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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.6 by root, Fri Mar 23 15:10:55 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 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_ALLOW_BLESSED 0x00000400UL
		30	#define F_CONV_BLESSED 0x00000800UL // NYI
		31	#define F_MAXDEPTH 0xf8000000UL
		32	#define S_MAXDEPTH 27
		33	#define F_MAXSIZE 0x01f00000UL
		34	#define S_MAXSIZE 20
		35
		36	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
		37	#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
18		38
19	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	39	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
20	#define F_DEFAULT 0	40	#define F_DEFAULT (9UL << S_MAXDEPTH)
21		41
22	#define INIT_SIZE 32 // initial scalar size to be allocated	42	#define INIT_SIZE 32 // initial scalar size to be allocated
		43	#define INDENT_STEP 3 // spaces per indentation level
		44
		45	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
23		46
24	#define SB do {	47	#define SB do {
25	#define SE } while (0)	48	#define SE } while (0)
26		49
27	static HV *json_stash;	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
		61	static HV json_stash, json_boolean_stash; // JSON::XS::
		62	static SV json_true, json_false;
		63
		64	/////////////////////////////////////////////////////////////////////////////
		65	// utility functions
		66
		67	static UV *
		68	SvJSON (SV *sv)
		69	{
		70	if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
		71	croak ("object is not of type JSON::XS");
		72
		73	return &SvUVX (SvRV (sv));
		74	}
		75
		76	static void
		77	shrink (SV *sv)
		78	{
		79	sv_utf8_downgrade (sv, 1);
		80	if (SvLEN (sv) > SvCUR (sv) + 1)
		81	{
		82	#ifdef SvPV_shrink_to_cur
		83	SvPV_shrink_to_cur (sv);
		84	#elif defined (SvPV_renew)
		85	SvPV_renew (sv, SvCUR (sv) + 1);
		86	#endif
		87	}
		88	}
		89
		90	// decode an utf-8 character and return it, or (UV)-1 in
		91	// case of an error.
		92	// we special-case "safe" characters from U+80 .. U+7FF,
		93	// but use the very good perl function to parse anything else.
		94	// note that we never call this function for a ascii codepoints
		95	inline UV
		96	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
		97	{
		98	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))
		99	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
		100	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
		101	{
		102	*clen = 2;
		103	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
		104	}
		105	else
		106	{
		107	*clen = (STRLEN)-1;
		108	return (UV)-1;
		109	}
		110	}
		111
		112	/////////////////////////////////////////////////////////////////////////////
		113	// encoder
28		114
29	// structure used for encoding JSON	115	// structure used for encoding JSON
30	typedef struct	116	typedef struct
31	{	117	{
32	char *cur;	118	char *cur; // SvPVX (sv) + current output position
33	STRLEN len; // SvLEN (sv)
34	char *end; // SvEND (sv)	119	char *end; // SvEND (sv)
35	SV *sv;	120	SV *sv; // result scalar
36	UV flags;	121	U32 flags; // F_*
37	int max_recurse;	122	U32 indent; // indentation level
38	int indent;	123	U32 maxdepth; // max. indentation/recursion level
39	} enc_t;	124	} enc_t;
40		125
41	// structure used for decoding JSON	126	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	/////////////////////////////////////////////////////////////////////////////
60
61	static void
62	need (enc_t *enc, STRLEN len)	127	need (enc_t *enc, STRLEN len)
63	{	128	{
64	if (enc->cur + len >= enc->end)	129	if (expect_false (enc->cur + len >= enc->end))
65	{	130	{
66	STRLEN cur = enc->cur - SvPVX (enc->sv);	131	STRLEN cur = enc->cur - SvPVX (enc->sv);
67	SvGROW (enc->sv, cur + len + 1);	132	SvGROW (enc->sv, cur + len + 1);
68	enc->cur = SvPVX (enc->sv) + cur;	133	enc->cur = SvPVX (enc->sv) + cur;
69	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);	134	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
70	}	135	}
71	}	136	}
72		137
73	static void	138	inline void
74	encode_ch (enc_t *enc, char ch)	139	encode_ch (enc_t *enc, char ch)
75	{	140	{
76	need (enc, 1);	141	need (enc, 1);
77	*enc->cur++ = ch;	142	*enc->cur++ = ch;
78	}	143	}
…		…
86		151
87	while (str < end)	152	while (str < end)
88	{	153	{
89	unsigned char ch = (unsigned char )str;	154	unsigned char ch = (unsigned char )str;
90		155
91	if (ch >= 0x20 && ch < 0x80) // most common case	156	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
92	{	157	{
93	if (ch == '"') // but with slow exceptions	158	if (expect_false (ch == '"')) // but with slow exceptions
94	{	159	{
95	need (enc, len += 1);	160	need (enc, len += 1);
96	*enc->cur++ = '\\';	161	*enc->cur++ = '\\';
97	*enc->cur++ = '"';	162	*enc->cur++ = '"';
98	}	163	}
99	else if (ch == '\\')	164	else if (expect_false (ch == '\\'))
100	{	165	{
101	need (enc, len += 1);	166	need (enc, len += 1);
102	*enc->cur++ = '\\';	167	*enc->cur++ = '\\';
103	*enc->cur++ = '\\';	168	*enc->cur++ = '\\';
104	}	169	}
…		…
122	STRLEN clen;	187	STRLEN clen;
123	UV uch;	188	UV uch;
124		189
125	if (is_utf8)	190	if (is_utf8)
126	{	191	{
127	uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);	192	uch = decode_utf8 (str, end - str, &clen);
128	if (clen == (STRLEN)-1)	193	if (clen == (STRLEN)-1)
129	croak ("malformed UTF-8 character in string, cannot convert to JSON");	194	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
130	}	195	}
131	else	196	else
132	{	197	{
133	uch = ch;	198	uch = ch;
134	clen = 1;	199	clen = 1;
135	}	200	}
136		201
137	if (uch < 0x80 \|\| enc->flags & F_ASCII)	202	if (uch > 0x10FFFFUL)
		203	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
		204
		205	if (uch < 0x80 \|\| enc->flags & F_ASCII \|\| (enc->flags & F_LATIN1 && uch > 0xFF))
138	{	206	{
139	if (uch > 0xFFFFUL)	207	if (uch > 0xFFFFUL)
140	{	208	{
141	need (enc, len += 11);	209	need (enc, len += 11);
142	sprintf (enc->cur, "\\u%04x\\u%04x",	210	sprintf (enc->cur, "\\u%04x\\u%04x",
143	(uch - 0x10000) / 0x400 + 0xD800,	211	(int)((uch - 0x10000) / 0x400 + 0xD800),
144	(uch - 0x10000) % 0x400 + 0xDC00);	212	(int)((uch - 0x10000) % 0x400 + 0xDC00));
145	enc->cur += 12;	213	enc->cur += 12;
146	}	214	}
147	else	215	else
148	{	216	{
149	static char hexdigit [16] = "0123456789abcdef";	217	static char hexdigit [16] = "0123456789abcdef";
…		…
156	*enc->cur++ = hexdigit [(uch >> 0) & 15];	224	*enc->cur++ = hexdigit [(uch >> 0) & 15];
157	}	225	}
158		226
159	str += clen;	227	str += clen;
160	}	228	}
		229	else if (enc->flags & F_LATIN1)
		230	{
		231	*enc->cur++ = uch;
		232	str += clen;
		233	}
161	else if (is_utf8)	234	else if (is_utf8)
162	{	235	{
163	need (enc, len += clen);	236	need (enc, len += clen);
164	do	237	do
165	{	238	{
…		…
167	}	240	}
168	while (--clen);	241	while (--clen);
169	}	242	}
170	else	243	else
171	{	244	{
172	need (enc, 10); // never more than 11 bytes needed	245	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
173	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	246	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
174	++str;	247	++str;
175	}	248	}
176	}	249	}
177	}	250	}
…		…
179		252
180	--len;	253	--len;
181	}	254	}
182	}	255	}
183		256
184	#define INDENT SB \	257	inline void
		258	encode_indent (enc_t *enc)
		259	{
185	if (enc->flags & F_INDENT) \	260	if (enc->flags & F_INDENT)
186	{ \	261	{
187	int i_; \	262	int spaces = enc->indent * INDENT_STEP;
188	need (enc, enc->indent); \
189	for (i_ = enc->indent * 3; i_--; )\
190	encode_ch (enc, ' '); \
191	} \
192	SE
193		263
194	#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE	264	need (enc, spaces);
195	#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE	265	memset (enc->cur, ' ', spaces);
196	#define COMMA SB \	266	enc->cur += spaces;
		267	}
		268	}
		269
		270	inline void
		271	encode_space (enc_t *enc)
		272	{
		273	need (enc, 1);
197	encode_ch (enc, ','); \	274	encode_ch (enc, ' ');
		275	}
		276
		277	inline void
		278	encode_nl (enc_t *enc)
		279	{
198	if (enc->flags & F_INDENT) \	280	if (enc->flags & F_INDENT)
199	NL; \	281	{
		282	need (enc, 1);
		283	encode_ch (enc, '\n');
		284	}
		285	}
		286
		287	inline void
		288	encode_comma (enc_t *enc)
		289	{
		290	encode_ch (enc, ',');
		291
		292	if (enc->flags & F_INDENT)
		293	encode_nl (enc);
200	else if (enc->flags & F_SPACE_AFTER) \	294	else if (enc->flags & F_SPACE_AFTER)
201	SPACE; \	295	encode_space (enc);
202	SE	296	}
203		297
204	static void encode_sv (enc_t enc, SV sv);	298	static void encode_sv (enc_t enc, SV sv);
205		299
206	static void	300	static void
207	encode_av (enc_t enc, AV av)	301	encode_av (enc_t enc, AV av)
208	{	302	{
209	int i, len = av_len (av);	303	int i, len = av_len (av);
210		304
211	encode_ch (enc, '['); NL;	305	if (enc->indent >= enc->maxdepth)
		306	croak ("data structure too deep (hit recursion limit)");
		307
		308	encode_ch (enc, '['); encode_nl (enc);
212	++enc->indent;	309	++enc->indent;
213		310
214	for (i = 0; i <= len; ++i)	311	for (i = 0; i <= len; ++i)
215	{	312	{
216	INDENT;	313	encode_indent (enc);
217	encode_sv (enc, *av_fetch (av, i, 0));	314	encode_sv (enc, *av_fetch (av, i, 0));
218		315
219	if (i < len)	316	if (i < len)
220	COMMA;	317	encode_comma (enc);
221	}	318	}
222		319
223	NL;	320	encode_nl (enc);
224		321
225	--enc->indent;	322	--enc->indent;
226	INDENT; encode_ch (enc, ']');	323	encode_indent (enc); encode_ch (enc, ']');
227	}	324	}
228		325
229	static void	326	static void
230	encode_he (enc_t enc, HE he)	327	encode_he (enc_t enc, HE he)
231	{	328	{
…		…
245	else	342	else
246	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));	343	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
247		344
248	encode_ch (enc, '"');	345	encode_ch (enc, '"');
249		346
250	if (enc->flags & F_SPACE_BEFORE) SPACE;	347	if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
251	encode_ch (enc, ':');	348	encode_ch (enc, ':');
252	if (enc->flags & F_SPACE_AFTER ) SPACE;	349	if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
253	encode_sv (enc, HeVAL (he));	350	encode_sv (enc, HeVAL (he));
254	}	351	}
255		352
256	// compare hash entries, used when all keys are bytestrings	353	// compare hash entries, used when all keys are bytestrings
257	static int	354	static int
…		…
263	HE b = (HE **)b_;	360	HE b = (HE **)b_;
264		361
265	STRLEN la = HeKLEN (a);	362	STRLEN la = HeKLEN (a);
266	STRLEN lb = HeKLEN (b);	363	STRLEN lb = HeKLEN (b);
267		364
268	if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))	365	if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
269	cmp = la < lb ? -1 : la == lb ? 0 : 1;	366	cmp = la - lb;
270		367
271	return cmp;	368	return cmp;
272	}	369	}
273		370
274	// compare hash entries, used when some keys are sv's or utf-x	371	// compare hash entries, used when some keys are sv's or utf-x
…		…
281	static void	378	static void
282	encode_hv (enc_t enc, HV hv)	379	encode_hv (enc_t enc, HV hv)
283	{	380	{
284	int count, i;	381	int count, i;
285		382
		383	if (enc->indent >= enc->maxdepth)
		384	croak ("data structure too deep (hit recursion limit)");
		385
286	encode_ch (enc, '{'); NL; ++enc->indent;	386	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
287		387
288	if ((count = hv_iterinit (hv)))	388	if ((count = hv_iterinit (hv)))
289	{	389	{
290	// for canonical output we have to sort by keys first	390	// for canonical output we have to sort by keys first
291	// actually, this is mostly due to the stupid so-called	391	// actually, this is mostly due to the stupid so-called
292	// security workaround added somewhere in 5.8.x.	392	// security workaround added somewhere in 5.8.x.
293	// that randomises hash orderings	393	// that randomises hash orderings
294	if (enc->flags & F_CANONICAL)	394	if (enc->flags & F_CANONICAL)
295	{	395	{
296	HE he, hes [count];
297	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
298		403
299	i = 0;	404	i = 0;
300	while ((he = hv_iternext (hv)))	405	while ((he = hv_iternext (hv)))
301	{	406	{
302	hes [i++] = he;	407	hes [i++] = he;
…		…
308		413
309	if (fast)	414	if (fast)
310	qsort (hes, count, sizeof (HE *), he_cmp_fast);	415	qsort (hes, count, sizeof (HE *), he_cmp_fast);
311	else	416	else
312	{	417	{
313	// hack to disable "use bytes"	418	// hack to forcefully disable "use bytes"
314	COP *oldcop = PL_curcop, cop;	419	COP cop = *PL_curcop;
315	cop.op_private = 0;	420	cop.op_private = 0;
		421
		422	ENTER;
		423	SAVETMPS;
		424
		425	SAVEVPTR (PL_curcop);
316	PL_curcop = &cop;	426	PL_curcop = &cop;
317		427
318	SAVETMPS;
319	qsort (hes, count, sizeof (HE *), he_cmp_slow);	428	qsort (hes, count, sizeof (HE *), he_cmp_slow);
		429
320	FREETMPS;	430	FREETMPS;
321		431	LEAVE;
322	PL_curcop = oldcop;
323	}	432	}
324		433
325	for (i = 0; i < count; ++i)	434	for (i = 0; i < count; ++i)
326	{	435	{
327	INDENT;	436	encode_indent (enc);
328	encode_he (enc, hes [i]);	437	encode_he (enc, hes [i]);
329		438
330	if (i < count - 1)	439	if (i < count - 1)
331	COMMA;	440	encode_comma (enc);
332	}	441	}
333		442
334	NL;	443	encode_nl (enc);
335	}	444	}
336	else	445	else
337	{	446	{
338	SV *sv;
339	HE *he = hv_iternext (hv);	447	HE *he = hv_iternext (hv);
340		448
341	for (;;)	449	for (;;)
342	{	450	{
343	INDENT;	451	encode_indent (enc);
344	encode_he (enc, he);	452	encode_he (enc, he);
345		453
346	if (!(he = hv_iternext (hv)))	454	if (!(he = hv_iternext (hv)))
347	break;	455	break;
348		456
349	COMMA;	457	encode_comma (enc);
350	}	458	}
351		459
352	NL;	460	encode_nl (enc);
353	}	461	}
354	}	462	}
355		463
356	--enc->indent; INDENT; encode_ch (enc, '}');	464	--enc->indent; encode_indent (enc); encode_ch (enc, '}');
		465	}
		466
		467	// encode objects, arrays and special \0=false and \1=true values.
		468	static void
		469	encode_rv (enc_t enc, SV sv)
		470	{
		471	svtype svt;
		472
		473	SvGETMAGIC (sv);
		474	svt = SvTYPE (sv);
		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	}
		530	else if (svt == SVt_PVHV)
		531	encode_hv (enc, (HV *)sv);
		532	else if (svt == SVt_PVAV)
		533	encode_av (enc, (AV *)sv);
		534	else if (svt < SVt_PVAV)
		535	{
		536	if (SvNIOK (sv) && SvIV (sv) == 0)
		537	encode_str (enc, "false", 5, 0);
		538	else if (SvNIOK (sv) && SvIV (sv) == 1)
		539	encode_str (enc, "true", 4, 0);
		540	else
		541	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
		542	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		543	}
		544	else
		545	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
		546	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
357	}	547	}
358		548
359	static void	549	static void
360	encode_sv (enc_t enc, SV sv)	550	encode_sv (enc_t enc, SV sv)
361	{	551	{
…		…
369	encode_str (enc, str, len, SvUTF8 (sv));	559	encode_str (enc, str, len, SvUTF8 (sv));
370	encode_ch (enc, '"');	560	encode_ch (enc, '"');
371	}	561	}
372	else if (SvNOKp (sv))	562	else if (SvNOKp (sv))
373	{	563	{
		564	// trust that perl will do the right thing w.r.t. JSON syntax.
374	need (enc, NV_DIG + 32);	565	need (enc, NV_DIG + 32);
375	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	566	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
376	enc->cur += strlen (enc->cur);	567	enc->cur += strlen (enc->cur);
377	}	568	}
378	else if (SvIOKp (sv))	569	else if (SvIOKp (sv))
379	{	570	{
380	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);
381	enc->cur +=	576	enc->cur +=
382	SvIsUV(sv)	577	SvIsUV(sv)
383	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	578	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
384	: 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	}
385	}	608	}
386	else if (SvROK (sv))	609	else if (SvROK (sv))
387	{	610	encode_rv (enc, SvRV (sv));
388	if (!--enc->max_recurse)
389	croak ("data structure too deep (hit recursion limit)");
390
391	sv = SvRV (sv);
392
393	switch (SvTYPE (sv))
394	{
395	case SVt_PVAV: encode_av (enc, (AV *)sv); break;
396	case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
397
398	default:
399	croak ("JSON can only represent references to arrays or hashes");
400	}
401	}
402	else if (!SvOK (sv))	611	else if (!SvOK (sv))
403	encode_str (enc, "null", 4, 0);	612	encode_str (enc, "null", 4, 0);
404	else	613	else
405	croak ("encountered perl type that JSON cannot handle");	614	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
		615	SvPV_nolen (sv), SvFLAGS (sv));
406	}	616	}
407		617
408	static SV *	618	static SV *
409	encode_json (SV *scalar, UV flags)	619	encode_json (SV *scalar, U32 flags)
410	{	620	{
		621	enc_t enc;
		622
411	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	623	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
412	croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)");	624	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
413		625
414	enc_t enc;
415	enc.flags = flags;	626	enc.flags = flags;
416	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	627	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
417	enc.cur = SvPVX (enc.sv);	628	enc.cur = SvPVX (enc.sv);
418	enc.end = SvEND (enc.sv);	629	enc.end = SvEND (enc.sv);
419	enc.max_recurse = 0;
420	enc.indent = 0;	630	enc.indent = 0;
		631	enc.maxdepth = DEC_DEPTH (flags);
421		632
422	SvPOK_only (enc.sv);	633	SvPOK_only (enc.sv);
423	encode_sv (&enc, scalar);	634	encode_sv (&enc, scalar);
424		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
425	if (!(flags & (F_ASCII \| F_UTF8)))	639	if (!(flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
426	SvUTF8_on (enc.sv);	640	SvUTF8_on (enc.sv);
427		641
428	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
429
430	#ifdef SvPV_shrink_to_cur
431	if (enc.flags & F_SHRINK)	642	if (enc.flags & F_SHRINK)
432	SvPV_shrink_to_cur (enc.sv);	643	shrink (enc.sv);
433	#endif	644
434	return enc.sv;	645	return enc.sv;
435	}	646	}
436		647
437	/////////////////////////////////////////////////////////////////////////////	648	/////////////////////////////////////////////////////////////////////////////
		649	// decoder
438		650
439	#define WS \	651	// structure used for decoding JSON
		652	typedef struct
		653	{
		654	char *cur; // current parser pointer
		655	char *end; // end of input string
		656	const char *err; // parse error, if != 0
		657	U32 flags; // F_*
		658	U32 depth; // recursion depth
		659	U32 maxdepth; // recursion depth limit
		660	} dec_t;
		661
		662	inline void
		663	decode_ws (dec_t *dec)
		664	{
440	for (;;) \	665	for (;;)
441	{ \	666	{
442	char ch = *dec->cur; \	667	char ch = *dec->cur;
		668
443	if (ch > 0x20 \	669	if (ch > 0x20
444	\|\| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \	670	\|\| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
445	break; \	671	break;
		672
446	++dec->cur; \	673	++dec->cur;
447	}	674	}
		675	}
448		676
449	#define ERR(reason) SB dec->err = reason; goto fail; SE	677	#define ERR(reason) SB dec->err = reason; goto fail; SE
		678
450	#define EXPECT_CH(ch) SB \	679	#define EXPECT_CH(ch) SB \
451	if (*dec->cur != ch) \	680	if (*dec->cur != ch) \
452	ERR (# ch " expected"); \	681	ERR (# ch " expected"); \
453	++dec->cur; \	682	++dec->cur; \
454	SE	683	SE
455		684
		685	#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
		686	#define DEC_DEC_DEPTH --dec->depth
		687
456	static SV decode_sv (dec_t dec);	688	static SV decode_sv (dec_t dec);
457		689
458	static signed char decode_hexdigit[256];	690	static signed char decode_hexdigit[256];
459		691
460	static UV	692	static UV
461	decode_4hex (dec_t *dec)	693	decode_4hex (dec_t *dec)
462	{	694	{
463	signed char d1, d2, d3, d4;	695	signed char d1, d2, d3, d4;
		696	unsigned char cur = (unsigned char )dec->cur;
464		697
465	d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];	698	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
466	if (d1 < 0) ERR ("four hexadecimal digits expected");	699	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
467	d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];	700	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
468	if (d2 < 0) ERR ("four hexadecimal digits expected");	701	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
469	d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
470	if (d3 < 0) ERR ("four hexadecimal digits expected");
471	d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
472	if (d4 < 0) ERR ("four hexadecimal digits expected");
473		702
474	dec->cur += 4;	703	dec->cur += 4;
475		704
476	return ((UV)d1) << 12	705	return ((UV)d1) << 12
477	\| ((UV)d2) << 8	706	\| ((UV)d2) << 8
…		…
480		709
481	fail:	710	fail:
482	return (UV)-1;	711	return (UV)-1;
483	}	712	}
484		713
485	#define APPEND_GROW(n) SB \
486	if (cur + (n) >= end) \
487	{ \
488	STRLEN ofs = cur - SvPVX (sv); \
489	SvGROW (sv, ofs + (n) + 1); \
490	cur = SvPVX (sv) + ofs; \
491	end = SvEND (sv); \
492	} \
493	SE
494
495	#define APPEND_CH(ch) SB \
496	APPEND_GROW (1); \
497	*cur++ = (ch); \
498	SE
499
500	static SV *	714	static SV *
501	decode_str (dec_t *dec)	715	decode_str (dec_t *dec)
502	{	716	{
503	SV *sv = NEWSV (0,2);	717	SV *sv = 0;
504	int utf8 = 0;	718	int utf8 = 0;
505	char *cur = SvPVX (sv);	719	char *dec_cur = dec->cur;
506	char *end = SvEND (sv);
507		720
508	for (;;)	721	do
509	{	722	{
510	unsigned char ch = (unsigned char )dec->cur;	723	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
		724	char *cur = buf;
511		725
512	if (ch == '"')	726	do
513	break;
514	else if (ch == '\\')
515	{	727	{
516	switch (*++dec->cur)	728	unsigned char ch = (unsigned char )dec_cur++;
		729
		730	if (expect_false (ch == '"'))
517	{	731	{
518	case '\\':	732	--dec_cur;
519	case '/':	733	break;
520	case '"': APPEND_CH (*dec->cur++); break;	734	}
521		735	else if (expect_false (ch == '\\'))
522	case 'b': APPEND_CH ('\010'); ++dec->cur; break;	736	{
523	case 't': APPEND_CH ('\011'); ++dec->cur; break;	737	switch (*dec_cur)
524	case 'n': APPEND_CH ('\012'); ++dec->cur; break;
525	case 'f': APPEND_CH ('\014'); ++dec->cur; break;
526	case 'r': APPEND_CH ('\015'); ++dec->cur; break;
527
528	case 'u':
529	{	738	{
530	UV lo, hi;	739	case '\\':
531	++dec->cur;	740	case '/':
		741	case '"': cur++ = dec_cur++; break;
532		742
533	hi = decode_4hex (dec);	743	case 'b': ++dec_cur; *cur++ = '\010'; break;
534	if (hi == (UV)-1)	744	case 't': ++dec_cur; *cur++ = '\011'; break;
535	goto fail;	745	case 'n': ++dec_cur; *cur++ = '\012'; break;
		746	case 'f': ++dec_cur; *cur++ = '\014'; break;
		747	case 'r': ++dec_cur; *cur++ = '\015'; break;
536		748
537	// possibly a surrogate pair	749	case 'u':
538	if (hi >= 0xd800 && hi < 0xdc00)
539	{	750	{
540	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')	751	UV lo, hi;
541	ERR ("missing low surrogate character in surrogate pair");	752	++dec_cur;
542		753
543	dec->cur += 2;	754	dec->cur = dec_cur;
544
545	lo = decode_4hex (dec);	755	hi = decode_4hex (dec);
		756	dec_cur = dec->cur;
546	if (lo == (UV)-1)	757	if (hi == (UV)-1)
547	goto fail;	758	goto fail;
548		759
		760	// possibly a surrogate pair
		761	if (hi >= 0xd800)
		762	if (hi < 0xdc00)
		763	{
		764	if (dec_cur [0] != '\\' \|\| dec_cur [1] != 'u')
		765	ERR ("missing low surrogate character in surrogate pair");
		766
		767	dec_cur += 2;
		768
		769	dec->cur = dec_cur;
		770	lo = decode_4hex (dec);
		771	dec_cur = dec->cur;
		772	if (lo == (UV)-1)
		773	goto fail;
		774
549	if (lo < 0xdc00 \|\| lo >= 0xe000)	775	if (lo < 0xdc00 \|\| lo >= 0xe000)
550	ERR ("surrogate pair expected");	776	ERR ("surrogate pair expected");
551		777
552	hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;	778	hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
		779	}
		780	else if (hi < 0xe000)
		781	ERR ("missing high surrogate character in surrogate pair");
		782
		783	if (hi >= 0x80)
		784	{
		785	utf8 = 1;
		786
		787	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
		788	}
		789	else
		790	*cur++ = hi;
553	}	791	}
554	else if (hi >= 0xdc00 && hi < 0xe000)
555	ERR ("missing high surrogate character in surrogate pair");
556
557	if (hi >= 0x80)
558	{	792	break;
559	utf8 = 1;
560		793
561	APPEND_GROW (4); // at most 4 bytes for 21 bits
562	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
563	}
564	else	794	default:
565	APPEND_CH (hi);	795	--dec_cur;
		796	ERR ("illegal backslash escape sequence in string");
566	}	797	}
567	break;	798	}
		799	else if (expect_true (ch >= 0x20 && ch <= 0x7f))
		800	*cur++ = ch;
		801	else if (ch >= 0x80)
		802	{
		803	STRLEN clen;
		804	UV uch;
568		805
569	default:
570	--dec->cur;	806	--dec_cur;
571	ERR ("illegal backslash escape sequence in string");	807
		808	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
		809	if (clen == (STRLEN)-1)
		810	ERR ("malformed UTF-8 character in JSON string");
		811
		812	do
		813	cur++ = dec_cur++;
		814	while (--clen);
		815
		816	utf8 = 1;
		817	}
		818	else
		819	{
		820	--dec_cur;
		821
		822	if (!ch)
		823	ERR ("unexpected end of string while parsing JSON string");
		824	else
		825	ERR ("invalid character encountered while parsing JSON string");
572	}	826	}
573	}	827	}
574	else if (ch >= 0x20 && ch <= 0x7f)	828	while (cur < buf + SHORT_STRING_LEN);
575	APPEND_CH (*dec->cur++);	829
576	else if (ch >= 0x80)
577	{	830	{
578	STRLEN clen;	831	STRLEN len = cur - buf;
579	UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
580	if (clen == (STRLEN)-1)
581	ERR ("malformed UTF-8 character in string, cannot convert to JSON");
582		832
583	APPEND_GROW (clen);	833	if (sv)
584	do
585	{	834	{
586	cur++ = dec->cur++;	835	SvGROW (sv, SvCUR (sv) + len + 1);
		836	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
		837	SvCUR_set (sv, SvCUR (sv) + len);
587	}	838	}
588	while (--clen);
589
590	utf8 = 1;
591	}
592	else if (dec->cur == dec->end)
593	ERR ("unexpected end of string while parsing json string");
594	else	839	else
595	ERR ("invalid character encountered");	840	sv = newSVpvn (buf, len);
596	}	841	}
		842	}
		843	while (*dec_cur != '"');
597		844
598	++dec->cur;	845	++dec_cur;
599		846
600	SvCUR_set (sv, cur - SvPVX (sv));	847	if (sv)
601		848	{
602	SvPOK_only (sv);	849	SvPOK_only (sv);
603	*SvEND (sv) = 0;	850	*SvEND (sv) = 0;
604		851
605	if (utf8)	852	if (utf8)
606	SvUTF8_on (sv);	853	SvUTF8_on (sv);
		854	}
		855	else
		856	sv = newSVpvn ("", 0);
607		857
608	#ifdef SvPV_shrink_to_cur	858	dec->cur = dec_cur;
609	if (dec->flags & F_SHRINK)
610	SvPV_shrink_to_cur (sv);
611	#endif
612
613	return sv;	859	return sv;
614		860
615	fail:	861	fail:
616	SvREFCNT_dec (sv);	862	dec->cur = dec_cur;
617	return 0;	863	return 0;
618	}	864	}
619		865
620	static SV *	866	static SV *
621	decode_num (dec_t *dec)	867	decode_num (dec_t *dec)
…		…
679	is_nv = 1;	925	is_nv = 1;
680	}	926	}
681		927
682	if (!is_nv)	928	if (!is_nv)
683	{	929	{
684	UV uv;	930	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
685	int numtype = grok_number (start, dec->cur - start, &uv);	931	if (*start == '-')
686	if (numtype & IS_NUMBER_IN_UV)	932	switch (dec->cur - start)
687	if (numtype & IS_NUMBER_NEG)
688	{	933	{
689	if (uv < (UV)IV_MIN)	934	case 2: return newSViv (-( start [1] - '0' * 1));
690	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));
691	}	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	}
692	else	957	else
693	return newSVuv (uv);	958	return newSVuv (uv);
		959
		960	// here would likely be the place for bigint support
694	}	961	}
		962	}
695		963
		964	// if we ever support bigint or bigfloat, this is the place for bigfloat
696	return newSVnv (Atof (start));	965	return newSVnv (Atof (start));
697		966
698	fail:	967	fail:
699	return 0;	968	return 0;
700	}	969	}
…		…
702	static SV *	971	static SV *
703	decode_av (dec_t *dec)	972	decode_av (dec_t *dec)
704	{	973	{
705	AV *av = newAV ();	974	AV *av = newAV ();
706		975
707	WS;	976	DEC_INC_DEPTH;
		977	decode_ws (dec);
		978
708	if (*dec->cur == ']')	979	if (*dec->cur == ']')
709	++dec->cur;	980	++dec->cur;
710	else	981	else
711	for (;;)	982	for (;;)
712	{	983	{
…		…
716	if (!value)	987	if (!value)
717	goto fail;	988	goto fail;
718		989
719	av_push (av, value);	990	av_push (av, value);
720		991
721	WS;	992	decode_ws (dec);
722		993
723	if (*dec->cur == ']')	994	if (*dec->cur == ']')
724	{	995	{
725	++dec->cur;	996	++dec->cur;
726	break;	997	break;
…		…
730	ERR (", or ] expected while parsing array");	1001	ERR (", or ] expected while parsing array");
731		1002
732	++dec->cur;	1003	++dec->cur;
733	}	1004	}
734		1005
		1006	DEC_DEC_DEPTH;
735	return newRV_noinc ((SV *)av);	1007	return newRV_noinc ((SV *)av);
736		1008
737	fail:	1009	fail:
738	SvREFCNT_dec (av);	1010	SvREFCNT_dec (av);
		1011	DEC_DEC_DEPTH;
739	return 0;	1012	return 0;
740	}	1013	}
741		1014
742	static SV *	1015	static SV *
743	decode_hv (dec_t *dec)	1016	decode_hv (dec_t *dec)
744	{	1017	{
745	HV *hv = newHV ();	1018	HV *hv = newHV ();
746		1019
747	WS;	1020	DEC_INC_DEPTH;
		1021	decode_ws (dec);
		1022
748	if (*dec->cur == '}')	1023	if (*dec->cur == '}')
749	++dec->cur;	1024	++dec->cur;
750	else	1025	else
751	for (;;)	1026	for (;;)
752	{	1027	{
753	SV key, value;	1028	SV key, value;
754		1029
755	WS; EXPECT_CH ('"');	1030	decode_ws (dec); EXPECT_CH ('"');
756		1031
757	key = decode_str (dec);	1032	key = decode_str (dec);
758	if (!key)	1033	if (!key)
759	goto fail;	1034	goto fail;
760		1035
761	WS; EXPECT_CH (':');	1036	decode_ws (dec); EXPECT_CH (':');
762		1037
763	value = decode_sv (dec);	1038	value = decode_sv (dec);
764	if (!value)	1039	if (!value)
765	{	1040	{
766	SvREFCNT_dec (key);	1041	SvREFCNT_dec (key);
767	goto fail;	1042	goto fail;
768	}	1043	}
769		1044
770	//TODO: optimise
771	hv_store_ent (hv, key, value, 0);	1045	hv_store_ent (hv, key, value, 0);
		1046	SvREFCNT_dec (key);
772		1047
773	WS;	1048	decode_ws (dec);
774		1049
775	if (*dec->cur == '}')	1050	if (*dec->cur == '}')
776	{	1051	{
777	++dec->cur;	1052	++dec->cur;
778	break;	1053	break;
…		…
782	ERR (", or } expected while parsing object/hash");	1057	ERR (", or } expected while parsing object/hash");
783		1058
784	++dec->cur;	1059	++dec->cur;
785	}	1060	}
786		1061
		1062	DEC_DEC_DEPTH;
787	return newRV_noinc ((SV *)hv);	1063	return newRV_noinc ((SV *)hv);
788		1064
789	fail:	1065	fail:
790	SvREFCNT_dec (hv);	1066	SvREFCNT_dec (hv);
		1067	DEC_DEC_DEPTH;
791	return 0;	1068	return 0;
792	}	1069	}
793		1070
794	static SV *	1071	static SV *
795	decode_sv (dec_t *dec)	1072	decode_sv (dec_t *dec)
796	{	1073	{
797	WS;	1074	decode_ws (dec);
		1075
		1076	// the beauty of JSON: you need exactly one character lookahead
		1077	// to parse anything.
798	switch (*dec->cur)	1078	switch (*dec->cur)
799	{	1079	{
800	case '"': ++dec->cur; return decode_str (dec);	1080	case '"': ++dec->cur; return decode_str (dec);
801	case '[': ++dec->cur; return decode_av (dec);	1081	case '[': ++dec->cur; return decode_av (dec);
802	case '{': ++dec->cur; return decode_hv (dec);	1082	case '{': ++dec->cur; return decode_hv (dec);
…		…
808		1088
809	case 't':	1089	case 't':
810	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1090	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
811	{	1091	{
812	dec->cur += 4;	1092	dec->cur += 4;
813	return newSViv (1);	1093	return SvREFCNT_inc (json_true);
814	}	1094	}
815	else	1095	else
816	ERR ("'true' expected");	1096	ERR ("'true' expected");
817		1097
818	break;	1098	break;
819		1099
820	case 'f':	1100	case 'f':
821	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1101	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
822	{	1102	{
823	dec->cur += 5;	1103	dec->cur += 5;
824	return newSViv (0);	1104	return SvREFCNT_inc (json_false);
825	}	1105	}
826	else	1106	else
827	ERR ("'false' expected");	1107	ERR ("'false' expected");
828		1108
829	break;	1109	break;
…		…
838	ERR ("'null' expected");	1118	ERR ("'null' expected");
839		1119
840	break;	1120	break;
841		1121
842	default:	1122	default:
843	ERR ("malformed json string");	1123	ERR ("malformed JSON string, neither array, object, number, string or atom");
844	break;	1124	break;
845	}	1125	}
846		1126
847	fail:	1127	fail:
848	return 0;	1128	return 0;
849	}	1129	}
850		1130
851	static SV *	1131	static SV *
852	decode_json (SV *string, UV flags)	1132	decode_json (SV string, U32 flags, UV offset_return)
853	{	1133	{
		1134	dec_t dec;
		1135	UV offset;
854	SV *sv;	1136	SV *sv;
		1137
		1138	SvGETMAGIC (string);
		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));
855		1144
856	if (flags & F_UTF8)	1145	if (flags & F_UTF8)
857	sv_utf8_downgrade (string, 0);	1146	sv_utf8_downgrade (string, 0);
858	else	1147	else
859	sv_utf8_upgrade (string);	1148	sv_utf8_upgrade (string);
860		1149
861	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1150	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
862		1151
863	dec_t dec;
864	dec.flags = flags;	1152	dec.flags = flags;
865	dec.cur = SvPVX (string);	1153	dec.cur = SvPVX (string);
866	dec.end = SvEND (string);	1154	dec.end = SvEND (string);
867	dec.err = 0;	1155	dec.err = 0;
		1156	dec.depth = 0;
		1157	dec.maxdepth = DEC_DEPTH (dec.flags);
868		1158
		1159	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
869	sv = decode_sv (&dec);	1160	sv = decode_sv (&dec);
870		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
871	if (!sv)	1185	if (!sv)
872	{	1186	{
873	IV offset = utf8_distance (dec.cur, SvPVX (string));
874	SV *uni = sv_newmortal ();	1187	SV *uni = sv_newmortal ();
		1188
875	// horrible hack to silence warning inside pv_uni_display	1189	// horrible hack to silence warning inside pv_uni_display
876	COP cop;	1190	COP cop = *PL_curcop;
877	memset (&cop, 0, sizeof (cop));
878	cop.cop_warnings = pWARN_NONE;	1191	cop.cop_warnings = pWARN_NONE;
		1192	ENTER;
879	SAVEVPTR (PL_curcop);	1193	SAVEVPTR (PL_curcop);
880	PL_curcop = &cop;	1194	PL_curcop = &cop;
881
882	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1195	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
		1196	LEAVE;
		1197
883	croak ("%s, at character offset %d (%s)",	1198	croak ("%s, at character offset %d [\"%s\"]",
884	dec.err,	1199	dec.err,
885	(int)offset,	1200	(int)offset,
886	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1201	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
887	}	1202	}
888		1203
889	sv = sv_2mortal (sv);	1204	sv = sv_2mortal (sv);
890		1205
891	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))	1206	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
892	croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)");	1207	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
893		1208
894	return sv;	1209	return sv;
895	}	1210	}
896		1211
		1212	/////////////////////////////////////////////////////////////////////////////
		1213	// XS interface functions
		1214
897	MODULE = JSON::XS PACKAGE = JSON::XS	1215	MODULE = JSON::XS PACKAGE = JSON::XS
898		1216
899	BOOT:	1217	BOOT:
900	{	1218	{
901	int i;	1219	int i;
902		1220
903	memset (decode_hexdigit, 0xff, 256);
904	for (i = 10; i--; )	1221	for (i = 0; i < 256; ++i)
905	decode_hexdigit ['0' + i] = i;	1222	decode_hexdigit [i] =
		1223	i >= '0' && i <= '9' ? i - '0'
		1224	: i >= 'a' && i <= 'f' ? i - 'a' + 10
		1225	: i >= 'A' && i <= 'F' ? i - 'A' + 10
		1226	: -1;
906		1227
907	for (i = 7; i--; )
908	{
909	decode_hexdigit ['a' + i] = 10 + i;
910	decode_hexdigit ['A' + i] = 10 + i;
911	}
912
913	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);
914	}	1233	}
915		1234
916	PROTOTYPES: DISABLE	1235	PROTOTYPES: DISABLE
917		1236
918	SV new (char dummy)	1237	SV new (char dummy)
…		…
921	OUTPUT:	1240	OUTPUT:
922	RETVAL	1241	RETVAL
923		1242
924	SV ascii (SV self, int enable = 1)	1243	SV ascii (SV self, int enable = 1)
925	ALIAS:	1244	ALIAS:
926	ascii = F_ASCII	1245	ascii = F_ASCII
		1246	latin1 = F_LATIN1
927	utf8 = F_UTF8	1247	utf8 = F_UTF8
928	indent = F_INDENT	1248	indent = F_INDENT
929	canonical = F_CANONICAL	1249	canonical = F_CANONICAL
930	space_before = F_SPACE_BEFORE	1250	space_before = F_SPACE_BEFORE
931	space_after = F_SPACE_AFTER	1251	space_after = F_SPACE_AFTER
932	json_rpc = F_JSON_RPC
933	pretty = F_PRETTY	1252	pretty = F_PRETTY
934	allow_nonref = F_ALLOW_NONREF	1253	allow_nonref = F_ALLOW_NONREF
935	shrink = F_SHRINK	1254	shrink = F_SHRINK
		1255	allow_blessed = F_ALLOW_BLESSED
		1256	convert_blessed = F_CONV_BLESSED
936	CODE:	1257	CODE:
937	{	1258	{
938	UV *uv = SvJSON (self);	1259	UV *uv = SvJSON (self);
939	if (enable)	1260	if (enable)
940	*uv \|= ix;	1261	*uv \|= ix;
…		…
944	RETVAL = newSVsv (self);	1265	RETVAL = newSVsv (self);
945	}	1266	}
946	OUTPUT:	1267	OUTPUT:
947	RETVAL	1268	RETVAL
948		1269
		1270	SV max_depth (SV self, UV max_depth = 0x80000000UL)
		1271	CODE:
		1272	{
		1273	UV *uv = SvJSON (self);
		1274	UV log2 = 0;
		1275
		1276	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
		1277
		1278	while ((1UL << log2) < max_depth)
		1279	++log2;
		1280
		1281	uv = uv & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
		1282
		1283	RETVAL = newSVsv (self);
		1284	}
		1285	OUTPUT:
		1286	RETVAL
		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
949	void encode (SV self, SV scalar)	1307	void encode (SV self, SV scalar)
950	PPCODE:	1308	PPCODE:
951	XPUSHs (encode_json (scalar, *SvJSON (self)));	1309	XPUSHs (encode_json (scalar, *SvJSON (self)));
952		1310
953	void decode (SV self, SV jsonstr)	1311	void decode (SV self, SV jsonstr)
954	PPCODE:	1312	PPCODE:
955	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	}
956		1323
957	PROTOTYPES: ENABLE	1324	PROTOTYPES: ENABLE
958		1325
959	void to_json (SV *scalar)	1326	void to_json (SV *scalar)
		1327	ALIAS:
		1328	objToJson = 0
960	PPCODE:	1329	PPCODE:
961	XPUSHs (encode_json (scalar, F_UTF8));	1330	XPUSHs (encode_json (scalar, F_DEFAULT \| F_UTF8));
962		1331
963	void from_json (SV *jsonstr)	1332	void from_json (SV *jsonstr)
		1333	ALIAS:
		1334	jsonToObj = 0
964	PPCODE:	1335	PPCODE:
965	XPUSHs (decode_json (jsonstr, F_UTF8));	1336	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8, 0));
966		1337

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.6 by root, Fri Mar 23 15:10:55 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 1.6 by root, Fri Mar 23 15:10:55 2007 UTC vs.
Revision 1.45 by root, Mon Jun 25 06:57:42 2007 UTC