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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.38 by root, Mon Jun 11 03:18:07 2007 UTC vs.
Revision 1.54 by root, Tue Jul 10 16:22:31 2007 UTC

…		…
9		9
10	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	10	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
11	# define snprintf _snprintf // C compilers have this in stdio.h	11	# define snprintf _snprintf // C compilers have this in stdio.h
12	#endif	12	#endif
13		13
		14	// some old perls do not have this, try to make it work, no
		15	// guarentees, though. if it breaks, you get to keep the pieces.
		16	#ifndef UTF8_MAXBYTES
		17	# define UTF8_MAXBYTES 13
		18	#endif
		19
14	#define F_ASCII 0x00000001UL	20	#define F_ASCII 0x00000001UL
15	#define F_LATIN1 0x00000002UL	21	#define F_LATIN1 0x00000002UL
16	#define F_UTF8 0x00000004UL	22	#define F_UTF8 0x00000004UL
17	#define F_INDENT 0x00000008UL	23	#define F_INDENT 0x00000008UL
18	#define F_CANONICAL 0x00000010UL	24	#define F_CANONICAL 0x00000010UL
19	#define F_SPACE_BEFORE 0x00000020UL	25	#define F_SPACE_BEFORE 0x00000020UL
20	#define F_SPACE_AFTER 0x00000040UL	26	#define F_SPACE_AFTER 0x00000040UL
21	#define F_ALLOW_NONREF 0x00000100UL	27	#define F_ALLOW_NONREF 0x00000100UL
22	#define F_SHRINK 0x00000200UL	28	#define F_SHRINK 0x00000200UL
		29	#define F_ALLOW_BLESSED 0x00000400UL
		30	#define F_CONV_BLESSED 0x00000800UL
23	#define F_MAXDEPTH 0xf8000000UL	31	#define F_MAXDEPTH 0xf8000000UL
24	#define S_MAXDEPTH 27	32	#define S_MAXDEPTH 27
		33	#define F_MAXSIZE 0x01f00000UL
		34	#define S_MAXSIZE 20
		35	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
25		36
26	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))	37	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27		38	#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
28	// F_SELFCONVERT? <=> to_json/toJson
29	// F_BLESSED? <=> { $__class__$ => }
30		39
31	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	40	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
32	#define F_DEFAULT (9UL << S_MAXDEPTH)	41	#define F_DEFAULT (9UL << S_MAXDEPTH)
33		42
34	#define INIT_SIZE 32 // initial scalar size to be allocated	43	#define INIT_SIZE 32 // initial scalar size to be allocated
…		…
48	#endif	57	#endif
49		58
50	#define expect_false(expr) expect ((expr) != 0, 0)	59	#define expect_false(expr) expect ((expr) != 0, 0)
51	#define expect_true(expr) expect ((expr) != 0, 1)	60	#define expect_true(expr) expect ((expr) != 0, 1)
52		61
53	static HV *json_stash; // JSON::XS::	62	static HV json_stash, json_boolean_stash; // JSON::XS::
		63	static SV json_true, json_false;
		64
		65	typedef struct {
		66	U32 flags;
		67	SV *cb_object;
		68	HV *cb_sk_object;
		69	} JSON;
54		70
55	/////////////////////////////////////////////////////////////////////////////	71	/////////////////////////////////////////////////////////////////////////////
56	// utility functions	72	// utility functions
57		73
58	static UV *	74	inline void
59	SvJSON (SV *sv)
60	{
61	if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
62	croak ("object is not of type JSON::XS");
63
64	return &SvUVX (SvRV (sv));
65	}
66
67	static void
68	shrink (SV *sv)	75	shrink (SV *sv)
69	{	76	{
70	sv_utf8_downgrade (sv, 1);	77	sv_utf8_downgrade (sv, 1);
71	if (SvLEN (sv) > SvCUR (sv) + 1)	78	if (SvLEN (sv) > SvCUR (sv) + 1)
72	{	79	{
…		…
107	typedef struct	114	typedef struct
108	{	115	{
109	char *cur; // SvPVX (sv) + current output position	116	char *cur; // SvPVX (sv) + current output position
110	char *end; // SvEND (sv)	117	char *end; // SvEND (sv)
111	SV *sv; // result scalar	118	SV *sv; // result scalar
112	U32 flags; // F_*	119	JSON json;
113	U32 indent; // indentation level	120	U32 indent; // indentation level
114	U32 maxdepth; // max. indentation/recursion level	121	U32 maxdepth; // max. indentation/recursion level
115	} enc_t;	122	} enc_t;
116		123
117	inline void	124	inline void
…		…
178	STRLEN clen;	185	STRLEN clen;
179	UV uch;	186	UV uch;
180		187
181	if (is_utf8)	188	if (is_utf8)
182	{	189	{
183	//uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184	uch = decode_utf8 (str, end - str, &clen);	190	uch = decode_utf8 (str, end - str, &clen);
185	if (clen == (STRLEN)-1)	191	if (clen == (STRLEN)-1)
186	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);	192	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187	}	193	}
188	else	194	else
…		…
192	}	198	}
193		199
194	if (uch > 0x10FFFFUL)	200	if (uch > 0x10FFFFUL)
195	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);	201	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196		202
197	if (uch < 0x80 \|\| enc->flags & F_ASCII \|\| (enc->flags & F_LATIN1 && uch > 0xFF))	203	if (uch < 0x80 \|\| enc->json.flags & F_ASCII \|\| (enc->json.flags & F_LATIN1 && uch > 0xFF))
198	{	204	{
199	if (uch > 0xFFFFUL)	205	if (uch > 0xFFFFUL)
200	{	206	{
201	need (enc, len += 11);	207	need (enc, len += 11);
202	sprintf (enc->cur, "\\u%04x\\u%04x",	208	sprintf (enc->cur, "\\u%04x\\u%04x",
…		…
216	*enc->cur++ = hexdigit [(uch >> 0) & 15];	222	*enc->cur++ = hexdigit [(uch >> 0) & 15];
217	}	223	}
218		224
219	str += clen;	225	str += clen;
220	}	226	}
221	else if (enc->flags & F_LATIN1)	227	else if (enc->json.flags & F_LATIN1)
222	{	228	{
223	*enc->cur++ = uch;	229	*enc->cur++ = uch;
224	str += clen;	230	str += clen;
225	}	231	}
226	else if (is_utf8)	232	else if (is_utf8)
…		…
247	}	253	}
248		254
249	inline void	255	inline void
250	encode_indent (enc_t *enc)	256	encode_indent (enc_t *enc)
251	{	257	{
252	if (enc->flags & F_INDENT)	258	if (enc->json.flags & F_INDENT)
253	{	259	{
254	int spaces = enc->indent * INDENT_STEP;	260	int spaces = enc->indent * INDENT_STEP;
255		261
256	need (enc, spaces);	262	need (enc, spaces);
257	memset (enc->cur, ' ', spaces);	263	memset (enc->cur, ' ', spaces);
…		…
267	}	273	}
268		274
269	inline void	275	inline void
270	encode_nl (enc_t *enc)	276	encode_nl (enc_t *enc)
271	{	277	{
272	if (enc->flags & F_INDENT)	278	if (enc->json.flags & F_INDENT)
273	{	279	{
274	need (enc, 1);	280	need (enc, 1);
275	encode_ch (enc, '\n');	281	encode_ch (enc, '\n');
276	}	282	}
277	}	283	}
…		…
279	inline void	285	inline void
280	encode_comma (enc_t *enc)	286	encode_comma (enc_t *enc)
281	{	287	{
282	encode_ch (enc, ',');	288	encode_ch (enc, ',');
283		289
284	if (enc->flags & F_INDENT)	290	if (enc->json.flags & F_INDENT)
285	encode_nl (enc);	291	encode_nl (enc);
286	else if (enc->flags & F_SPACE_AFTER)	292	else if (enc->json.flags & F_SPACE_AFTER)
287	encode_space (enc);	293	encode_space (enc);
288	}	294	}
289		295
290	static void encode_sv (enc_t enc, SV sv);	296	static void encode_sv (enc_t enc, SV sv);
291		297
…		…
334	else	340	else
335	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));	341	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336		342
337	encode_ch (enc, '"');	343	encode_ch (enc, '"');
338		344
339	if (enc->flags & F_SPACE_BEFORE) encode_space (enc);	345	if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340	encode_ch (enc, ':');	346	encode_ch (enc, ':');
341	if (enc->flags & F_SPACE_AFTER ) encode_space (enc);	347	if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342	encode_sv (enc, HeVAL (he));	348	encode_sv (enc, HeVAL (he));
343	}	349	}
344		350
345	// compare hash entries, used when all keys are bytestrings	351	// compare hash entries, used when all keys are bytestrings
346	static int	352	static int
…		…
381	{	387	{
382	// for canonical output we have to sort by keys first	388	// for canonical output we have to sort by keys first
383	// actually, this is mostly due to the stupid so-called	389	// actually, this is mostly due to the stupid so-called
384	// security workaround added somewhere in 5.8.x.	390	// security workaround added somewhere in 5.8.x.
385	// that randomises hash orderings	391	// that randomises hash orderings
386	if (enc->flags & F_CANONICAL)	392	if (enc->json.flags & F_CANONICAL)
387	{	393	{
388	int fast = 1;	394	int fast = 1;
389	HE *he;	395	HE *he;
390	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	396	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
391	HE *hes = _alloca (count sizeof (HE));	397	HE *hes = _alloca (count sizeof (HE));
…		…
463	svtype svt;	469	svtype svt;
464		470
465	SvGETMAGIC (sv);	471	SvGETMAGIC (sv);
466	svt = SvTYPE (sv);	472	svt = SvTYPE (sv);
467		473
		474	if (expect_false (SvOBJECT (sv)))
		475	{
		476	if (SvSTASH (sv) == json_boolean_stash)
		477	{
		478	if (SvIV (sv))
		479	encode_str (enc, "true", 4, 0);
		480	else
		481	encode_str (enc, "false", 5, 0);
		482	}
		483	else
		484	{
		485	#if 0
		486	if (0 && sv_derived_from (rv, "JSON::Literal"))
		487	{
		488	// not yet
		489	}
		490	#endif
		491	if (enc->json.flags & F_CONV_BLESSED)
		492	{
		493	// we re-bless the reference to get overload and other niceties right
		494	GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
		495
		496	if (to_json)
		497	{
		498	int count;
		499	dSP;
		500
		501	ENTER; SAVETMPS; PUSHMARK (SP);
		502	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
		503
		504	// calling with G_SCALAR ensures that we always get a 1 return value
		505	PUTBACK;
		506	call_sv ((SV *)GvCV (to_json), G_SCALAR);
		507	SPAGAIN;
		508
		509	// catch this surprisingly common error
		510	if (SvROK (TOPs) && SvRV (TOPs) == sv)
		511	croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
		512
		513	sv = POPs;
		514	PUTBACK;
		515
		516	encode_sv (enc, sv);
		517
		518	FREETMPS; LEAVE;
		519	}
		520	else if (enc->json.flags & F_ALLOW_BLESSED)
		521	encode_str (enc, "null", 4, 0);
		522	else
		523	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
		524	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		525	}
		526	else if (enc->json.flags & F_ALLOW_BLESSED)
		527	encode_str (enc, "null", 4, 0);
		528	else
		529	croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
		530	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		531	}
		532	}
468	if (svt == SVt_PVHV)	533	else if (svt == SVt_PVHV)
469	encode_hv (enc, (HV *)sv);	534	encode_hv (enc, (HV *)sv);
470	else if (svt == SVt_PVAV)	535	else if (svt == SVt_PVAV)
471	encode_av (enc, (AV *)sv);	536	encode_av (enc, (AV *)sv);
472	else if (svt < SVt_PVAV)	537	else if (svt < SVt_PVAV)
473	{	538	{
474	if (SvNIOK (sv) && SvIV (sv) == 0)	539	STRLEN len = 0;
		540	char *pv = svt ? SvPV (sv, len) : 0;
		541
		542	if (len == 1 && *pv == '1')
		543	encode_str (enc, "true", 4, 0);
		544	else if (len == 1 && *pv == '0')
475	encode_str (enc, "false", 5, 0);	545	encode_str (enc, "false", 5, 0);
476	else if (SvNIOK (sv) && SvIV (sv) == 1)
477	encode_str (enc, "true", 4, 0);
478	else	546	else
479	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",	547	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	548	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481	}	549	}
482	else	550	else
…		…
497	encode_str (enc, str, len, SvUTF8 (sv));	565	encode_str (enc, str, len, SvUTF8 (sv));
498	encode_ch (enc, '"');	566	encode_ch (enc, '"');
499	}	567	}
500	else if (SvNOKp (sv))	568	else if (SvNOKp (sv))
501	{	569	{
		570	// trust that perl will do the right thing w.r.t. JSON syntax.
502	need (enc, NV_DIG + 32);	571	need (enc, NV_DIG + 32);
503	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	572	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
504	enc->cur += strlen (enc->cur);	573	enc->cur += strlen (enc->cur);
505	}	574	}
506	else if (SvIOKp (sv))	575	else if (SvIOKp (sv))
507	{	576	{
508	// we assume we can always read an IV as a UV	577	// we assume we can always read an IV as a UV
509	if (SvUV (sv) & ~(UV)0x7fff)	578	if (SvUV (sv) & ~(UV)0x7fff)
510	{	579	{
		580	// large integer, use the (rather slow) snprintf way.
511	need (enc, sizeof (UV) * 3);	581	need (enc, sizeof (UV) * 3);
512	enc->cur +=	582	enc->cur +=
513	SvIsUV(sv)	583	SvIsUV(sv)
514	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))	584	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
515	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));	585	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
…		…
518	{	588	{
519	// optimise the "small number case"	589	// optimise the "small number case"
520	// code will likely be branchless and use only a single multiplication	590	// code will likely be branchless and use only a single multiplication
521	I32 i = SvIV (sv);	591	I32 i = SvIV (sv);
522	U32 u;	592	U32 u;
		593	char digit, nz = 0;
523		594
524	need (enc, 6);	595	need (enc, 6);
525		596
526	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;	597	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
527	u = i < 0 ? -i : i;	598	u = i < 0 ? -i : i;
528		599
529	// convert to 4.28 fixed-point representation	600	// convert to 4.28 fixed-point representation
530	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits	601	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
531		602
532	char digit, nz = 0;	603	// now output digit by digit, each time masking out the integer part
533		604	// and multiplying by 5 while moving the decimal point one to the right,
		605	// resulting in a net multiplication by 10.
		606	// we always write the digit to memory but conditionally increment
		607	// the pointer, to ease the usage of conditional move instructions.
534	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffff) 5;	608	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffff) 5;
535	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffff) 5;	609	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffff) 5;
536	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffff) 5;	610	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffff) 5;
537	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffff) 5;	611	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffff) 5;
538	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1;	612	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
539	}	613	}
540	}	614	}
541	else if (SvROK (sv))	615	else if (SvROK (sv))
542	encode_rv (enc, SvRV (sv));	616	encode_rv (enc, SvRV (sv));
543	else if (!SvOK (sv))	617	else if (!SvOK (sv))
…		…
546	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	620	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
547	SvPV_nolen (sv), SvFLAGS (sv));	621	SvPV_nolen (sv), SvFLAGS (sv));
548	}	622	}
549		623
550	static SV *	624	static SV *
551	encode_json (SV *scalar, U32 flags)	625	encode_json (SV scalar, JSON json)
552	{	626	{
553	enc_t enc;	627	enc_t enc;
554		628
555	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	629	if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
556	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");	630	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
557		631
558	enc.flags = flags;	632	enc.json = *json;
559	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	633	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
560	enc.cur = SvPVX (enc.sv);	634	enc.cur = SvPVX (enc.sv);
561	enc.end = SvEND (enc.sv);	635	enc.end = SvEND (enc.sv);
562	enc.indent = 0;	636	enc.indent = 0;
563	enc.maxdepth = DEC_DEPTH (flags);	637	enc.maxdepth = DEC_DEPTH (enc.json.flags);
564		638
565	SvPOK_only (enc.sv);	639	SvPOK_only (enc.sv);
566	encode_sv (&enc, scalar);	640	encode_sv (&enc, scalar);
567		641
568	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	642	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
569	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	643	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
570		644
571	if (!(flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))	645	if (!(enc.json.flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
572	SvUTF8_on (enc.sv);	646	SvUTF8_on (enc.sv);
573		647
574	if (enc.flags & F_SHRINK)	648	if (enc.json.flags & F_SHRINK)
575	shrink (enc.sv);	649	shrink (enc.sv);
576		650
577	return enc.sv;	651	return enc.sv;
578	}	652	}
579		653
…		…
584	typedef struct	658	typedef struct
585	{	659	{
586	char *cur; // current parser pointer	660	char *cur; // current parser pointer
587	char *end; // end of input string	661	char *end; // end of input string
588	const char *err; // parse error, if != 0	662	const char *err; // parse error, if != 0
589	U32 flags; // F_*	663	JSON json;
590	U32 depth; // recursion depth	664	U32 depth; // recursion depth
591	U32 maxdepth; // recursion depth limit	665	U32 maxdepth; // recursion depth limit
592	} dec_t;	666	} dec_t;
593		667
594	inline void	668	inline void
…		…
625	decode_4hex (dec_t *dec)	699	decode_4hex (dec_t *dec)
626	{	700	{
627	signed char d1, d2, d3, d4;	701	signed char d1, d2, d3, d4;
628	unsigned char cur = (unsigned char )dec->cur;	702	unsigned char cur = (unsigned char )dec->cur;
629		703
630	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected");	704	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
631	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected");	705	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
632	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected");	706	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
633	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected");	707	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
634		708
635	dec->cur += 4;	709	dec->cur += 4;
636		710
637	return ((UV)d1) << 12	711	return ((UV)d1) << 12
638	\| ((UV)d2) << 8	712	\| ((UV)d2) << 8
…		…
861	{	935	{
862	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so	936	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
863	if (*start == '-')	937	if (*start == '-')
864	switch (dec->cur - start)	938	switch (dec->cur - start)
865	{	939	{
866	case 2: return newSViv (-( start [1] - '0' ));	940	case 2: return newSViv (-( start [1] - '0' * 1));
867	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));	941	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
868	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));	942	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
869	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));	943	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
870	}	944	}
871	else	945	else
872	switch (dec->cur - start)	946	switch (dec->cur - start)
873	{	947	{
874	case 1: return newSViv ( start [0] - '0' );	948	case 1: return newSViv ( start [0] - '0' * 1);
875	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);	949	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
876	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);	950	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
877	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);	951	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
878	}	952	}
879		953
…		…
886	if (uv < (UV)IV_MIN)	960	if (uv < (UV)IV_MIN)
887	return newSViv (-(IV)uv);	961	return newSViv (-(IV)uv);
888	}	962	}
889	else	963	else
890	return newSVuv (uv);	964	return newSVuv (uv);
		965
		966	// here would likely be the place for bigint support
891	}	967	}
892	}	968	}
893		969
		970	// if we ever support bigint or bigfloat, this is the place for bigfloat
894	return newSVnv (Atof (start));	971	return newSVnv (Atof (start));
895		972
896	fail:	973	fail:
897	return 0;	974	return 0;
898	}	975	}
…		…
942	}	1019	}
943		1020
944	static SV *	1021	static SV *
945	decode_hv (dec_t *dec)	1022	decode_hv (dec_t *dec)
946	{	1023	{
		1024	SV *sv;
947	HV *hv = newHV ();	1025	HV *hv = newHV ();
948		1026
949	DEC_INC_DEPTH;	1027	DEC_INC_DEPTH;
950	decode_ws (dec);	1028	decode_ws (dec);
951		1029
952	if (*dec->cur == '}')	1030	if (*dec->cur == '}')
953	++dec->cur;	1031	++dec->cur;
954	else	1032	else
955	for (;;)	1033	for (;;)
956	{	1034	{
957	SV key, value;
958
959	decode_ws (dec); EXPECT_CH ('"');	1035	decode_ws (dec); EXPECT_CH ('"');
960		1036
961	key = decode_str (dec);	1037	// heuristic: assume that
962	if (!key)	1038	// a) decode_str + hv_store_ent are abysmally slow.
963	goto fail;	1039	// b) most hash keys are short, simple ascii text.
		1040	// => try to "fast-match" such strings to avoid
		1041	// the overhead of decode_str + hv_store_ent.
		1042	{
		1043	SV *value;
		1044	char *p = dec->cur;
		1045	char *e = p + 24; // only try up to 24 bytes
964		1046
965	decode_ws (dec); EXPECT_CH (':');	1047	for (;;)
966
967	value = decode_sv (dec);
968	if (!value)
969	{	1048	{
		1049	// the >= 0x80 is true on most architectures
		1050	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
		1051	{
		1052	// slow path, back up and use decode_str
		1053	SV *key = decode_str (dec);
		1054	if (!key)
		1055	goto fail;
		1056
		1057	decode_ws (dec); EXPECT_CH (':');
		1058
		1059	value = decode_sv (dec);
		1060	if (!value)
		1061	{
		1062	SvREFCNT_dec (key);
		1063	goto fail;
		1064	}
		1065
		1066	hv_store_ent (hv, key, value, 0);
970	SvREFCNT_dec (key);	1067	SvREFCNT_dec (key);
		1068
		1069	break;
		1070	}
		1071	else if (*p == '"')
		1072	{
		1073	// fast path, got a simple key
		1074	char *key = dec->cur;
		1075	int len = p - key;
		1076	dec->cur = p + 1;
		1077
		1078	decode_ws (dec); EXPECT_CH (':');
		1079
		1080	value = decode_sv (dec);
		1081	if (!value)
971	goto fail;	1082	goto fail;
		1083
		1084	hv_store (hv, key, len, value, 0);
		1085
		1086	break;
		1087	}
		1088
		1089	++p;
972	}	1090	}
973		1091	}
974	hv_store_ent (hv, key, value, 0);
975	SvREFCNT_dec (key);
976		1092
977	decode_ws (dec);	1093	decode_ws (dec);
978		1094
979	if (*dec->cur == '}')	1095	if (*dec->cur == '}')
980	{	1096	{
…		…
987		1103
988	++dec->cur;	1104	++dec->cur;
989	}	1105	}
990		1106
991	DEC_DEC_DEPTH;	1107	DEC_DEC_DEPTH;
992	return newRV_noinc ((SV *)hv);	1108	sv = newRV_noinc ((SV *)hv);
		1109
		1110	// check filter callbacks
		1111	if (dec->json.flags & F_HOOK)
		1112	{
		1113	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
		1114	{
		1115	HE cb, he;
		1116
		1117	hv_iterinit (hv);
		1118	he = hv_iternext (hv);
		1119	hv_iterinit (hv);
		1120
		1121	// the next line creates a mortal sv each time its called.
		1122	// might want to optimise this for common cases.
		1123	cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
		1124
		1125	if (cb)
		1126	{
		1127	dSP;
		1128	int count;
		1129
		1130	ENTER; SAVETMPS; PUSHMARK (SP);
		1131	XPUSHs (HeVAL (he));
		1132
		1133	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
		1134
		1135	if (count == 1)
		1136	{
		1137	sv = newSVsv (POPs);
		1138	FREETMPS; LEAVE;
		1139	return sv;
		1140	}
		1141
		1142	FREETMPS; LEAVE;
		1143	}
		1144	}
		1145
		1146	if (dec->json.cb_object)
		1147	{
		1148	dSP;
		1149	int count;
		1150
		1151	ENTER; SAVETMPS; PUSHMARK (SP);
		1152	XPUSHs (sv_2mortal (sv));
		1153
		1154	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
		1155
		1156	if (count == 1)
		1157	{
		1158	sv = newSVsv (POPs);
		1159	FREETMPS; LEAVE;
		1160	return sv;
		1161	}
		1162
		1163	SvREFCNT_inc (sv);
		1164	FREETMPS; LEAVE;
		1165	}
		1166	}
		1167
		1168	return sv;
993		1169
994	fail:	1170	fail:
995	SvREFCNT_dec (hv);	1171	SvREFCNT_dec (hv);
996	DEC_DEC_DEPTH;	1172	DEC_DEC_DEPTH;
997	return 0;	1173	return 0;
…		…
999		1175
1000	static SV *	1176	static SV *
1001	decode_sv (dec_t *dec)	1177	decode_sv (dec_t *dec)
1002	{	1178	{
1003	decode_ws (dec);	1179	decode_ws (dec);
		1180
		1181	// the beauty of JSON: you need exactly one character lookahead
		1182	// to parse anything.
1004	switch (*dec->cur)	1183	switch (*dec->cur)
1005	{	1184	{
1006	case '"': ++dec->cur; return decode_str (dec);	1185	case '"': ++dec->cur; return decode_str (dec);
1007	case '[': ++dec->cur; return decode_av (dec);	1186	case '[': ++dec->cur; return decode_av (dec);
1008	case '{': ++dec->cur; return decode_hv (dec);	1187	case '{': ++dec->cur; return decode_hv (dec);
…		…
1014		1193
1015	case 't':	1194	case 't':
1016	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1195	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1017	{	1196	{
1018	dec->cur += 4;	1197	dec->cur += 4;
1019	return newSViv (1);	1198	return SvREFCNT_inc (json_true);
1020	}	1199	}
1021	else	1200	else
1022	ERR ("'true' expected");	1201	ERR ("'true' expected");
1023		1202
1024	break;	1203	break;
1025		1204
1026	case 'f':	1205	case 'f':
1027	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1206	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1028	{	1207	{
1029	dec->cur += 5;	1208	dec->cur += 5;
1030	return newSViv (0);	1209	return SvREFCNT_inc (json_false);
1031	}	1210	}
1032	else	1211	else
1033	ERR ("'false' expected");	1212	ERR ("'false' expected");
1034		1213
1035	break;	1214	break;
…		…
1053	fail:	1232	fail:
1054	return 0;	1233	return 0;
1055	}	1234	}
1056		1235
1057	static SV *	1236	static SV *
1058	decode_json (SV string, U32 flags, UV offset_return)	1237	decode_json (SV string, JSON json, UV *offset_return)
1059	{	1238	{
1060	dec_t dec;	1239	dec_t dec;
1061	UV offset;	1240	UV offset;
1062	SV *sv;	1241	SV *sv;
1063		1242
1064	SvGETMAGIC (string);	1243	SvGETMAGIC (string);
1065	SvUPGRADE (string, SVt_PV);	1244	SvUPGRADE (string, SVt_PV);
1066		1245
		1246	if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
		1247	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1248	(unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
		1249
1067	if (flags & F_UTF8)	1250	if (json->flags & F_UTF8)
1068	sv_utf8_downgrade (string, 0);	1251	sv_utf8_downgrade (string, 0);
1069	else	1252	else
1070	sv_utf8_upgrade (string);	1253	sv_utf8_upgrade (string);
1071		1254
1072	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1255	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1073		1256
1074	dec.flags = flags;	1257	dec.json = *json;
1075	dec.cur = SvPVX (string);	1258	dec.cur = SvPVX (string);
1076	dec.end = SvEND (string);	1259	dec.end = SvEND (string);
1077	dec.err = 0;	1260	dec.err = 0;
1078	dec.depth = 0;	1261	dec.depth = 0;
1079	dec.maxdepth = DEC_DEPTH (dec.flags);	1262	dec.maxdepth = DEC_DEPTH (dec.json.flags);
		1263
		1264	if (dec.json.cb_object \|\| dec.json.cb_sk_object)
		1265	dec.json.flags \|= F_HOOK;
1080		1266
1081	*dec.end = 0; // this should basically be a nop, too, but make sure it's there	1267	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
1082	sv = decode_sv (&dec);	1268	sv = decode_sv (&dec);
1083		1269
1084	if (!(offset_return \|\| !sv))	1270	if (!(offset_return \|\| !sv))
…		…
1094	}	1280	}
1095	}	1281	}
1096		1282
1097	if (offset_return \|\| !sv)	1283	if (offset_return \|\| !sv)
1098	{	1284	{
1099	offset = dec.flags & F_UTF8	1285	offset = dec.json.flags & F_UTF8
1100	? dec.cur - SvPVX (string)	1286	? dec.cur - SvPVX (string)
1101	: utf8_distance (dec.cur, SvPVX (string));	1287	: utf8_distance (dec.cur, SvPVX (string));
1102		1288
1103	if (offset_return)	1289	if (offset_return)
1104	*offset_return = offset;	1290	*offset_return = offset;
…		…
1123	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1309	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1124	}	1310	}
1125		1311
1126	sv = sv_2mortal (sv);	1312	sv = sv_2mortal (sv);
1127		1313
1128	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))	1314	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1129	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");	1315	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1130		1316
1131	return sv;	1317	return sv;
1132	}	1318	}
1133		1319
…		…
1137	MODULE = JSON::XS PACKAGE = JSON::XS	1323	MODULE = JSON::XS PACKAGE = JSON::XS
1138		1324
1139	BOOT:	1325	BOOT:
1140	{	1326	{
1141	int i;	1327	int i;
1142
1143	memset (decode_hexdigit, 0xff, 256);
1144		1328
1145	for (i = 0; i < 256; ++i)	1329	for (i = 0; i < 256; ++i)
1146	decode_hexdigit [i] =	1330	decode_hexdigit [i] =
1147	i >= '0' && i <= '9' ? i - '0'	1331	i >= '0' && i <= '9' ? i - '0'
1148	: i >= 'a' && i <= 'f' ? i - 'a' + 10	1332	: i >= 'a' && i <= 'f' ? i - 'a' + 10
1149	: i >= 'A' && i <= 'F' ? i - 'A' + 10	1333	: i >= 'A' && i <= 'F' ? i - 'A' + 10
1150	: -1;	1334	: -1;
1151		1335
1152	json_stash = gv_stashpv ("JSON::XS", 1);	1336	json_stash = gv_stashpv ("JSON::XS" , 1);
		1337	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
		1338
		1339	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
		1340	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1153	}	1341	}
1154		1342
1155	PROTOTYPES: DISABLE	1343	PROTOTYPES: DISABLE
1156		1344
1157	SV new (char dummy)	1345	void new (char *klass)
1158	CODE:	1346	PPCODE:
1159	RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);	1347	{
1160	OUTPUT:	1348	SV *pv = NEWSV (0, sizeof (JSON));
1161	RETVAL	1349	SvPOK_only (pv);
		1350	Zero (SvPVX (pv), 1, JSON);
		1351	((JSON *)SvPVX (pv))->flags = F_DEFAULT;
		1352	XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
		1353	}
1162		1354
1163	SV ascii (SV self, int enable = 1)	1355	void ascii (JSON *self, int enable = 1)
1164	ALIAS:	1356	ALIAS:
1165	ascii = F_ASCII	1357	ascii = F_ASCII
1166	latin1 = F_LATIN1	1358	latin1 = F_LATIN1
1167	utf8 = F_UTF8	1359	utf8 = F_UTF8
1168	indent = F_INDENT	1360	indent = F_INDENT
1169	canonical = F_CANONICAL	1361	canonical = F_CANONICAL
1170	space_before = F_SPACE_BEFORE	1362	space_before = F_SPACE_BEFORE
1171	space_after = F_SPACE_AFTER	1363	space_after = F_SPACE_AFTER
1172	pretty = F_PRETTY	1364	pretty = F_PRETTY
1173	allow_nonref = F_ALLOW_NONREF	1365	allow_nonref = F_ALLOW_NONREF
1174	shrink = F_SHRINK	1366	shrink = F_SHRINK
		1367	allow_blessed = F_ALLOW_BLESSED
		1368	convert_blessed = F_CONV_BLESSED
1175	CODE:	1369	PPCODE:
1176	{	1370	{
1177	UV *uv = SvJSON (self);
1178	if (enable)	1371	if (enable)
1179	*uv \|= ix;	1372	self->flags \|= ix;
1180	else	1373	else
1181	*uv &= ~ix;	1374	self->flags &= ~ix;
1182		1375
1183	RETVAL = newSVsv (self);	1376	XPUSHs (ST (0));
1184	}	1377	}
1185	OUTPUT:
1186	RETVAL
1187		1378
1188	SV max_depth (SV self, UV max_depth = 0x80000000UL)	1379	void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1189	CODE:	1380	PPCODE:
1190	{	1381	{
1191	UV *uv = SvJSON (self);
1192	UV log2 = 0;	1382	UV log2 = 0;
1193		1383
1194	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;	1384	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1195		1385
1196	while ((1UL << log2) < max_depth)	1386	while ((1UL << log2) < max_depth)
1197	++log2;	1387	++log2;
1198		1388
1199	uv = uv & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);	1389	self->flags = self->flags & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
1200		1390
1201	RETVAL = newSVsv (self);	1391	XPUSHs (ST (0));
1202	}	1392	}
1203	OUTPUT:
1204	RETVAL
1205		1393
1206	void encode (SV self, SV scalar)	1394	void max_size (JSON *self, UV max_size = 0)
1207	PPCODE:	1395	PPCODE:
1208	XPUSHs (encode_json (scalar, *SvJSON (self)));	1396	{
		1397	UV log2 = 0;
1209		1398
1210	void decode (SV self, SV jsonstr)	1399	if (max_size > 0x80000000UL) max_size = 0x80000000UL;
		1400	if (max_size == 1) max_size = 2;
		1401
		1402	while ((1UL << log2) < max_size)
		1403	++log2;
		1404
		1405	self->flags = self->flags & ~F_MAXSIZE \| (log2 << S_MAXSIZE);
		1406
		1407	XPUSHs (ST (0));
		1408	}
		1409
		1410	void filter_json_object (JSON self, SV cb = &PL_sv_undef)
1211	PPCODE:	1411	PPCODE:
		1412	{
		1413	SvREFCNT_dec (self->cb_object);
		1414	self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
		1415
		1416	XPUSHs (ST (0));
		1417	}
		1418
		1419	void filter_json_single_key_object (JSON self, SV key, SV *cb = &PL_sv_undef)
		1420	PPCODE:
		1421	{
		1422	if (!self->cb_sk_object)
		1423	self->cb_sk_object = newHV ();
		1424
		1425	if (SvOK (cb))
		1426	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
		1427	else
		1428	{
		1429	hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
		1430
		1431	if (!HvKEYS (self->cb_sk_object))
		1432	{
		1433	SvREFCNT_dec (self->cb_sk_object);
		1434	self->cb_sk_object = 0;
		1435	}
		1436	}
		1437
		1438	XPUSHs (ST (0));
		1439	}
		1440
		1441	void encode (JSON self, SV scalar)
		1442	PPCODE:
		1443	XPUSHs (encode_json (scalar, self));
		1444
		1445	void decode (JSON self, SV jsonstr)
		1446	PPCODE:
1212	XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));	1447	XPUSHs (decode_json (jsonstr, self, 0));
1213		1448
1214	void decode_prefix (SV self, SV jsonstr)	1449	void decode_prefix (JSON self, SV jsonstr)
1215	PPCODE:	1450	PPCODE:
1216	{	1451	{
1217	UV offset;	1452	UV offset;
1218	EXTEND (SP, 2);	1453	EXTEND (SP, 2);
1219	PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));	1454	PUSHs (decode_json (jsonstr, self, &offset));
1220	PUSHs (sv_2mortal (newSVuv (offset)));	1455	PUSHs (sv_2mortal (newSVuv (offset)));
1221	}	1456	}
1222		1457
		1458	void DESTROY (JSON *self)
		1459	CODE:
		1460	SvREFCNT_dec (self->cb_sk_object);
		1461	SvREFCNT_dec (self->cb_object);
		1462
1223	PROTOTYPES: ENABLE	1463	PROTOTYPES: ENABLE
1224		1464
1225	void to_json (SV *scalar)	1465	void to_json (SV *scalar)
1226	ALIAS:
1227	objToJson = 0
1228	PPCODE:	1466	PPCODE:
		1467	{
		1468	JSON json = { F_DEFAULT \| F_UTF8 };
1229	XPUSHs (encode_json (scalar, F_DEFAULT \| F_UTF8));	1469	XPUSHs (encode_json (scalar, &json));
		1470	}
1230		1471
1231	void from_json (SV *jsonstr)	1472	void from_json (SV *jsonstr)
1232	ALIAS:
1233	jsonToObj = 0
1234	PPCODE:	1473	PPCODE:
		1474	{
		1475	JSON json = { F_DEFAULT \| F_UTF8 };
1235	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8, 0));	1476	XPUSHs (decode_json (jsonstr, &json, 0));
		1477	}
1236		1478

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.38 by root, Mon Jun 11 03:18:07 2007 UTC vs. Revision 1.54 by root, Tue Jul 10 16:22:31 2007 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.38 by root, Mon Jun 11 03:18:07 2007 UTC vs.
Revision 1.54 by root, Tue Jul 10 16:22:31 2007 UTC