[ViewVC] Diff of: cvs/JSON-XS/XS.xs

Comparing JSON-XS/XS.xs (file contents):
Revision 1.17 by root, Sun Mar 25 02:46:41 2007 UTC vs.
Revision 1.26 by root, Fri Apr 6 21:17:09 2007 UTC

…		…
3	#include "XSUB.h"	3	#include "XSUB.h"
4		4
5	#include "assert.h"	5	#include "assert.h"
6	#include "string.h"	6	#include "string.h"
7	#include "stdlib.h"	7	#include "stdlib.h"
		8	#include "stdio.h"
8		9
		10	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		11	# define snprintf _snprintf // C compilers have this in stdio.h
		12	#endif
		13
9	#define F_ASCII 0x00000001	14	#define F_ASCII 0x00000001UL
10	#define F_UTF8 0x00000002	15	#define F_UTF8 0x00000002UL
11	#define F_INDENT 0x00000004	16	#define F_INDENT 0x00000004UL
12	#define F_CANONICAL 0x00000008	17	#define F_CANONICAL 0x00000008UL
13	#define F_SPACE_BEFORE 0x00000010	18	#define F_SPACE_BEFORE 0x00000010UL
14	#define F_SPACE_AFTER 0x00000020	19	#define F_SPACE_AFTER 0x00000020UL
15	#define F_ALLOW_NONREF 0x00000080	20	#define F_ALLOW_NONREF 0x00000080UL
16	#define F_SHRINK 0x00000100	21	#define F_SHRINK 0x00000100UL
		22	#define F_MAXDEPTH 0xf8000000UL
		23	#define S_MAXDEPTH 27
17		24
18	// F_SKIPINVALID?	25	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
19	// F_EXECCODEREF?	26
20	// F_SELFCONVERT? <=> { &__class__ => }	27	// F_SELFCONVERT? <=> to_json/toJson
		28	// F_BLESSED? <=> { $__class__$ => }
21		29
22	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	30	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
23	#define F_DEFAULT 0	31	#define F_DEFAULT (9UL << S_MAXDEPTH)
24		32
25	#define INIT_SIZE 32 // initial scalar size to be allocated	33	#define INIT_SIZE 32 // initial scalar size to be allocated
26	#define INDENT_STEP 3 // spaces per indentation level	34	#define INDENT_STEP 3 // spaces per indentation level
27		35
28	#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character	36	#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
…		…
73	{	81	{
74	*clen = 2;	82	*clen = 2;
75	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);	83	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
76	}	84	}
77	else	85	else
		86	{
		87	*clen = (STRLEN)-1;
78	return (UV)-1;	88	return (UV)-1;
		89	}
79	}	90	}
80		91
81	/////////////////////////////////////////////////////////////////////////////	92	/////////////////////////////////////////////////////////////////////////////
82	// encoder	93	// encoder
83		94
…		…
85	typedef struct	96	typedef struct
86	{	97	{
87	char *cur; // SvPVX (sv) + current output position	98	char *cur; // SvPVX (sv) + current output position
88	char *end; // SvEND (sv)	99	char *end; // SvEND (sv)
89	SV *sv; // result scalar	100	SV *sv; // result scalar
90	UV flags; // F_*	101	U32 flags; // F_*
91	int indent; // indentation level	102	U32 indent; // indentation level
92	int max_depth; // max. recursion level	103	U32 maxdepth; // max. indentation/recursion level
93	} enc_t;	104	} enc_t;
94		105
95	static void	106	static void
96	need (enc_t *enc, STRLEN len)	107	need (enc_t *enc, STRLEN len)
97	{	108	{
…		…
265	static void	276	static void
266	encode_av (enc_t enc, AV av)	277	encode_av (enc_t enc, AV av)
267	{	278	{
268	int i, len = av_len (av);	279	int i, len = av_len (av);
269		280
		281	if (enc->indent >= enc->maxdepth)
		282	croak ("data structure too deep (hit recursion limit)");
		283
270	encode_ch (enc, '['); encode_nl (enc);	284	encode_ch (enc, '['); encode_nl (enc);
271	++enc->indent;	285	++enc->indent;
272		286
273	for (i = 0; i <= len; ++i)	287	for (i = 0; i <= len; ++i)
274	{	288	{
…		…
339		353
340	static void	354	static void
341	encode_hv (enc_t enc, HV hv)	355	encode_hv (enc_t enc, HV hv)
342	{	356	{
343	int count, i;	357	int count, i;
		358
		359	if (enc->indent >= enc->maxdepth)
		360	croak ("data structure too deep (hit recursion limit)");
344		361
345	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;	362	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
346		363
347	if ((count = hv_iterinit (hv)))	364	if ((count = hv_iterinit (hv)))
348	{	365	{
…		…
396		413
397	encode_nl (enc);	414	encode_nl (enc);
398	}	415	}
399	else	416	else
400	{	417	{
401	SV *sv;
402	HE *he = hv_iternext (hv);	418	HE *he = hv_iternext (hv);
403		419
404	for (;;)	420	for (;;)
405	{	421	{
406	encode_indent (enc);	422	encode_indent (enc);
…		…
415	encode_nl (enc);	431	encode_nl (enc);
416	}	432	}
417	}	433	}
418		434
419	--enc->indent; encode_indent (enc); encode_ch (enc, '}');	435	--enc->indent; encode_indent (enc); encode_ch (enc, '}');
		436	}
		437
		438	// encode objects, arrays and special \0=false and \1=true values.
		439	static void
		440	encode_rv (enc_t enc, SV sv)
		441	{
		442	svtype svt;
		443
		444	SvGETMAGIC (sv);
		445	svt = SvTYPE (sv);
		446
		447	if (svt == SVt_PVHV)
		448	encode_hv (enc, (HV *)sv);
		449	else if (svt == SVt_PVAV)
		450	encode_av (enc, (AV *)sv);
		451	else if (svt < SVt_PVAV)
		452	{
		453	if (SvNIOK (sv) && SvIV (sv) == 0)
		454	encode_str (enc, "false", 5, 0);
		455	else if (SvNIOK (sv) && SvIV (sv) == 1)
		456	encode_str (enc, "true", 4, 0);
		457	else
		458	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
		459	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		460	}
		461	else
		462	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
		463	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
420	}	464	}
421		465
422	static void	466	static void
423	encode_sv (enc_t enc, SV sv)	467	encode_sv (enc_t enc, SV sv)
424	{	468	{
…		…
445	SvIsUV(sv)	489	SvIsUV(sv)
446	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	490	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
447	: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));	491	: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
448	}	492	}
449	else if (SvROK (sv))	493	else if (SvROK (sv))
450	{	494	encode_rv (enc, SvRV (sv));
451	SV *rv = SvRV (sv);
452
453	if (enc->indent >= enc->max_depth)
454	croak ("data structure too deep (hit recursion limit)");
455
456	switch (SvTYPE (rv))
457	{
458	case SVt_PVAV: encode_av (enc, (AV *)rv); break;
459	case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
460
461	default:
462	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
463	SvPV_nolen (sv));
464	}
465	}
466	else if (!SvOK (sv))	495	else if (!SvOK (sv))
467	encode_str (enc, "null", 4, 0);	496	encode_str (enc, "null", 4, 0);
468	else	497	else
469	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	498	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
470	SvPV_nolen (sv), SvFLAGS (sv));	499	SvPV_nolen (sv), SvFLAGS (sv));
471	}	500	}
472		501
473	static SV *	502	static SV *
474	encode_json (SV *scalar, UV flags)	503	encode_json (SV *scalar, U32 flags)
475	{	504	{
		505	enc_t enc;
		506
476	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	507	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
477	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");	508	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
478		509
479	enc_t enc;
480	enc.flags = flags;	510	enc.flags = flags;
481	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	511	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
482	enc.cur = SvPVX (enc.sv);	512	enc.cur = SvPVX (enc.sv);
483	enc.end = SvEND (enc.sv);	513	enc.end = SvEND (enc.sv);
484	enc.indent = 0;	514	enc.indent = 0;
485	enc.max_depth = 0x7fffffffUL;	515	enc.maxdepth = DEC_DEPTH (flags);
486		516
487	SvPOK_only (enc.sv);	517	SvPOK_only (enc.sv);
488	encode_sv (&enc, scalar);	518	encode_sv (&enc, scalar);
489		519
490	if (!(flags & (F_ASCII \| F_UTF8)))	520	if (!(flags & (F_ASCII \| F_UTF8)))
…		…
505	typedef struct	535	typedef struct
506	{	536	{
507	char *cur; // current parser pointer	537	char *cur; // current parser pointer
508	char *end; // end of input string	538	char *end; // end of input string
509	const char *err; // parse error, if != 0	539	const char *err; // parse error, if != 0
510	UV flags; // F_*	540	U32 flags; // F_*
		541	U32 depth; // recursion depth
		542	U32 maxdepth; // recursion depth limit
511	} dec_t;	543	} dec_t;
512		544
513	static void	545	static void
514	decode_ws (dec_t *dec)	546	decode_ws (dec_t *dec)
515	{	547	{
…		…
524	++dec->cur;	556	++dec->cur;
525	}	557	}
526	}	558	}
527		559
528	#define ERR(reason) SB dec->err = reason; goto fail; SE	560	#define ERR(reason) SB dec->err = reason; goto fail; SE
		561
529	#define EXPECT_CH(ch) SB \	562	#define EXPECT_CH(ch) SB \
530	if (*dec->cur != ch) \	563	if (*dec->cur != ch) \
531	ERR (# ch " expected"); \	564	ERR (# ch " expected"); \
532	++dec->cur; \	565	++dec->cur; \
533	SE	566	SE
		567
		568	#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
		569	#define DEC_DEC_DEPTH --dec->depth
534		570
535	static SV decode_sv (dec_t dec);	571	static SV decode_sv (dec_t dec);
536		572
537	static signed char decode_hexdigit[256];	573	static signed char decode_hexdigit[256];
538		574
…		…
640	}	676	}
641	else if (ch >= 0x20 && ch <= 0x7f)	677	else if (ch >= 0x20 && ch <= 0x7f)
642	*cur++ = ch;	678	*cur++ = ch;
643	else if (ch >= 0x80)	679	else if (ch >= 0x80)
644	{	680	{
		681	STRLEN clen;
		682	UV uch;
		683
645	--dec->cur;	684	--dec->cur;
646		685
647	STRLEN clen;
648	UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);	686	uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
649	if (clen == (STRLEN)-1)	687	if (clen == (STRLEN)-1)
650	ERR ("malformed UTF-8 character in JSON string");	688	ERR ("malformed UTF-8 character in JSON string");
651		689
652	do	690	do
653	{
654	cur++ = dec->cur++;	691	cur++ = dec->cur++;
655	}
656	while (--clen);	692	while (--clen);
657		693
658	utf8 = 1;	694	utf8 = 1;
659	}	695	}
660	else if (!ch)
661	ERR ("unexpected end of string while parsing json string");
662	else	696	else
		697	{
		698	--dec->cur;
		699
		700	if (!ch)
		701	ERR ("unexpected end of string while parsing JSON string");
		702	else
663	ERR ("invalid character encountered");	703	ERR ("invalid character encountered while parsing JSON string");
664		704	}
665	}	705	}
666	while (cur < buf + SHORT_STRING_LEN);	706	while (cur < buf + SHORT_STRING_LEN);
667		707
		708	{
668	STRLEN len = cur - buf;	709	STRLEN len = cur - buf;
669		710
670	if (sv)	711	if (sv)
671	{	712	{
672	SvGROW (sv, SvCUR (sv) + len + 1);	713	SvGROW (sv, SvCUR (sv) + len + 1);
673	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	714	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
674	SvCUR_set (sv, SvCUR (sv) + len);	715	SvCUR_set (sv, SvCUR (sv) + len);
675	}	716	}
676	else	717	else
677	sv = newSVpvn (buf, len);	718	sv = newSVpvn (buf, len);
		719	}
678	}	720	}
679	while (*dec->cur != '"');	721	while (*dec->cur != '"');
680		722
681	++dec->cur;	723	++dec->cur;
682		724
…		…
782	static SV *	824	static SV *
783	decode_av (dec_t *dec)	825	decode_av (dec_t *dec)
784	{	826	{
785	AV *av = newAV ();	827	AV *av = newAV ();
786		828
		829	DEC_INC_DEPTH;
787	decode_ws (dec);	830	decode_ws (dec);
		831
788	if (*dec->cur == ']')	832	if (*dec->cur == ']')
789	++dec->cur;	833	++dec->cur;
790	else	834	else
791	for (;;)	835	for (;;)
792	{	836	{
…		…
810	ERR (", or ] expected while parsing array");	854	ERR (", or ] expected while parsing array");
811		855
812	++dec->cur;	856	++dec->cur;
813	}	857	}
814		858
		859	DEC_DEC_DEPTH;
815	return newRV_noinc ((SV *)av);	860	return newRV_noinc ((SV *)av);
816		861
817	fail:	862	fail:
818	SvREFCNT_dec (av);	863	SvREFCNT_dec (av);
		864	DEC_DEC_DEPTH;
819	return 0;	865	return 0;
820	}	866	}
821		867
822	static SV *	868	static SV *
823	decode_hv (dec_t *dec)	869	decode_hv (dec_t *dec)
824	{	870	{
825	HV *hv = newHV ();	871	HV *hv = newHV ();
826		872
		873	DEC_INC_DEPTH;
827	decode_ws (dec);	874	decode_ws (dec);
		875
828	if (*dec->cur == '}')	876	if (*dec->cur == '}')
829	++dec->cur;	877	++dec->cur;
830	else	878	else
831	for (;;)	879	for (;;)
832	{	880	{
…		…
845	{	893	{
846	SvREFCNT_dec (key);	894	SvREFCNT_dec (key);
847	goto fail;	895	goto fail;
848	}	896	}
849		897
850	//TODO: optimise
851	hv_store_ent (hv, key, value, 0);	898	hv_store_ent (hv, key, value, 0);
		899	SvREFCNT_dec (key);
852		900
853	decode_ws (dec);	901	decode_ws (dec);
854		902
855	if (*dec->cur == '}')	903	if (*dec->cur == '}')
856	{	904	{
…		…
862	ERR (", or } expected while parsing object/hash");	910	ERR (", or } expected while parsing object/hash");
863		911
864	++dec->cur;	912	++dec->cur;
865	}	913	}
866		914
		915	DEC_DEC_DEPTH;
867	return newRV_noinc ((SV *)hv);	916	return newRV_noinc ((SV *)hv);
868		917
869	fail:	918	fail:
870	SvREFCNT_dec (hv);	919	SvREFCNT_dec (hv);
		920	DEC_DEC_DEPTH;
871	return 0;	921	return 0;
872	}	922	}
873		923
874	static SV *	924	static SV *
875	decode_sv (dec_t *dec)	925	decode_sv (dec_t *dec)
…		…
918	ERR ("'null' expected");	968	ERR ("'null' expected");
919		969
920	break;	970	break;
921		971
922	default:	972	default:
923	ERR ("malformed json string, neither array, object, number, string or atom");	973	ERR ("malformed JSON string, neither array, object, number, string or atom");
924	break;	974	break;
925	}	975	}
926		976
927	fail:	977	fail:
928	return 0;	978	return 0;
929	}	979	}
930		980
931	static SV *	981	static SV *
932	decode_json (SV *string, UV flags)	982	decode_json (SV *string, U32 flags)
933	{	983	{
		984	dec_t dec;
934	SV *sv;	985	SV *sv;
		986
		987	SvUPGRADE (string, SVt_PV);
935		988
936	if (flags & F_UTF8)	989	if (flags & F_UTF8)
937	sv_utf8_downgrade (string, 0);	990	sv_utf8_downgrade (string, 0);
938	else	991	else
939	sv_utf8_upgrade (string);	992	sv_utf8_upgrade (string);
940		993
941	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	994	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
942		995
943	dec_t dec;
944	dec.flags = flags;	996	dec.flags = flags;
945	dec.cur = SvPVX (string);	997	dec.cur = SvPVX (string);
946	dec.end = SvEND (string);	998	dec.end = SvEND (string);
947	dec.err = 0;	999	dec.err = 0;
		1000	dec.depth = 0;
		1001	dec.maxdepth = DEC_DEPTH (dec.flags);
948		1002
		1003	*dec.end = 0; // this should basically be a nop, too, but make sure its there
949	sv = decode_sv (&dec);	1004	sv = decode_sv (&dec);
950		1005
951	if (!sv)	1006	if (!sv)
952	{	1007	{
953	IV offset = dec.flags & F_UTF8	1008	IV offset = dec.flags & F_UTF8
…		…
962	SAVEVPTR (PL_curcop);	1017	SAVEVPTR (PL_curcop);
963	PL_curcop = &cop;	1018	PL_curcop = &cop;
964	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1019	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
965	LEAVE;	1020	LEAVE;
966		1021
967	croak ("%s, at character offset %d (%s)",	1022	croak ("%s, at character offset %d [\"%s\"]",
968	dec.err,	1023	dec.err,
969	(int)offset,	1024	(int)offset,
970	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1025	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
971	}	1026	}
972		1027
…		…
986	BOOT:	1041	BOOT:
987	{	1042	{
988	int i;	1043	int i;
989		1044
990	memset (decode_hexdigit, 0xff, 256);	1045	memset (decode_hexdigit, 0xff, 256);
		1046
991	for (i = 10; i--; )	1047	for (i = 0; i < 256; ++i)
992	decode_hexdigit ['0' + i] = i;	1048	decode_hexdigit [i] =
993		1049	i >= '0' && i <= '9' ? i - '0'
994	for (i = 7; i--; )	1050	: i >= 'a' && i <= 'f' ? i - 'a' + 10
995	{	1051	: i >= 'A' && i <= 'F' ? i - 'A' + 10
996	decode_hexdigit ['a' + i] = 10 + i;	1052	: -1;
997	decode_hexdigit ['A' + i] = 10 + i;
998	}
999		1053
1000	json_stash = gv_stashpv ("JSON::XS", 1);	1054	json_stash = gv_stashpv ("JSON::XS", 1);
1001	}	1055	}
1002		1056
1003	PROTOTYPES: DISABLE	1057	PROTOTYPES: DISABLE
…		…
1030	RETVAL = newSVsv (self);	1084	RETVAL = newSVsv (self);
1031	}	1085	}
1032	OUTPUT:	1086	OUTPUT:
1033	RETVAL	1087	RETVAL
1034		1088
		1089	SV max_depth (SV self, UV max_depth = 0x80000000UL)
		1090	CODE:
		1091	{
		1092	UV *uv = SvJSON (self);
		1093	UV log2 = 0;
		1094
		1095	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
		1096
		1097	while ((1UL << log2) < max_depth)
		1098	++log2;
		1099
		1100	uv = uv & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
		1101
		1102	RETVAL = newSVsv (self);
		1103	}
		1104	OUTPUT:
		1105	RETVAL
		1106
1035	void encode (SV self, SV scalar)	1107	void encode (SV self, SV scalar)
1036	PPCODE:	1108	PPCODE:
1037	XPUSHs (encode_json (scalar, *SvJSON (self)));	1109	XPUSHs (encode_json (scalar, *SvJSON (self)));
1038		1110
1039	void decode (SV self, SV jsonstr)	1111	void decode (SV self, SV jsonstr)
…		…
1044		1116
1045	void to_json (SV *scalar)	1117	void to_json (SV *scalar)
1046	ALIAS:	1118	ALIAS:
1047	objToJson = 0	1119	objToJson = 0
1048	PPCODE:	1120	PPCODE:
1049	XPUSHs (encode_json (scalar, F_UTF8));	1121	XPUSHs (encode_json (scalar, F_DEFAULT \| F_UTF8));
1050		1122
1051	void from_json (SV *jsonstr)	1123	void from_json (SV *jsonstr)
1052	ALIAS:	1124	ALIAS:
1053	jsonToObj = 0	1125	jsonToObj = 0
1054	PPCODE:	1126	PPCODE:
1055	XPUSHs (decode_json (jsonstr, F_UTF8));	1127	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8));
1056		1128

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.17 by root, Sun Mar 25 02:46:41 2007 UTC vs. Revision 1.26 by root, Fri Apr 6 21:17:09 2007 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.17 by root, Sun Mar 25 02:46:41 2007 UTC vs.
Revision 1.26 by root, Fri Apr 6 21:17:09 2007 UTC