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 1.101 by root, Tue Sep 8 18:00:03 2009 UTC vs.
Revision 1.124 by root, Mon Nov 24 18:42:51 2014 UTC

17// guarantees, 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
184 return SvUTF8 (sv) 204 return SvUTF8 (sv)
185 ? utf8_distance (offset, SvPVX (sv)) 205 ? utf8_distance (offset, SvPVX (sv))
186 : offset - SvPVX (sv); 206 : offset - SvPVX (sv);
187} 207}
188 208
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}
189///////////////////////////////////////////////////////////////////////////// 304/////////////////////////////////////////////////////////////////////////////
190// encoder 305// encoder
191 306
192// structure used for encoding JSON 307// structure used for encoding JSON
193typedef struct 308typedef struct
380 495
381 if (enc->indent >= enc->json.max_depth) 496 if (enc->indent >= enc->json.max_depth)
382 croak (ERR_NESTING_EXCEEDED); 497 croak (ERR_NESTING_EXCEEDED);
383 498
384 encode_ch (enc, '['); 499 encode_ch (enc, '[');
385 500
386 if (len >= 0) 501 if (len >= 0)
387 { 502 {
388 encode_nl (enc); ++enc->indent; 503 encode_nl (enc); ++enc->indent;
389 504
390 for (i = 0; i <= len; ++i) 505 for (i = 0; i <= len; ++i)
402 encode_comma (enc); 517 encode_comma (enc);
403 } 518 }
404 519
405 encode_nl (enc); --enc->indent; encode_indent (enc); 520 encode_nl (enc); --enc->indent; encode_indent (enc);
406 } 521 }
407 522
408 encode_ch (enc, ']'); 523 encode_ch (enc, ']');
409} 524}
410 525
411static void 526static void
412encode_hk (enc_t *enc, HE *he) 527encode_hk (enc_t *enc, HE *he)
416 if (HeKLEN (he) == HEf_SVKEY) 531 if (HeKLEN (he) == HEf_SVKEY)
417 { 532 {
418 SV *sv = HeSVKEY (he); 533 SV *sv = HeSVKEY (he);
419 STRLEN len; 534 STRLEN len;
420 char *str; 535 char *str;
421 536
422 SvGETMAGIC (sv); 537 SvGETMAGIC (sv);
423 str = SvPV (sv, len); 538 str = SvPV (sv, len);
424 539
425 encode_str (enc, str, len, SvUTF8 (sv)); 540 encode_str (enc, str, len, SvUTF8 (sv));
426 } 541 }
492 } 607 }
493 608
494 if (count) 609 if (count)
495 { 610 {
496 int i, fast = 1; 611 int i, fast = 1;
497#if defined(__BORLANDC__) || defined(_MSC_VER) 612 HE *hes_stack [STACK_HES];
498 HE **hes = _alloca (count * sizeof (HE)); 613 HE **hes = hes_stack;
499#else 614
500 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 615 // allocate larger arrays on the heap
501#endif 616 if (count > STACK_HES)
617 {
618 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
619 hes = (HE **)SvPVX (sv);
620 }
502 621
503 i = 0; 622 i = 0;
504 while ((he = hv_iternext (hv))) 623 while ((he = hv_iternext (hv)))
505 { 624 {
506 hes [i++] = he; 625 hes [i++] = he;
575// encode objects, arrays and special \0=false and \1=true values. 694// encode objects, arrays and special \0=false and \1=true values.
576static void 695static void
577encode_rv (enc_t *enc, SV *sv) 696encode_rv (enc_t *enc, SV *sv)
578{ 697{
579 svtype svt; 698 svtype svt;
699 GV *method;
580 700
581 SvGETMAGIC (sv); 701 SvGETMAGIC (sv);
582 svt = SvTYPE (sv); 702 svt = SvTYPE (sv);
583 703
584 if (expect_false (SvOBJECT (sv))) 704 if (expect_false (SvOBJECT (sv)))
585 { 705 {
586 HV *stash = !JSON_SLOW || json_boolean_stash 706 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
587 ? json_boolean_stash 707 ? types_boolean_stash
588 : gv_stashpv ("JSON::XS::Boolean", 1); 708 : gv_stashpv ("Types::Serialiser::Boolean", 1);
709 HV *stash = SvSTASH (sv);
589 710
590 if (SvSTASH (sv) == stash) 711 if (stash == boolean_stash)
591 { 712 {
592 if (SvIV (sv)) 713 if (SvIV (sv))
593 encode_str (enc, "true", 4, 0); 714 encode_str (enc, "true", 4, 0);
594 else 715 else
595 encode_str (enc, "false", 5, 0); 716 encode_str (enc, "false", 5, 0);
596 } 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);
597 else 784 else
598 {
599#if 0
600 if (0 && sv_derived_from (rv, "JSON::Literal"))
601 {
602 // not yet
603 }
604#endif
605 if (enc->json.flags & F_CONV_BLESSED)
606 {
607 // we re-bless the reference to get overload and other niceties right
608 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
609
610 if (to_json)
611 {
612 dSP;
613
614 ENTER; SAVETMPS; PUSHMARK (SP);
615 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
616
617 // calling with G_SCALAR ensures that we always get a 1 return value
618 PUTBACK;
619 call_sv ((SV *)GvCV (to_json), G_SCALAR);
620 SPAGAIN;
621
622 // catch this surprisingly common error
623 if (SvROK (TOPs) && SvRV (TOPs) == sv)
624 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
625
626 sv = POPs;
627 PUTBACK;
628
629 encode_sv (enc, sv);
630
631 FREETMPS; LEAVE;
632 }
633 else if (enc->json.flags & F_ALLOW_BLESSED)
634 encode_str (enc, "null", 4, 0);
635 else
636 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
637 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
638 }
639 else if (enc->json.flags & F_ALLOW_BLESSED)
640 encode_str (enc, "null", 4, 0);
641 else
642 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)",
643 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 786 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
644 }
645 } 787 }
646 else if (svt == SVt_PVHV) 788 else if (svt == SVt_PVHV)
647 encode_hv (enc, (HV *)sv); 789 encode_hv (enc, (HV *)sv);
648 else if (svt == SVt_PVAV) 790 else if (svt == SVt_PVAV)
649 encode_av (enc, (AV *)sv); 791 encode_av (enc, (AV *)sv);
725 } 867 }
726 else 868 else
727 { 869 {
728 // large integer, use the (rather slow) snprintf way. 870 // large integer, use the (rather slow) snprintf way.
729 need (enc, IVUV_MAXCHARS); 871 need (enc, IVUV_MAXCHARS);
730 enc->cur += 872 enc->cur +=
731 SvIsUV(sv) 873 SvIsUV(sv)
732 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) 874 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
733 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 875 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
734 } 876 }
735 } 877 }
736 else if (SvROK (sv)) 878 else if (SvROK (sv))
737 encode_rv (enc, SvRV (sv)); 879 encode_rv (enc, SvRV (sv));
738 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 880 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
739 encode_str (enc, "null", 4, 0); 881 encode_str (enc, "null", 4, 0);
740 else 882 else
741 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",
742 SvPV_nolen (sv), SvFLAGS (sv)); 884 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
743} 885}
744 886
745static SV * 887static SV *
746encode_json (SV *scalar, JSON *json) 888encode_json (SV *scalar, JSON *json)
747{ 889{
962 *cur++ = *dec_cur++; 1104 *cur++ = *dec_cur++;
963 while (--clen); 1105 while (--clen);
964 1106
965 utf8 = 1; 1107 utf8 = 1;
966 } 1108 }
1109 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1110 *cur++ = ch;
967 else 1111 else
968 { 1112 {
969 --dec_cur; 1113 --dec_cur;
970 1114
971 if (!ch) 1115 if (!ch)
1116 } 1260 }
1117 1261
1118 len -= *start == '-' ? 1 : 0; 1262 len -= *start == '-' ? 1 : 0;
1119 1263
1120 // does not fit into IV or UV, try NV 1264 // does not fit into IV or UV, try NV
1121 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1265 if (len <= NV_DIG)
1122 #if defined (LDBL_DIG)
1123 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1124 #endif
1125 )
1126 // fits into NV without loss of precision 1266 // fits into NV without loss of precision
1127 return newSVnv (Atof (start)); 1267 return newSVnv (json_atof (start));
1128 1268
1129 // everything else fails, convert it to a string 1269 // everything else fails, convert it to a string
1130 return newSVpvn (start, dec->cur - start); 1270 return newSVpvn (start, dec->cur - start);
1131 } 1271 }
1132 1272
1133 // loss of precision here 1273 // loss of precision here
1134 return newSVnv (Atof (start)); 1274 return newSVnv (json_atof (start));
1135 1275
1136fail: 1276fail:
1137 return 0; 1277 return 0;
1138} 1278}
1139 1279
1163 if (*dec->cur == ']') 1303 if (*dec->cur == ']')
1164 { 1304 {
1165 ++dec->cur; 1305 ++dec->cur;
1166 break; 1306 break;
1167 } 1307 }
1168 1308
1169 if (*dec->cur != ',') 1309 if (*dec->cur != ',')
1170 ERR (", or ] expected while parsing array"); 1310 ERR (", or ] expected while parsing array");
1171 1311
1172 ++dec->cur; 1312 ++dec->cur;
1173 1313
1308 dSP; 1448 dSP;
1309 int count; 1449 int count;
1310 1450
1311 ENTER; SAVETMPS; PUSHMARK (SP); 1451 ENTER; SAVETMPS; PUSHMARK (SP);
1312 XPUSHs (HeVAL (he)); 1452 XPUSHs (HeVAL (he));
1453 sv_2mortal (sv);
1313 1454
1314 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1455 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1315 1456
1316 if (count == 1) 1457 if (count == 1)
1317 { 1458 {
1318 sv = newSVsv (POPs); 1459 sv = newSVsv (POPs);
1319 FREETMPS; LEAVE; 1460 FREETMPS; LEAVE;
1320 return sv; 1461 return sv;
1321 } 1462 }
1322 1463
1464 SvREFCNT_inc (sv);
1323 FREETMPS; LEAVE; 1465 FREETMPS; LEAVE;
1324 } 1466 }
1325 } 1467 }
1326 1468
1327 if (dec->json.cb_object) 1469 if (dec->json.cb_object)
1353 DEC_DEC_DEPTH; 1495 DEC_DEC_DEPTH;
1354 return 0; 1496 return 0;
1355} 1497}
1356 1498
1357static 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 *
1358decode_sv (dec_t *dec) 1582decode_sv (dec_t *dec)
1359{ 1583{
1360 // the beauty of JSON: you need exactly one character lookahead 1584 // the beauty of JSON: you need exactly one character lookahead
1361 // to parse everything. 1585 // to parse everything.
1362 switch (*dec->cur) 1586 switch (*dec->cur)
1363 { 1587 {
1364 case '"': ++dec->cur; return decode_str (dec); 1588 case '"': ++dec->cur; return decode_str (dec);
1365 case '[': ++dec->cur; return decode_av (dec); 1589 case '[': ++dec->cur; return decode_av (dec);
1366 case '{': ++dec->cur; return decode_hv (dec); 1590 case '{': ++dec->cur; return decode_hv (dec);
1591 case '(': return decode_tag (dec);
1367 1592
1368 case '-': 1593 case '-':
1369 case '0': case '1': case '2': case '3': case '4': 1594 case '0': case '1': case '2': case '3': case '4':
1370 case '5': case '6': case '7': case '8': case '9': 1595 case '5': case '6': case '7': case '8': case '9':
1371 return decode_num (dec); 1596 return decode_num (dec);
1373 case 't': 1598 case 't':
1374 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1599 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1375 { 1600 {
1376 dec->cur += 4; 1601 dec->cur += 4;
1377#if JSON_SLOW 1602#if JSON_SLOW
1378 json_true = get_bool ("JSON::XS::true"); 1603 types_true = get_bool ("Types::Serialiser::true");
1379#endif 1604#endif
1380 return newSVsv (json_true); 1605 return newSVsv (types_true);
1381 } 1606 }
1382 else 1607 else
1383 ERR ("'true' expected"); 1608 ERR ("'true' expected");
1384 1609
1385 break; 1610 break;
1387 case 'f': 1612 case 'f':
1388 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1613 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1389 { 1614 {
1390 dec->cur += 5; 1615 dec->cur += 5;
1391#if JSON_SLOW 1616#if JSON_SLOW
1392 json_false = get_bool ("JSON::XS::false"); 1617 types_false = get_bool ("Types::Serialiser::false");
1393#endif 1618#endif
1394 return newSVsv (json_false); 1619 return newSVsv (types_false);
1395 } 1620 }
1396 else 1621 else
1397 ERR ("'false' expected"); 1622 ERR ("'false' expected");
1398 1623
1399 break; 1624 break;
1408 ERR ("'null' expected"); 1633 ERR ("'null' expected");
1409 1634
1410 break; 1635 break;
1411 1636
1412 default: 1637 default:
1413 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1638 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1414 break; 1639 break;
1415 } 1640 }
1416 1641
1417fail: 1642fail:
1418 return 0; 1643 return 0;
1423{ 1648{
1424 dec_t dec; 1649 dec_t dec;
1425 SV *sv; 1650 SV *sv;
1426 1651
1427 /* work around bugs in 5.10 where manipulating magic values 1652 /* work around bugs in 5.10 where manipulating magic values
1428 * will perl ignore the magic in subsequent accesses 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).
1429 */ 1656 */
1430 /*SvGETMAGIC (string);*/ 1657 /*SvGETMAGIC (string);*/
1431 if (SvMAGICAL (string)) 1658 if (SvMAGICAL (string) || !SvPOK (string))
1432 string = sv_2mortal (newSVsv (string)); 1659 string = sv_2mortal (newSVsv (string));
1433 1660
1434 SvUPGRADE (string, SVt_PV); 1661 SvUPGRADE (string, SVt_PV);
1435 1662
1436 /* 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
1453 if (offset > json->max_size && json->max_size) 1680 if (offset > json->max_size && json->max_size)
1454 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",
1455 (unsigned long)SvCUR (string), (unsigned long)json->max_size); 1682 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1456 } 1683 }
1457 1684
1458 if (json->flags & F_UTF8) 1685 if (DECODE_WANTS_OCTETS (json))
1459 sv_utf8_downgrade (string, 0); 1686 sv_utf8_downgrade (string, 0);
1460 else 1687 else
1461 sv_utf8_upgrade (string); 1688 sv_utf8_upgrade (string);
1462 1689
1463 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1690 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1505 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);
1506 LEAVE; 1733 LEAVE;
1507 1734
1508 croak ("%s, at character offset %d (before \"%s\")", 1735 croak ("%s, at character offset %d (before \"%s\")",
1509 dec.err, 1736 dec.err,
1510 ptr_to_index (string, dec.cur), 1737 (int)ptr_to_index (string, dec.cur),
1511 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1738 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1512 } 1739 }
1513 1740
1514 sv = sv_2mortal (sv); 1741 sv = sv_2mortal (sv);
1515 1742
1524 1751
1525static void 1752static void
1526incr_parse (JSON *self) 1753incr_parse (JSON *self)
1527{ 1754{
1528 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...
1529 1759
1530 for (;;) 1760 for (;;)
1531 { 1761 {
1532 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D 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
1533 switch (self->incr_mode) 1763 switch (self->incr_mode)
1534 { 1764 {
1535 // only used for intiial whitespace skipping 1765 // only used for initial whitespace skipping
1536 case INCR_M_WS: 1766 case INCR_M_WS:
1537 for (;;) 1767 for (;;)
1538 { 1768 {
1539 if (*p > 0x20) 1769 if (*p > 0x20)
1540 { 1770 {
1771 if (*p == '#')
1772 {
1773 self->incr_mode = INCR_M_C0;
1774 goto incr_m_c;
1775 }
1776 else
1777 {
1541 self->incr_mode = INCR_M_JSON; 1778 self->incr_mode = INCR_M_JSON;
1542 goto incr_m_json; 1779 goto incr_m_json;
1780 }
1543 } 1781 }
1544 else if (!*p) 1782 else if (!*p)
1545 goto interrupt; 1783 goto interrupt;
1546 1784
1547 ++p; 1785 ++p;
1553 goto interrupt; 1791 goto interrupt;
1554 1792
1555 ++p; 1793 ++p;
1556 self->incr_mode = INCR_M_STR; 1794 self->incr_mode = INCR_M_STR;
1557 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;
1558 1815
1559 // inside a string 1816 // inside a string
1560 case INCR_M_STR: 1817 case INCR_M_STR:
1561 incr_m_str: 1818 incr_m_str:
1562 for (;;) 1819 for (;;)
1613 self->incr_mode = INCR_M_STR; 1870 self->incr_mode = INCR_M_STR;
1614 goto incr_m_str; 1871 goto incr_m_str;
1615 1872
1616 case '[': 1873 case '[':
1617 case '{': 1874 case '{':
1875 case '(':
1618 if (++self->incr_nest > self->max_depth) 1876 if (++self->incr_nest > self->max_depth)
1619 croak (ERR_NESTING_EXCEEDED); 1877 croak (ERR_NESTING_EXCEEDED);
1620 break; 1878 break;
1621 1879
1622 case ']': 1880 case ']':
1623 case '}': 1881 case '}':
1624 if (--self->incr_nest <= 0) 1882 if (--self->incr_nest <= 0)
1625 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;
1626 } 1893 }
1627 } 1894 }
1628 } 1895 }
1629 1896
1630 modechange: 1897 modechange:
1631 ; 1898 ;
1632 } 1899 }
1633 1900
1634interrupt: 1901interrupt:
1635 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
1636 //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
1637} 1905}
1638 1906
1639///////////////////////////////////////////////////////////////////////////// 1907/////////////////////////////////////////////////////////////////////////////
1640// XS interface functions 1908// XS interface functions
1650 i >= '0' && i <= '9' ? i - '0' 1918 i >= '0' && i <= '9' ? i - '0'
1651 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1919 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1652 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1920 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1653 : -1; 1921 : -1;
1654 1922
1655 json_stash = gv_stashpv ("JSON::XS" , 1); 1923 json_stash = gv_stashpv ("JSON::XS" , 1);
1656 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1924 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1657 1925
1658 json_true = get_bool ("JSON::XS::true"); 1926 types_true = get_bool ("Types::Serialiser::true");
1659 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);
1660 1931
1661 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ 1932 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1662} 1933}
1663 1934
1664PROTOTYPES: DISABLE 1935PROTOTYPES: DISABLE
1665 1936
1666void CLONE (...) 1937void CLONE (...)
1667 CODE: 1938 CODE:
1668 json_stash = 0; 1939 json_stash = 0;
1669 json_boolean_stash = 0; 1940 types_boolean_stash = 0;
1670 1941
1671void new (char *klass) 1942void new (char *klass)
1672 PPCODE: 1943 PPCODE:
1673{ 1944{
1674 SV *pv = NEWSV (0, sizeof (JSON)); 1945 SV *pv = NEWSV (0, sizeof (JSON));
1675 SvPOK_only (pv); 1946 SvPOK_only (pv);
1676 json_init ((JSON *)SvPVX (pv)); 1947 json_init ((JSON *)SvPVX (pv));
1677 XPUSHs (sv_2mortal (sv_bless ( 1948 XPUSHs (sv_2mortal (sv_bless (
1678 newRV_noinc (pv), 1949 newRV_noinc (pv),
1679 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 1950 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1694 shrink = F_SHRINK 1965 shrink = F_SHRINK
1695 allow_blessed = F_ALLOW_BLESSED 1966 allow_blessed = F_ALLOW_BLESSED
1696 convert_blessed = F_CONV_BLESSED 1967 convert_blessed = F_CONV_BLESSED
1697 relaxed = F_RELAXED 1968 relaxed = F_RELAXED
1698 allow_unknown = F_ALLOW_UNKNOWN 1969 allow_unknown = F_ALLOW_UNKNOWN
1970 allow_tags = F_ALLOW_TAGS
1699 PPCODE: 1971 PPCODE:
1700{ 1972{
1701 if (enable) 1973 if (enable)
1702 self->flags |= ix; 1974 self->flags |= ix;
1703 else 1975 else
1719 get_shrink = F_SHRINK 1991 get_shrink = F_SHRINK
1720 get_allow_blessed = F_ALLOW_BLESSED 1992 get_allow_blessed = F_ALLOW_BLESSED
1721 get_convert_blessed = F_CONV_BLESSED 1993 get_convert_blessed = F_CONV_BLESSED
1722 get_relaxed = F_RELAXED 1994 get_relaxed = F_RELAXED
1723 get_allow_unknown = F_ALLOW_UNKNOWN 1995 get_allow_unknown = F_ALLOW_UNKNOWN
1996 get_allow_tags = F_ALLOW_TAGS
1724 PPCODE: 1997 PPCODE:
1725 XPUSHs (boolSV (self->flags & ix)); 1998 XPUSHs (boolSV (self->flags & ix));
1726 1999
1727void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 2000void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1728 PPCODE: 2001 PPCODE:
1756} 2029}
1757 2030
1758void 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)
1759 PPCODE: 2032 PPCODE:
1760{ 2033{
1761 if (!self->cb_sk_object) 2034 if (!self->cb_sk_object)
1762 self->cb_sk_object = newHV (); 2035 self->cb_sk_object = newHV ();
1763 2036
1764 if (SvOK (cb)) 2037 if (SvOK (cb))
1765 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2038 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1766 else 2039 else
1777 XPUSHs (ST (0)); 2050 XPUSHs (ST (0));
1778} 2051}
1779 2052
1780void encode (JSON *self, SV *scalar) 2053void encode (JSON *self, SV *scalar)
1781 PPCODE: 2054 PPCODE:
1782 XPUSHs (encode_json (scalar, self)); 2055 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2056 XPUSHs (scalar);
1783 2057
1784void decode (JSON *self, SV *jsonstr) 2058void decode (JSON *self, SV *jsonstr)
1785 PPCODE: 2059 PPCODE:
1786 XPUSHs (decode_json (jsonstr, self, 0)); 2060 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2061 XPUSHs (jsonstr);
1787 2062
1788void decode_prefix (JSON *self, SV *jsonstr) 2063void decode_prefix (JSON *self, SV *jsonstr)
1789 PPCODE: 2064 PPCODE:
1790{ 2065{
2066 SV *sv;
1791 char *offset; 2067 char *offset;
2068 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1792 EXTEND (SP, 2); 2069 EXTEND (SP, 2);
1793 PUSHs (decode_json (jsonstr, self, &offset)); 2070 PUSHs (sv);
1794 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2071 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1795} 2072}
1796 2073
1797void incr_parse (JSON *self, SV *jsonstr = 0) 2074void incr_parse (JSON *self, SV *jsonstr = 0)
1798 PPCODE: 2075 PPCODE:
1799{ 2076{
1800 if (!self->incr_text) 2077 if (!self->incr_text)
1801 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 }
1802 2098
1803 // append data, if any 2099 // append data, if any
1804 if (jsonstr) 2100 if (jsonstr)
1805 { 2101 {
2102 /* make sure both strings have same encoding */
2103 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1806 if (SvUTF8 (jsonstr)) 2104 if (SvUTF8 (jsonstr))
2105 sv_utf8_downgrade (jsonstr, 0);
1807 { 2106 else
1808 if (!SvUTF8 (self->incr_text))
1809 {
1810 /* utf-8-ness differs, need to upgrade */
1811 sv_utf8_upgrade (self->incr_text);
1812
1813 if (self->incr_pos)
1814 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1815 - (U8 *)SvPVX (self->incr_text);
1816 }
1817 }
1818 else if (SvUTF8 (self->incr_text))
1819 sv_utf8_upgrade (jsonstr); 2107 sv_utf8_upgrade (jsonstr);
1820 2108
2109 /* and then just blindly append */
1821 { 2110 {
1822 STRLEN len; 2111 STRLEN len;
1823 const char *str = SvPV (jsonstr, len); 2112 const char *str = SvPV (jsonstr, len);
1824 STRLEN cur = SvCUR (self->incr_text); 2113 STRLEN cur = SvCUR (self->incr_text);
1825 2114
1833 } 2122 }
1834 2123
1835 if (GIMME_V != G_VOID) 2124 if (GIMME_V != G_VOID)
1836 do 2125 do
1837 { 2126 {
2127 SV *sv;
1838 char *offset; 2128 char *offset;
1839 2129
1840 if (!INCR_DONE (self)) 2130 if (!INCR_DONE (self))
1841 { 2131 {
1842 incr_parse (self); 2132 incr_parse (self);
1844 if (self->incr_pos > self->max_size && self->max_size) 2134 if (self->incr_pos > self->max_size && self->max_size)
1845 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 2135 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1846 (unsigned long)self->incr_pos, (unsigned long)self->max_size); 2136 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1847 2137
1848 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
1849 break; 2147 break;
2148 }
1850 } 2149 }
1851 2150
1852 XPUSHs (decode_json (self->incr_text, self, &offset)); 2151 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2152 XPUSHs (sv);
1853 2153
1854 self->incr_pos -= offset - SvPVX (self->incr_text); 2154 self->incr_pos -= offset - SvPVX (self->incr_text);
1855 self->incr_nest = 0; 2155 self->incr_nest = 0;
1856 self->incr_mode = 0; 2156 self->incr_mode = 0;
1857 2157
1901 SvREFCNT_dec (self->incr_text); 2201 SvREFCNT_dec (self->incr_text);
1902 2202
1903PROTOTYPES: ENABLE 2203PROTOTYPES: ENABLE
1904 2204
1905void encode_json (SV *scalar) 2205void encode_json (SV *scalar)
1906 ALIAS:
1907 to_json_ = 0
1908 encode_json = F_UTF8
1909 PPCODE: 2206 PPCODE:
1910{ 2207{
1911 JSON json; 2208 JSON json;
1912 json_init (&json); 2209 json_init (&json);
1913 json.flags |= ix; 2210 json.flags |= F_UTF8;
1914 XPUSHs (encode_json (scalar, &json)); 2211 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2212 XPUSHs (scalar);
1915} 2213}
1916 2214
1917void decode_json (SV *jsonstr) 2215void decode_json (SV *jsonstr)
1918 ALIAS:
1919 from_json_ = 0
1920 decode_json = F_UTF8
1921 PPCODE: 2216 PPCODE:
1922{ 2217{
1923 JSON json; 2218 JSON json;
1924 json_init (&json); 2219 json_init (&json);
1925 json.flags |= ix; 2220 json.flags |= F_UTF8;
1926 XPUSHs (decode_json (jsonstr, &json, 0)); 2221 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2222 XPUSHs (jsonstr);
1927} 2223}
1928 2224
1929

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines