0
31#endif
32
33// three extra for rounding, sign, and end of string
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) 34#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
23 35
24#define F_ASCII 0x00000001UL 36#define F_ASCII 0x00000001UL
25#define F_LATIN1 0x00000002UL 37#define F_LATIN1 0x00000002UL
26#define F_UTF8 0x00000004UL 38#define F_UTF8 0x00000004UL
27#define F_INDENT 0x00000008UL 39#define F_INDENT 0x00000008UL
… …
32#define F_SHRINK 0x00000200UL 44#define F_SHRINK 0x00000200UL
33#define F_ALLOW_BLESSED 0x00000400UL 45#define F_ALLOW_BLESSED 0x00000400UL
34#define F_CONV_BLESSED 0x00000800UL 46#define F_CONV_BLESSED 0x00000800UL
35#define F_RELAXED 0x00001000UL 47#define F_RELAXED 0x00001000UL
36#define F_ALLOW_UNKNOWN 0x00002000UL 48#define F_ALLOW_UNKNOWN 0x00002000UL
49#define F_ALLOW_TAGS 0x00004000UL
37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 50#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
38 51
39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 52#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
40 53
41#define INIT_SIZE 32 // initial scalar size to be allocated 54#define INIT_SIZE 32 // initial scalar size to be allocated
42#define INDENT_STEP 3 // spaces per indentation level 55#define INDENT_STEP 3 // spaces per indentation level
43 56
44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 57#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
58
59#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
45 60
46#define SB do { 61#define SB do {
47#define SE } while (0) 62#define SE } while (0)
48 63
49#if __GNUC__ >= 3 64#if __GNUC__ >= 3
… …
69#else 84#else
70# define JSON_SLOW 0 85# define JSON_SLOW 0
71# define JSON_STASH json_stash 86# define JSON_STASH json_stash
72#endif 87#endif
73 88
89// the amount of HEs to allocate on the stack, when sorting keys
90#define STACK_HES 64
91
74static HV *json_stash, *json_boolean_stash; // JSON::XS:: 92static HV *json_stash, *types_boolean_stash; // JSON::XS::
75static SV *json_true, *json_false; 93static SV *types_true, *types_false, *sv_json;
76 94
77enum { 95enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0 96 INCR_M_WS = 0, // initial whitespace skipping, must be 0
79 INCR_M_STR, // inside string 97 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash 98 INCR_M_BS, // inside backslash
99 INCR_M_C0, // inside comment in initial whitespace sequence
100 INCR_M_C1, // inside comment in other places
81 INCR_M_JSON // outside anything, count nesting 101 INCR_M_JSON // outside anything, count nesting
82}; 102};
83 103
84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) 104#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85 105
… …
175 *s++ = 0x80 | ( ch & 0x3f); 195 *s++ = 0x80 | ( ch & 0x3f);
176 196
177 return s; 197 return s;
178} 198}
179 199
200// convert offset pointer to character index, sv must be string
201static STRLEN
202ptr_to_index (SV *sv, char *offset)
203{
204 return SvUTF8 (sv)
205 ? utf8_distance (offset, SvPVX (sv))
206 : offset - SvPVX (sv);
207}
208
209/////////////////////////////////////////////////////////////////////////////
210// fp hell
211
212// scan a group of digits, and a trailing exponent
213static void
214json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
215{
216 UV uaccum = 0;
217 int eaccum = 0;
218
219 // if we recurse too deep, skip all remaining digits
220 // to avoid a stack overflow attack
221 if (expect_false (--maxdepth <= 0))
222 while (((U8)*s - '0') < 10)
223 ++s;
224
225 for (;;)
226 {
227 U8 dig = (U8)*s - '0';
228
229 if (expect_false (dig >= 10))
230 {
231 if (dig == (U8)((U8)'.' - (U8)'0'))
232 {
233 ++s;
234 json_atof_scan1 (s, accum, expo, 1, maxdepth);
235 }
236 else if ((dig | ' ') == 'e' - '0')
237 {
238 int exp2 = 0;
239 int neg = 0;
240
241 ++s;
242
243 if (*s == '-')
244 {
245 ++s;
246 neg = 1;
247 }
248 else if (*s == '+')
249 ++s;
250
251 while ((dig = (U8)*s - '0') < 10)
252 exp2 = exp2 * 10 + *s++ - '0';
253
254 *expo += neg ? -exp2 : exp2;
255 }
256
257 break;
258 }
259
260 ++s;
261
262 uaccum = uaccum * 10 + dig;
263 ++eaccum;
264
265 // if we have too many digits, then recurse for more
266 // we actually do this for rather few digits
267 if (uaccum >= (UV_MAX - 9) / 10)
268 {
269 if (postdp) *expo -= eaccum;
270 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
271 if (postdp) *expo += eaccum;
272
273 break;
274 }
275 }
276
277 // this relies greatly on the quality of the pow ()
278 // implementation of the platform, but a good
279 // implementation is hard to beat.
280 // (IEEE 754 conformant ones are required to be exact)
281 if (postdp) *expo -= eaccum;
282 *accum += uaccum * Perl_pow (10., *expo);
283 *expo += eaccum;
284}
285
286static NV
287json_atof (const char *s)
288{
289 NV accum = 0.;
290 int expo = 0;
291 int neg = 0;
292
293 if (*s == '-')
294 {
295 ++s;
296 neg = 1;
297 }
298
299 // a recursion depth of ten gives us >>500 bits
300 json_atof_scan1 (s, &accum, &expo, 0, 10);
301
302 return neg ? -accum : accum;
303}
180///////////////////////////////////////////////////////////////////////////// 304/////////////////////////////////////////////////////////////////////////////
181// encoder 305// encoder
182 306
183// structure used for encoding JSON 307// structure used for encoding JSON
184typedef struct 308typedef struct
… …
194INLINE void 318INLINE void
195need (enc_t *enc, STRLEN len) 319need (enc_t *enc, STRLEN len)
196{ 320{
197 if (expect_false (enc->cur + len >= enc->end)) 321 if (expect_false (enc->cur + len >= enc->end))
198 { 322 {
199 STRLEN cur = enc->cur - SvPVX (enc->sv); 323 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
200 SvGROW (enc->sv, cur + len + 1); 324 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
201 enc->cur = SvPVX (enc->sv) + cur; 325 enc->cur = SvPVX (enc->sv) + cur;
202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 326 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
203 } 327 }
204} 328}
205 329
… …
280 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 404 (int)((uch - 0x10000) % 0x400 + 0xDC00));
281 enc->cur += 12; 405 enc->cur += 12;
282 } 406 }
283 else 407 else
284 { 408 {
285 static char hexdigit [16] = "0123456789abcdef";
286 need (enc, len += 5); 409 need (enc, len += 5);
287 *enc->cur++ = '\\'; 410 *enc->cur++ = '\\';
288 *enc->cur++ = 'u'; 411 *enc->cur++ = 'u';
289 *enc->cur++ = hexdigit [ uch >> 12 ]; 412 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
290 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 413 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
291 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 414 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
292 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 415 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
293 } 416 }
294 417
295 str += clen; 418 str += clen;
296 } 419 }
297 else if (enc->json.flags & F_LATIN1) 420 else if (enc->json.flags & F_LATIN1)
… …
372 495
373 if (enc->indent >= enc->json.max_depth) 496 if (enc->indent >= enc->json.max_depth)
374 croak (ERR_NESTING_EXCEEDED); 497 croak (ERR_NESTING_EXCEEDED);
375 498
376 encode_ch (enc, '['); 499 encode_ch (enc, '[');
377 500
378 if (len >= 0) 501 if (len >= 0)
379 { 502 {
380 encode_nl (enc); ++enc->indent; 503 encode_nl (enc); ++enc->indent;
381 504
382 for (i = 0; i <= len; ++i) 505 for (i = 0; i <= len; ++i)
… …
394 encode_comma (enc); 517 encode_comma (enc);
395 } 518 }
396 519
397 encode_nl (enc); --enc->indent; encode_indent (enc); 520 encode_nl (enc); --enc->indent; encode_indent (enc);
398 } 521 }
399 522
400 encode_ch (enc, ']'); 523 encode_ch (enc, ']');
401} 524}
402 525
403static void 526static void
404encode_hk (enc_t *enc, HE *he) 527encode_hk (enc_t *enc, HE *he)
… …
408 if (HeKLEN (he) == HEf_SVKEY) 531 if (HeKLEN (he) == HEf_SVKEY)
409 { 532 {
410 SV *sv = HeSVKEY (he); 533 SV *sv = HeSVKEY (he);
411 STRLEN len; 534 STRLEN len;
412 char *str; 535 char *str;
413 536
414 SvGETMAGIC (sv); 537 SvGETMAGIC (sv);
415 str = SvPV (sv, len); 538 str = SvPV (sv, len);
416 539
417 encode_str (enc, str, len, SvUTF8 (sv)); 540 encode_str (enc, str, len, SvUTF8 (sv));
418 } 541 }
… …
461 584
462 encode_ch (enc, '{'); 585 encode_ch (enc, '{');
463 586
464 // for canonical output we have to sort by keys first 587 // for canonical output we have to sort by keys first
465 // actually, this is mostly due to the stupid so-called 588 // actually, this is mostly due to the stupid so-called
466 // security workaround added somewhere in 5.8.x. 589 // security workaround added somewhere in 5.8.x
467 // that randomises hash orderings 590 // that randomises hash orderings
468 if (enc->json.flags & F_CANONICAL) 591 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
469 { 592 {
470 int count = hv_iterinit (hv); 593 int count = hv_iterinit (hv);
471 594
472 if (SvMAGICAL (hv)) 595 if (SvMAGICAL (hv))
473 { 596 {
… …
484 } 607 }
485 608
486 if (count) 609 if (count)
487 { 610 {
488 int i, fast = 1; 611 int i, fast = 1;
489#if defined(__BORLANDC__) || defined(_MSC_VER) 612 HE *hes_stack [STACK_HES];
490 HE **hes = _alloca (count * sizeof (HE)); 613 HE **hes = hes_stack;
491#else 614
492 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 615 // allocate larger arrays on the heap
493#endif 616 if (count > STACK_HES)
617 {
618 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
619 hes = (HE **)SvPVX (sv);
620 }
494 621
495 i = 0; 622 i = 0;
496 while ((he = hv_iternext (hv))) 623 while ((he = hv_iternext (hv)))
497 { 624 {
498 hes [i++] = he; 625 hes [i++] = he;
… …
567// encode objects, arrays and special \0=false and \1=true values. 694// encode objects, arrays and special \0=false and \1=true values.
568static void 695static void
569encode_rv (enc_t *enc, SV *sv) 696encode_rv (enc_t *enc, SV *sv)
570{ 697{
571 svtype svt; 698 svtype svt;
699 GV *method;
572 700
573 SvGETMAGIC (sv); 701 SvGETMAGIC (sv);
574 svt = SvTYPE (sv); 702 svt = SvTYPE (sv);
575 703
576 if (expect_false (SvOBJECT (sv))) 704 if (expect_false (SvOBJECT (sv)))
577 { 705 {
578 HV *stash = !JSON_SLOW || json_boolean_stash 706 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
579 ? json_boolean_stash 707 ? types_boolean_stash
580 : gv_stashpv ("JSON::XS::Boolean", 1); 708 : gv_stashpv ("Types::Serialiser::Boolean", 1);
709 HV *stash = SvSTASH (sv);
581 710
582 if (SvSTASH (sv) == stash) 711 if (stash == boolean_stash)
583 { 712 {
584 if (SvIV (sv)) 713 if (SvIV (sv))
585 encode_str (enc, "true", 4, 0); 714 encode_str (enc, "true", 4, 0);
586 else 715 else
587 encode_str (enc, "false", 5, 0); 716 encode_str (enc, "false", 5, 0);
588 } 717 }
718 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
719 {
720 int count;
721 dSP;
722
723 ENTER; SAVETMPS; PUSHMARK (SP);
724 EXTEND (SP, 2);
725 // we re-bless the reference to get overload and other niceties right
726 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
727 PUSHs (sv_json);
728
729 PUTBACK;
730 count = call_sv ((SV *)GvCV (method), G_ARRAY);
731 SPAGAIN;
732
733 // catch this surprisingly common error
734 if (SvROK (TOPs) && SvRV (TOPs) == sv)
735 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
736
737 encode_ch (enc, '(');
738 encode_ch (enc, '"');
739 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
740 encode_ch (enc, '"');
741 encode_ch (enc, ')');
742 encode_ch (enc, '[');
743
744 while (count)
745 {
746 encode_sv (enc, SP[1 - count--]);
747
748 if (count)
749 encode_ch (enc, ',');
750 }
751
752 encode_ch (enc, ']');
753
754 PUTBACK;
755
756 FREETMPS; LEAVE;
757 }
758 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
759 {
760 dSP;
761
762 ENTER; SAVETMPS; PUSHMARK (SP);
763 // we re-bless the reference to get overload and other niceties right
764 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
765
766 // calling with G_SCALAR ensures that we always get a 1 return value
767 PUTBACK;
768 call_sv ((SV *)GvCV (method), G_SCALAR);
769 SPAGAIN;
770
771 // catch this surprisingly common error
772 if (SvROK (TOPs) && SvRV (TOPs) == sv)
773 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
774
775 sv = POPs;
776 PUTBACK;
777
778 encode_sv (enc, sv);
779
780 FREETMPS; LEAVE;
781 }
782 else if (enc->json.flags & F_ALLOW_BLESSED)
783 encode_str (enc, "null", 4, 0);
589 else 784 else
590 {
591#if 0
592 if (0 && sv_derived_from (rv, "JSON::Literal"))
593 {
594 // not yet
595 }
596#endif
597 if (enc->json.flags & F_CONV_BLESSED)
598 {
599 // we re-bless the reference to get overload and other niceties right
600 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
601
602 if (to_json)
603 {
604 dSP;
605
606 ENTER; SAVETMPS; PUSHMARK (SP);
607 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
608
609 // calling with G_SCALAR ensures that we always get a 1 return value
610 PUTBACK;
611 call_sv ((SV *)GvCV (to_json), G_SCALAR);
612 SPAGAIN;
613
614 // catch this surprisingly common error
615 if (SvROK (TOPs) && SvRV (TOPs) == sv)
616 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
617
618 sv = POPs;
619 PUTBACK;
620
621 encode_sv (enc, sv);
622
623 FREETMPS; LEAVE;
624 }
625 else if (enc->json.flags & F_ALLOW_BLESSED)
626 encode_str (enc, "null", 4, 0);
627 else
628 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
629 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
630 }
631 else if (enc->json.flags & F_ALLOW_BLESSED)
632 encode_str (enc, "null", 4, 0);
633 else
634 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 785 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
635 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 786 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
636 }
637 } 787 }
638 else if (svt == SVt_PVHV) 788 else if (svt == SVt_PVHV)
639 encode_hv (enc, (HV *)sv); 789 encode_hv (enc, (HV *)sv);
640 else if (svt == SVt_PVAV) 790 else if (svt == SVt_PVAV)
641 encode_av (enc, (AV *)sv); 791 encode_av (enc, (AV *)sv);
… …
717 } 867 }
718 else 868 else
719 { 869 {
720 // large integer, use the (rather slow) snprintf way. 870 // large integer, use the (rather slow) snprintf way.
721 need (enc, IVUV_MAXCHARS); 871 need (enc, IVUV_MAXCHARS);
722 enc->cur += 872 enc->cur +=
723 SvIsUV(sv) 873 SvIsUV(sv)
724 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) 874 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
725 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 875 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
726 } 876 }
727 } 877 }
728 else if (SvROK (sv)) 878 else if (SvROK (sv))
729 encode_rv (enc, SvRV (sv)); 879 encode_rv (enc, SvRV (sv));
730 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 880 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
731 encode_str (enc, "null", 4, 0); 881 encode_str (enc, "null", 4, 0);
732 else 882 else
733 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 883 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
734 SvPV_nolen (sv), SvFLAGS (sv)); 884 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
735} 885}
736 886
737static SV * 887static SV *
738encode_json (SV *scalar, JSON *json) 888encode_json (SV *scalar, JSON *json)
739{ 889{
… …
751 : enc.json.flags & F_LATIN1 ? 0x000100UL 901 : enc.json.flags & F_LATIN1 ? 0x000100UL
752 : 0x110000UL; 902 : 0x110000UL;
753 903
754 SvPOK_only (enc.sv); 904 SvPOK_only (enc.sv);
755 encode_sv (&enc, scalar); 905 encode_sv (&enc, scalar);
906 encode_nl (&enc);
756 907
757 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 908 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
758 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 909 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
759 910
760 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 911 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
940 else if (expect_true (ch >= 0x20 && ch < 0x80)) 1091 else if (expect_true (ch >= 0x20 && ch < 0x80))
941 *cur++ = ch; 1092 *cur++ = ch;
942 else if (ch >= 0x80) 1093 else if (ch >= 0x80)
943 { 1094 {
944 STRLEN clen; 1095 STRLEN clen;
945 UV uch;
946 1096
947 --dec_cur; 1097 --dec_cur;
948 1098
949 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1099 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
950 if (clen == (STRLEN)-1) 1100 if (clen == (STRLEN)-1)
951 ERR ("malformed UTF-8 character in JSON string"); 1101 ERR ("malformed UTF-8 character in JSON string");
952 1102
953 do 1103 do
954 *cur++ = *dec_cur++; 1104 *cur++ = *dec_cur++;
955 while (--clen); 1105 while (--clen);
956 1106
957 utf8 = 1; 1107 utf8 = 1;
958 } 1108 }
1109 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1110 *cur++ = ch;
959 else 1111 else
960 { 1112 {
961 --dec_cur; 1113 --dec_cur;
962 1114
963 if (!ch) 1115 if (!ch)
… …
971 { 1123 {
972 STRLEN len = cur - buf; 1124 STRLEN len = cur - buf;
973 1125
974 if (sv) 1126 if (sv)
975 { 1127 {
976 SvGROW (sv, SvCUR (sv) + len + 1); 1128 STRLEN cur = SvCUR (sv);
1129
1130 if (SvLEN (sv) <= cur + len)
1131 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1132
977 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1133 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
978 SvCUR_set (sv, SvCUR (sv) + len); 1134 SvCUR_set (sv, SvCUR (sv) + len);
979 } 1135 }
980 else 1136 else
981 sv = newSVpvn (buf, len); 1137 sv = newSVpvn (buf, len);
… …
1072 1228
1073 // special case the rather common 1..5-digit-int case 1229 // special case the rather common 1..5-digit-int case
1074 if (*start == '-') 1230 if (*start == '-')
1075 switch (len) 1231 switch (len)
1076 { 1232 {
1077 case 2: return newSViv (-( start [1] - '0' * 1)); 1233 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1078 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1234 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1079 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1235 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1080 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1236 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1081 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); 1237 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1082 } 1238 }
1083 else 1239 else
1084 switch (len) 1240 switch (len)
1085 { 1241 {
1086 case 1: return newSViv ( start [0] - '0' * 1); 1242 case 1: return newSViv ( start [0] - '0' * 1);
1087 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1243 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1088 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1244 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1089 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1245 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1090 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); 1246 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1091 } 1247 }
1092 1248
1093 { 1249 {
1094 UV uv; 1250 UV uv;
1095 int numtype = grok_number (start, len, &uv); 1251 int numtype = grok_number (start, len, &uv);
… …
1104 } 1260 }
1105 1261
1106 len -= *start == '-' ? 1 : 0; 1262 len -= *start == '-' ? 1 : 0;
1107 1263
1108 // does not fit into IV or UV, try NV 1264 // does not fit into IV or UV, try NV
1109 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1265 if (len <= NV_DIG)
1110 #if defined (LDBL_DIG)
1111 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1112 #endif
1113 )
1114 // fits into NV without loss of precision 1266 // fits into NV without loss of precision
1115 return newSVnv (Atof (start)); 1267 return newSVnv (json_atof (start));
1116 1268
1117 // everything else fails, convert it to a string 1269 // everything else fails, convert it to a string
1118 return newSVpvn (start, dec->cur - start); 1270 return newSVpvn (start, dec->cur - start);
1119 } 1271 }
1120 1272
1121 // loss of precision here 1273 // loss of precision here
1122 return newSVnv (Atof (start)); 1274 return newSVnv (json_atof (start));
1123 1275
1124fail: 1276fail:
1125 return 0; 1277 return 0;
1126} 1278}
1127 1279
… …
1151 if (*dec->cur == ']') 1303 if (*dec->cur == ']')
1152 { 1304 {
1153 ++dec->cur; 1305 ++dec->cur;
1154 break; 1306 break;
1155 } 1307 }
1156 1308
1157 if (*dec->cur != ',') 1309 if (*dec->cur != ',')
1158 ERR (", or ] expected while parsing array"); 1310 ERR (", or ] expected while parsing array");
1159 1311
1160 ++dec->cur; 1312 ++dec->cur;
1161 1313
… …
1296 dSP; 1448 dSP;
1297 int count; 1449 int count;
1298 1450
1299 ENTER; SAVETMPS; PUSHMARK (SP); 1451 ENTER; SAVETMPS; PUSHMARK (SP);
1300 XPUSHs (HeVAL (he)); 1452 XPUSHs (HeVAL (he));
1453 sv_2mortal (sv);
1301 1454
1302 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1455 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1303 1456
1304 if (count == 1) 1457 if (count == 1)
1305 { 1458 {
1306 sv = newSVsv (POPs); 1459 sv = newSVsv (POPs);
1307 FREETMPS; LEAVE; 1460 FREETMPS; LEAVE;
1308 return sv; 1461 return sv;
1309 } 1462 }
1310 1463
1464 SvREFCNT_inc (sv);
1311 FREETMPS; LEAVE; 1465 FREETMPS; LEAVE;
1312 } 1466 }
1313 } 1467 }
1314 1468
1315 if (dec->json.cb_object) 1469 if (dec->json.cb_object)
… …
1341 DEC_DEC_DEPTH; 1495 DEC_DEC_DEPTH;
1342 return 0; 1496 return 0;
1343} 1497}
1344 1498
1345static SV * 1499static SV *
1500decode_tag (dec_t *dec)
1501{
1502 SV *tag = 0;
1503 SV *val = 0;
1504
1505 if (!(dec->json.flags & F_ALLOW_TAGS))
1506 ERR ("malformed JSON string, neither array, object, number, string or atom");
1507
1508 ++dec->cur;
1509
1510 decode_ws (dec);
1511
1512 tag = decode_sv (dec);
1513 if (!tag)
1514 goto fail;
1515
1516 if (!SvPOK (tag))
1517 ERR ("malformed JSON string, (tag) must be a string");
1518
1519 decode_ws (dec);
1520
1521 if (*dec->cur != ')')
1522 ERR (") expected after tag");
1523
1524 ++dec->cur;
1525
1526 decode_ws (dec);
1527
1528 val = decode_sv (dec);
1529 if (!val)
1530 goto fail;
1531
1532 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1533 ERR ("malformed JSON string, tag value must be an array");
1534
1535 {
1536 AV *av = (AV *)SvRV (val);
1537 int i, len = av_len (av) + 1;
1538 HV *stash = gv_stashsv (tag, 0);
1539 SV *sv;
1540
1541 if (!stash)
1542 ERR ("cannot decode perl-object (package does not exist)");
1543
1544 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1545
1546 if (!method)
1547 ERR ("cannot decode perl-object (package does not have a THAW method)");
1548
1549 dSP;
1550
1551 ENTER; SAVETMPS; PUSHMARK (SP);
1552 EXTEND (SP, len + 2);
1553 // we re-bless the reference to get overload and other niceties right
1554 PUSHs (tag);
1555 PUSHs (sv_json);
1556
1557 for (i = 0; i < len; ++i)
1558 PUSHs (*av_fetch (av, i, 1));
1559
1560 PUTBACK;
1561 call_sv ((SV *)GvCV (method), G_SCALAR);
1562 SPAGAIN;
1563
1564 SvREFCNT_dec (tag);
1565 SvREFCNT_dec (val);
1566 sv = SvREFCNT_inc (POPs);
1567
1568 PUTBACK;
1569
1570 FREETMPS; LEAVE;
1571
1572 return sv;
1573 }
1574
1575fail:
1576 SvREFCNT_dec (tag);
1577 SvREFCNT_dec (val);
1578 return 0;
1579}
1580
1581static SV *
1346decode_sv (dec_t *dec) 1582decode_sv (dec_t *dec)
1347{ 1583{
1348 // the beauty of JSON: you need exactly one character lookahead 1584 // the beauty of JSON: you need exactly one character lookahead
1349 // to parse everything. 1585 // to parse everything.
1350 switch (*dec->cur) 1586 switch (*dec->cur)
1351 { 1587 {
1352 case '"': ++dec->cur; return decode_str (dec); 1588 case '"': ++dec->cur; return decode_str (dec);
1353 case '[': ++dec->cur; return decode_av (dec); 1589 case '[': ++dec->cur; return decode_av (dec);
1354 case '{': ++dec->cur; return decode_hv (dec); 1590 case '{': ++dec->cur; return decode_hv (dec);
1591 case '(': return decode_tag (dec);
1355 1592
1356 case '-': 1593 case '-':
1357 case '0': case '1': case '2': case '3': case '4': 1594 case '0': case '1': case '2': case '3': case '4':
1358 case '5': case '6': case '7': case '8': case '9': 1595 case '5': case '6': case '7': case '8': case '9':
1359 return decode_num (dec); 1596 return decode_num (dec);
… …
1361 case 't': 1598 case 't':
1362 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1599 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1363 { 1600 {
1364 dec->cur += 4; 1601 dec->cur += 4;
1365#if JSON_SLOW 1602#if JSON_SLOW
1366 json_true = get_bool ("JSON::XS::true"); 1603 types_true = get_bool ("Types::Serialiser::true");
1367#endif 1604#endif
1368 return newSVsv (json_true); 1605 return newSVsv (types_true);
1369 } 1606 }
1370 else 1607 else
1371 ERR ("'true' expected"); 1608 ERR ("'true' expected");
1372 1609
1373 break; 1610 break;
… …
1375 case 'f': 1612 case 'f':
1376 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1613 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1377 { 1614 {
1378 dec->cur += 5; 1615 dec->cur += 5;
1379#if JSON_SLOW 1616#if JSON_SLOW
1380 json_false = get_bool ("JSON::XS::false"); 1617 types_false = get_bool ("Types::Serialiser::false");
1381#endif 1618#endif
1382 return newSVsv (json_false); 1619 return newSVsv (types_false);
1383 } 1620 }
1384 else 1621 else
1385 ERR ("'false' expected"); 1622 ERR ("'false' expected");
1386 1623
1387 break; 1624 break;
… …
1396 ERR ("'null' expected"); 1633 ERR ("'null' expected");
1397 1634
1398 break; 1635 break;
1399 1636
1400 default: 1637 default:
1401 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1638 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1402 break; 1639 break;
1403 } 1640 }
1404 1641
1405fail: 1642fail:
1406 return 0; 1643 return 0;
1407} 1644}
1408 1645
1409static SV * 1646static SV *
1410decode_json (SV *string, JSON *json, STRLEN *offset_return) 1647decode_json (SV *string, JSON *json, char **offset_return)
1411{ 1648{
1412 dec_t dec; 1649 dec_t dec;
1413 STRLEN offset;
1414 SV *sv; 1650 SV *sv;
1415 1651
1652 /* work around bugs in 5.10 where manipulating magic values
1653 * makes perl ignore the magic in subsequent accesses.
1654 * also make a copy of non-PV values, to get them into a clean
1655 * state (SvPV should do that, but it's buggy, see below).
1656 */
1416 SvGETMAGIC (string); 1657 /*SvGETMAGIC (string);*/
1658 if (SvMAGICAL (string) || !SvPOK (string))
1659 string = sv_2mortal (newSVsv (string));
1660
1417 SvUPGRADE (string, SVt_PV); 1661 SvUPGRADE (string, SVt_PV);
1418 1662
1419 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1663 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1420 * assertion with -DDEBUGGING, although SvCUR is documented to 1664 * assertion with -DDEBUGGING, although SvCUR is documented to
1421 * return the xpv_cur field which certainly exists after upgrading. 1665 * return the xpv_cur field which certainly exists after upgrading.
… …
1424 * and hope for the best. 1668 * and hope for the best.
1425 * Damnit, SvPV_nolen still trips over yet another assertion. This 1669 * Damnit, SvPV_nolen still trips over yet another assertion. This
1426 * assertion business is seriously broken, try yet another workaround 1670 * assertion business is seriously broken, try yet another workaround
1427 * for the broken -DDEBUGGING. 1671 * for the broken -DDEBUGGING.
1428 */ 1672 */
1673 {
1429#ifdef DEBUGGING 1674#ifdef DEBUGGING
1430 offset = SvOK (string) ? sv_len (string) : 0; 1675 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1431#else 1676#else
1432 offset = SvCUR (string); 1677 STRLEN offset = SvCUR (string);
1433#endif 1678#endif
1434 1679
1435 if (offset > json->max_size && json->max_size) 1680 if (offset > json->max_size && json->max_size)
1436 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1681 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1437 (unsigned long)SvCUR (string), (unsigned long)json->max_size); 1682 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1683 }
1438 1684
1439 if (json->flags & F_UTF8) 1685 if (DECODE_WANTS_OCTETS (json))
1440 sv_utf8_downgrade (string, 0); 1686 sv_utf8_downgrade (string, 0);
1441 else 1687 else
1442 sv_utf8_upgrade (string); 1688 sv_utf8_upgrade (string);
1443 1689
1444 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1690 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
… …
1454 1700
1455 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1701 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1456 1702
1457 decode_ws (&dec); 1703 decode_ws (&dec);
1458 sv = decode_sv (&dec); 1704 sv = decode_sv (&dec);
1705
1706 if (offset_return)
1707 *offset_return = dec.cur;
1459 1708
1460 if (!(offset_return || !sv)) 1709 if (!(offset_return || !sv))
1461 { 1710 {
1462 // check for trailing garbage 1711 // check for trailing garbage
1463 decode_ws (&dec); 1712 decode_ws (&dec);
… …
1466 { 1715 {
1467 dec.err = "garbage after JSON object"; 1716 dec.err = "garbage after JSON object";
1468 SvREFCNT_dec (sv); 1717 SvREFCNT_dec (sv);
1469 sv = 0; 1718 sv = 0;
1470 } 1719 }
1471 }
1472
1473 if (offset_return || !sv)
1474 {
1475 offset = dec.json.flags & F_UTF8
1476 ? dec.cur - SvPVX (string)
1477 : utf8_distance (dec.cur, SvPVX (string));
1478
1479 if (offset_return)
1480 *offset_return = offset;
1481 } 1720 }
1482 1721
1483 if (!sv) 1722 if (!sv)
1484 { 1723 {
1485 SV *uni = sv_newmortal (); 1724 SV *uni = sv_newmortal ();
… …
1491 SAVEVPTR (PL_curcop); 1730 SAVEVPTR (PL_curcop);
1492 PL_curcop = &cop; 1731 PL_curcop = &cop;
1493 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1732 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1494 LEAVE; 1733 LEAVE;
1495 1734
1496 croak ("%s, at character offset %d [\"%s\"]", 1735 croak ("%s, at character offset %d (before \"%s\")",
1497 dec.err, 1736 dec.err,
1498 (int)offset, 1737 (int)ptr_to_index (string, dec.cur),
1499 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1738 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1500 } 1739 }
1501 1740
1502 sv = sv_2mortal (sv); 1741 sv = sv_2mortal (sv);
1503 1742
… …
1512 1751
1513static void 1752static void
1514incr_parse (JSON *self) 1753incr_parse (JSON *self)
1515{ 1754{
1516 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1755 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1756
1757 // the state machine here is a bit convoluted and could be simplified a lot
1758 // but this would make it slower, so...
1517 1759
1518 for (;;) 1760 for (;;)
1519 { 1761 {
1520 //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 1762 //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
1521 switch (self->incr_mode) 1763 switch (self->incr_mode)
1522 { 1764 {
1523 // only used for intiial whitespace skipping 1765 // only used for initial whitespace skipping
1524 case INCR_M_WS: 1766 case INCR_M_WS:
1525 for (;;) 1767 for (;;)
1526 { 1768 {
1527 if (*p > 0x20) 1769 if (*p > 0x20)
1528 { 1770 {
1771 if (*p == '#')
1772 {
1773 self->incr_mode = INCR_M_C0;
1774 goto incr_m_c;
1775 }
1776 else
1777 {
1529 self->incr_mode = INCR_M_JSON; 1778 self->incr_mode = INCR_M_JSON;
1530 goto incr_m_json; 1779 goto incr_m_json;
1780 }
1531 } 1781 }
1532 else if (!*p) 1782 else if (!*p)
1533 goto interrupt; 1783 goto interrupt;
1534 1784
1535 ++p; 1785 ++p;
… …
1541 goto interrupt; 1791 goto interrupt;
1542 1792
1543 ++p; 1793 ++p;
1544 self->incr_mode = INCR_M_STR; 1794 self->incr_mode = INCR_M_STR;
1545 goto incr_m_str; 1795 goto incr_m_str;
1796
1797 // inside #-style comments
1798 case INCR_M_C0:
1799 case INCR_M_C1:
1800 incr_m_c:
1801 for (;;)
1802 {
1803 if (*p == '\n')
1804 {
1805 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1806 break;
1807 }
1808 else if (!*p)
1809 goto interrupt;
1810
1811 ++p;
1812 }
1813
1814 break;
1546 1815
1547 // inside a string 1816 // inside a string
1548 case INCR_M_STR: 1817 case INCR_M_STR:
1549 incr_m_str: 1818 incr_m_str:
1550 for (;;) 1819 for (;;)
… …
1601 self->incr_mode = INCR_M_STR; 1870 self->incr_mode = INCR_M_STR;
1602 goto incr_m_str; 1871 goto incr_m_str;
1603 1872
1604 case '[': 1873 case '[':
1605 case '{': 1874 case '{':
1875 case '(':
1606 if (++self->incr_nest > self->max_depth) 1876 if (++self->incr_nest > self->max_depth)
1607 croak (ERR_NESTING_EXCEEDED); 1877 croak (ERR_NESTING_EXCEEDED);
1608 break; 1878 break;
1609 1879
1610 case ']': 1880 case ']':
1611 case '}': 1881 case '}':
1612 if (--self->incr_nest <= 0) 1882 if (--self->incr_nest <= 0)
1613 goto interrupt; 1883 goto interrupt;
1884 break;
1885
1886 case ')':
1887 --self->incr_nest;
1888 break;
1889
1890 case '#':
1891 self->incr_mode = INCR_M_C1;
1892 goto incr_m_c;
1614 } 1893 }
1615 } 1894 }
1616 } 1895 }
1617 1896
1618 modechange: 1897 modechange:
1619 ; 1898 ;
1620 } 1899 }
1621 1900
1622interrupt: 1901interrupt:
1623 self->incr_pos = p - SvPVX (self->incr_text); 1902 self->incr_pos = p - SvPVX (self->incr_text);
1903 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1624 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D 1904 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1625} 1905}
1626 1906
1627///////////////////////////////////////////////////////////////////////////// 1907/////////////////////////////////////////////////////////////////////////////
1628// XS interface functions 1908// XS interface functions
… …
1638 i >= '0' && i <= '9' ? i - '0' 1918 i >= '0' && i <= '9' ? i - '0'
1639 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1919 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1640 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1920 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1641 : -1; 1921 : -1;
1642 1922
1643 json_stash = gv_stashpv ("JSON::XS" , 1); 1923 json_stash = gv_stashpv ("JSON::XS" , 1);
1644 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1924 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1645 1925
1646 json_true = get_bool ("JSON::XS::true"); 1926 types_true = get_bool ("Types::Serialiser::true");
1647 json_false = get_bool ("JSON::XS::false"); 1927 types_false = get_bool ("Types::Serialiser::false");
1928
1929 sv_json = newSVpv ("JSON", 0);
1930 SvREADONLY_on (sv_json);
1931
1932 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1648} 1933}
1649 1934
1650PROTOTYPES: DISABLE 1935PROTOTYPES: DISABLE
1651 1936
1652void CLONE (...) 1937void CLONE (...)
1653 CODE: 1938 CODE:
1654 json_stash = 0; 1939 json_stash = 0;
1655 json_boolean_stash = 0; 1940 types_boolean_stash = 0;
1656 1941
1657void new (char *klass) 1942void new (char *klass)
1658 PPCODE: 1943 PPCODE:
1659{ 1944{
1660 SV *pv = NEWSV (0, sizeof (JSON)); 1945 SV *pv = NEWSV (0, sizeof (JSON));
1661 SvPOK_only (pv); 1946 SvPOK_only (pv);
1662 json_init ((JSON *)SvPVX (pv)); 1947 json_init ((JSON *)SvPVX (pv));
1663 XPUSHs (sv_2mortal (sv_bless ( 1948 XPUSHs (sv_2mortal (sv_bless (
1664 newRV_noinc (pv), 1949 newRV_noinc (pv),
1665 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1950 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
… …
1680 shrink = F_SHRINK 1965 shrink = F_SHRINK
1681 allow_blessed = F_ALLOW_BLESSED 1966 allow_blessed = F_ALLOW_BLESSED
1682 convert_blessed = F_CONV_BLESSED 1967 convert_blessed = F_CONV_BLESSED
1683 relaxed = F_RELAXED 1968 relaxed = F_RELAXED
1684 allow_unknown = F_ALLOW_UNKNOWN 1969 allow_unknown = F_ALLOW_UNKNOWN
1970 allow_tags = F_ALLOW_TAGS
1685 PPCODE: 1971 PPCODE:
1686{ 1972{
1687 if (enable) 1973 if (enable)
1688 self->flags |= ix; 1974 self->flags |= ix;
1689 else 1975 else
… …
1705 get_shrink = F_SHRINK 1991 get_shrink = F_SHRINK
1706 get_allow_blessed = F_ALLOW_BLESSED 1992 get_allow_blessed = F_ALLOW_BLESSED
1707 get_convert_blessed = F_CONV_BLESSED 1993 get_convert_blessed = F_CONV_BLESSED
1708 get_relaxed = F_RELAXED 1994 get_relaxed = F_RELAXED
1709 get_allow_unknown = F_ALLOW_UNKNOWN 1995 get_allow_unknown = F_ALLOW_UNKNOWN
1996 get_allow_tags = F_ALLOW_TAGS
1710 PPCODE: 1997 PPCODE:
1711 XPUSHs (boolSV (self->flags & ix)); 1998 XPUSHs (boolSV (self->flags & ix));
1712 1999
1713void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 2000void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1714 PPCODE: 2001 PPCODE:
… …
1742} 2029}
1743 2030
1744void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2031void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1745 PPCODE: 2032 PPCODE:
1746{ 2033{
1747 if (!self->cb_sk_object) 2034 if (!self->cb_sk_object)
1748 self->cb_sk_object = newHV (); 2035 self->cb_sk_object = newHV ();
1749 2036
1750 if (SvOK (cb)) 2037 if (SvOK (cb))
1751 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2038 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1752 else 2039 else
… …
1763 XPUSHs (ST (0)); 2050 XPUSHs (ST (0));
1764} 2051}
1765 2052
1766void encode (JSON *self, SV *scalar) 2053void encode (JSON *self, SV *scalar)
1767 PPCODE: 2054 PPCODE:
1768 XPUSHs (encode_json (scalar, self)); 2055 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2056 XPUSHs (scalar);
1769 2057
1770void decode (JSON *self, SV *jsonstr) 2058void decode (JSON *self, SV *jsonstr)
1771 PPCODE: 2059 PPCODE:
1772 XPUSHs (decode_json (jsonstr, self, 0)); 2060 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2061 XPUSHs (jsonstr);
1773 2062
1774void decode_prefix (JSON *self, SV *jsonstr) 2063void decode_prefix (JSON *self, SV *jsonstr)
1775 PPCODE: 2064 PPCODE:
1776{ 2065{
1777 STRLEN offset; 2066 SV *sv;
2067 char *offset;
2068 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1778 EXTEND (SP, 2); 2069 EXTEND (SP, 2);
1779 PUSHs (decode_json (jsonstr, self, &offset)); 2070 PUSHs (sv);
1780 PUSHs (sv_2mortal (newSVuv (offset))); 2071 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1781} 2072}
1782 2073
1783void incr_parse (JSON *self, SV *jsonstr = 0) 2074void incr_parse (JSON *self, SV *jsonstr = 0)
1784 PPCODE: 2075 PPCODE:
1785{ 2076{
1786 if (!self->incr_text) 2077 if (!self->incr_text)
1787 self->incr_text = newSVpvn ("", 0); 2078 self->incr_text = newSVpvn ("", 0);
2079
2080 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2081 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2082 if (DECODE_WANTS_OCTETS (self))
2083 {
2084 if (self->incr_pos)
2085 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2086 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2087
2088 sv_utf8_downgrade (self->incr_text, 0);
2089 }
2090 else
2091 {
2092 sv_utf8_upgrade (self->incr_text);
2093
2094 if (self->incr_pos)
2095 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2096 - (U8 *)SvPVX (self->incr_text);
2097 }
1788 2098
1789 // append data, if any 2099 // append data, if any
1790 if (jsonstr) 2100 if (jsonstr)
1791 { 2101 {
2102 /* make sure both strings have same encoding */
1792 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 2103 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2104 if (SvUTF8 (jsonstr))
2105 sv_utf8_downgrade (jsonstr, 0);
1793 { 2106 else
1794 /* utf-8-ness differs, need to upgrade */
1795 sv_utf8_upgrade (self->incr_text); 2107 sv_utf8_upgrade (jsonstr);
1796 2108
1797 if (self->incr_pos) 2109 /* and then just blindly append */
1798 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1799 - (U8 *)SvPVX (self->incr_text);
1800 }
1801
1802 { 2110 {
1803 STRLEN len; 2111 STRLEN len;
1804 const char *str = SvPV (jsonstr, len); 2112 const char *str = SvPV (jsonstr, len);
1805 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 2113 STRLEN cur = SvCUR (self->incr_text);
2114
2115 if (SvLEN (self->incr_text) <= cur + len)
2116 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2117
1806 Move (str, SvEND (self->incr_text), len, char); 2118 Move (str, SvEND (self->incr_text), len, char);
1807 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2119 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1808 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 2120 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1809 } 2121 }
1810 } 2122 }
1811 2123
1812 if (GIMME_V != G_VOID) 2124 if (GIMME_V != G_VOID)
1813 do 2125 do
1814 { 2126 {
2127 SV *sv;
1815 STRLEN offset; 2128 char *offset;
1816 2129
1817 if (!INCR_DONE (self)) 2130 if (!INCR_DONE (self))
1818 { 2131 {
1819 incr_parse (self); 2132 incr_parse (self);
1820 2133
1821 if (self->incr_pos > self->max_size && self->max_size) 2134 if (self->incr_pos > self->max_size && self->max_size)
1822 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 2135 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1823 (unsigned long)self->incr_pos, (unsigned long)self->max_size); 2136 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1824 2137
1825 if (!INCR_DONE (self)) 2138 if (!INCR_DONE (self))
2139 {
2140 // as an optimisation, do not accumulate white space in the incr buffer
2141 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2142 {
2143 self->incr_pos = 0;
2144 SvCUR_set (self->incr_text, 0);
2145 }
2146
1826 break; 2147 break;
2148 }
1827 } 2149 }
1828 2150
1829 XPUSHs (decode_json (self->incr_text, self, &offset)); 2151 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2152 XPUSHs (sv);
1830 2153
1831 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1832 self->incr_pos -= offset; 2154 self->incr_pos -= offset - SvPVX (self->incr_text);
1833 self->incr_nest = 0; 2155 self->incr_nest = 0;
1834 self->incr_mode = 0; 2156 self->incr_mode = 0;
2157
2158 sv_chop (self->incr_text, offset);
1835 } 2159 }
1836 while (GIMME_V == G_ARRAY); 2160 while (GIMME_V == G_ARRAY);
1837} 2161}
1838 2162
1839SV *incr_text (JSON *self) 2163SV *incr_text (JSON *self)
… …
1877 SvREFCNT_dec (self->incr_text); 2201 SvREFCNT_dec (self->incr_text);
1878 2202
1879PROTOTYPES: ENABLE 2203PROTOTYPES: ENABLE
1880 2204
1881void encode_json (SV *scalar) 2205void encode_json (SV *scalar)
1882 ALIAS:
1883 to_json_ = 0
1884 encode_json = F_UTF8
1885 PPCODE: 2206 PPCODE:
1886{ 2207{
1887 JSON json; 2208 JSON json;
1888 json_init (&json); 2209 json_init (&json);
1889 json.flags |= ix; 2210 json.flags |= F_UTF8;
1890 XPUSHs (encode_json (scalar, &json)); 2211 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2212 XPUSHs (scalar);
1891} 2213}
1892 2214
1893void decode_json (SV *jsonstr) 2215void decode_json (SV *jsonstr)
1894 ALIAS:
1895 from_json_ = 0
1896 decode_json = F_UTF8
1897 PPCODE: 2216 PPCODE:
1898{ 2217{
1899 JSON json; 2218 JSON json;
1900 json_init (&json); 2219 json_init (&json);
1901 json.flags |= ix; 2220 json.flags |= F_UTF8;
1902 XPUSHs (decode_json (jsonstr, &json, 0)); 2221 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2222 XPUSHs (jsonstr);
1903} 2223}
1904 2224
1905

Diff Legend

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