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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.25 by root, Fri Apr 6 20:37:22 2007 UTC vs.
Revision 1.45 by root, Mon Jun 25 06:57:42 2007 UTC

…		…
3	#include "XSUB.h"	3	#include "XSUB.h"
4		4
5	#include "assert.h"	5	#include "assert.h"
6	#include "string.h"	6	#include "string.h"
7	#include "stdlib.h"	7	#include "stdlib.h"
		8	#include "stdio.h"
8		9
		10	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		11	# define snprintf _snprintf // C compilers have this in stdio.h
		12	#endif
		13
		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
9	#define F_ASCII 0x00000001UL	20	#define F_ASCII 0x00000001UL
		21	#define F_LATIN1 0x00000002UL
10	#define F_UTF8 0x00000002UL	22	#define F_UTF8 0x00000004UL
11	#define F_INDENT 0x00000004UL	23	#define F_INDENT 0x00000008UL
12	#define F_CANONICAL 0x00000008UL	24	#define F_CANONICAL 0x00000010UL
13	#define F_SPACE_BEFORE 0x00000010UL	25	#define F_SPACE_BEFORE 0x00000020UL
14	#define F_SPACE_AFTER 0x00000020UL	26	#define F_SPACE_AFTER 0x00000040UL
15	#define F_ALLOW_NONREF 0x00000080UL	27	#define F_ALLOW_NONREF 0x00000100UL
16	#define F_SHRINK 0x00000100UL	28	#define F_SHRINK 0x00000200UL
		29	#define F_ALLOW_BLESSED 0x00000400UL
		30	#define F_CONV_BLESSED 0x00000800UL // NYI
17	#define F_MAXDEPTH 0xf8000000UL	31	#define F_MAXDEPTH 0xf8000000UL
18	#define S_MAXDEPTH 27	32	#define S_MAXDEPTH 27
		33	#define F_MAXSIZE 0x01f00000UL
		34	#define S_MAXSIZE 20
19		35
20	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))	36	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21		37	#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
22	// F_SELFCONVERT? <=> to_json/toJson
23	// F_BLESSED? <=> { $__class__$ => }
24		38
25	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	39	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
26	#define F_DEFAULT (9UL << S_MAXDEPTH)	40	#define F_DEFAULT (9UL << S_MAXDEPTH)
27		41
28	#define INIT_SIZE 32 // initial scalar size to be allocated	42	#define INIT_SIZE 32 // initial scalar size to be allocated
29	#define INDENT_STEP 3 // spaces per indentation level	43	#define INDENT_STEP 3 // spaces per indentation level
30		44
31	#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32	#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
33		46
34	#define SB do {	47	#define SB do {
35	#define SE } while (0)	48	#define SE } while (0)
36		49
		50	#if __GNUC__ >= 3
		51	# define expect(expr,value) __builtin_expect ((expr),(value))
		52	# define inline inline
		53	#else
		54	# define expect(expr,value) (expr)
		55	# define inline static
		56	#endif
		57
		58	#define expect_false(expr) expect ((expr) != 0, 0)
		59	#define expect_true(expr) expect ((expr) != 0, 1)
		60
37	static HV *json_stash; // JSON::XS::	61	static HV json_stash, json_boolean_stash; // JSON::XS::
		62	static SV json_true, json_false;
38		63
39	/////////////////////////////////////////////////////////////////////////////	64	/////////////////////////////////////////////////////////////////////////////
40	// utility functions	65	// utility functions
41		66
42	static UV *	67	static UV *
…		…
65	// decode an utf-8 character and return it, or (UV)-1 in	90	// decode an utf-8 character and return it, or (UV)-1 in
66	// case of an error.	91	// case of an error.
67	// we special-case "safe" characters from U+80 .. U+7FF,	92	// we special-case "safe" characters from U+80 .. U+7FF,
68	// but use the very good perl function to parse anything else.	93	// but use the very good perl function to parse anything else.
69	// note that we never call this function for a ascii codepoints	94	// note that we never call this function for a ascii codepoints
70	static UV	95	inline UV
71	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)	96	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
72	{	97	{
73	if (s[0] > 0xdf \|\| s[0] < 0xc2)	98	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))
74	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);	99	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)	100	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76	{	101	{
77	*clen = 2;	102	*clen = 2;
78	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);	103	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
…		…
96	U32 flags; // F_*	121	U32 flags; // F_*
97	U32 indent; // indentation level	122	U32 indent; // indentation level
98	U32 maxdepth; // max. indentation/recursion level	123	U32 maxdepth; // max. indentation/recursion level
99	} enc_t;	124	} enc_t;
100		125
101	static void	126	inline void
102	need (enc_t *enc, STRLEN len)	127	need (enc_t *enc, STRLEN len)
103	{	128	{
104	if (enc->cur + len >= enc->end)	129	if (expect_false (enc->cur + len >= enc->end))
105	{	130	{
106	STRLEN cur = enc->cur - SvPVX (enc->sv);	131	STRLEN cur = enc->cur - SvPVX (enc->sv);
107	SvGROW (enc->sv, cur + len + 1);	132	SvGROW (enc->sv, cur + len + 1);
108	enc->cur = SvPVX (enc->sv) + cur;	133	enc->cur = SvPVX (enc->sv) + cur;
109	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);	134	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
110	}	135	}
111	}	136	}
112		137
113	static void	138	inline void
114	encode_ch (enc_t *enc, char ch)	139	encode_ch (enc_t *enc, char ch)
115	{	140	{
116	need (enc, 1);	141	need (enc, 1);
117	*enc->cur++ = ch;	142	*enc->cur++ = ch;
118	}	143	}
…		…
126		151
127	while (str < end)	152	while (str < end)
128	{	153	{
129	unsigned char ch = (unsigned char )str;	154	unsigned char ch = (unsigned char )str;
130		155
131	if (ch >= 0x20 && ch < 0x80) // most common case	156	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
132	{	157	{
133	if (ch == '"') // but with slow exceptions	158	if (expect_false (ch == '"')) // but with slow exceptions
134	{	159	{
135	need (enc, len += 1);	160	need (enc, len += 1);
136	*enc->cur++ = '\\';	161	*enc->cur++ = '\\';
137	*enc->cur++ = '"';	162	*enc->cur++ = '"';
138	}	163	}
139	else if (ch == '\\')	164	else if (expect_false (ch == '\\'))
140	{	165	{
141	need (enc, len += 1);	166	need (enc, len += 1);
142	*enc->cur++ = '\\';	167	*enc->cur++ = '\\';
143	*enc->cur++ = '\\';	168	*enc->cur++ = '\\';
144	}	169	}
…		…
162	STRLEN clen;	187	STRLEN clen;
163	UV uch;	188	UV uch;
164		189
165	if (is_utf8)	190	if (is_utf8)
166	{	191	{
167	//uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
168	uch = decode_utf8 (str, end - str, &clen);	192	uch = decode_utf8 (str, end - str, &clen);
169	if (clen == (STRLEN)-1)	193	if (clen == (STRLEN)-1)
170	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);	194	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
171	}	195	}
172	else	196	else
…		…
176	}	200	}
177		201
178	if (uch > 0x10FFFFUL)	202	if (uch > 0x10FFFFUL)
179	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);	203	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
180		204
181	if (uch < 0x80 \|\| enc->flags & F_ASCII)	205	if (uch < 0x80 \|\| enc->flags & F_ASCII \|\| (enc->flags & F_LATIN1 && uch > 0xFF))
182	{	206	{
183	if (uch > 0xFFFFUL)	207	if (uch > 0xFFFFUL)
184	{	208	{
185	need (enc, len += 11);	209	need (enc, len += 11);
186	sprintf (enc->cur, "\\u%04x\\u%04x",	210	sprintf (enc->cur, "\\u%04x\\u%04x",
…		…
200	*enc->cur++ = hexdigit [(uch >> 0) & 15];	224	*enc->cur++ = hexdigit [(uch >> 0) & 15];
201	}	225	}
202		226
203	str += clen;	227	str += clen;
204	}	228	}
		229	else if (enc->flags & F_LATIN1)
		230	{
		231	*enc->cur++ = uch;
		232	str += clen;
		233	}
205	else if (is_utf8)	234	else if (is_utf8)
206	{	235	{
207	need (enc, len += clen);	236	need (enc, len += clen);
208	do	237	do
209	{	238	{
…		…
211	}	240	}
212	while (--clen);	241	while (--clen);
213	}	242	}
214	else	243	else
215	{	244	{
216	need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed	245	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
217	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	246	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
218	++str;	247	++str;
219	}	248	}
220	}	249	}
221	}	250	}
…		…
223		252
224	--len;	253	--len;
225	}	254	}
226	}	255	}
227		256
228	static void	257	inline void
229	encode_indent (enc_t *enc)	258	encode_indent (enc_t *enc)
230	{	259	{
231	if (enc->flags & F_INDENT)	260	if (enc->flags & F_INDENT)
232	{	261	{
233	int spaces = enc->indent * INDENT_STEP;	262	int spaces = enc->indent * INDENT_STEP;
…		…
236	memset (enc->cur, ' ', spaces);	265	memset (enc->cur, ' ', spaces);
237	enc->cur += spaces;	266	enc->cur += spaces;
238	}	267	}
239	}	268	}
240		269
241	static void	270	inline void
242	encode_space (enc_t *enc)	271	encode_space (enc_t *enc)
243	{	272	{
244	need (enc, 1);	273	need (enc, 1);
245	encode_ch (enc, ' ');	274	encode_ch (enc, ' ');
246	}	275	}
247		276
248	static void	277	inline void
249	encode_nl (enc_t *enc)	278	encode_nl (enc_t *enc)
250	{	279	{
251	if (enc->flags & F_INDENT)	280	if (enc->flags & F_INDENT)
252	{	281	{
253	need (enc, 1);	282	need (enc, 1);
254	encode_ch (enc, '\n');	283	encode_ch (enc, '\n');
255	}	284	}
256	}	285	}
257		286
258	static void	287	inline void
259	encode_comma (enc_t *enc)	288	encode_comma (enc_t *enc)
260	{	289	{
261	encode_ch (enc, ',');	290	encode_ch (enc, ',');
262		291
263	if (enc->flags & F_INDENT)	292	if (enc->flags & F_INDENT)
…		…
362	// actually, this is mostly due to the stupid so-called	391	// actually, this is mostly due to the stupid so-called
363	// security workaround added somewhere in 5.8.x.	392	// security workaround added somewhere in 5.8.x.
364	// that randomises hash orderings	393	// that randomises hash orderings
365	if (enc->flags & F_CANONICAL)	394	if (enc->flags & F_CANONICAL)
366	{	395	{
367	HE he, hes [count]; // if your compiler dies here, you need to enable C99 mode
368	int fast = 1;	396	int fast = 1;
		397	HE *he;
		398	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		399	HE *hes = _alloca (count sizeof (HE));
		400	#else
		401	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
		402	#endif
369		403
370	i = 0;	404	i = 0;
371	while ((he = hv_iternext (hv)))	405	while ((he = hv_iternext (hv)))
372	{	406	{
373	hes [i++] = he;	407	hes [i++] = he;
…		…
437	svtype svt;	471	svtype svt;
438		472
439	SvGETMAGIC (sv);	473	SvGETMAGIC (sv);
440	svt = SvTYPE (sv);	474	svt = SvTYPE (sv);
441		475
		476	if (expect_false (SvOBJECT (sv)))
		477	{
		478	if (SvSTASH (sv) == json_boolean_stash)
		479	{
		480	if (SvIV (sv) == 0)
		481	encode_str (enc, "false", 5, 0);
		482	else
		483	encode_str (enc, "true", 4, 0);
		484	}
		485	else
		486	{
		487	#if 0
		488	if (0 && sv_derived_from (rv, "JSON::Literal"))
		489	{
		490	// not yet
		491	}
		492	#endif
		493	if (enc->flags & F_CONV_BLESSED)
		494	{
		495	// we re-bless the reference to get overload and other niceties right
		496	GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
		497
		498	if (to_json)
		499	{
		500	dSP;
		501	ENTER;
		502	SAVETMPS;
		503	PUSHMARK (SP);
		504	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
		505
		506	// calling with G_SCALAR ensures that we always get a 1 reutrn value
		507	// check anyways.
		508	PUTBACK;
		509	assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
		510	SPAGAIN;
		511
		512	encode_sv (enc, POPs);
		513
		514	FREETMPS;
		515	LEAVE;
		516	}
		517	else if (enc->flags & F_ALLOW_BLESSED)
		518	encode_str (enc, "null", 4, 0);
		519	else
		520	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
		521	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		522	}
		523	else if (enc->flags & F_ALLOW_BLESSED)
		524	encode_str (enc, "null", 4, 0);
		525	else
		526	croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
		527	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		528	}
		529	}
442	if (svt == SVt_PVHV)	530	else if (svt == SVt_PVHV)
443	encode_hv (enc, (HV *)sv);	531	encode_hv (enc, (HV *)sv);
444	else if (svt == SVt_PVAV)	532	else if (svt == SVt_PVAV)
445	encode_av (enc, (AV *)sv);	533	encode_av (enc, (AV *)sv);
446	else if (svt < SVt_PVAV)	534	else if (svt < SVt_PVAV)
447	{	535	{
…		…
471	encode_str (enc, str, len, SvUTF8 (sv));	559	encode_str (enc, str, len, SvUTF8 (sv));
472	encode_ch (enc, '"');	560	encode_ch (enc, '"');
473	}	561	}
474	else if (SvNOKp (sv))	562	else if (SvNOKp (sv))
475	{	563	{
		564	// trust that perl will do the right thing w.r.t. JSON syntax.
476	need (enc, NV_DIG + 32);	565	need (enc, NV_DIG + 32);
477	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	566	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
478	enc->cur += strlen (enc->cur);	567	enc->cur += strlen (enc->cur);
479	}	568	}
480	else if (SvIOKp (sv))	569	else if (SvIOKp (sv))
481	{	570	{
482	need (enc, 64);	571	// we assume we can always read an IV as a UV
		572	if (SvUV (sv) & ~(UV)0x7fff)
		573	{
		574	// large integer, use the (rather slow) snprintf way.
		575	need (enc, sizeof (UV) * 3);
483	enc->cur +=	576	enc->cur +=
484	SvIsUV(sv)	577	SvIsUV(sv)
485	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	578	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
486	: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));	579	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
		580	}
		581	else
		582	{
		583	// optimise the "small number case"
		584	// code will likely be branchless and use only a single multiplication
		585	I32 i = SvIV (sv);
		586	U32 u;
		587	char digit, nz = 0;
		588
		589	need (enc, 6);
		590
		591	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
		592	u = i < 0 ? -i : i;
		593
		594	// convert to 4.28 fixed-point representation
		595	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
		596
		597	// now output digit by digit, each time masking out the integer part
		598	// and multiplying by 5 while moving the decimal point one to the right,
		599	// resulting in a net multiplication by 10.
		600	// we always write the digit to memory but conditionally increment
		601	// the pointer, to ease the usage of conditional move instructions.
		602	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffff) 5;
		603	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffff) 5;
		604	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffff) 5;
		605	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffff) 5;
		606	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
		607	}
487	}	608	}
488	else if (SvROK (sv))	609	else if (SvROK (sv))
489	encode_rv (enc, SvRV (sv));	610	encode_rv (enc, SvRV (sv));
490	else if (!SvOK (sv))	611	else if (!SvOK (sv))
491	encode_str (enc, "null", 4, 0);	612	encode_str (enc, "null", 4, 0);
…		…
510	enc.maxdepth = DEC_DEPTH (flags);	631	enc.maxdepth = DEC_DEPTH (flags);
511		632
512	SvPOK_only (enc.sv);	633	SvPOK_only (enc.sv);
513	encode_sv (&enc, scalar);	634	encode_sv (&enc, scalar);
514		635
		636	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
		637	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
		638
515	if (!(flags & (F_ASCII \| F_UTF8)))	639	if (!(flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
516	SvUTF8_on (enc.sv);	640	SvUTF8_on (enc.sv);
517
518	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
519		641
520	if (enc.flags & F_SHRINK)	642	if (enc.flags & F_SHRINK)
521	shrink (enc.sv);	643	shrink (enc.sv);
522		644
523	return enc.sv;	645	return enc.sv;
…		…
535	U32 flags; // F_*	657	U32 flags; // F_*
536	U32 depth; // recursion depth	658	U32 depth; // recursion depth
537	U32 maxdepth; // recursion depth limit	659	U32 maxdepth; // recursion depth limit
538	} dec_t;	660	} dec_t;
539		661
540	static void	662	inline void
541	decode_ws (dec_t *dec)	663	decode_ws (dec_t *dec)
542	{	664	{
543	for (;;)	665	for (;;)
544	{	666	{
545	char ch = *dec->cur;	667	char ch = *dec->cur;
…		…
571	decode_4hex (dec_t *dec)	693	decode_4hex (dec_t *dec)
572	{	694	{
573	signed char d1, d2, d3, d4;	695	signed char d1, d2, d3, d4;
574	unsigned char cur = (unsigned char )dec->cur;	696	unsigned char cur = (unsigned char )dec->cur;
575		697
576	d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");	698	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
577	d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");	699	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
578	d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");	700	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
579	d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");	701	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
580		702
581	dec->cur += 4;	703	dec->cur += 4;
582		704
583	return ((UV)d1) << 12	705	return ((UV)d1) << 12
584	\| ((UV)d2) << 8	706	\| ((UV)d2) << 8
…		…
592	static SV *	714	static SV *
593	decode_str (dec_t *dec)	715	decode_str (dec_t *dec)
594	{	716	{
595	SV *sv = 0;	717	SV *sv = 0;
596	int utf8 = 0;	718	int utf8 = 0;
		719	char *dec_cur = dec->cur;
597		720
598	do	721	do
599	{	722	{
600	char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];	723	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
601	char *cur = buf;	724	char *cur = buf;
602		725
603	do	726	do
604	{	727	{
605	unsigned char ch = (unsigned char )dec->cur++;	728	unsigned char ch = (unsigned char )dec_cur++;
606		729
607	if (ch == '"')	730	if (expect_false (ch == '"'))
608	{	731	{
609	--dec->cur;	732	--dec_cur;
610	break;	733	break;
611	}	734	}
612	else if (ch == '\\')	735	else if (expect_false (ch == '\\'))
613	{	736	{
614	switch (*dec->cur)	737	switch (*dec_cur)
615	{	738	{
616	case '\\':	739	case '\\':
617	case '/':	740	case '/':
618	case '"': cur++ = dec->cur++; break;	741	case '"': cur++ = dec_cur++; break;
619		742
620	case 'b': ++dec->cur; *cur++ = '\010'; break;	743	case 'b': ++dec_cur; *cur++ = '\010'; break;
621	case 't': ++dec->cur; *cur++ = '\011'; break;	744	case 't': ++dec_cur; *cur++ = '\011'; break;
622	case 'n': ++dec->cur; *cur++ = '\012'; break;	745	case 'n': ++dec_cur; *cur++ = '\012'; break;
623	case 'f': ++dec->cur; *cur++ = '\014'; break;	746	case 'f': ++dec_cur; *cur++ = '\014'; break;
624	case 'r': ++dec->cur; *cur++ = '\015'; break;	747	case 'r': ++dec_cur; *cur++ = '\015'; break;
625		748
626	case 'u':	749	case 'u':
627	{	750	{
628	UV lo, hi;	751	UV lo, hi;
629	++dec->cur;	752	++dec_cur;
630		753
		754	dec->cur = dec_cur;
631	hi = decode_4hex (dec);	755	hi = decode_4hex (dec);
		756	dec_cur = dec->cur;
632	if (hi == (UV)-1)	757	if (hi == (UV)-1)
633	goto fail;	758	goto fail;
634		759
635	// possibly a surrogate pair	760	// possibly a surrogate pair
636	if (hi >= 0xd800)	761	if (hi >= 0xd800)
637	if (hi < 0xdc00)	762	if (hi < 0xdc00)
638	{	763	{
639	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')	764	if (dec_cur [0] != '\\' \|\| dec_cur [1] != 'u')
640	ERR ("missing low surrogate character in surrogate pair");	765	ERR ("missing low surrogate character in surrogate pair");
641		766
642	dec->cur += 2;	767	dec_cur += 2;
643		768
		769	dec->cur = dec_cur;
644	lo = decode_4hex (dec);	770	lo = decode_4hex (dec);
		771	dec_cur = dec->cur;
645	if (lo == (UV)-1)	772	if (lo == (UV)-1)
646	goto fail;	773	goto fail;
647		774
648	if (lo < 0xdc00 \|\| lo >= 0xe000)	775	if (lo < 0xdc00 \|\| lo >= 0xe000)
649	ERR ("surrogate pair expected");	776	ERR ("surrogate pair expected");
…		…
663	*cur++ = hi;	790	*cur++ = hi;
664	}	791	}
665	break;	792	break;
666		793
667	default:	794	default:
668	--dec->cur;	795	--dec_cur;
669	ERR ("illegal backslash escape sequence in string");	796	ERR ("illegal backslash escape sequence in string");
670	}	797	}
671	}	798	}
672	else if (ch >= 0x20 && ch <= 0x7f)	799	else if (expect_true (ch >= 0x20 && ch <= 0x7f))
673	*cur++ = ch;	800	*cur++ = ch;
674	else if (ch >= 0x80)	801	else if (ch >= 0x80)
675	{	802	{
676	STRLEN clen;	803	STRLEN clen;
677	UV uch;	804	UV uch;
678		805
679	--dec->cur;	806	--dec_cur;
680		807
681	uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);	808	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
682	if (clen == (STRLEN)-1)	809	if (clen == (STRLEN)-1)
683	ERR ("malformed UTF-8 character in JSON string");	810	ERR ("malformed UTF-8 character in JSON string");
684		811
685	do	812	do
686	cur++ = dec->cur++;	813	cur++ = dec_cur++;
687	while (--clen);	814	while (--clen);
688		815
689	utf8 = 1;	816	utf8 = 1;
690	}	817	}
691	else	818	else
692	{	819	{
693	--dec->cur;	820	--dec_cur;
694		821
695	if (!ch)	822	if (!ch)
696	ERR ("unexpected end of string while parsing JSON string");	823	ERR ("unexpected end of string while parsing JSON string");
697	else	824	else
698	ERR ("invalid character encountered while parsing JSON string");	825	ERR ("invalid character encountered while parsing JSON string");
…		…
711	}	838	}
712	else	839	else
713	sv = newSVpvn (buf, len);	840	sv = newSVpvn (buf, len);
714	}	841	}
715	}	842	}
716	while (*dec->cur != '"');	843	while (*dec_cur != '"');
717		844
718	++dec->cur;	845	++dec_cur;
719		846
720	if (sv)	847	if (sv)
721	{	848	{
722	SvPOK_only (sv);	849	SvPOK_only (sv);
723	*SvEND (sv) = 0;	850	*SvEND (sv) = 0;
…		…
726	SvUTF8_on (sv);	853	SvUTF8_on (sv);
727	}	854	}
728	else	855	else
729	sv = newSVpvn ("", 0);	856	sv = newSVpvn ("", 0);
730		857
		858	dec->cur = dec_cur;
731	return sv;	859	return sv;
732		860
733	fail:	861	fail:
		862	dec->cur = dec_cur;
734	return 0;	863	return 0;
735	}	864	}
736		865
737	static SV *	866	static SV *
738	decode_num (dec_t *dec)	867	decode_num (dec_t *dec)
…		…
796	is_nv = 1;	925	is_nv = 1;
797	}	926	}
798		927
799	if (!is_nv)	928	if (!is_nv)
800	{	929	{
801	UV uv;	930	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
802	int numtype = grok_number (start, dec->cur - start, &uv);	931	if (*start == '-')
803	if (numtype & IS_NUMBER_IN_UV)	932	switch (dec->cur - start)
804	if (numtype & IS_NUMBER_NEG)
805	{	933	{
806	if (uv < (UV)IV_MIN)	934	case 2: return newSViv (-( start [1] - '0' * 1));
807	return newSViv (-(IV)uv);	935	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
		936	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
		937	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
808	}	938	}
		939	else
		940	switch (dec->cur - start)
		941	{
		942	case 1: return newSViv ( start [0] - '0' * 1);
		943	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
		944	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
		945	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		946	}
		947
		948	{
		949	UV uv;
		950	int numtype = grok_number (start, dec->cur - start, &uv);
		951	if (numtype & IS_NUMBER_IN_UV)
		952	if (numtype & IS_NUMBER_NEG)
		953	{
		954	if (uv < (UV)IV_MIN)
		955	return newSViv (-(IV)uv);
		956	}
809	else	957	else
810	return newSVuv (uv);	958	return newSVuv (uv);
		959
		960	// here would likely be the place for bigint support
811	}	961	}
		962	}
812		963
		964	// if we ever support bigint or bigfloat, this is the place for bigfloat
813	return newSVnv (Atof (start));	965	return newSVnv (Atof (start));
814		966
815	fail:	967	fail:
816	return 0;	968	return 0;
817	}	969	}
…		…
918		1070
919	static SV *	1071	static SV *
920	decode_sv (dec_t *dec)	1072	decode_sv (dec_t *dec)
921	{	1073	{
922	decode_ws (dec);	1074	decode_ws (dec);
		1075
		1076	// the beauty of JSON: you need exactly one character lookahead
		1077	// to parse anything.
923	switch (*dec->cur)	1078	switch (*dec->cur)
924	{	1079	{
925	case '"': ++dec->cur; return decode_str (dec);	1080	case '"': ++dec->cur; return decode_str (dec);
926	case '[': ++dec->cur; return decode_av (dec);	1081	case '[': ++dec->cur; return decode_av (dec);
927	case '{': ++dec->cur; return decode_hv (dec);	1082	case '{': ++dec->cur; return decode_hv (dec);
…		…
933		1088
934	case 't':	1089	case 't':
935	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1090	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
936	{	1091	{
937	dec->cur += 4;	1092	dec->cur += 4;
938	return newSViv (1);	1093	return SvREFCNT_inc (json_true);
939	}	1094	}
940	else	1095	else
941	ERR ("'true' expected");	1096	ERR ("'true' expected");
942		1097
943	break;	1098	break;
944		1099
945	case 'f':	1100	case 'f':
946	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1101	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
947	{	1102	{
948	dec->cur += 5;	1103	dec->cur += 5;
949	return newSViv (0);	1104	return SvREFCNT_inc (json_false);
950	}	1105	}
951	else	1106	else
952	ERR ("'false' expected");	1107	ERR ("'false' expected");
953		1108
954	break;	1109	break;
…		…
972	fail:	1127	fail:
973	return 0;	1128	return 0;
974	}	1129	}
975		1130
976	static SV *	1131	static SV *
977	decode_json (SV *string, U32 flags)	1132	decode_json (SV string, U32 flags, UV offset_return)
978	{	1133	{
979	dec_t dec;	1134	dec_t dec;
		1135	UV offset;
980	SV *sv;	1136	SV *sv;
981		1137
		1138	SvGETMAGIC (string);
982	SvUPGRADE (string, SVt_PV);	1139	SvUPGRADE (string, SVt_PV);
		1140
		1141	if (flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (flags))
		1142	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1143	(unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (flags));
983		1144
984	if (flags & F_UTF8)	1145	if (flags & F_UTF8)
985	sv_utf8_downgrade (string, 0);	1146	sv_utf8_downgrade (string, 0);
986	else	1147	else
987	sv_utf8_upgrade (string);	1148	sv_utf8_upgrade (string);
…		…
993	dec.end = SvEND (string);	1154	dec.end = SvEND (string);
994	dec.err = 0;	1155	dec.err = 0;
995	dec.depth = 0;	1156	dec.depth = 0;
996	dec.maxdepth = DEC_DEPTH (dec.flags);	1157	dec.maxdepth = DEC_DEPTH (dec.flags);
997		1158
998	*dec.end = 0; // this should basically be a nop, too, but make sure its there	1159	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
999	sv = decode_sv (&dec);	1160	sv = decode_sv (&dec);
1000		1161
		1162	if (!(offset_return \|\| !sv))
		1163	{
		1164	// check for trailing garbage
		1165	decode_ws (&dec);
		1166
		1167	if (*dec.cur)
		1168	{
		1169	dec.err = "garbage after JSON object";
		1170	SvREFCNT_dec (sv);
		1171	sv = 0;
		1172	}
		1173	}
		1174
		1175	if (offset_return \|\| !sv)
		1176	{
		1177	offset = dec.flags & F_UTF8
		1178	? dec.cur - SvPVX (string)
		1179	: utf8_distance (dec.cur, SvPVX (string));
		1180
		1181	if (offset_return)
		1182	*offset_return = offset;
		1183	}
		1184
1001	if (!sv)	1185	if (!sv)
1002	{	1186	{
1003	IV offset = dec.flags & F_UTF8
1004	? dec.cur - SvPVX (string)
1005	: utf8_distance (dec.cur, SvPVX (string));
1006	SV *uni = sv_newmortal ();	1187	SV *uni = sv_newmortal ();
1007		1188
1008	// horrible hack to silence warning inside pv_uni_display	1189	// horrible hack to silence warning inside pv_uni_display
1009	COP cop = *PL_curcop;	1190	COP cop = *PL_curcop;
1010	cop.cop_warnings = pWARN_NONE;	1191	cop.cop_warnings = pWARN_NONE;
…		…
1035		1216
1036	BOOT:	1217	BOOT:
1037	{	1218	{
1038	int i;	1219	int i;
1039		1220
1040	memset (decode_hexdigit, 0xff, 256);
1041
1042	for (i = 0; i < 256; ++i)	1221	for (i = 0; i < 256; ++i)
1043	decode_hexdigit [i] =	1222	decode_hexdigit [i] =
1044	i >= '0' && i <= '9' ? i - '0'	1223	i >= '0' && i <= '9' ? i - '0'
1045	: i >= 'a' && i <= 'f' ? i - 'a' + 10	1224	: i >= 'a' && i <= 'f' ? i - 'a' + 10
1046	: i >= 'A' && i <= 'F' ? i - 'A' + 10	1225	: i >= 'A' && i <= 'F' ? i - 'A' + 10
1047	: -1;	1226	: -1;
1048		1227
1049	json_stash = gv_stashpv ("JSON::XS", 1);	1228	json_stash = gv_stashpv ("JSON::XS" , 1);
		1229	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
		1230
		1231	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
		1232	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1050	}	1233	}
1051		1234
1052	PROTOTYPES: DISABLE	1235	PROTOTYPES: DISABLE
1053		1236
1054	SV new (char dummy)	1237	SV new (char dummy)
…		…
1057	OUTPUT:	1240	OUTPUT:
1058	RETVAL	1241	RETVAL
1059		1242
1060	SV ascii (SV self, int enable = 1)	1243	SV ascii (SV self, int enable = 1)
1061	ALIAS:	1244	ALIAS:
1062	ascii = F_ASCII	1245	ascii = F_ASCII
		1246	latin1 = F_LATIN1
1063	utf8 = F_UTF8	1247	utf8 = F_UTF8
1064	indent = F_INDENT	1248	indent = F_INDENT
1065	canonical = F_CANONICAL	1249	canonical = F_CANONICAL
1066	space_before = F_SPACE_BEFORE	1250	space_before = F_SPACE_BEFORE
1067	space_after = F_SPACE_AFTER	1251	space_after = F_SPACE_AFTER
1068	pretty = F_PRETTY	1252	pretty = F_PRETTY
1069	allow_nonref = F_ALLOW_NONREF	1253	allow_nonref = F_ALLOW_NONREF
1070	shrink = F_SHRINK	1254	shrink = F_SHRINK
		1255	allow_blessed = F_ALLOW_BLESSED
		1256	convert_blessed = F_CONV_BLESSED
1071	CODE:	1257	CODE:
1072	{	1258	{
1073	UV *uv = SvJSON (self);	1259	UV *uv = SvJSON (self);
1074	if (enable)	1260	if (enable)
1075	*uv \|= ix;	1261	*uv \|= ix;
…		…
1097	RETVAL = newSVsv (self);	1283	RETVAL = newSVsv (self);
1098	}	1284	}
1099	OUTPUT:	1285	OUTPUT:
1100	RETVAL	1286	RETVAL
1101		1287
		1288	SV max_size (SV self, UV max_size = 0)
		1289	CODE:
		1290	{
		1291	UV *uv = SvJSON (self);
		1292	UV log2 = 0;
		1293
		1294	if (max_size > 0x80000000UL) max_size = 0x80000000UL;
		1295	if (max_size == 1) max_size = 2;
		1296
		1297	while ((1UL << log2) < max_size)
		1298	++log2;
		1299
		1300	uv = uv & ~F_MAXSIZE \| (log2 << S_MAXSIZE);
		1301
		1302	RETVAL = newSVsv (self);
		1303	}
		1304	OUTPUT:
		1305	RETVAL
		1306
1102	void encode (SV self, SV scalar)	1307	void encode (SV self, SV scalar)
1103	PPCODE:	1308	PPCODE:
1104	XPUSHs (encode_json (scalar, *SvJSON (self)));	1309	XPUSHs (encode_json (scalar, *SvJSON (self)));
1105		1310
1106	void decode (SV self, SV jsonstr)	1311	void decode (SV self, SV jsonstr)
1107	PPCODE:	1312	PPCODE:
1108	XPUSHs (decode_json (jsonstr, *SvJSON (self)));	1313	XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
		1314
		1315	void decode_prefix (SV self, SV jsonstr)
		1316	PPCODE:
		1317	{
		1318	UV offset;
		1319	EXTEND (SP, 2);
		1320	PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
		1321	PUSHs (sv_2mortal (newSVuv (offset)));
		1322	}
1109		1323
1110	PROTOTYPES: ENABLE	1324	PROTOTYPES: ENABLE
1111		1325
1112	void to_json (SV *scalar)	1326	void to_json (SV *scalar)
1113	ALIAS:	1327	ALIAS:
…		…
1117		1331
1118	void from_json (SV *jsonstr)	1332	void from_json (SV *jsonstr)
1119	ALIAS:	1333	ALIAS:
1120	jsonToObj = 0	1334	jsonToObj = 0
1121	PPCODE:	1335	PPCODE:
1122	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8));	1336	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8, 0));
1123		1337

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.25 by root, Fri Apr 6 20:37:22 2007 UTC vs. Revision 1.45 by root, Mon Jun 25 06:57:42 2007 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.25 by root, Fri Apr 6 20:37:22 2007 UTC vs.
Revision 1.45 by root, Mon Jun 25 06:57:42 2007 UTC