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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.37 by root, Wed Jun 6 17:49:01 2007 UTC vs.
Revision 1.47 by root, Sun Jul 1 14:08:03 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 // NYI
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
25		35
26	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))	36	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27		37	#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
28	// F_SELFCONVERT? <=> to_json/toJson
29	// F_BLESSED? <=> { $__class__$ => }
30		38
31	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	39	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
32	#define F_DEFAULT (9UL << S_MAXDEPTH)	40	#define F_DEFAULT (9UL << S_MAXDEPTH)
33		41
34	#define INIT_SIZE 32 // initial scalar size to be allocated	42	#define INIT_SIZE 32 // initial scalar size to be allocated
35	#define INDENT_STEP 3 // spaces per indentation level	43	#define INDENT_STEP 3 // spaces per indentation level
36		44
37	#define SHORT_STRING_LEN 512 // special-case strings of up to this size	45	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
38		46
39	#define SB do {	47	#define SB do {
40	#define SE } while (0)	48	#define SE } while (0)
41		49
42	#if __GNUC__ >= 3	50	#if __GNUC__ >= 3
…		…
48	#endif	56	#endif
49		57
50	#define expect_false(expr) expect ((expr) != 0, 0)	58	#define expect_false(expr) expect ((expr) != 0, 0)
51	#define expect_true(expr) expect ((expr) != 0, 1)	59	#define expect_true(expr) expect ((expr) != 0, 1)
52		60
53	static HV *json_stash; // JSON::XS::	61	static HV json_stash, json_boolean_stash; // JSON::XS::
		62	static SV json_true, json_false;
		63
		64	typedef struct json {
		65	U32 flags;
		66	} JSON__XS;
54		67
55	/////////////////////////////////////////////////////////////////////////////	68	/////////////////////////////////////////////////////////////////////////////
56	// utility functions	69	// utility functions
57		70
58	static UV *	71	static UV *
…		…
107	typedef struct	120	typedef struct
108	{	121	{
109	char *cur; // SvPVX (sv) + current output position	122	char *cur; // SvPVX (sv) + current output position
110	char *end; // SvEND (sv)	123	char *end; // SvEND (sv)
111	SV *sv; // result scalar	124	SV *sv; // result scalar
112	U32 flags; // F_*	125	struct json json;
113	U32 indent; // indentation level	126	U32 indent; // indentation level
114	U32 maxdepth; // max. indentation/recursion level	127	U32 maxdepth; // max. indentation/recursion level
115	} enc_t;	128	} enc_t;
116		129
117	inline void	130	inline void
…		…
178	STRLEN clen;	191	STRLEN clen;
179	UV uch;	192	UV uch;
180		193
181	if (is_utf8)	194	if (is_utf8)
182	{	195	{
183	//uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184	uch = decode_utf8 (str, end - str, &clen);	196	uch = decode_utf8 (str, end - str, &clen);
185	if (clen == (STRLEN)-1)	197	if (clen == (STRLEN)-1)
186	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);	198	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187	}	199	}
188	else	200	else
…		…
192	}	204	}
193		205
194	if (uch > 0x10FFFFUL)	206	if (uch > 0x10FFFFUL)
195	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);	207	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196		208
197	if (uch < 0x80 \|\| enc->flags & F_ASCII \|\| (enc->flags & F_LATIN1 && uch > 0xFF))	209	if (uch < 0x80 \|\| enc->json.flags & F_ASCII \|\| (enc->json.flags & F_LATIN1 && uch > 0xFF))
198	{	210	{
199	if (uch > 0xFFFFUL)	211	if (uch > 0xFFFFUL)
200	{	212	{
201	need (enc, len += 11);	213	need (enc, len += 11);
202	sprintf (enc->cur, "\\u%04x\\u%04x",	214	sprintf (enc->cur, "\\u%04x\\u%04x",
…		…
216	*enc->cur++ = hexdigit [(uch >> 0) & 15];	228	*enc->cur++ = hexdigit [(uch >> 0) & 15];
217	}	229	}
218		230
219	str += clen;	231	str += clen;
220	}	232	}
221	else if (enc->flags & F_LATIN1)	233	else if (enc->json.flags & F_LATIN1)
222	{	234	{
223	*enc->cur++ = uch;	235	*enc->cur++ = uch;
224	str += clen;	236	str += clen;
225	}	237	}
226	else if (is_utf8)	238	else if (is_utf8)
…		…
247	}	259	}
248		260
249	inline void	261	inline void
250	encode_indent (enc_t *enc)	262	encode_indent (enc_t *enc)
251	{	263	{
252	if (enc->flags & F_INDENT)	264	if (enc->json.flags & F_INDENT)
253	{	265	{
254	int spaces = enc->indent * INDENT_STEP;	266	int spaces = enc->indent * INDENT_STEP;
255		267
256	need (enc, spaces);	268	need (enc, spaces);
257	memset (enc->cur, ' ', spaces);	269	memset (enc->cur, ' ', spaces);
…		…
267	}	279	}
268		280
269	inline void	281	inline void
270	encode_nl (enc_t *enc)	282	encode_nl (enc_t *enc)
271	{	283	{
272	if (enc->flags & F_INDENT)	284	if (enc->json.flags & F_INDENT)
273	{	285	{
274	need (enc, 1);	286	need (enc, 1);
275	encode_ch (enc, '\n');	287	encode_ch (enc, '\n');
276	}	288	}
277	}	289	}
…		…
279	inline void	291	inline void
280	encode_comma (enc_t *enc)	292	encode_comma (enc_t *enc)
281	{	293	{
282	encode_ch (enc, ',');	294	encode_ch (enc, ',');
283		295
284	if (enc->flags & F_INDENT)	296	if (enc->json.flags & F_INDENT)
285	encode_nl (enc);	297	encode_nl (enc);
286	else if (enc->flags & F_SPACE_AFTER)	298	else if (enc->json.flags & F_SPACE_AFTER)
287	encode_space (enc);	299	encode_space (enc);
288	}	300	}
289		301
290	static void encode_sv (enc_t enc, SV sv);	302	static void encode_sv (enc_t enc, SV sv);
291		303
…		…
334	else	346	else
335	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));	347	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336		348
337	encode_ch (enc, '"');	349	encode_ch (enc, '"');
338		350
339	if (enc->flags & F_SPACE_BEFORE) encode_space (enc);	351	if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340	encode_ch (enc, ':');	352	encode_ch (enc, ':');
341	if (enc->flags & F_SPACE_AFTER ) encode_space (enc);	353	if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342	encode_sv (enc, HeVAL (he));	354	encode_sv (enc, HeVAL (he));
343	}	355	}
344		356
345	// compare hash entries, used when all keys are bytestrings	357	// compare hash entries, used when all keys are bytestrings
346	static int	358	static int
…		…
381	{	393	{
382	// for canonical output we have to sort by keys first	394	// for canonical output we have to sort by keys first
383	// actually, this is mostly due to the stupid so-called	395	// actually, this is mostly due to the stupid so-called
384	// security workaround added somewhere in 5.8.x.	396	// security workaround added somewhere in 5.8.x.
385	// that randomises hash orderings	397	// that randomises hash orderings
386	if (enc->flags & F_CANONICAL)	398	if (enc->json.flags & F_CANONICAL)
387	{	399	{
388	int fast = 1;	400	int fast = 1;
389	HE *he;	401	HE *he;
390	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	402	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
391	HE *hes = _alloca (count sizeof (HE));	403	HE *hes = _alloca (count sizeof (HE));
…		…
463	svtype svt;	475	svtype svt;
464		476
465	SvGETMAGIC (sv);	477	SvGETMAGIC (sv);
466	svt = SvTYPE (sv);	478	svt = SvTYPE (sv);
467		479
		480	if (expect_false (SvOBJECT (sv)))
		481	{
		482	if (SvSTASH (sv) == json_boolean_stash)
		483	{
		484	if (SvIV (sv) == 0)
		485	encode_str (enc, "false", 5, 0);
		486	else
		487	encode_str (enc, "true", 4, 0);
		488	}
		489	else
		490	{
		491	#if 0
		492	if (0 && sv_derived_from (rv, "JSON::Literal"))
		493	{
		494	// not yet
		495	}
		496	#endif
		497	if (enc->json.flags & F_CONV_BLESSED)
		498	{
		499	// we re-bless the reference to get overload and other niceties right
		500	GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
		501
		502	if (to_json)
		503	{
		504	dSP;
		505	ENTER;
		506	SAVETMPS;
		507	PUSHMARK (SP);
		508	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
		509
		510	// calling with G_SCALAR ensures that we always get a 1 reutrn value
		511	// check anyways.
		512	PUTBACK;
		513	assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
		514	SPAGAIN;
		515
		516	encode_sv (enc, POPs);
		517
		518	FREETMPS;
		519	LEAVE;
		520	}
		521	else if (enc->json.flags & F_ALLOW_BLESSED)
		522	encode_str (enc, "null", 4, 0);
		523	else
		524	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
		525	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		526	}
		527	else if (enc->json.flags & F_ALLOW_BLESSED)
		528	encode_str (enc, "null", 4, 0);
		529	else
		530	croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
		531	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		532	}
		533	}
468	if (svt == SVt_PVHV)	534	else if (svt == SVt_PVHV)
469	encode_hv (enc, (HV *)sv);	535	encode_hv (enc, (HV *)sv);
470	else if (svt == SVt_PVAV)	536	else if (svt == SVt_PVAV)
471	encode_av (enc, (AV *)sv);	537	encode_av (enc, (AV *)sv);
472	else if (svt < SVt_PVAV)	538	else if (svt < SVt_PVAV)
473	{	539	{
…		…
497	encode_str (enc, str, len, SvUTF8 (sv));	563	encode_str (enc, str, len, SvUTF8 (sv));
498	encode_ch (enc, '"');	564	encode_ch (enc, '"');
499	}	565	}
500	else if (SvNOKp (sv))	566	else if (SvNOKp (sv))
501	{	567	{
		568	// trust that perl will do the right thing w.r.t. JSON syntax.
502	need (enc, NV_DIG + 32);	569	need (enc, NV_DIG + 32);
503	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	570	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
504	enc->cur += strlen (enc->cur);	571	enc->cur += strlen (enc->cur);
505	}	572	}
506	else if (SvIOKp (sv))	573	else if (SvIOKp (sv))
507	{	574	{
508	// we assume we can always read an IV as a UV	575	// we assume we can always read an IV as a UV
509	if (SvUV (sv) & ~(UV)0x7fff)	576	if (SvUV (sv) & ~(UV)0x7fff)
510	{	577	{
		578	// large integer, use the (rather slow) snprintf way.
511	need (enc, sizeof (UV) * 3);	579	need (enc, sizeof (UV) * 3);
512	enc->cur +=	580	enc->cur +=
513	SvIsUV(sv)	581	SvIsUV(sv)
514	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))	582	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
515	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));	583	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
…		…
518	{	586	{
519	// optimise the "small number case"	587	// optimise the "small number case"
520	// code will likely be branchless and use only a single multiplication	588	// code will likely be branchless and use only a single multiplication
521	I32 i = SvIV (sv);	589	I32 i = SvIV (sv);
522	U32 u;	590	U32 u;
		591	char digit, nz = 0;
523		592
524	need (enc, 6);	593	need (enc, 6);
525		594
526	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;	595	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
527	u = i < 0 ? -i : i;	596	u = i < 0 ? -i : i;
528		597
529	// convert to 4.28 fixed-point representation	598	// convert to 4.28 fixed-point representation
530	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits	599	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
531		600
532	char digit, nz = 0;	601	// now output digit by digit, each time masking out the integer part
533		602	// and multiplying by 5 while moving the decimal point one to the right,
		603	// resulting in a net multiplication by 10.
		604	// we always write the digit to memory but conditionally increment
		605	// 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;	606	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;	607	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;	608	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;	609	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;	610	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
539	}	611	}
540	}	612	}
541	else if (SvROK (sv))	613	else if (SvROK (sv))
542	encode_rv (enc, SvRV (sv));	614	encode_rv (enc, SvRV (sv));
543	else if (!SvOK (sv))	615	else if (!SvOK (sv))
…		…
546	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	618	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
547	SvPV_nolen (sv), SvFLAGS (sv));	619	SvPV_nolen (sv), SvFLAGS (sv));
548	}	620	}
549		621
550	static SV *	622	static SV *
551	encode_json (SV *scalar, U32 flags)	623	encode_json (SV scalar, struct json json)
552	{	624	{
553	enc_t enc;	625	enc_t enc;
554		626
555	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	627	if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
556	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");	628	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
557		629
558	enc.flags = flags;	630	enc.json = *json;
559	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	631	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
560	enc.cur = SvPVX (enc.sv);	632	enc.cur = SvPVX (enc.sv);
561	enc.end = SvEND (enc.sv);	633	enc.end = SvEND (enc.sv);
562	enc.indent = 0;	634	enc.indent = 0;
563	enc.maxdepth = DEC_DEPTH (flags);	635	enc.maxdepth = DEC_DEPTH (enc.json.flags);
564		636
565	SvPOK_only (enc.sv);	637	SvPOK_only (enc.sv);
566	encode_sv (&enc, scalar);	638	encode_sv (&enc, scalar);
567		639
568	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	640	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
569	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	641	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
570		642
571	if (!(flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))	643	if (!(enc.json.flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
572	SvUTF8_on (enc.sv);	644	SvUTF8_on (enc.sv);
573		645
574	if (enc.flags & F_SHRINK)	646	if (enc.json.flags & F_SHRINK)
575	shrink (enc.sv);	647	shrink (enc.sv);
576		648
577	return enc.sv;	649	return enc.sv;
578	}	650	}
579		651
…		…
584	typedef struct	656	typedef struct
585	{	657	{
586	char *cur; // current parser pointer	658	char *cur; // current parser pointer
587	char *end; // end of input string	659	char *end; // end of input string
588	const char *err; // parse error, if != 0	660	const char *err; // parse error, if != 0
589	U32 flags; // F_*	661	struct json json;
590	U32 depth; // recursion depth	662	U32 depth; // recursion depth
591	U32 maxdepth; // recursion depth limit	663	U32 maxdepth; // recursion depth limit
592	} dec_t;	664	} dec_t;
593		665
594	inline void	666	inline void
…		…
625	decode_4hex (dec_t *dec)	697	decode_4hex (dec_t *dec)
626	{	698	{
627	signed char d1, d2, d3, d4;	699	signed char d1, d2, d3, d4;
628	unsigned char cur = (unsigned char )dec->cur;	700	unsigned char cur = (unsigned char )dec->cur;
629		701
630	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected");	702	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");	703	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");	704	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");	705	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
634		706
635	dec->cur += 4;	707	dec->cur += 4;
636		708
637	return ((UV)d1) << 12	709	return ((UV)d1) << 12
638	\| ((UV)d2) << 8	710	\| ((UV)d2) << 8
…		…
646	static SV *	718	static SV *
647	decode_str (dec_t *dec)	719	decode_str (dec_t *dec)
648	{	720	{
649	SV *sv = 0;	721	SV *sv = 0;
650	int utf8 = 0;	722	int utf8 = 0;
		723	char *dec_cur = dec->cur;
651		724
652	do	725	do
653	{	726	{
654	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];	727	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
655	char *cur = buf;	728	char *cur = buf;
656		729
657	do	730	do
658	{	731	{
659	unsigned char ch = (unsigned char )dec->cur++;	732	unsigned char ch = (unsigned char )dec_cur++;
660		733
661	if (expect_false (ch == '"'))	734	if (expect_false (ch == '"'))
662	{	735	{
663	--dec->cur;	736	--dec_cur;
664	break;	737	break;
665	}	738	}
666	else if (expect_false (ch == '\\'))	739	else if (expect_false (ch == '\\'))
667	{	740	{
668	switch (*dec->cur)	741	switch (*dec_cur)
669	{	742	{
670	case '\\':	743	case '\\':
671	case '/':	744	case '/':
672	case '"': cur++ = dec->cur++; break;	745	case '"': cur++ = dec_cur++; break;
673		746
674	case 'b': ++dec->cur; *cur++ = '\010'; break;	747	case 'b': ++dec_cur; *cur++ = '\010'; break;
675	case 't': ++dec->cur; *cur++ = '\011'; break;	748	case 't': ++dec_cur; *cur++ = '\011'; break;
676	case 'n': ++dec->cur; *cur++ = '\012'; break;	749	case 'n': ++dec_cur; *cur++ = '\012'; break;
677	case 'f': ++dec->cur; *cur++ = '\014'; break;	750	case 'f': ++dec_cur; *cur++ = '\014'; break;
678	case 'r': ++dec->cur; *cur++ = '\015'; break;	751	case 'r': ++dec_cur; *cur++ = '\015'; break;
679		752
680	case 'u':	753	case 'u':
681	{	754	{
682	UV lo, hi;	755	UV lo, hi;
683	++dec->cur;	756	++dec_cur;
684		757
		758	dec->cur = dec_cur;
685	hi = decode_4hex (dec);	759	hi = decode_4hex (dec);
		760	dec_cur = dec->cur;
686	if (hi == (UV)-1)	761	if (hi == (UV)-1)
687	goto fail;	762	goto fail;
688		763
689	// possibly a surrogate pair	764	// possibly a surrogate pair
690	if (hi >= 0xd800)	765	if (hi >= 0xd800)
691	if (hi < 0xdc00)	766	if (hi < 0xdc00)
692	{	767	{
693	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')	768	if (dec_cur [0] != '\\' \|\| dec_cur [1] != 'u')
694	ERR ("missing low surrogate character in surrogate pair");	769	ERR ("missing low surrogate character in surrogate pair");
695		770
696	dec->cur += 2;	771	dec_cur += 2;
697		772
		773	dec->cur = dec_cur;
698	lo = decode_4hex (dec);	774	lo = decode_4hex (dec);
		775	dec_cur = dec->cur;
699	if (lo == (UV)-1)	776	if (lo == (UV)-1)
700	goto fail;	777	goto fail;
701		778
702	if (lo < 0xdc00 \|\| lo >= 0xe000)	779	if (lo < 0xdc00 \|\| lo >= 0xe000)
703	ERR ("surrogate pair expected");	780	ERR ("surrogate pair expected");
…		…
717	*cur++ = hi;	794	*cur++ = hi;
718	}	795	}
719	break;	796	break;
720		797
721	default:	798	default:
722	--dec->cur;	799	--dec_cur;
723	ERR ("illegal backslash escape sequence in string");	800	ERR ("illegal backslash escape sequence in string");
724	}	801	}
725	}	802	}
726	else if (expect_true (ch >= 0x20 && ch <= 0x7f))	803	else if (expect_true (ch >= 0x20 && ch <= 0x7f))
727	*cur++ = ch;	804	*cur++ = ch;
728	else if (ch >= 0x80)	805	else if (ch >= 0x80)
729	{	806	{
730	STRLEN clen;	807	STRLEN clen;
731	UV uch;	808	UV uch;
732		809
733	--dec->cur;	810	--dec_cur;
734		811
735	uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);	812	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
736	if (clen == (STRLEN)-1)	813	if (clen == (STRLEN)-1)
737	ERR ("malformed UTF-8 character in JSON string");	814	ERR ("malformed UTF-8 character in JSON string");
738		815
739	do	816	do
740	cur++ = dec->cur++;	817	cur++ = dec_cur++;
741	while (--clen);	818	while (--clen);
742		819
743	utf8 = 1;	820	utf8 = 1;
744	}	821	}
745	else	822	else
746	{	823	{
747	--dec->cur;	824	--dec_cur;
748		825
749	if (!ch)	826	if (!ch)
750	ERR ("unexpected end of string while parsing JSON string");	827	ERR ("unexpected end of string while parsing JSON string");
751	else	828	else
752	ERR ("invalid character encountered while parsing JSON string");	829	ERR ("invalid character encountered while parsing JSON string");
…		…
765	}	842	}
766	else	843	else
767	sv = newSVpvn (buf, len);	844	sv = newSVpvn (buf, len);
768	}	845	}
769	}	846	}
770	while (*dec->cur != '"');	847	while (*dec_cur != '"');
771		848
772	++dec->cur;	849	++dec_cur;
773		850
774	if (sv)	851	if (sv)
775	{	852	{
776	SvPOK_only (sv);	853	SvPOK_only (sv);
777	*SvEND (sv) = 0;	854	*SvEND (sv) = 0;
…		…
780	SvUTF8_on (sv);	857	SvUTF8_on (sv);
781	}	858	}
782	else	859	else
783	sv = newSVpvn ("", 0);	860	sv = newSVpvn ("", 0);
784		861
		862	dec->cur = dec_cur;
785	return sv;	863	return sv;
786		864
787	fail:	865	fail:
		866	dec->cur = dec_cur;
788	return 0;	867	return 0;
789	}	868	}
790		869
791	static SV *	870	static SV *
792	decode_num (dec_t *dec)	871	decode_num (dec_t *dec)
…		…
854	{	933	{
855	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so	934	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
856	if (*start == '-')	935	if (*start == '-')
857	switch (dec->cur - start)	936	switch (dec->cur - start)
858	{	937	{
859	case 2: return newSViv (-( start [1] - '0' ));	938	case 2: return newSViv (-( start [1] - '0' * 1));
860	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));	939	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
861	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));	940	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
862	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));	941	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
863	}	942	}
864	else	943	else
865	switch (dec->cur - start)	944	switch (dec->cur - start)
866	{	945	{
867	case 1: return newSViv ( start [0] - '0' );	946	case 1: return newSViv ( start [0] - '0' * 1);
868	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);	947	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
869	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);	948	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
870	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);	949	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
871	}	950	}
872		951
…		…
879	if (uv < (UV)IV_MIN)	958	if (uv < (UV)IV_MIN)
880	return newSViv (-(IV)uv);	959	return newSViv (-(IV)uv);
881	}	960	}
882	else	961	else
883	return newSVuv (uv);	962	return newSVuv (uv);
		963
		964	// here would likely be the place for bigint support
884	}	965	}
885	}	966	}
886		967
		968	// if we ever support bigint or bigfloat, this is the place for bigfloat
887	return newSVnv (Atof (start));	969	return newSVnv (Atof (start));
888		970
889	fail:	971	fail:
890	return 0;	972	return 0;
891	}	973	}
…		…
945	if (*dec->cur == '}')	1027	if (*dec->cur == '}')
946	++dec->cur;	1028	++dec->cur;
947	else	1029	else
948	for (;;)	1030	for (;;)
949	{	1031	{
950	SV key, value;
951
952	decode_ws (dec); EXPECT_CH ('"');	1032	decode_ws (dec); EXPECT_CH ('"');
953		1033
954	key = decode_str (dec);	1034	// heuristic: assume that
955	if (!key)	1035	// a) decode_str + hv_store_ent are abysmally slow.
956	goto fail;	1036	// b) most hash keys are short, simple ascii text.
		1037	// => try to "fast-match" such strings to avoid
		1038	// the overhead of decode_str + hv_store_ent.
		1039	{
		1040	SV *value;
		1041	char *p = dec->cur;
		1042	char *e = p + 24; // only try up to 24 bytes
957		1043
958	decode_ws (dec); EXPECT_CH (':');	1044	for (;;)
959
960	value = decode_sv (dec);
961	if (!value)
962	{	1045	{
		1046	// the >= 0x80 is true on most architectures
		1047	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
		1048	{
		1049	// slow path, back up and use decode_str
		1050	SV *key = decode_str (dec);
		1051	if (!key)
		1052	goto fail;
		1053
		1054	decode_ws (dec); EXPECT_CH (':');
		1055
		1056	value = decode_sv (dec);
		1057	if (!value)
		1058	{
		1059	SvREFCNT_dec (key);
		1060	goto fail;
		1061	}
		1062
		1063	hv_store_ent (hv, key, value, 0);
963	SvREFCNT_dec (key);	1064	SvREFCNT_dec (key);
		1065
		1066	break;
		1067	}
		1068	else if (*p == '"')
		1069	{
		1070	// fast path, got a simple key
		1071	char *key = dec->cur;
		1072	int len = p - key;
		1073	dec->cur = p + 1;
		1074
		1075	decode_ws (dec); EXPECT_CH (':');
		1076
		1077	value = decode_sv (dec);
		1078	if (!value)
964	goto fail;	1079	goto fail;
		1080
		1081	hv_store (hv, key, len, value, 0);
		1082
		1083	break;
		1084	}
		1085
		1086	++p;
965	}	1087	}
966		1088	}
967	hv_store_ent (hv, key, value, 0);
968	SvREFCNT_dec (key);
969		1089
970	decode_ws (dec);	1090	decode_ws (dec);
971		1091
972	if (*dec->cur == '}')	1092	if (*dec->cur == '}')
973	{	1093	{
…		…
992		1112
993	static SV *	1113	static SV *
994	decode_sv (dec_t *dec)	1114	decode_sv (dec_t *dec)
995	{	1115	{
996	decode_ws (dec);	1116	decode_ws (dec);
		1117
		1118	// the beauty of JSON: you need exactly one character lookahead
		1119	// to parse anything.
997	switch (*dec->cur)	1120	switch (*dec->cur)
998	{	1121	{
999	case '"': ++dec->cur; return decode_str (dec);	1122	case '"': ++dec->cur; return decode_str (dec);
1000	case '[': ++dec->cur; return decode_av (dec);	1123	case '[': ++dec->cur; return decode_av (dec);
1001	case '{': ++dec->cur; return decode_hv (dec);	1124	case '{': ++dec->cur; return decode_hv (dec);
…		…
1007		1130
1008	case 't':	1131	case 't':
1009	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1132	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1010	{	1133	{
1011	dec->cur += 4;	1134	dec->cur += 4;
1012	return newSViv (1);	1135	return SvREFCNT_inc (json_true);
1013	}	1136	}
1014	else	1137	else
1015	ERR ("'true' expected");	1138	ERR ("'true' expected");
1016		1139
1017	break;	1140	break;
1018		1141
1019	case 'f':	1142	case 'f':
1020	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1143	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1021	{	1144	{
1022	dec->cur += 5;	1145	dec->cur += 5;
1023	return newSViv (0);	1146	return SvREFCNT_inc (json_false);
1024	}	1147	}
1025	else	1148	else
1026	ERR ("'false' expected");	1149	ERR ("'false' expected");
1027		1150
1028	break;	1151	break;
…		…
1046	fail:	1169	fail:
1047	return 0;	1170	return 0;
1048	}	1171	}
1049		1172
1050	static SV *	1173	static SV *
1051	decode_json (SV string, U32 flags, UV offset_return)	1174	decode_json (SV string, struct json json, UV *offset_return)
1052	{	1175	{
1053	dec_t dec;	1176	dec_t dec;
1054	UV offset;	1177	UV offset;
1055	SV *sv;	1178	SV *sv;
1056		1179
1057	SvGETMAGIC (string);	1180	SvGETMAGIC (string);
1058	SvUPGRADE (string, SVt_PV);	1181	SvUPGRADE (string, SVt_PV);
1059		1182
		1183	if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
		1184	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1185	(unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
		1186
1060	if (flags & F_UTF8)	1187	if (json->flags & F_UTF8)
1061	sv_utf8_downgrade (string, 0);	1188	sv_utf8_downgrade (string, 0);
1062	else	1189	else
1063	sv_utf8_upgrade (string);	1190	sv_utf8_upgrade (string);
1064		1191
1065	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1192	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1066		1193
1067	dec.flags = flags;	1194	dec.json = *json;
1068	dec.cur = SvPVX (string);	1195	dec.cur = SvPVX (string);
1069	dec.end = SvEND (string);	1196	dec.end = SvEND (string);
1070	dec.err = 0;	1197	dec.err = 0;
1071	dec.depth = 0;	1198	dec.depth = 0;
1072	dec.maxdepth = DEC_DEPTH (dec.flags);	1199	dec.maxdepth = DEC_DEPTH (dec.json.flags);
1073		1200
1074	*dec.end = 0; // this should basically be a nop, too, but make sure it's there	1201	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
1075	sv = decode_sv (&dec);	1202	sv = decode_sv (&dec);
1076		1203
1077	if (!(offset_return \|\| !sv))	1204	if (!(offset_return \|\| !sv))
…		…
1087	}	1214	}
1088	}	1215	}
1089		1216
1090	if (offset_return \|\| !sv)	1217	if (offset_return \|\| !sv)
1091	{	1218	{
1092	offset = dec.flags & F_UTF8	1219	offset = dec.json.flags & F_UTF8
1093	? dec.cur - SvPVX (string)	1220	? dec.cur - SvPVX (string)
1094	: utf8_distance (dec.cur, SvPVX (string));	1221	: utf8_distance (dec.cur, SvPVX (string));
1095		1222
1096	if (offset_return)	1223	if (offset_return)
1097	*offset_return = offset;	1224	*offset_return = offset;
…		…
1116	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1243	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1117	}	1244	}
1118		1245
1119	sv = sv_2mortal (sv);	1246	sv = sv_2mortal (sv);
1120		1247
1121	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))	1248	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1122	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");	1249	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1123		1250
1124	return sv;	1251	return sv;
1125	}	1252	}
1126		1253
…		…
1130	MODULE = JSON::XS PACKAGE = JSON::XS	1257	MODULE = JSON::XS PACKAGE = JSON::XS
1131		1258
1132	BOOT:	1259	BOOT:
1133	{	1260	{
1134	int i;	1261	int i;
1135
1136	memset (decode_hexdigit, 0xff, 256);
1137		1262
1138	for (i = 0; i < 256; ++i)	1263	for (i = 0; i < 256; ++i)
1139	decode_hexdigit [i] =	1264	decode_hexdigit [i] =
1140	i >= '0' && i <= '9' ? i - '0'	1265	i >= '0' && i <= '9' ? i - '0'
1141	: i >= 'a' && i <= 'f' ? i - 'a' + 10	1266	: i >= 'a' && i <= 'f' ? i - 'a' + 10
1142	: i >= 'A' && i <= 'F' ? i - 'A' + 10	1267	: i >= 'A' && i <= 'F' ? i - 'A' + 10
1143	: -1;	1268	: -1;
1144		1269
1145	json_stash = gv_stashpv ("JSON::XS", 1);	1270	json_stash = gv_stashpv ("JSON::XS" , 1);
		1271	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
		1272
		1273	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
		1274	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1146	}	1275	}
1147		1276
1148	PROTOTYPES: DISABLE	1277	PROTOTYPES: DISABLE
1149		1278
1150	SV new (char dummy)	1279	SV new (char dummy)
…		…
1153	OUTPUT:	1282	OUTPUT:
1154	RETVAL	1283	RETVAL
1155		1284
1156	SV ascii (SV self, int enable = 1)	1285	SV ascii (SV self, int enable = 1)
1157	ALIAS:	1286	ALIAS:
1158	ascii = F_ASCII	1287	ascii = F_ASCII
1159	latin1 = F_LATIN1	1288	latin1 = F_LATIN1
1160	utf8 = F_UTF8	1289	utf8 = F_UTF8
1161	indent = F_INDENT	1290	indent = F_INDENT
1162	canonical = F_CANONICAL	1291	canonical = F_CANONICAL
1163	space_before = F_SPACE_BEFORE	1292	space_before = F_SPACE_BEFORE
1164	space_after = F_SPACE_AFTER	1293	space_after = F_SPACE_AFTER
1165	pretty = F_PRETTY	1294	pretty = F_PRETTY
1166	allow_nonref = F_ALLOW_NONREF	1295	allow_nonref = F_ALLOW_NONREF
1167	shrink = F_SHRINK	1296	shrink = F_SHRINK
		1297	allow_blessed = F_ALLOW_BLESSED
		1298	convert_blessed = F_CONV_BLESSED
1168	CODE:	1299	CODE:
1169	{	1300	{
1170	UV *uv = SvJSON (self);	1301	UV *uv = SvJSON (self);
1171	if (enable)	1302	if (enable)
1172	*uv \|= ix;	1303	*uv \|= ix;
…		…
1194	RETVAL = newSVsv (self);	1325	RETVAL = newSVsv (self);
1195	}	1326	}
1196	OUTPUT:	1327	OUTPUT:
1197	RETVAL	1328	RETVAL
1198		1329
		1330	SV max_size (SV self, UV max_size = 0)
		1331	CODE:
		1332	{
		1333	UV *uv = SvJSON (self);
		1334	UV log2 = 0;
		1335
		1336	if (max_size > 0x80000000UL) max_size = 0x80000000UL;
		1337	if (max_size == 1) max_size = 2;
		1338
		1339	while ((1UL << log2) < max_size)
		1340	++log2;
		1341
		1342	uv = uv & ~F_MAXSIZE \| (log2 << S_MAXSIZE);
		1343
		1344	RETVAL = newSVsv (self);
		1345	}
		1346	OUTPUT:
		1347	RETVAL
		1348
1199	void encode (SV self, SV scalar)	1349	void encode (SV self, SV scalar)
1200	PPCODE:	1350	PPCODE:
		1351	{
		1352	struct json json = { *SvJSON (self) };
1201	XPUSHs (encode_json (scalar, *SvJSON (self)));	1353	XPUSHs (encode_json (scalar, &json));
		1354	}
1202		1355
1203	void decode (SV self, SV jsonstr)	1356	void decode (SV self, SV jsonstr)
1204	PPCODE:	1357	PPCODE:
		1358	{
		1359	struct json json = { *SvJSON (self) };
1205	XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));	1360	XPUSHs (decode_json (jsonstr, &json, 0));
		1361	}
1206		1362
1207	void decode_prefix (SV self, SV jsonstr)	1363	void decode_prefix (SV self, SV jsonstr)
1208	PPCODE:	1364	PPCODE:
1209	{	1365	{
1210	UV offset;	1366	UV offset;
		1367	struct json json = { *SvJSON (self) };
1211	EXTEND (SP, 2);	1368	EXTEND (SP, 2);
1212	PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));	1369	PUSHs (decode_json (jsonstr, &json, &offset));
1213	PUSHs (sv_2mortal (newSVuv (offset)));	1370	PUSHs (sv_2mortal (newSVuv (offset)));
1214	}	1371	}
1215		1372
1216	PROTOTYPES: ENABLE	1373	PROTOTYPES: ENABLE
1217		1374
1218	void to_json (SV *scalar)	1375	void to_json (SV *scalar)
1219	ALIAS:
1220	objToJson = 0
1221	PPCODE:	1376	PPCODE:
		1377	{
		1378	struct json json = { F_DEFAULT \| F_UTF8 };
1222	XPUSHs (encode_json (scalar, F_DEFAULT \| F_UTF8));	1379	XPUSHs (encode_json (scalar, &json));
		1380	}
1223		1381
1224	void from_json (SV *jsonstr)	1382	void from_json (SV *jsonstr)
1225	ALIAS:
1226	jsonToObj = 0
1227	PPCODE:	1383	PPCODE:
		1384	{
		1385	struct json json = { F_DEFAULT \| F_UTF8 };
1228	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8, 0));	1386	XPUSHs (decode_json (jsonstr, &json, 0));
		1387	}
1229		1388

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.37 by root, Wed Jun 6 17:49:01 2007 UTC vs. Revision 1.47 by root, Sun Jul 1 14:08:03 2007 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.37 by root, Wed Jun 6 17:49:01 2007 UTC vs.
Revision 1.47 by root, Sun Jul 1 14:08:03 2007 UTC