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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.8 by root, Fri Mar 23 16:13:59 2007 UTC vs.
Revision 1.43 by root, Sat Jun 23 23:49:29 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 0x00000001	20	#define F_ASCII 0x00000001UL
		21	#define F_LATIN1 0x00000002UL
10	#define F_UTF8 0x00000002	22	#define F_UTF8 0x00000004UL
11	#define F_INDENT 0x00000004	23	#define F_INDENT 0x00000008UL
12	#define F_CANONICAL 0x00000008	24	#define F_CANONICAL 0x00000010UL
13	#define F_SPACE_BEFORE 0x00000010	25	#define F_SPACE_BEFORE 0x00000020UL
14	#define F_SPACE_AFTER 0x00000020	26	#define F_SPACE_AFTER 0x00000040UL
15	#define F_JSON_RPC 0x00000040
16	#define F_ALLOW_NONREF 0x00000080	27	#define F_ALLOW_NONREF 0x00000100UL
17	#define F_SHRINK 0x00000100	28	#define F_SHRINK 0x00000200UL
		29	#define F_MAXDEPTH 0xf8000000UL
		30	#define S_MAXDEPTH 27
		31
		32	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
		33
		34	// F_SELFCONVERT? <=> to_json/toJson
		35	// F_BLESSED? <=> { $__class__$ => }
18		36
19	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	37	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
20	#define F_DEFAULT 0	38	#define F_DEFAULT (9UL << S_MAXDEPTH)
21		39
22	#define INIT_SIZE 32 // initial scalar size to be allocated	40	#define INIT_SIZE 32 // initial scalar size to be allocated
		41	#define INDENT_STEP 3 // spaces per indentation level
		42
		43	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
23		44
24	#define SB do {	45	#define SB do {
25	#define SE } while (0)	46	#define SE } while (0)
26		47
		48	#if __GNUC__ >= 3
		49	# define expect(expr,value) __builtin_expect ((expr),(value))
		50	# define inline inline
		51	#else
		52	# define expect(expr,value) (expr)
		53	# define inline static
		54	#endif
		55
		56	#define expect_false(expr) expect ((expr) != 0, 0)
		57	#define expect_true(expr) expect ((expr) != 0, 1)
		58
27	static HV *json_stash;	59	static HV *json_stash; // JSON::XS::
		60	static SV json_true, json_false;
		61
		62	/////////////////////////////////////////////////////////////////////////////
		63	// utility functions
		64
		65	static UV *
		66	SvJSON (SV *sv)
		67	{
		68	if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
		69	croak ("object is not of type JSON::XS");
		70
		71	return &SvUVX (SvRV (sv));
		72	}
		73
		74	static void
		75	shrink (SV *sv)
		76	{
		77	sv_utf8_downgrade (sv, 1);
		78	if (SvLEN (sv) > SvCUR (sv) + 1)
		79	{
		80	#ifdef SvPV_shrink_to_cur
		81	SvPV_shrink_to_cur (sv);
		82	#elif defined (SvPV_renew)
		83	SvPV_renew (sv, SvCUR (sv) + 1);
		84	#endif
		85	}
		86	}
		87
		88	// decode an utf-8 character and return it, or (UV)-1 in
		89	// case of an error.
		90	// we special-case "safe" characters from U+80 .. U+7FF,
		91	// but use the very good perl function to parse anything else.
		92	// note that we never call this function for a ascii codepoints
		93	inline UV
		94	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
		95	{
		96	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))
		97	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
		98	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
		99	{
		100	*clen = 2;
		101	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
		102	}
		103	else
		104	{
		105	*clen = (STRLEN)-1;
		106	return (UV)-1;
		107	}
		108	}
		109
		110	/////////////////////////////////////////////////////////////////////////////
		111	// encoder
28		112
29	// structure used for encoding JSON	113	// structure used for encoding JSON
30	typedef struct	114	typedef struct
31	{	115	{
32	char *cur;	116	char *cur; // SvPVX (sv) + current output position
33	STRLEN len; // SvLEN (sv)
34	char *end; // SvEND (sv)	117	char *end; // SvEND (sv)
35	SV *sv;	118	SV *sv; // result scalar
36	UV flags;	119	U32 flags; // F_*
37	int max_recurse;	120	U32 indent; // indentation level
38	int indent;	121	U32 maxdepth; // max. indentation/recursion level
39	} enc_t;	122	} enc_t;
40		123
41	// structure used for decoding JSON	124	inline void
42	typedef struct
43	{
44	char *cur;
45	char *end;
46	const char *err;
47	UV flags;
48	} dec_t;
49
50	static UV *
51	SvJSON (SV *sv)
52	{
53	if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
54	croak ("object is not of type JSON::XS");
55
56	return &SvUVX (SvRV (sv));
57	}
58
59	static void
60	shrink (SV *sv)
61	{
62	sv_utf8_downgrade (sv, 1);
63	#ifdef SvPV_shrink_to_cur
64	SvPV_shrink_to_cur (sv);
65	#endif
66	}
67
68	/////////////////////////////////////////////////////////////////////////////
69
70	static void
71	need (enc_t *enc, STRLEN len)	125	need (enc_t *enc, STRLEN len)
72	{	126	{
73	if (enc->cur + len >= enc->end)	127	if (expect_false (enc->cur + len >= enc->end))
74	{	128	{
75	STRLEN cur = enc->cur - SvPVX (enc->sv);	129	STRLEN cur = enc->cur - SvPVX (enc->sv);
76	SvGROW (enc->sv, cur + len + 1);	130	SvGROW (enc->sv, cur + len + 1);
77	enc->cur = SvPVX (enc->sv) + cur;	131	enc->cur = SvPVX (enc->sv) + cur;
78	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);	132	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
79	}	133	}
80	}	134	}
81		135
82	static void	136	inline void
83	encode_ch (enc_t *enc, char ch)	137	encode_ch (enc_t *enc, char ch)
84	{	138	{
85	need (enc, 1);	139	need (enc, 1);
86	*enc->cur++ = ch;	140	*enc->cur++ = ch;
87	}	141	}
…		…
95		149
96	while (str < end)	150	while (str < end)
97	{	151	{
98	unsigned char ch = (unsigned char )str;	152	unsigned char ch = (unsigned char )str;
99		153
100	if (ch >= 0x20 && ch < 0x80) // most common case	154	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
101	{	155	{
102	if (ch == '"') // but with slow exceptions	156	if (expect_false (ch == '"')) // but with slow exceptions
103	{	157	{
104	need (enc, len += 1);	158	need (enc, len += 1);
105	*enc->cur++ = '\\';	159	*enc->cur++ = '\\';
106	*enc->cur++ = '"';	160	*enc->cur++ = '"';
107	}	161	}
108	else if (ch == '\\')	162	else if (expect_false (ch == '\\'))
109	{	163	{
110	need (enc, len += 1);	164	need (enc, len += 1);
111	*enc->cur++ = '\\';	165	*enc->cur++ = '\\';
112	*enc->cur++ = '\\';	166	*enc->cur++ = '\\';
113	}	167	}
…		…
131	STRLEN clen;	185	STRLEN clen;
132	UV uch;	186	UV uch;
133		187
134	if (is_utf8)	188	if (is_utf8)
135	{	189	{
136	uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);	190	uch = decode_utf8 (str, end - str, &clen);
137	if (clen == (STRLEN)-1)	191	if (clen == (STRLEN)-1)
138	croak ("malformed UTF-8 character in string, cannot convert to JSON");	192	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
139	}	193	}
140	else	194	else
141	{	195	{
142	uch = ch;	196	uch = ch;
143	clen = 1;	197	clen = 1;
144	}	198	}
145		199
146	if (uch < 0x80 \|\| enc->flags & F_ASCII)	200	if (uch > 0x10FFFFUL)
		201	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
		202
		203	if (uch < 0x80 \|\| enc->flags & F_ASCII \|\| (enc->flags & F_LATIN1 && uch > 0xFF))
147	{	204	{
148	if (uch > 0xFFFFUL)	205	if (uch > 0xFFFFUL)
149	{	206	{
150	need (enc, len += 11);	207	need (enc, len += 11);
151	sprintf (enc->cur, "\\u%04x\\u%04x",	208	sprintf (enc->cur, "\\u%04x\\u%04x",
152	(uch - 0x10000) / 0x400 + 0xD800,	209	(int)((uch - 0x10000) / 0x400 + 0xD800),
153	(uch - 0x10000) % 0x400 + 0xDC00);	210	(int)((uch - 0x10000) % 0x400 + 0xDC00));
154	enc->cur += 12;	211	enc->cur += 12;
155	}	212	}
156	else	213	else
157	{	214	{
158	static char hexdigit [16] = "0123456789abcdef";	215	static char hexdigit [16] = "0123456789abcdef";
…		…
165	*enc->cur++ = hexdigit [(uch >> 0) & 15];	222	*enc->cur++ = hexdigit [(uch >> 0) & 15];
166	}	223	}
167		224
168	str += clen;	225	str += clen;
169	}	226	}
		227	else if (enc->flags & F_LATIN1)
		228	{
		229	*enc->cur++ = uch;
		230	str += clen;
		231	}
170	else if (is_utf8)	232	else if (is_utf8)
171	{	233	{
172	need (enc, len += clen);	234	need (enc, len += clen);
173	do	235	do
174	{	236	{
…		…
176	}	238	}
177	while (--clen);	239	while (--clen);
178	}	240	}
179	else	241	else
180	{	242	{
181	need (enc, len += 10); // never more than 11 bytes needed	243	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
182	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	244	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
183	++str;	245	++str;
184	}	246	}
185	}	247	}
186	}	248	}
…		…
188		250
189	--len;	251	--len;
190	}	252	}
191	}	253	}
192		254
193	#define INDENT SB \	255	inline void
		256	encode_indent (enc_t *enc)
		257	{
194	if (enc->flags & F_INDENT) \	258	if (enc->flags & F_INDENT)
195	{ \	259	{
196	int i_; \	260	int spaces = enc->indent * INDENT_STEP;
197	need (enc, enc->indent); \
198	for (i_ = enc->indent * 3; i_--; )\
199	encode_ch (enc, ' '); \
200	} \
201	SE
202		261
203	#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE	262	need (enc, spaces);
204	#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE	263	memset (enc->cur, ' ', spaces);
205	#define COMMA SB \	264	enc->cur += spaces;
		265	}
		266	}
		267
		268	inline void
		269	encode_space (enc_t *enc)
		270	{
		271	need (enc, 1);
206	encode_ch (enc, ','); \	272	encode_ch (enc, ' ');
		273	}
		274
		275	inline void
		276	encode_nl (enc_t *enc)
		277	{
207	if (enc->flags & F_INDENT) \	278	if (enc->flags & F_INDENT)
208	NL; \	279	{
		280	need (enc, 1);
		281	encode_ch (enc, '\n');
		282	}
		283	}
		284
		285	inline void
		286	encode_comma (enc_t *enc)
		287	{
		288	encode_ch (enc, ',');
		289
		290	if (enc->flags & F_INDENT)
		291	encode_nl (enc);
209	else if (enc->flags & F_SPACE_AFTER) \	292	else if (enc->flags & F_SPACE_AFTER)
210	SPACE; \	293	encode_space (enc);
211	SE	294	}
212		295
213	static void encode_sv (enc_t enc, SV sv);	296	static void encode_sv (enc_t enc, SV sv);
214		297
215	static void	298	static void
216	encode_av (enc_t enc, AV av)	299	encode_av (enc_t enc, AV av)
217	{	300	{
218	int i, len = av_len (av);	301	int i, len = av_len (av);
219		302
220	encode_ch (enc, '['); NL;	303	if (enc->indent >= enc->maxdepth)
		304	croak ("data structure too deep (hit recursion limit)");
		305
		306	encode_ch (enc, '['); encode_nl (enc);
221	++enc->indent;	307	++enc->indent;
222		308
223	for (i = 0; i <= len; ++i)	309	for (i = 0; i <= len; ++i)
224	{	310	{
225	INDENT;	311	encode_indent (enc);
226	encode_sv (enc, *av_fetch (av, i, 0));	312	encode_sv (enc, *av_fetch (av, i, 0));
227		313
228	if (i < len)	314	if (i < len)
229	COMMA;	315	encode_comma (enc);
230	}	316	}
231		317
232	NL;	318	encode_nl (enc);
233		319
234	--enc->indent;	320	--enc->indent;
235	INDENT; encode_ch (enc, ']');	321	encode_indent (enc); encode_ch (enc, ']');
236	}	322	}
237		323
238	static void	324	static void
239	encode_he (enc_t enc, HE he)	325	encode_he (enc_t enc, HE he)
240	{	326	{
…		…
254	else	340	else
255	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));	341	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
256		342
257	encode_ch (enc, '"');	343	encode_ch (enc, '"');
258		344
259	if (enc->flags & F_SPACE_BEFORE) SPACE;	345	if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
260	encode_ch (enc, ':');	346	encode_ch (enc, ':');
261	if (enc->flags & F_SPACE_AFTER ) SPACE;	347	if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
262	encode_sv (enc, HeVAL (he));	348	encode_sv (enc, HeVAL (he));
263	}	349	}
264		350
265	// compare hash entries, used when all keys are bytestrings	351	// compare hash entries, used when all keys are bytestrings
266	static int	352	static int
…		…
272	HE b = (HE **)b_;	358	HE b = (HE **)b_;
273		359
274	STRLEN la = HeKLEN (a);	360	STRLEN la = HeKLEN (a);
275	STRLEN lb = HeKLEN (b);	361	STRLEN lb = HeKLEN (b);
276		362
277	if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))	363	if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
278	cmp = la < lb ? -1 : la == lb ? 0 : 1;	364	cmp = la - lb;
279		365
280	return cmp;	366	return cmp;
281	}	367	}
282		368
283	// compare hash entries, used when some keys are sv's or utf-x	369	// compare hash entries, used when some keys are sv's or utf-x
…		…
290	static void	376	static void
291	encode_hv (enc_t enc, HV hv)	377	encode_hv (enc_t enc, HV hv)
292	{	378	{
293	int count, i;	379	int count, i;
294		380
		381	if (enc->indent >= enc->maxdepth)
		382	croak ("data structure too deep (hit recursion limit)");
		383
295	encode_ch (enc, '{'); NL; ++enc->indent;	384	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
296		385
297	if ((count = hv_iterinit (hv)))	386	if ((count = hv_iterinit (hv)))
298	{	387	{
299	// for canonical output we have to sort by keys first	388	// for canonical output we have to sort by keys first
300	// actually, this is mostly due to the stupid so-called	389	// actually, this is mostly due to the stupid so-called
301	// security workaround added somewhere in 5.8.x.	390	// security workaround added somewhere in 5.8.x.
302	// that randomises hash orderings	391	// that randomises hash orderings
303	if (enc->flags & F_CANONICAL)	392	if (enc->flags & F_CANONICAL)
304	{	393	{
305	HE he, hes [count];
306	int fast = 1;	394	int fast = 1;
		395	HE *he;
		396	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		397	HE *hes = _alloca (count sizeof (HE));
		398	#else
		399	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
		400	#endif
307		401
308	i = 0;	402	i = 0;
309	while ((he = hv_iternext (hv)))	403	while ((he = hv_iternext (hv)))
310	{	404	{
311	hes [i++] = he;	405	hes [i++] = he;
…		…
335	LEAVE;	429	LEAVE;
336	}	430	}
337		431
338	for (i = 0; i < count; ++i)	432	for (i = 0; i < count; ++i)
339	{	433	{
340	INDENT;	434	encode_indent (enc);
341	encode_he (enc, hes [i]);	435	encode_he (enc, hes [i]);
342		436
343	if (i < count - 1)	437	if (i < count - 1)
344	COMMA;	438	encode_comma (enc);
345	}	439	}
346		440
347	NL;	441	encode_nl (enc);
348	}	442	}
349	else	443	else
350	{	444	{
351	SV *sv;
352	HE *he = hv_iternext (hv);	445	HE *he = hv_iternext (hv);
353		446
354	for (;;)	447	for (;;)
355	{	448	{
356	INDENT;	449	encode_indent (enc);
357	encode_he (enc, he);	450	encode_he (enc, he);
358		451
359	if (!(he = hv_iternext (hv)))	452	if (!(he = hv_iternext (hv)))
360	break;	453	break;
361		454
362	COMMA;	455	encode_comma (enc);
363	}	456	}
364		457
365	NL;	458	encode_nl (enc);
366	}	459	}
367	}	460	}
368		461
369	--enc->indent; INDENT; encode_ch (enc, '}');	462	--enc->indent; encode_indent (enc); encode_ch (enc, '}');
		463	}
		464
		465	// encode objects, arrays and special \0=false and \1=true values.
		466	static void
		467	encode_rv (enc_t enc, SV sv)
		468	{
		469	svtype svt;
		470
		471	SvGETMAGIC (sv);
		472	svt = SvTYPE (sv);
		473
		474	if (svt == SVt_PVHV)
		475	encode_hv (enc, (HV *)sv);
		476	else if (svt == SVt_PVAV)
		477	encode_av (enc, (AV *)sv);
		478	else if (svt < SVt_PVAV)
		479	{
		480	if (SvNIOK (sv) && SvIV (sv) == 0)
		481	encode_str (enc, "false", 5, 0);
		482	else if (SvNIOK (sv) && SvIV (sv) == 1)
		483	encode_str (enc, "true", 4, 0);
		484	else
		485	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
		486	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		487	}
		488	else
		489	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
		490	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
370	}	491	}
371		492
372	static void	493	static void
373	encode_sv (enc_t enc, SV sv)	494	encode_sv (enc_t enc, SV sv)
374	{	495	{
…		…
382	encode_str (enc, str, len, SvUTF8 (sv));	503	encode_str (enc, str, len, SvUTF8 (sv));
383	encode_ch (enc, '"');	504	encode_ch (enc, '"');
384	}	505	}
385	else if (SvNOKp (sv))	506	else if (SvNOKp (sv))
386	{	507	{
		508	// trust that perl will do the right thing w.r.t. JSON syntax.
387	need (enc, NV_DIG + 32);	509	need (enc, NV_DIG + 32);
388	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	510	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
389	enc->cur += strlen (enc->cur);	511	enc->cur += strlen (enc->cur);
390	}	512	}
391	else if (SvIOKp (sv))	513	else if (SvIOKp (sv))
392	{	514	{
393	need (enc, 64);	515	// we assume we can always read an IV as a UV
		516	if (SvUV (sv) & ~(UV)0x7fff)
		517	{
		518	// large integer, use the (rather slow) snprintf way.
		519	need (enc, sizeof (UV) * 3);
394	enc->cur +=	520	enc->cur +=
395	SvIsUV(sv)	521	SvIsUV(sv)
396	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	522	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
397	: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));	523	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
		524	}
		525	else
		526	{
		527	// optimise the "small number case"
		528	// code will likely be branchless and use only a single multiplication
		529	I32 i = SvIV (sv);
		530	U32 u;
		531	char digit, nz = 0;
		532
		533	need (enc, 6);
		534
		535	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
		536	u = i < 0 ? -i : i;
		537
		538	// convert to 4.28 fixed-point representation
		539	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
		540
		541	// now output digit by digit, each time masking out the integer part
		542	// and multiplying by 5 while moving the decimal point one to the right,
		543	// resulting in a net multiplication by 10.
		544	// we always write the digit to memory but conditionally increment
		545	// the pointer, to ease the usage of conditional move instructions.
		546	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffff) 5;
		547	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffff) 5;
		548	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffff) 5;
		549	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffff) 5;
		550	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
		551	}
398	}	552	}
399	else if (SvROK (sv))	553	else if (SvROK (sv))
400	{	554	encode_rv (enc, SvRV (sv));
401	if (!--enc->max_recurse)
402	croak ("data structure too deep (hit recursion limit)");
403
404	sv = SvRV (sv);
405
406	switch (SvTYPE (sv))
407	{
408	case SVt_PVAV: encode_av (enc, (AV *)sv); break;
409	case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
410
411	default:
412	croak ("JSON can only represent references to arrays or hashes");
413	}
414	}
415	else if (!SvOK (sv))	555	else if (!SvOK (sv))
416	encode_str (enc, "null", 4, 0);	556	encode_str (enc, "null", 4, 0);
417	else	557	else
418	croak ("encountered perl type that JSON cannot handle");	558	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
		559	SvPV_nolen (sv), SvFLAGS (sv));
419	}	560	}
420		561
421	static SV *	562	static SV *
422	encode_json (SV *scalar, UV flags)	563	encode_json (SV *scalar, U32 flags)
423	{	564	{
		565	enc_t enc;
		566
424	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	567	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
425	croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)");	568	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
426		569
427	enc_t enc;
428	enc.flags = flags;	570	enc.flags = flags;
429	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	571	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
430	enc.cur = SvPVX (enc.sv);	572	enc.cur = SvPVX (enc.sv);
431	enc.end = SvEND (enc.sv);	573	enc.end = SvEND (enc.sv);
432	enc.max_recurse = 0;
433	enc.indent = 0;	574	enc.indent = 0;
		575	enc.maxdepth = DEC_DEPTH (flags);
434		576
435	SvPOK_only (enc.sv);	577	SvPOK_only (enc.sv);
436	encode_sv (&enc, scalar);	578	encode_sv (&enc, scalar);
437		579
		580	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
		581	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
		582
438	if (!(flags & (F_ASCII \| F_UTF8)))	583	if (!(flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
439	SvUTF8_on (enc.sv);	584	SvUTF8_on (enc.sv);
440
441	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
442		585
443	if (enc.flags & F_SHRINK)	586	if (enc.flags & F_SHRINK)
444	shrink (enc.sv);	587	shrink (enc.sv);
445		588
446	return enc.sv;	589	return enc.sv;
447	}	590	}
448		591
449	/////////////////////////////////////////////////////////////////////////////	592	/////////////////////////////////////////////////////////////////////////////
		593	// decoder
450		594
451	#define WS \	595	// structure used for decoding JSON
		596	typedef struct
		597	{
		598	char *cur; // current parser pointer
		599	char *end; // end of input string
		600	const char *err; // parse error, if != 0
		601	U32 flags; // F_*
		602	U32 depth; // recursion depth
		603	U32 maxdepth; // recursion depth limit
		604	} dec_t;
		605
		606	inline void
		607	decode_ws (dec_t *dec)
		608	{
452	for (;;) \	609	for (;;)
453	{ \	610	{
454	char ch = *dec->cur; \	611	char ch = *dec->cur;
		612
455	if (ch > 0x20 \	613	if (ch > 0x20
456	\|\| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \	614	\|\| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
457	break; \	615	break;
		616
458	++dec->cur; \	617	++dec->cur;
459	}	618	}
		619	}
460		620
461	#define ERR(reason) SB dec->err = reason; goto fail; SE	621	#define ERR(reason) SB dec->err = reason; goto fail; SE
		622
462	#define EXPECT_CH(ch) SB \	623	#define EXPECT_CH(ch) SB \
463	if (*dec->cur != ch) \	624	if (*dec->cur != ch) \
464	ERR (# ch " expected"); \	625	ERR (# ch " expected"); \
465	++dec->cur; \	626	++dec->cur; \
466	SE	627	SE
467		628
		629	#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
		630	#define DEC_DEC_DEPTH --dec->depth
		631
468	static SV decode_sv (dec_t dec);	632	static SV decode_sv (dec_t dec);
469		633
470	static signed char decode_hexdigit[256];	634	static signed char decode_hexdigit[256];
471		635
472	static UV	636	static UV
473	decode_4hex (dec_t *dec)	637	decode_4hex (dec_t *dec)
474	{	638	{
475	signed char d1, d2, d3, d4;	639	signed char d1, d2, d3, d4;
		640	unsigned char cur = (unsigned char )dec->cur;
476		641
477	d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];	642	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
478	if (d1 < 0) ERR ("four hexadecimal digits expected");	643	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
479	d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];	644	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
480	if (d2 < 0) ERR ("four hexadecimal digits expected");	645	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
481	d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
482	if (d3 < 0) ERR ("four hexadecimal digits expected");
483	d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
484	if (d4 < 0) ERR ("four hexadecimal digits expected");
485		646
486	dec->cur += 4;	647	dec->cur += 4;
487		648
488	return ((UV)d1) << 12	649	return ((UV)d1) << 12
489	\| ((UV)d2) << 8	650	\| ((UV)d2) << 8
…		…
492		653
493	fail:	654	fail:
494	return (UV)-1;	655	return (UV)-1;
495	}	656	}
496		657
497	#define APPEND_GROW(n) SB \
498	if (cur + (n) >= end) \
499	{ \
500	STRLEN ofs = cur - SvPVX (sv); \
501	SvGROW (sv, ofs + (n) + 1); \
502	cur = SvPVX (sv) + ofs; \
503	end = SvEND (sv); \
504	} \
505	SE
506
507	#define APPEND_CH(ch) SB \
508	APPEND_GROW (1); \
509	*cur++ = (ch); \
510	SE
511
512	static SV *	658	static SV *
513	decode_str (dec_t *dec)	659	decode_str (dec_t *dec)
514	{	660	{
515	SV *sv = NEWSV (0,2);	661	SV *sv = 0;
516	int utf8 = 0;	662	int utf8 = 0;
517	char *cur = SvPVX (sv);	663	char *dec_cur = dec->cur;
518	char *end = SvEND (sv);
519		664
520	for (;;)	665	do
521	{	666	{
522	unsigned char ch = (unsigned char )dec->cur;	667	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
		668	char *cur = buf;
523		669
524	if (ch == '"')	670	do
525	break;
526	else if (ch == '\\')
527	{	671	{
528	switch (*++dec->cur)	672	unsigned char ch = (unsigned char )dec_cur++;
		673
		674	if (expect_false (ch == '"'))
529	{	675	{
530	case '\\':	676	--dec_cur;
531	case '/':	677	break;
532	case '"': APPEND_CH (*dec->cur++); break;	678	}
533		679	else if (expect_false (ch == '\\'))
534	case 'b': APPEND_CH ('\010'); ++dec->cur; break;	680	{
535	case 't': APPEND_CH ('\011'); ++dec->cur; break;	681	switch (*dec_cur)
536	case 'n': APPEND_CH ('\012'); ++dec->cur; break;
537	case 'f': APPEND_CH ('\014'); ++dec->cur; break;
538	case 'r': APPEND_CH ('\015'); ++dec->cur; break;
539
540	case 'u':
541	{	682	{
542	UV lo, hi;	683	case '\\':
543	++dec->cur;	684	case '/':
		685	case '"': cur++ = dec_cur++; break;
544		686
545	hi = decode_4hex (dec);	687	case 'b': ++dec_cur; *cur++ = '\010'; break;
546	if (hi == (UV)-1)	688	case 't': ++dec_cur; *cur++ = '\011'; break;
547	goto fail;	689	case 'n': ++dec_cur; *cur++ = '\012'; break;
		690	case 'f': ++dec_cur; *cur++ = '\014'; break;
		691	case 'r': ++dec_cur; *cur++ = '\015'; break;
548		692
549	// possibly a surrogate pair	693	case 'u':
550	if (hi >= 0xd800 && hi < 0xdc00)
551	{	694	{
552	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')	695	UV lo, hi;
553	ERR ("missing low surrogate character in surrogate pair");	696	++dec_cur;
554		697
555	dec->cur += 2;	698	dec->cur = dec_cur;
556
557	lo = decode_4hex (dec);	699	hi = decode_4hex (dec);
		700	dec_cur = dec->cur;
558	if (lo == (UV)-1)	701	if (hi == (UV)-1)
559	goto fail;	702	goto fail;
560		703
		704	// possibly a surrogate pair
		705	if (hi >= 0xd800)
		706	if (hi < 0xdc00)
		707	{
		708	if (dec_cur [0] != '\\' \|\| dec_cur [1] != 'u')
		709	ERR ("missing low surrogate character in surrogate pair");
		710
		711	dec_cur += 2;
		712
		713	dec->cur = dec_cur;
		714	lo = decode_4hex (dec);
		715	dec_cur = dec->cur;
		716	if (lo == (UV)-1)
		717	goto fail;
		718
561	if (lo < 0xdc00 \|\| lo >= 0xe000)	719	if (lo < 0xdc00 \|\| lo >= 0xe000)
562	ERR ("surrogate pair expected");	720	ERR ("surrogate pair expected");
563		721
564	hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;	722	hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
		723	}
		724	else if (hi < 0xe000)
		725	ERR ("missing high surrogate character in surrogate pair");
		726
		727	if (hi >= 0x80)
		728	{
		729	utf8 = 1;
		730
		731	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
		732	}
		733	else
		734	*cur++ = hi;
565	}	735	}
566	else if (hi >= 0xdc00 && hi < 0xe000)
567	ERR ("missing high surrogate character in surrogate pair");
568
569	if (hi >= 0x80)
570	{	736	break;
571	utf8 = 1;
572		737
573	APPEND_GROW (4); // at most 4 bytes for 21 bits
574	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
575	}
576	else	738	default:
577	APPEND_CH (hi);	739	--dec_cur;
		740	ERR ("illegal backslash escape sequence in string");
578	}	741	}
579	break;	742	}
		743	else if (expect_true (ch >= 0x20 && ch <= 0x7f))
		744	*cur++ = ch;
		745	else if (ch >= 0x80)
		746	{
		747	STRLEN clen;
		748	UV uch;
580		749
581	default:
582	--dec->cur;	750	--dec_cur;
583	ERR ("illegal backslash escape sequence in string");	751
		752	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
		753	if (clen == (STRLEN)-1)
		754	ERR ("malformed UTF-8 character in JSON string");
		755
		756	do
		757	cur++ = dec_cur++;
		758	while (--clen);
		759
		760	utf8 = 1;
		761	}
		762	else
		763	{
		764	--dec_cur;
		765
		766	if (!ch)
		767	ERR ("unexpected end of string while parsing JSON string");
		768	else
		769	ERR ("invalid character encountered while parsing JSON string");
584	}	770	}
585	}	771	}
586	else if (ch >= 0x20 && ch <= 0x7f)	772	while (cur < buf + SHORT_STRING_LEN);
587	APPEND_CH (*dec->cur++);	773
588	else if (ch >= 0x80)
589	{	774	{
590	STRLEN clen;	775	STRLEN len = cur - buf;
591	UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
592	if (clen == (STRLEN)-1)
593	ERR ("malformed UTF-8 character in JSON string");
594		776
595	APPEND_GROW (clen);	777	if (sv)
596	do
597	{	778	{
598	cur++ = dec->cur++;	779	SvGROW (sv, SvCUR (sv) + len + 1);
		780	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
		781	SvCUR_set (sv, SvCUR (sv) + len);
599	}	782	}
600	while (--clen);
601
602	utf8 = 1;
603	}
604	else if (dec->cur == dec->end)
605	ERR ("unexpected end of string while parsing json string");
606	else	783	else
607	ERR ("invalid character encountered");	784	sv = newSVpvn (buf, len);
608	}	785	}
		786	}
		787	while (*dec_cur != '"');
609		788
610	++dec->cur;	789	++dec_cur;
611		790
612	SvCUR_set (sv, cur - SvPVX (sv));	791	if (sv)
613		792	{
614	SvPOK_only (sv);	793	SvPOK_only (sv);
615	*SvEND (sv) = 0;	794	*SvEND (sv) = 0;
616		795
617	if (utf8)	796	if (utf8)
618	SvUTF8_on (sv);	797	SvUTF8_on (sv);
		798	}
		799	else
		800	sv = newSVpvn ("", 0);
619		801
620	if (dec->flags & F_SHRINK)	802	dec->cur = dec_cur;
621	shrink (sv);
622
623	return sv;	803	return sv;
624		804
625	fail:	805	fail:
626	SvREFCNT_dec (sv);	806	dec->cur = dec_cur;
627	return 0;	807	return 0;
628	}	808	}
629		809
630	static SV *	810	static SV *
631	decode_num (dec_t *dec)	811	decode_num (dec_t *dec)
…		…
689	is_nv = 1;	869	is_nv = 1;
690	}	870	}
691		871
692	if (!is_nv)	872	if (!is_nv)
693	{	873	{
694	UV uv;	874	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
695	int numtype = grok_number (start, dec->cur - start, &uv);	875	if (*start == '-')
696	if (numtype & IS_NUMBER_IN_UV)	876	switch (dec->cur - start)
697	if (numtype & IS_NUMBER_NEG)
698	{	877	{
699	if (uv < (UV)IV_MIN)	878	case 2: return newSViv (-( start [1] - '0' * 1));
700	return newSViv (-(IV)uv);	879	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
		880	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
		881	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
701	}	882	}
		883	else
		884	switch (dec->cur - start)
		885	{
		886	case 1: return newSViv ( start [0] - '0' * 1);
		887	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
		888	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
		889	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		890	}
		891
		892	{
		893	UV uv;
		894	int numtype = grok_number (start, dec->cur - start, &uv);
		895	if (numtype & IS_NUMBER_IN_UV)
		896	if (numtype & IS_NUMBER_NEG)
		897	{
		898	if (uv < (UV)IV_MIN)
		899	return newSViv (-(IV)uv);
		900	}
702	else	901	else
703	return newSVuv (uv);	902	return newSVuv (uv);
		903
		904	// here would likely be the place for bigint support
704	}	905	}
		906	}
705		907
		908	// if we ever support bigint or bigfloat, this is the place for bigfloat
706	return newSVnv (Atof (start));	909	return newSVnv (Atof (start));
707		910
708	fail:	911	fail:
709	return 0;	912	return 0;
710	}	913	}
…		…
712	static SV *	915	static SV *
713	decode_av (dec_t *dec)	916	decode_av (dec_t *dec)
714	{	917	{
715	AV *av = newAV ();	918	AV *av = newAV ();
716		919
717	WS;	920	DEC_INC_DEPTH;
		921	decode_ws (dec);
		922
718	if (*dec->cur == ']')	923	if (*dec->cur == ']')
719	++dec->cur;	924	++dec->cur;
720	else	925	else
721	for (;;)	926	for (;;)
722	{	927	{
…		…
726	if (!value)	931	if (!value)
727	goto fail;	932	goto fail;
728		933
729	av_push (av, value);	934	av_push (av, value);
730		935
731	WS;	936	decode_ws (dec);
732		937
733	if (*dec->cur == ']')	938	if (*dec->cur == ']')
734	{	939	{
735	++dec->cur;	940	++dec->cur;
736	break;	941	break;
…		…
740	ERR (", or ] expected while parsing array");	945	ERR (", or ] expected while parsing array");
741		946
742	++dec->cur;	947	++dec->cur;
743	}	948	}
744		949
		950	DEC_DEC_DEPTH;
745	return newRV_noinc ((SV *)av);	951	return newRV_noinc ((SV *)av);
746		952
747	fail:	953	fail:
748	SvREFCNT_dec (av);	954	SvREFCNT_dec (av);
		955	DEC_DEC_DEPTH;
749	return 0;	956	return 0;
750	}	957	}
751		958
752	static SV *	959	static SV *
753	decode_hv (dec_t *dec)	960	decode_hv (dec_t *dec)
754	{	961	{
755	HV *hv = newHV ();	962	HV *hv = newHV ();
756		963
757	WS;	964	DEC_INC_DEPTH;
		965	decode_ws (dec);
		966
758	if (*dec->cur == '}')	967	if (*dec->cur == '}')
759	++dec->cur;	968	++dec->cur;
760	else	969	else
761	for (;;)	970	for (;;)
762	{	971	{
763	SV key, value;	972	SV key, value;
764		973
765	WS; EXPECT_CH ('"');	974	decode_ws (dec); EXPECT_CH ('"');
766		975
767	key = decode_str (dec);	976	key = decode_str (dec);
768	if (!key)	977	if (!key)
769	goto fail;	978	goto fail;
770		979
771	WS; EXPECT_CH (':');	980	decode_ws (dec); EXPECT_CH (':');
772		981
773	value = decode_sv (dec);	982	value = decode_sv (dec);
774	if (!value)	983	if (!value)
775	{	984	{
776	SvREFCNT_dec (key);	985	SvREFCNT_dec (key);
777	goto fail;	986	goto fail;
778	}	987	}
779		988
780	//TODO: optimise
781	hv_store_ent (hv, key, value, 0);	989	hv_store_ent (hv, key, value, 0);
		990	SvREFCNT_dec (key);
782		991
783	WS;	992	decode_ws (dec);
784		993
785	if (*dec->cur == '}')	994	if (*dec->cur == '}')
786	{	995	{
787	++dec->cur;	996	++dec->cur;
788	break;	997	break;
…		…
792	ERR (", or } expected while parsing object/hash");	1001	ERR (", or } expected while parsing object/hash");
793		1002
794	++dec->cur;	1003	++dec->cur;
795	}	1004	}
796		1005
		1006	DEC_DEC_DEPTH;
797	return newRV_noinc ((SV *)hv);	1007	return newRV_noinc ((SV *)hv);
798		1008
799	fail:	1009	fail:
800	SvREFCNT_dec (hv);	1010	SvREFCNT_dec (hv);
		1011	DEC_DEC_DEPTH;
801	return 0;	1012	return 0;
802	}	1013	}
803		1014
804	static SV *	1015	static SV *
805	decode_sv (dec_t *dec)	1016	decode_sv (dec_t *dec)
806	{	1017	{
807	WS;	1018	decode_ws (dec);
		1019
		1020	// the beauty of JSON: you need exactly one character lookahead
		1021	// to parse anything.
808	switch (*dec->cur)	1022	switch (*dec->cur)
809	{	1023	{
810	case '"': ++dec->cur; return decode_str (dec);	1024	case '"': ++dec->cur; return decode_str (dec);
811	case '[': ++dec->cur; return decode_av (dec);	1025	case '[': ++dec->cur; return decode_av (dec);
812	case '{': ++dec->cur; return decode_hv (dec);	1026	case '{': ++dec->cur; return decode_hv (dec);
…		…
818		1032
819	case 't':	1033	case 't':
820	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1034	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
821	{	1035	{
822	dec->cur += 4;	1036	dec->cur += 4;
823	return newSViv (1);	1037	return SvREFCNT_inc (json_true);
824	}	1038	}
825	else	1039	else
826	ERR ("'true' expected");	1040	ERR ("'true' expected");
827		1041
828	break;	1042	break;
829		1043
830	case 'f':	1044	case 'f':
831	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1045	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
832	{	1046	{
833	dec->cur += 5;	1047	dec->cur += 5;
834	return newSViv (0);	1048	return SvREFCNT_inc (json_false);
835	}	1049	}
836	else	1050	else
837	ERR ("'false' expected");	1051	ERR ("'false' expected");
838		1052
839	break;	1053	break;
…		…
848	ERR ("'null' expected");	1062	ERR ("'null' expected");
849		1063
850	break;	1064	break;
851		1065
852	default:	1066	default:
853	ERR ("malformed json string, neither array, object, number, string or atom");	1067	ERR ("malformed JSON string, neither array, object, number, string or atom");
854	break;	1068	break;
855	}	1069	}
856		1070
857	fail:	1071	fail:
858	return 0;	1072	return 0;
859	}	1073	}
860		1074
861	static SV *	1075	static SV *
862	decode_json (SV *string, UV flags)	1076	decode_json (SV string, U32 flags, UV offset_return)
863	{	1077	{
		1078	dec_t dec;
		1079	UV offset;
864	SV *sv;	1080	SV *sv;
		1081
		1082	SvGETMAGIC (string);
		1083	SvUPGRADE (string, SVt_PV);
865		1084
866	if (flags & F_UTF8)	1085	if (flags & F_UTF8)
867	sv_utf8_downgrade (string, 0);	1086	sv_utf8_downgrade (string, 0);
868	else	1087	else
869	sv_utf8_upgrade (string);	1088	sv_utf8_upgrade (string);
870		1089
871	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1090	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
872		1091
873	dec_t dec;
874	dec.flags = flags;	1092	dec.flags = flags;
875	dec.cur = SvPVX (string);	1093	dec.cur = SvPVX (string);
876	dec.end = SvEND (string);	1094	dec.end = SvEND (string);
877	dec.err = 0;	1095	dec.err = 0;
		1096	dec.depth = 0;
		1097	dec.maxdepth = DEC_DEPTH (dec.flags);
878		1098
		1099	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
879	sv = decode_sv (&dec);	1100	sv = decode_sv (&dec);
880		1101
		1102	if (!(offset_return \|\| !sv))
		1103	{
		1104	// check for trailing garbage
		1105	decode_ws (&dec);
		1106
		1107	if (*dec.cur)
		1108	{
		1109	dec.err = "garbage after JSON object";
		1110	SvREFCNT_dec (sv);
		1111	sv = 0;
		1112	}
		1113	}
		1114
		1115	if (offset_return \|\| !sv)
		1116	{
		1117	offset = dec.flags & F_UTF8
		1118	? dec.cur - SvPVX (string)
		1119	: utf8_distance (dec.cur, SvPVX (string));
		1120
		1121	if (offset_return)
		1122	*offset_return = offset;
		1123	}
		1124
881	if (!sv)	1125	if (!sv)
882	{	1126	{
883	IV offset = dec.flags & F_UTF8
884	? dec.cur - SvPVX (string)
885	: utf8_distance (dec.cur, SvPVX (string));
886	SV *uni = sv_newmortal ();	1127	SV *uni = sv_newmortal ();
887		1128
888	// horrible hack to silence warning inside pv_uni_display	1129	// horrible hack to silence warning inside pv_uni_display
889	COP cop = *PL_curcop;	1130	COP cop = *PL_curcop;
890	cop.cop_warnings = pWARN_NONE;	1131	cop.cop_warnings = pWARN_NONE;
…		…
892	SAVEVPTR (PL_curcop);	1133	SAVEVPTR (PL_curcop);
893	PL_curcop = &cop;	1134	PL_curcop = &cop;
894	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1135	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
895	LEAVE;	1136	LEAVE;
896		1137
897	croak ("%s, at character offset %d (%s)",	1138	croak ("%s, at character offset %d [\"%s\"]",
898	dec.err,	1139	dec.err,
899	(int)offset,	1140	(int)offset,
900	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1141	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
901	}	1142	}
902		1143
903	sv = sv_2mortal (sv);	1144	sv = sv_2mortal (sv);
904		1145
905	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))	1146	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
906	croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)");	1147	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
907		1148
908	return sv;	1149	return sv;
909	}	1150	}
910		1151
		1152	/////////////////////////////////////////////////////////////////////////////
		1153	// XS interface functions
		1154
911	MODULE = JSON::XS PACKAGE = JSON::XS	1155	MODULE = JSON::XS PACKAGE = JSON::XS
912		1156
913	BOOT:	1157	BOOT:
914	{	1158	{
915	int i;	1159	int i;
916		1160
917	memset (decode_hexdigit, 0xff, 256);
918	for (i = 10; i--; )	1161	for (i = 0; i < 256; ++i)
919	decode_hexdigit ['0' + i] = i;	1162	decode_hexdigit [i] =
920		1163	i >= '0' && i <= '9' ? i - '0'
921	for (i = 7; i--; )	1164	: i >= 'a' && i <= 'f' ? i - 'a' + 10
922	{	1165	: i >= 'A' && i <= 'F' ? i - 'A' + 10
923	decode_hexdigit ['a' + i] = 10 + i;	1166	: -1;
924	decode_hexdigit ['A' + i] = 10 + i;
925	}
926		1167
927	json_stash = gv_stashpv ("JSON::XS", 1);	1168	json_stash = gv_stashpv ("JSON::XS", 1);
		1169
		1170	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
		1171	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
928	}	1172	}
929		1173
930	PROTOTYPES: DISABLE	1174	PROTOTYPES: DISABLE
931		1175
932	SV new (char dummy)	1176	SV new (char dummy)
…		…
936	RETVAL	1180	RETVAL
937		1181
938	SV ascii (SV self, int enable = 1)	1182	SV ascii (SV self, int enable = 1)
939	ALIAS:	1183	ALIAS:
940	ascii = F_ASCII	1184	ascii = F_ASCII
		1185	latin1 = F_LATIN1
941	utf8 = F_UTF8	1186	utf8 = F_UTF8
942	indent = F_INDENT	1187	indent = F_INDENT
943	canonical = F_CANONICAL	1188	canonical = F_CANONICAL
944	space_before = F_SPACE_BEFORE	1189	space_before = F_SPACE_BEFORE
945	space_after = F_SPACE_AFTER	1190	space_after = F_SPACE_AFTER
946	json_rpc = F_JSON_RPC
947	pretty = F_PRETTY	1191	pretty = F_PRETTY
948	allow_nonref = F_ALLOW_NONREF	1192	allow_nonref = F_ALLOW_NONREF
949	shrink = F_SHRINK	1193	shrink = F_SHRINK
950	CODE:	1194	CODE:
951	{	1195	{
…		…
958	RETVAL = newSVsv (self);	1202	RETVAL = newSVsv (self);
959	}	1203	}
960	OUTPUT:	1204	OUTPUT:
961	RETVAL	1205	RETVAL
962		1206
		1207	SV max_depth (SV self, UV max_depth = 0x80000000UL)
		1208	CODE:
		1209	{
		1210	UV *uv = SvJSON (self);
		1211	UV log2 = 0;
		1212
		1213	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
		1214
		1215	while ((1UL << log2) < max_depth)
		1216	++log2;
		1217
		1218	uv = uv & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
		1219
		1220	RETVAL = newSVsv (self);
		1221	}
		1222	OUTPUT:
		1223	RETVAL
		1224
963	void encode (SV self, SV scalar)	1225	void encode (SV self, SV scalar)
964	PPCODE:	1226	PPCODE:
965	XPUSHs (encode_json (scalar, *SvJSON (self)));	1227	XPUSHs (encode_json (scalar, *SvJSON (self)));
966		1228
967	void decode (SV self, SV jsonstr)	1229	void decode (SV self, SV jsonstr)
968	PPCODE:	1230	PPCODE:
969	XPUSHs (decode_json (jsonstr, *SvJSON (self)));	1231	XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
		1232
		1233	void decode_prefix (SV self, SV jsonstr)
		1234	PPCODE:
		1235	{
		1236	UV offset;
		1237	EXTEND (SP, 2);
		1238	PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
		1239	PUSHs (sv_2mortal (newSVuv (offset)));
		1240	}
970		1241
971	PROTOTYPES: ENABLE	1242	PROTOTYPES: ENABLE
972		1243
973	void to_json (SV *scalar)	1244	void to_json (SV *scalar)
		1245	ALIAS:
		1246	objToJson = 0
974	PPCODE:	1247	PPCODE:
975	XPUSHs (encode_json (scalar, F_UTF8));	1248	XPUSHs (encode_json (scalar, F_DEFAULT \| F_UTF8));
976		1249
977	void from_json (SV *jsonstr)	1250	void from_json (SV *jsonstr)
		1251	ALIAS:
		1252	jsonToObj = 0
978	PPCODE:	1253	PPCODE:
979	XPUSHs (decode_json (jsonstr, F_UTF8));	1254	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8, 0));
980		1255

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.8 by root, Fri Mar 23 16:13:59 2007 UTC vs. Revision 1.43 by root, Sat Jun 23 23:49:29 2007 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.8 by root, Fri Mar 23 16:13:59 2007 UTC vs.
Revision 1.43 by root, Sat Jun 23 23:49:29 2007 UTC