[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 1.52 by root, Mon Jul 2 02:57:11 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", 1);
		495
		496	if (to_json)
		497	{
		498	dSP; ENTER; SAVETMPS; PUSHMARK (SP);
		499	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
		500
		501	// calling with G_SCALAR ensures that we always get a 1 reutrn value
		502	// check anyways.
		503	PUTBACK;
		504	assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
		505	SPAGAIN;
		506
		507	encode_sv (enc, POPs);
		508
		509	FREETMPS; LEAVE;
		510	}
		511	else if (enc->json.flags & F_ALLOW_BLESSED)
		512	encode_str (enc, "null", 4, 0);
		513	else
		514	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
		515	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		516	}
		517	else if (enc->json.flags & F_ALLOW_BLESSED)
		518	encode_str (enc, "null", 4, 0);
		519	else
		520	croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
		521	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		522	}
		523	}
468	if (svt == SVt_PVHV)	524	else if (svt == SVt_PVHV)
469	encode_hv (enc, (HV *)sv);	525	encode_hv (enc, (HV *)sv);
470	else if (svt == SVt_PVAV)	526	else if (svt == SVt_PVAV)
471	encode_av (enc, (AV *)sv);	527	encode_av (enc, (AV *)sv);
472	else if (svt < SVt_PVAV)	528	else if (svt < SVt_PVAV)
473	{	529	{
474	if (SvNIOK (sv) && SvIV (sv) == 0)	530	STRLEN len = 0;
		531	char *pv = svt ? SvPV (sv, len) : 0;
		532
		533	if (len == 1 && *pv == '1')
		534	encode_str (enc, "true", 4, 0);
		535	else if (len == 1 && *pv == '0')
475	encode_str (enc, "false", 5, 0);	536	encode_str (enc, "false", 5, 0);
476	else if (SvNIOK (sv) && SvIV (sv) == 1)
477	encode_str (enc, "true", 4, 0);
478	else	537	else
479	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",	538	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	539	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481	}	540	}
482	else	541	else
…		…
497	encode_str (enc, str, len, SvUTF8 (sv));	556	encode_str (enc, str, len, SvUTF8 (sv));
498	encode_ch (enc, '"');	557	encode_ch (enc, '"');
499	}	558	}
500	else if (SvNOKp (sv))	559	else if (SvNOKp (sv))
501	{	560	{
		561	// trust that perl will do the right thing w.r.t. JSON syntax.
502	need (enc, NV_DIG + 32);	562	need (enc, NV_DIG + 32);
503	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	563	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
504	enc->cur += strlen (enc->cur);	564	enc->cur += strlen (enc->cur);
505	}	565	}
506	else if (SvIOKp (sv))	566	else if (SvIOKp (sv))
507	{	567	{
508	// we assume we can always read an IV as a UV	568	// we assume we can always read an IV as a UV
509	if (SvUV (sv) & ~(UV)0x7fff)	569	if (SvUV (sv) & ~(UV)0x7fff)
510	{	570	{
		571	// large integer, use the (rather slow) snprintf way.
511	need (enc, sizeof (UV) * 3);	572	need (enc, sizeof (UV) * 3);
512	enc->cur +=	573	enc->cur +=
513	SvIsUV(sv)	574	SvIsUV(sv)
514	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))	575	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
515	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));	576	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
…		…
518	{	579	{
519	// optimise the "small number case"	580	// optimise the "small number case"
520	// code will likely be branchless and use only a single multiplication	581	// code will likely be branchless and use only a single multiplication
521	I32 i = SvIV (sv);	582	I32 i = SvIV (sv);
522	U32 u;	583	U32 u;
		584	char digit, nz = 0;
523		585
524	need (enc, 6);	586	need (enc, 6);
525		587
526	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;	588	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
527	u = i < 0 ? -i : i;	589	u = i < 0 ? -i : i;
528		590
529	// convert to 4.28 fixed-point representation	591	// convert to 4.28 fixed-point representation
530	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits	592	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
531		593
532	char digit, nz = 0;	594	// now output digit by digit, each time masking out the integer part
533		595	// and multiplying by 5 while moving the decimal point one to the right,
		596	// resulting in a net multiplication by 10.
		597	// we always write the digit to memory but conditionally increment
		598	// 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;	599	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;	600	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;	601	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;	602	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;	603	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
539	}	604	}
540	}	605	}
541	else if (SvROK (sv))	606	else if (SvROK (sv))
542	encode_rv (enc, SvRV (sv));	607	encode_rv (enc, SvRV (sv));
543	else if (!SvOK (sv))	608	else if (!SvOK (sv))
…		…
546	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	611	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
547	SvPV_nolen (sv), SvFLAGS (sv));	612	SvPV_nolen (sv), SvFLAGS (sv));
548	}	613	}
549		614
550	static SV *	615	static SV *
551	encode_json (SV *scalar, U32 flags)	616	encode_json (SV scalar, JSON json)
552	{	617	{
553	enc_t enc;	618	enc_t enc;
554		619
555	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	620	if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
556	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");	621	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
557		622
558	enc.flags = flags;	623	enc.json = *json;
559	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	624	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
560	enc.cur = SvPVX (enc.sv);	625	enc.cur = SvPVX (enc.sv);
561	enc.end = SvEND (enc.sv);	626	enc.end = SvEND (enc.sv);
562	enc.indent = 0;	627	enc.indent = 0;
563	enc.maxdepth = DEC_DEPTH (flags);	628	enc.maxdepth = DEC_DEPTH (enc.json.flags);
564		629
565	SvPOK_only (enc.sv);	630	SvPOK_only (enc.sv);
566	encode_sv (&enc, scalar);	631	encode_sv (&enc, scalar);
567		632
568	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	633	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
569	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	634	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
570		635
571	if (!(flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))	636	if (!(enc.json.flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
572	SvUTF8_on (enc.sv);	637	SvUTF8_on (enc.sv);
573		638
574	if (enc.flags & F_SHRINK)	639	if (enc.json.flags & F_SHRINK)
575	shrink (enc.sv);	640	shrink (enc.sv);
576		641
577	return enc.sv;	642	return enc.sv;
578	}	643	}
579		644
…		…
584	typedef struct	649	typedef struct
585	{	650	{
586	char *cur; // current parser pointer	651	char *cur; // current parser pointer
587	char *end; // end of input string	652	char *end; // end of input string
588	const char *err; // parse error, if != 0	653	const char *err; // parse error, if != 0
589	U32 flags; // F_*	654	JSON json;
590	U32 depth; // recursion depth	655	U32 depth; // recursion depth
591	U32 maxdepth; // recursion depth limit	656	U32 maxdepth; // recursion depth limit
592	} dec_t;	657	} dec_t;
593		658
594	inline void	659	inline void
…		…
625	decode_4hex (dec_t *dec)	690	decode_4hex (dec_t *dec)
626	{	691	{
627	signed char d1, d2, d3, d4;	692	signed char d1, d2, d3, d4;
628	unsigned char cur = (unsigned char )dec->cur;	693	unsigned char cur = (unsigned char )dec->cur;
629		694
630	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected");	695	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");	696	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");	697	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");	698	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
634		699
635	dec->cur += 4;	700	dec->cur += 4;
636		701
637	return ((UV)d1) << 12	702	return ((UV)d1) << 12
638	\| ((UV)d2) << 8	703	\| ((UV)d2) << 8
…		…
861	{	926	{
862	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so	927	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
863	if (*start == '-')	928	if (*start == '-')
864	switch (dec->cur - start)	929	switch (dec->cur - start)
865	{	930	{
866	case 2: return newSViv (-( start [1] - '0' ));	931	case 2: return newSViv (-( start [1] - '0' * 1));
867	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));	932	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));	933	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));	934	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
870	}	935	}
871	else	936	else
872	switch (dec->cur - start)	937	switch (dec->cur - start)
873	{	938	{
874	case 1: return newSViv ( start [0] - '0' );	939	case 1: return newSViv ( start [0] - '0' * 1);
875	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);	940	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);	941	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);	942	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
878	}	943	}
879		944
…		…
886	if (uv < (UV)IV_MIN)	951	if (uv < (UV)IV_MIN)
887	return newSViv (-(IV)uv);	952	return newSViv (-(IV)uv);
888	}	953	}
889	else	954	else
890	return newSVuv (uv);	955	return newSVuv (uv);
		956
		957	// here would likely be the place for bigint support
891	}	958	}
892	}	959	}
893		960
		961	// if we ever support bigint or bigfloat, this is the place for bigfloat
894	return newSVnv (Atof (start));	962	return newSVnv (Atof (start));
895		963
896	fail:	964	fail:
897	return 0;	965	return 0;
898	}	966	}
…		…
942	}	1010	}
943		1011
944	static SV *	1012	static SV *
945	decode_hv (dec_t *dec)	1013	decode_hv (dec_t *dec)
946	{	1014	{
		1015	SV *sv;
947	HV *hv = newHV ();	1016	HV *hv = newHV ();
948		1017
949	DEC_INC_DEPTH;	1018	DEC_INC_DEPTH;
950	decode_ws (dec);	1019	decode_ws (dec);
951		1020
952	if (*dec->cur == '}')	1021	if (*dec->cur == '}')
953	++dec->cur;	1022	++dec->cur;
954	else	1023	else
955	for (;;)	1024	for (;;)
956	{	1025	{
957	SV key, value;
958
959	decode_ws (dec); EXPECT_CH ('"');	1026	decode_ws (dec); EXPECT_CH ('"');
960		1027
961	key = decode_str (dec);	1028	// heuristic: assume that
962	if (!key)	1029	// a) decode_str + hv_store_ent are abysmally slow.
963	goto fail;	1030	// b) most hash keys are short, simple ascii text.
		1031	// => try to "fast-match" such strings to avoid
		1032	// the overhead of decode_str + hv_store_ent.
		1033	{
		1034	SV *value;
		1035	char *p = dec->cur;
		1036	char *e = p + 24; // only try up to 24 bytes
964		1037
965	decode_ws (dec); EXPECT_CH (':');	1038	for (;;)
966
967	value = decode_sv (dec);
968	if (!value)
969	{	1039	{
		1040	// the >= 0x80 is true on most architectures
		1041	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
		1042	{
		1043	// slow path, back up and use decode_str
		1044	SV *key = decode_str (dec);
		1045	if (!key)
		1046	goto fail;
		1047
		1048	decode_ws (dec); EXPECT_CH (':');
		1049
		1050	value = decode_sv (dec);
		1051	if (!value)
		1052	{
		1053	SvREFCNT_dec (key);
		1054	goto fail;
		1055	}
		1056
		1057	hv_store_ent (hv, key, value, 0);
970	SvREFCNT_dec (key);	1058	SvREFCNT_dec (key);
		1059
		1060	break;
		1061	}
		1062	else if (*p == '"')
		1063	{
		1064	// fast path, got a simple key
		1065	char *key = dec->cur;
		1066	int len = p - key;
		1067	dec->cur = p + 1;
		1068
		1069	decode_ws (dec); EXPECT_CH (':');
		1070
		1071	value = decode_sv (dec);
		1072	if (!value)
971	goto fail;	1073	goto fail;
		1074
		1075	hv_store (hv, key, len, value, 0);
		1076
		1077	break;
		1078	}
		1079
		1080	++p;
972	}	1081	}
973		1082	}
974	hv_store_ent (hv, key, value, 0);
975	SvREFCNT_dec (key);
976		1083
977	decode_ws (dec);	1084	decode_ws (dec);
978		1085
979	if (*dec->cur == '}')	1086	if (*dec->cur == '}')
980	{	1087	{
…		…
987		1094
988	++dec->cur;	1095	++dec->cur;
989	}	1096	}
990		1097
991	DEC_DEC_DEPTH;	1098	DEC_DEC_DEPTH;
992	return newRV_noinc ((SV *)hv);	1099	sv = newRV_noinc ((SV *)hv);
		1100
		1101	// check filter callbacks
		1102	if (dec->json.flags & F_HOOK)
		1103	{
		1104	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
		1105	{
		1106	HE cb, he;
		1107
		1108	hv_iterinit (hv);
		1109	he = hv_iternext (hv);
		1110	hv_iterinit (hv);
		1111
		1112	// the next line creates a mortal sv each time its called.
		1113	// might want to optimise this for common cases.
		1114	cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
		1115
		1116	if (cb)
		1117	{
		1118	int count;
		1119	ENTER; SAVETMPS;
		1120
		1121	dSP; PUSHMARK (SP);
		1122	XPUSHs (HeVAL (he));
		1123
		1124	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
		1125
		1126	if (count == 1)
		1127	{
		1128	sv = newSVsv (POPs);
		1129	FREETMPS; LEAVE;
		1130	return sv;
		1131	}
		1132
		1133	FREETMPS; LEAVE;
		1134	}
		1135	}
		1136
		1137	if (dec->json.cb_object)
		1138	{
		1139	int count;
		1140	ENTER; SAVETMPS;
		1141
		1142	dSP; ENTER; SAVETMPS; PUSHMARK (SP);
		1143	XPUSHs (sv_2mortal (sv));
		1144
		1145	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
		1146
		1147	if (count == 1)
		1148	{
		1149	sv = newSVsv (POPs);
		1150	FREETMPS; LEAVE;
		1151	return sv;
		1152	}
		1153
		1154	SvREFCNT_inc (sv);
		1155	FREETMPS; LEAVE;
		1156	}
		1157	}
		1158
		1159	return sv;
993		1160
994	fail:	1161	fail:
995	SvREFCNT_dec (hv);	1162	SvREFCNT_dec (hv);
996	DEC_DEC_DEPTH;	1163	DEC_DEC_DEPTH;
997	return 0;	1164	return 0;
…		…
999		1166
1000	static SV *	1167	static SV *
1001	decode_sv (dec_t *dec)	1168	decode_sv (dec_t *dec)
1002	{	1169	{
1003	decode_ws (dec);	1170	decode_ws (dec);
		1171
		1172	// the beauty of JSON: you need exactly one character lookahead
		1173	// to parse anything.
1004	switch (*dec->cur)	1174	switch (*dec->cur)
1005	{	1175	{
1006	case '"': ++dec->cur; return decode_str (dec);	1176	case '"': ++dec->cur; return decode_str (dec);
1007	case '[': ++dec->cur; return decode_av (dec);	1177	case '[': ++dec->cur; return decode_av (dec);
1008	case '{': ++dec->cur; return decode_hv (dec);	1178	case '{': ++dec->cur; return decode_hv (dec);
…		…
1014		1184
1015	case 't':	1185	case 't':
1016	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1186	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1017	{	1187	{
1018	dec->cur += 4;	1188	dec->cur += 4;
1019	return newSViv (1);	1189	return SvREFCNT_inc (json_true);
1020	}	1190	}
1021	else	1191	else
1022	ERR ("'true' expected");	1192	ERR ("'true' expected");
1023		1193
1024	break;	1194	break;
1025		1195
1026	case 'f':	1196	case 'f':
1027	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1197	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1028	{	1198	{
1029	dec->cur += 5;	1199	dec->cur += 5;
1030	return newSViv (0);	1200	return SvREFCNT_inc (json_false);
1031	}	1201	}
1032	else	1202	else
1033	ERR ("'false' expected");	1203	ERR ("'false' expected");
1034		1204
1035	break;	1205	break;
…		…
1053	fail:	1223	fail:
1054	return 0;	1224	return 0;
1055	}	1225	}
1056		1226
1057	static SV *	1227	static SV *
1058	decode_json (SV string, U32 flags, UV offset_return)	1228	decode_json (SV string, JSON json, UV *offset_return)
1059	{	1229	{
1060	dec_t dec;	1230	dec_t dec;
1061	UV offset;	1231	UV offset;
1062	SV *sv;	1232	SV *sv;
1063		1233
1064	SvGETMAGIC (string);	1234	SvGETMAGIC (string);
1065	SvUPGRADE (string, SVt_PV);	1235	SvUPGRADE (string, SVt_PV);
1066		1236
		1237	if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
		1238	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1239	(unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
		1240
1067	if (flags & F_UTF8)	1241	if (json->flags & F_UTF8)
1068	sv_utf8_downgrade (string, 0);	1242	sv_utf8_downgrade (string, 0);
1069	else	1243	else
1070	sv_utf8_upgrade (string);	1244	sv_utf8_upgrade (string);
1071		1245
1072	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1246	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1073		1247
1074	dec.flags = flags;	1248	dec.json = *json;
1075	dec.cur = SvPVX (string);	1249	dec.cur = SvPVX (string);
1076	dec.end = SvEND (string);	1250	dec.end = SvEND (string);
1077	dec.err = 0;	1251	dec.err = 0;
1078	dec.depth = 0;	1252	dec.depth = 0;
1079	dec.maxdepth = DEC_DEPTH (dec.flags);	1253	dec.maxdepth = DEC_DEPTH (dec.json.flags);
		1254
		1255	if (dec.json.cb_object \|\| dec.json.cb_sk_object)
		1256	dec.json.flags \|= F_HOOK;
1080		1257
1081	*dec.end = 0; // this should basically be a nop, too, but make sure it's there	1258	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
1082	sv = decode_sv (&dec);	1259	sv = decode_sv (&dec);
1083		1260
1084	if (!(offset_return \|\| !sv))	1261	if (!(offset_return \|\| !sv))
…		…
1094	}	1271	}
1095	}	1272	}
1096		1273
1097	if (offset_return \|\| !sv)	1274	if (offset_return \|\| !sv)
1098	{	1275	{
1099	offset = dec.flags & F_UTF8	1276	offset = dec.json.flags & F_UTF8
1100	? dec.cur - SvPVX (string)	1277	? dec.cur - SvPVX (string)
1101	: utf8_distance (dec.cur, SvPVX (string));	1278	: utf8_distance (dec.cur, SvPVX (string));
1102		1279
1103	if (offset_return)	1280	if (offset_return)
1104	*offset_return = offset;	1281	*offset_return = offset;
…		…
1123	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1300	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1124	}	1301	}
1125		1302
1126	sv = sv_2mortal (sv);	1303	sv = sv_2mortal (sv);
1127		1304
1128	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))	1305	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)");	1306	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1130		1307
1131	return sv;	1308	return sv;
1132	}	1309	}
1133		1310
…		…
1137	MODULE = JSON::XS PACKAGE = JSON::XS	1314	MODULE = JSON::XS PACKAGE = JSON::XS
1138		1315
1139	BOOT:	1316	BOOT:
1140	{	1317	{
1141	int i;	1318	int i;
1142
1143	memset (decode_hexdigit, 0xff, 256);
1144		1319
1145	for (i = 0; i < 256; ++i)	1320	for (i = 0; i < 256; ++i)
1146	decode_hexdigit [i] =	1321	decode_hexdigit [i] =
1147	i >= '0' && i <= '9' ? i - '0'	1322	i >= '0' && i <= '9' ? i - '0'
1148	: i >= 'a' && i <= 'f' ? i - 'a' + 10	1323	: i >= 'a' && i <= 'f' ? i - 'a' + 10
1149	: i >= 'A' && i <= 'F' ? i - 'A' + 10	1324	: i >= 'A' && i <= 'F' ? i - 'A' + 10
1150	: -1;	1325	: -1;
1151		1326
1152	json_stash = gv_stashpv ("JSON::XS", 1);	1327	json_stash = gv_stashpv ("JSON::XS" , 1);
		1328	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
		1329
		1330	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
		1331	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1153	}	1332	}
1154		1333
1155	PROTOTYPES: DISABLE	1334	PROTOTYPES: DISABLE
1156		1335
1157	SV new (char dummy)	1336	void new (char *klass)
1158	CODE:	1337	PPCODE:
1159	RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);	1338	{
1160	OUTPUT:	1339	SV *pv = NEWSV (0, sizeof (JSON));
1161	RETVAL	1340	SvPOK_only (pv);
		1341	Zero (SvPVX (pv), 1, JSON);
		1342	((JSON *)SvPVX (pv))->flags = F_DEFAULT;
		1343	XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
		1344	}
1162		1345
1163	SV ascii (SV self, int enable = 1)	1346	void ascii (JSON *self, int enable = 1)
1164	ALIAS:	1347	ALIAS:
1165	ascii = F_ASCII	1348	ascii = F_ASCII
1166	latin1 = F_LATIN1	1349	latin1 = F_LATIN1
1167	utf8 = F_UTF8	1350	utf8 = F_UTF8
1168	indent = F_INDENT	1351	indent = F_INDENT
1169	canonical = F_CANONICAL	1352	canonical = F_CANONICAL
1170	space_before = F_SPACE_BEFORE	1353	space_before = F_SPACE_BEFORE
1171	space_after = F_SPACE_AFTER	1354	space_after = F_SPACE_AFTER
1172	pretty = F_PRETTY	1355	pretty = F_PRETTY
1173	allow_nonref = F_ALLOW_NONREF	1356	allow_nonref = F_ALLOW_NONREF
1174	shrink = F_SHRINK	1357	shrink = F_SHRINK
		1358	allow_blessed = F_ALLOW_BLESSED
		1359	convert_blessed = F_CONV_BLESSED
1175	CODE:	1360	PPCODE:
1176	{	1361	{
1177	UV *uv = SvJSON (self);
1178	if (enable)	1362	if (enable)
1179	*uv \|= ix;	1363	self->flags \|= ix;
1180	else	1364	else
1181	*uv &= ~ix;	1365	self->flags &= ~ix;
1182		1366
1183	RETVAL = newSVsv (self);	1367	XPUSHs (ST (0));
1184	}	1368	}
1185	OUTPUT:
1186	RETVAL
1187		1369
1188	SV max_depth (SV self, UV max_depth = 0x80000000UL)	1370	void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1189	CODE:	1371	PPCODE:
1190	{	1372	{
1191	UV *uv = SvJSON (self);
1192	UV log2 = 0;	1373	UV log2 = 0;
1193		1374
1194	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;	1375	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1195		1376
1196	while ((1UL << log2) < max_depth)	1377	while ((1UL << log2) < max_depth)
1197	++log2;	1378	++log2;
1198		1379
1199	uv = uv & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);	1380	self->flags = self->flags & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
1200		1381
1201	RETVAL = newSVsv (self);	1382	XPUSHs (ST (0));
1202	}	1383	}
1203	OUTPUT:
1204	RETVAL
1205		1384
1206	void encode (SV self, SV scalar)	1385	void max_size (JSON *self, UV max_size = 0)
1207	PPCODE:	1386	PPCODE:
1208	XPUSHs (encode_json (scalar, *SvJSON (self)));	1387	{
		1388	UV log2 = 0;
1209		1389
1210	void decode (SV self, SV jsonstr)	1390	if (max_size > 0x80000000UL) max_size = 0x80000000UL;
		1391	if (max_size == 1) max_size = 2;
		1392
		1393	while ((1UL << log2) < max_size)
		1394	++log2;
		1395
		1396	self->flags = self->flags & ~F_MAXSIZE \| (log2 << S_MAXSIZE);
		1397
		1398	XPUSHs (ST (0));
		1399	}
		1400
		1401	void filter_json_object (JSON self, SV cb = &PL_sv_undef)
1211	PPCODE:	1402	PPCODE:
		1403	{
		1404	SvREFCNT_dec (self->cb_object);
		1405	self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
		1406
		1407	XPUSHs (ST (0));
		1408	}
		1409
		1410	void filter_json_single_key_object (JSON self, SV key, SV *cb = &PL_sv_undef)
		1411	PPCODE:
		1412	{
		1413	if (!self->cb_sk_object)
		1414	self->cb_sk_object = newHV ();
		1415
		1416	if (SvOK (cb))
		1417	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
		1418	else
		1419	{
		1420	hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
		1421
		1422	if (!HvKEYS (self->cb_sk_object))
		1423	{
		1424	SvREFCNT_dec (self->cb_sk_object);
		1425	self->cb_sk_object = 0;
		1426	}
		1427	}
		1428
		1429	XPUSHs (ST (0));
		1430	}
		1431
		1432	void encode (JSON self, SV scalar)
		1433	PPCODE:
		1434	XPUSHs (encode_json (scalar, self));
		1435
		1436	void decode (JSON self, SV jsonstr)
		1437	PPCODE:
1212	XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));	1438	XPUSHs (decode_json (jsonstr, self, 0));
1213		1439
1214	void decode_prefix (SV self, SV jsonstr)	1440	void decode_prefix (JSON self, SV jsonstr)
1215	PPCODE:	1441	PPCODE:
1216	{	1442	{
1217	UV offset;	1443	UV offset;
1218	EXTEND (SP, 2);	1444	EXTEND (SP, 2);
1219	PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));	1445	PUSHs (decode_json (jsonstr, self, &offset));
1220	PUSHs (sv_2mortal (newSVuv (offset)));	1446	PUSHs (sv_2mortal (newSVuv (offset)));
1221	}	1447	}
1222		1448
		1449	void DESTROY (JSON *self)
		1450	CODE:
		1451	SvREFCNT_dec (self->cb_sk_object);
		1452	SvREFCNT_dec (self->cb_object);
		1453
1223	PROTOTYPES: ENABLE	1454	PROTOTYPES: ENABLE
1224		1455
1225	void to_json (SV *scalar)	1456	void to_json (SV *scalar)
1226	ALIAS:
1227	objToJson = 0
1228	PPCODE:	1457	PPCODE:
		1458	{
		1459	JSON json = { F_DEFAULT \| F_UTF8 };
1229	XPUSHs (encode_json (scalar, F_DEFAULT \| F_UTF8));	1460	XPUSHs (encode_json (scalar, &json));
		1461	}
1230		1462
1231	void from_json (SV *jsonstr)	1463	void from_json (SV *jsonstr)
1232	ALIAS:
1233	jsonToObj = 0
1234	PPCODE:	1464	PPCODE:
		1465	{
		1466	JSON json = { F_DEFAULT \| F_UTF8 };
1235	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8, 0));	1467	XPUSHs (decode_json (jsonstr, &json, 0));
		1468	}
1236		1469

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 1.52 by root, Mon Jul 2 02:57:11 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 1.52 by root, Mon Jul 2 02:57:11 2007 UTC