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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.5 by root, Thu Mar 22 23:24:18 2007 UTC vs.
Revision 1.35 by root, Wed Jun 6 14:52:49 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
9	#define F_ASCII 0x00000001	14	#define F_ASCII 0x00000001UL
		15	#define F_LATIN1 0x00000002UL
10	#define F_UTF8 0x00000002	16	#define F_UTF8 0x00000004UL
11	#define F_INDENT 0x00000004	17	#define F_INDENT 0x00000008UL
12	#define F_CANONICAL 0x00000008	18	#define F_CANONICAL 0x00000010UL
13	#define F_SPACE_BEFORE 0x00000010	19	#define F_SPACE_BEFORE 0x00000020UL
14	#define F_SPACE_AFTER 0x00000020	20	#define F_SPACE_AFTER 0x00000040UL
15	#define F_JSON_RPC 0x00000040
16	#define F_ALLOW_NONREF 0x00000080	21	#define F_ALLOW_NONREF 0x00000100UL
		22	#define F_SHRINK 0x00000200UL
		23	#define F_MAXDEPTH 0xf8000000UL
		24	#define S_MAXDEPTH 27
		25
		26	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
		27
		28	// F_SELFCONVERT? <=> to_json/toJson
		29	// F_BLESSED? <=> { $__class__$ => }
17		30
18	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	31	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
19	#define F_DEFAULT 0	32	#define F_DEFAULT (9UL << S_MAXDEPTH)
20		33
21	#define INIT_SIZE 32 // initial scalar size to be allocated	34	#define INIT_SIZE 32 // initial scalar size to be allocated
		35	#define INDENT_STEP 3 // spaces per indentation level
		36
		37	#define SHORT_STRING_LEN 512 // special-case strings of up to this size
22		38
23	#define SB do {	39	#define SB do {
24	#define SE } while (0)	40	#define SE } while (0)
25		41
		42	#if __GNUC__ >= 3
		43	# define expect(expr,value) __builtin_expect ((expr),(value))
		44	# define inline inline
		45	#else
		46	# define expect(expr,value) (expr)
		47	# define inline static
		48	#endif
		49
		50	#define expect_false(expr) expect ((expr) != 0, 0)
		51	#define expect_true(expr) expect ((expr) != 0, 1)
		52
26	static HV *json_stash;	53	static HV *json_stash; // JSON::XS::
		54
		55	/////////////////////////////////////////////////////////////////////////////
		56	// utility functions
		57
		58	static UV *
		59	SvJSON (SV *sv)
		60	{
		61	if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
		62	croak ("object is not of type JSON::XS");
		63
		64	return &SvUVX (SvRV (sv));
		65	}
		66
		67	static void
		68	shrink (SV *sv)
		69	{
		70	sv_utf8_downgrade (sv, 1);
		71	if (SvLEN (sv) > SvCUR (sv) + 1)
		72	{
		73	#ifdef SvPV_shrink_to_cur
		74	SvPV_shrink_to_cur (sv);
		75	#elif defined (SvPV_renew)
		76	SvPV_renew (sv, SvCUR (sv) + 1);
		77	#endif
		78	}
		79	}
		80
		81	// decode an utf-8 character and return it, or (UV)-1 in
		82	// case of an error.
		83	// we special-case "safe" characters from U+80 .. U+7FF,
		84	// but use the very good perl function to parse anything else.
		85	// note that we never call this function for a ascii codepoints
		86	inline UV
		87	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
		88	{
		89	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))
		90	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
		91	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
		92	{
		93	*clen = 2;
		94	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
		95	}
		96	else
		97	{
		98	*clen = (STRLEN)-1;
		99	return (UV)-1;
		100	}
		101	}
		102
		103	/////////////////////////////////////////////////////////////////////////////
		104	// encoder
27		105
28	// structure used for encoding JSON	106	// structure used for encoding JSON
29	typedef struct	107	typedef struct
30	{	108	{
31	char *cur;	109	char *cur; // SvPVX (sv) + current output position
32	STRLEN len; // SvLEN (sv)
33	char *end; // SvEND (sv)	110	char *end; // SvEND (sv)
34	SV *sv;	111	SV *sv; // result scalar
35	UV flags;	112	U32 flags; // F_*
36	int max_recurse;	113	U32 indent; // indentation level
37	int indent;	114	U32 maxdepth; // max. indentation/recursion level
38	} enc_t;	115	} enc_t;
39		116
40	// structure used for decoding JSON	117	inline void
41	typedef struct
42	{
43	char *cur;
44	char *end;
45	const char *err;
46	UV flags;
47	} dec_t;
48
49	static UV *
50	SvJSON (SV *sv)
51	{
52	if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
53	croak ("object is not of type JSON::XS");
54
55	return &SvUVX (SvRV (sv));
56	}
57
58	/////////////////////////////////////////////////////////////////////////////
59
60	static void
61	need (enc_t *enc, STRLEN len)	118	need (enc_t *enc, STRLEN len)
62	{	119	{
63	if (enc->cur + len >= enc->end)	120	if (expect_false (enc->cur + len >= enc->end))
64	{	121	{
65	STRLEN cur = enc->cur - SvPVX (enc->sv);	122	STRLEN cur = enc->cur - SvPVX (enc->sv);
66	SvGROW (enc->sv, cur + len + 1);	123	SvGROW (enc->sv, cur + len + 1);
67	enc->cur = SvPVX (enc->sv) + cur;	124	enc->cur = SvPVX (enc->sv) + cur;
68	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);	125	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
69	}	126	}
70	}	127	}
71		128
72	static void	129	inline void
73	encode_ch (enc_t *enc, char ch)	130	encode_ch (enc_t *enc, char ch)
74	{	131	{
75	need (enc, 1);	132	need (enc, 1);
76	*enc->cur++ = ch;	133	*enc->cur++ = ch;
77	}	134	}
…		…
85		142
86	while (str < end)	143	while (str < end)
87	{	144	{
88	unsigned char ch = (unsigned char )str;	145	unsigned char ch = (unsigned char )str;
89		146
90	if (ch == '"')	147	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
91	{	148	{
		149	if (expect_false (ch == '"')) // but with slow exceptions
		150	{
92	need (enc, len += 1);	151	need (enc, len += 1);
93	*enc->cur++ = '\\';	152	*enc->cur++ = '\\';
94	*enc->cur++ = '"';	153	*enc->cur++ = '"';
95	++str;
96	}	154	}
97	else if (ch == '\\')	155	else if (expect_false (ch == '\\'))
98	{	156	{
99	need (enc, len += 1);	157	need (enc, len += 1);
100	*enc->cur++ = '\\';	158	*enc->cur++ = '\\';
101	*enc->cur++ = '\\';	159	*enc->cur++ = '\\';
102	++str;
103	}	160	}
104	else if (ch >= 0x20 && ch < 0x80) // most common case	161	else
105	{
106	*enc->cur++ = ch;	162	*enc->cur++ = ch;
107	++str;	163
108	}
109	else if (ch == '\015')
110	{
111	need (enc, len += 1);
112	*enc->cur++ = '\\';
113	*enc->cur++ = 'r';
114	++str;
115	}
116	else if (ch == '\012')
117	{
118	need (enc, len += 1);
119	*enc->cur++ = '\\';
120	*enc->cur++ = 'n';
121	++str;	164	++str;
122	}	165	}
123	else	166	else
124	{	167	{
125	STRLEN clen;	168	switch (ch)
126	UV uch;
127
128	if (is_utf8)
129	{	169	{
130	uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);	170	case '\010': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 'b'; ++str; break;
131	if (clen == (STRLEN)-1)	171	case '\011': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 't'; ++str; break;
132	croak ("malformed UTF-8 character in string, cannot convert to JSON");	172	case '\012': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 'n'; ++str; break;
133	}	173	case '\014': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 'f'; ++str; break;
134	else	174	case '\015': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 'r'; ++str; break;
135	{
136	uch = ch;
137	clen = 1;
138	}
139		175
140	if (uch < 0x80 \|\| enc->flags & F_ASCII)	176	default:
141	{
142	if (uch > 0xFFFFUL)
143	{	177	{
		178	STRLEN clen;
		179	UV uch;
		180
		181	if (is_utf8)
		182	{
		183	//uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
		184	uch = decode_utf8 (str, end - str, &clen);
		185	if (clen == (STRLEN)-1)
		186	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
		187	}
		188	else
		189	{
		190	uch = ch;
		191	clen = 1;
		192	}
		193
		194	if (uch > 0x10FFFFUL)
		195	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
		196
		197	if (uch < 0x80 \|\| enc->flags & F_ASCII \|\| (enc->flags & F_LATIN1 && uch > 0xFF))
		198	{
		199	if (uch > 0xFFFFUL)
		200	{
144	need (enc, len += 11);	201	need (enc, len += 11);
145	sprintf (enc->cur, "\\u%04x\\u%04x",	202	sprintf (enc->cur, "\\u%04x\\u%04x",
146	(uch - 0x10000) / 0x400 + 0xD800,	203	(int)((uch - 0x10000) / 0x400 + 0xD800),
147	(uch - 0x10000) % 0x400 + 0xDC00);	204	(int)((uch - 0x10000) % 0x400 + 0xDC00));
148	enc->cur += 12;	205	enc->cur += 12;
		206	}
		207	else
		208	{
		209	static char hexdigit [16] = "0123456789abcdef";
		210	need (enc, len += 5);
		211	*enc->cur++ = '\\';
		212	*enc->cur++ = 'u';
		213	*enc->cur++ = hexdigit [ uch >> 12 ];
		214	*enc->cur++ = hexdigit [(uch >> 8) & 15];
		215	*enc->cur++ = hexdigit [(uch >> 4) & 15];
		216	*enc->cur++ = hexdigit [(uch >> 0) & 15];
		217	}
		218
		219	str += clen;
		220	}
		221	else if (enc->flags & F_LATIN1)
		222	{
		223	*enc->cur++ = uch;
		224	str += clen;
		225	}
		226	else if (is_utf8)
		227	{
		228	need (enc, len += clen);
		229	do
		230	{
		231	enc->cur++ = str++;
		232	}
		233	while (--clen);
		234	}
		235	else
		236	{
		237	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
		238	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
		239	++str;
		240	}
149	}	241	}
150	else
151	{
152	static char hexdigit [16] = "0123456789abcdef";
153	need (enc, len += 5);
154	*enc->cur++ = '\\';
155	*enc->cur++ = 'u';
156	*enc->cur++ = hexdigit [ uch >> 12 ];
157	*enc->cur++ = hexdigit [(uch >> 8) & 15];
158	*enc->cur++ = hexdigit [(uch >> 4) & 15];
159	*enc->cur++ = hexdigit [(uch >> 0) & 15];
160	}
161
162	str += clen;
163	}
164	else if (is_utf8)
165	{
166	need (enc, len += clen);
167	do
168	{
169	enc->cur++ = str++;
170	}
171	while (--clen);
172	}
173	else
174	{
175	need (enc, 10); // never more than 11 bytes needed
176	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
177	++str;
178	}	242	}
179	}	243	}
180		244
181	--len;	245	--len;
182	}	246	}
183	}	247	}
184		248
185	#define INDENT SB \	249	inline void
		250	encode_indent (enc_t *enc)
		251	{
186	if (enc->flags & F_INDENT) \	252	if (enc->flags & F_INDENT)
187	{ \	253	{
188	int i_; \	254	int spaces = enc->indent * INDENT_STEP;
189	need (enc, enc->indent); \
190	for (i_ = enc->indent * 3; i_--; )\
191	encode_ch (enc, ' '); \
192	} \
193	SE
194		255
195	#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE	256	need (enc, spaces);
196	#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE	257	memset (enc->cur, ' ', spaces);
197	#define COMMA SB \	258	enc->cur += spaces;
		259	}
		260	}
		261
		262	inline void
		263	encode_space (enc_t *enc)
		264	{
		265	need (enc, 1);
198	encode_ch (enc, ','); \	266	encode_ch (enc, ' ');
		267	}
		268
		269	inline void
		270	encode_nl (enc_t *enc)
		271	{
199	if (enc->flags & F_INDENT) \	272	if (enc->flags & F_INDENT)
200	NL; \	273	{
		274	need (enc, 1);
		275	encode_ch (enc, '\n');
		276	}
		277	}
		278
		279	inline void
		280	encode_comma (enc_t *enc)
		281	{
		282	encode_ch (enc, ',');
		283
		284	if (enc->flags & F_INDENT)
		285	encode_nl (enc);
201	else if (enc->flags & F_SPACE_AFTER) \	286	else if (enc->flags & F_SPACE_AFTER)
202	SPACE; \	287	encode_space (enc);
203	SE	288	}
204		289
205	static void encode_sv (enc_t enc, SV sv);	290	static void encode_sv (enc_t enc, SV sv);
206		291
207	static void	292	static void
208	encode_av (enc_t enc, AV av)	293	encode_av (enc_t enc, AV av)
209	{	294	{
210	int i, len = av_len (av);	295	int i, len = av_len (av);
211		296
212	encode_ch (enc, '['); NL;	297	if (enc->indent >= enc->maxdepth)
		298	croak ("data structure too deep (hit recursion limit)");
		299
		300	encode_ch (enc, '['); encode_nl (enc);
213	++enc->indent;	301	++enc->indent;
214		302
215	for (i = 0; i <= len; ++i)	303	for (i = 0; i <= len; ++i)
216	{	304	{
217	INDENT;	305	encode_indent (enc);
218	encode_sv (enc, *av_fetch (av, i, 0));	306	encode_sv (enc, *av_fetch (av, i, 0));
219		307
220	if (i < len)	308	if (i < len)
221	COMMA;	309	encode_comma (enc);
222	}	310	}
223		311
224	NL;	312	encode_nl (enc);
225		313
226	--enc->indent;	314	--enc->indent;
227	INDENT; encode_ch (enc, ']');	315	encode_indent (enc); encode_ch (enc, ']');
228	}	316	}
229		317
230	static void	318	static void
231	encode_he (enc_t enc, HE he)	319	encode_he (enc_t enc, HE he)
232	{	320	{
…		…
246	else	334	else
247	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));	335	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
248		336
249	encode_ch (enc, '"');	337	encode_ch (enc, '"');
250		338
251	if (enc->flags & F_SPACE_BEFORE) SPACE;	339	if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
252	encode_ch (enc, ':');	340	encode_ch (enc, ':');
253	if (enc->flags & F_SPACE_AFTER ) SPACE;	341	if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
254	encode_sv (enc, HeVAL (he));	342	encode_sv (enc, HeVAL (he));
255	}	343	}
256		344
257	// compare hash entries, used when all keys are bytestrings	345	// compare hash entries, used when all keys are bytestrings
258	static int	346	static int
…		…
264	HE b = (HE **)b_;	352	HE b = (HE **)b_;
265		353
266	STRLEN la = HeKLEN (a);	354	STRLEN la = HeKLEN (a);
267	STRLEN lb = HeKLEN (b);	355	STRLEN lb = HeKLEN (b);
268		356
269	if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))	357	if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
270	cmp = la < lb ? -1 : la == lb ? 0 : 1;	358	cmp = la - lb;
271		359
272	return cmp;	360	return cmp;
273	}	361	}
274		362
275	// compare hash entries, used when some keys are sv's or utf-x	363	// compare hash entries, used when some keys are sv's or utf-x
…		…
282	static void	370	static void
283	encode_hv (enc_t enc, HV hv)	371	encode_hv (enc_t enc, HV hv)
284	{	372	{
285	int count, i;	373	int count, i;
286		374
		375	if (enc->indent >= enc->maxdepth)
		376	croak ("data structure too deep (hit recursion limit)");
		377
287	encode_ch (enc, '{'); NL; ++enc->indent;	378	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
288		379
289	if ((count = hv_iterinit (hv)))	380	if ((count = hv_iterinit (hv)))
290	{	381	{
291	// for canonical output we have to sort by keys first	382	// for canonical output we have to sort by keys first
292	// actually, this is mostly due to the stupid so-called	383	// actually, this is mostly due to the stupid so-called
293	// security workaround added somewhere in 5.8.x.	384	// security workaround added somewhere in 5.8.x.
294	// that randomises hash orderings	385	// that randomises hash orderings
295	if (enc->flags & F_CANONICAL)	386	if (enc->flags & F_CANONICAL)
296	{	387	{
297	HE he, hes [count];
298	int fast = 1;	388	int fast = 1;
		389	HE *he;
		390	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		391	HE *hes = _alloca (count sizeof (HE));
		392	#else
		393	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
		394	#endif
299		395
300	i = 0;	396	i = 0;
301	while ((he = hv_iternext (hv)))	397	while ((he = hv_iternext (hv)))
302	{	398	{
303	hes [i++] = he;	399	hes [i++] = he;
…		…
309		405
310	if (fast)	406	if (fast)
311	qsort (hes, count, sizeof (HE *), he_cmp_fast);	407	qsort (hes, count, sizeof (HE *), he_cmp_fast);
312	else	408	else
313	{	409	{
314	// hack to disable "use bytes"	410	// hack to forcefully disable "use bytes"
315	COP *oldcop = PL_curcop, cop;	411	COP cop = *PL_curcop;
316	cop.op_private = 0;	412	cop.op_private = 0;
		413
		414	ENTER;
		415	SAVETMPS;
		416
		417	SAVEVPTR (PL_curcop);
317	PL_curcop = &cop;	418	PL_curcop = &cop;
318		419
319	SAVETMPS;
320	qsort (hes, count, sizeof (HE *), he_cmp_slow);	420	qsort (hes, count, sizeof (HE *), he_cmp_slow);
		421
321	FREETMPS;	422	FREETMPS;
322		423	LEAVE;
323	PL_curcop = oldcop;
324	}	424	}
325		425
326	for (i = 0; i < count; ++i)	426	for (i = 0; i < count; ++i)
327	{	427	{
328	INDENT;	428	encode_indent (enc);
329	encode_he (enc, hes [i]);	429	encode_he (enc, hes [i]);
330		430
331	if (i < count - 1)	431	if (i < count - 1)
332	COMMA;	432	encode_comma (enc);
333	}	433	}
334		434
335	NL;	435	encode_nl (enc);
336	}	436	}
337	else	437	else
338	{	438	{
339	SV *sv;
340	HE *he = hv_iternext (hv);	439	HE *he = hv_iternext (hv);
341		440
342	for (;;)	441	for (;;)
343	{	442	{
344	INDENT;	443	encode_indent (enc);
345	encode_he (enc, he);	444	encode_he (enc, he);
346		445
347	if (!(he = hv_iternext (hv)))	446	if (!(he = hv_iternext (hv)))
348	break;	447	break;
349		448
350	COMMA;	449	encode_comma (enc);
351	}	450	}
352		451
353	NL;	452	encode_nl (enc);
354	}	453	}
355	}	454	}
356		455
357	--enc->indent; INDENT; encode_ch (enc, '}');	456	--enc->indent; encode_indent (enc); encode_ch (enc, '}');
		457	}
		458
		459	// encode objects, arrays and special \0=false and \1=true values.
		460	static void
		461	encode_rv (enc_t enc, SV sv)
		462	{
		463	svtype svt;
		464
		465	SvGETMAGIC (sv);
		466	svt = SvTYPE (sv);
		467
		468	if (svt == SVt_PVHV)
		469	encode_hv (enc, (HV *)sv);
		470	else if (svt == SVt_PVAV)
		471	encode_av (enc, (AV *)sv);
		472	else if (svt < SVt_PVAV)
		473	{
		474	if (SvNIOK (sv) && SvIV (sv) == 0)
		475	encode_str (enc, "false", 5, 0);
		476	else if (SvNIOK (sv) && SvIV (sv) == 1)
		477	encode_str (enc, "true", 4, 0);
		478	else
		479	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
		480	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		481	}
		482	else
		483	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
		484	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
358	}	485	}
359		486
360	static void	487	static void
361	encode_sv (enc_t enc, SV sv)	488	encode_sv (enc_t enc, SV sv)
362	{	489	{
…		…
376	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	503	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
377	enc->cur += strlen (enc->cur);	504	enc->cur += strlen (enc->cur);
378	}	505	}
379	else if (SvIOKp (sv))	506	else if (SvIOKp (sv))
380	{	507	{
		508	// we assume we can always read an IV as a UV
		509	if (SvUV (sv) & ~(UV)0x7fff)
		510	{
381	need (enc, 64);	511	need (enc, 32);
382	enc->cur +=	512	enc->cur +=
383	SvIsUV(sv)	513	SvIsUV(sv)
384	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	514	? snprintf (enc->cur, 32, "%"UVuf, (UV)SvUVX (sv))
385	: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));	515	: snprintf (enc->cur, 32, "%"IVdf, (IV)SvIVX (sv));
		516	}
		517	else
		518	{
		519	// optimise the "small number case"
		520	// code will likely be branchless and use only a single multiplication
		521	I32 i = SvIV (sv);
		522	U32 u;
		523
		524	need (enc, 6);
		525
		526	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
		527	u = i < 0 ? -i : i;
		528
		529	// convert to 4.28 fixed-point representation
		530	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
		531
		532	char digit, nz = 0;
		533
		534	digit = u >> 28; enc->cur = digit + '0'; nz \|= digit; enc->cur += nz ? 1 : 0; u = (u & 0xfffffff) 5;
		535	digit = u >> 27; enc->cur = digit + '0'; nz \|= digit; enc->cur += nz ? 1 : 0; u = (u & 0x7ffffff) 5;
		536	digit = u >> 26; enc->cur = digit + '0'; nz \|= digit; enc->cur += nz ? 1 : 0; u = (u & 0x3ffffff) 5;
		537	digit = u >> 25; enc->cur = digit + '0'; nz \|= digit; enc->cur += nz ? 1 : 0; u = (u & 0x1ffffff) 5;
		538	digit = u >> 24; *enc->cur = digit + '0'; nz \|= digit; enc->cur += 1;
		539	}
386	}	540	}
387	else if (SvROK (sv))	541	else if (SvROK (sv))
388	{	542	encode_rv (enc, SvRV (sv));
389	if (!--enc->max_recurse)
390	croak ("data structure too deep (hit recursion limit)");
391
392	sv = SvRV (sv);
393
394	switch (SvTYPE (sv))
395	{
396	case SVt_PVAV: encode_av (enc, (AV *)sv); break;
397	case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
398
399	default:
400	croak ("JSON can only represent references to arrays or hashes");
401	}
402	}
403	else if (!SvOK (sv))	543	else if (!SvOK (sv))
404	encode_str (enc, "null", 4, 0);	544	encode_str (enc, "null", 4, 0);
405	else	545	else
406	croak ("encountered perl type that JSON cannot handle");	546	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
		547	SvPV_nolen (sv), SvFLAGS (sv));
407	}	548	}
408		549
409	static SV *	550	static SV *
410	encode_json (SV *scalar, UV flags)	551	encode_json (SV *scalar, U32 flags)
411	{	552	{
		553	enc_t enc;
		554
412	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	555	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
413	croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)");	556	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
414		557
415	enc_t enc;
416	enc.flags = flags;	558	enc.flags = flags;
417	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	559	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
418	enc.cur = SvPVX (enc.sv);	560	enc.cur = SvPVX (enc.sv);
419	enc.end = SvEND (enc.sv);	561	enc.end = SvEND (enc.sv);
420	enc.max_recurse = 0;
421	enc.indent = 0;	562	enc.indent = 0;
		563	enc.maxdepth = DEC_DEPTH (flags);
422		564
423	SvPOK_only (enc.sv);	565	SvPOK_only (enc.sv);
424	encode_sv (&enc, scalar);	566	encode_sv (&enc, scalar);
425		567
		568	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
		569	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
		570
426	if (!(flags & (F_ASCII \| F_UTF8)))	571	if (!(flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
427	SvUTF8_on (enc.sv);	572	SvUTF8_on (enc.sv);
428		573
429	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	574	if (enc.flags & F_SHRINK)
		575	shrink (enc.sv);
		576
430	return enc.sv;	577	return enc.sv;
431	}	578	}
432		579
433	/////////////////////////////////////////////////////////////////////////////	580	/////////////////////////////////////////////////////////////////////////////
		581	// decoder
434		582
435	#define WS \	583	// structure used for decoding JSON
		584	typedef struct
		585	{
		586	char *cur; // current parser pointer
		587	char *end; // end of input string
		588	const char *err; // parse error, if != 0
		589	U32 flags; // F_*
		590	U32 depth; // recursion depth
		591	U32 maxdepth; // recursion depth limit
		592	} dec_t;
		593
		594	inline void
		595	decode_ws (dec_t *dec)
		596	{
436	for (;;) \	597	for (;;)
437	{ \	598	{
438	char ch = *dec->cur; \	599	char ch = *dec->cur;
		600
439	if (ch > 0x20 \	601	if (ch > 0x20
440	\|\| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \	602	\|\| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
441	break; \	603	break;
		604
442	++dec->cur; \	605	++dec->cur;
443	}	606	}
		607	}
444		608
445	#define ERR(reason) SB dec->err = reason; goto fail; SE	609	#define ERR(reason) SB dec->err = reason; goto fail; SE
		610
446	#define EXPECT_CH(ch) SB \	611	#define EXPECT_CH(ch) SB \
447	if (*dec->cur != ch) \	612	if (*dec->cur != ch) \
448	ERR (# ch " expected"); \	613	ERR (# ch " expected"); \
449	++dec->cur; \	614	++dec->cur; \
450	SE	615	SE
451		616
		617	#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
		618	#define DEC_DEC_DEPTH --dec->depth
		619
452	static SV decode_sv (dec_t dec);	620	static SV decode_sv (dec_t dec);
453		621
454	static signed char decode_hexdigit[256];	622	static signed char decode_hexdigit[256];
455		623
456	static UV	624	static UV
457	decode_4hex (dec_t *dec)	625	decode_4hex (dec_t *dec)
458	{	626	{
459	signed char d1, d2, d3, d4;	627	signed char d1, d2, d3, d4;
		628	unsigned char cur = (unsigned char )dec->cur;
460		629
461	d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];	630	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected");
462	if (d1 < 0) ERR ("four hexadecimal digits expected");	631	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected");
463	d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];	632	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected");
464	if (d2 < 0) ERR ("four hexadecimal digits expected");	633	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected");
465	d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
466	if (d3 < 0) ERR ("four hexadecimal digits expected");
467	d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
468	if (d4 < 0) ERR ("four hexadecimal digits expected");
469		634
470	dec->cur += 4;	635	dec->cur += 4;
471		636
472	return ((UV)d1) << 12	637	return ((UV)d1) << 12
473	\| ((UV)d2) << 8	638	\| ((UV)d2) << 8
…		…
476		641
477	fail:	642	fail:
478	return (UV)-1;	643	return (UV)-1;
479	}	644	}
480		645
481	#define APPEND_GROW(n) SB \
482	if (cur + (n) >= end) \
483	{ \
484	STRLEN ofs = cur - SvPVX (sv); \
485	SvGROW (sv, ofs + (n) + 1); \
486	cur = SvPVX (sv) + ofs; \
487	end = SvEND (sv); \
488	} \
489	SE
490
491	#define APPEND_CH(ch) SB \
492	APPEND_GROW (1); \
493	*cur++ = (ch); \
494	SE
495
496	static SV *	646	static SV *
497	decode_str (dec_t *dec)	647	decode_str (dec_t *dec)
498	{	648	{
499	SV *sv = NEWSV (0,2);	649	SV *sv = 0;
500	int utf8 = 0;	650	int utf8 = 0;
501	char *cur = SvPVX (sv);
502	char *end = SvEND (sv);
503		651
504	for (;;)	652	do
505	{	653	{
506	unsigned char ch = (unsigned char )dec->cur;	654	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
		655	char *cur = buf;
507		656
508	if (ch == '"')	657	do
509	break;
510	else if (ch == '\\')
511	{	658	{
512	switch (*++dec->cur)	659	unsigned char ch = (unsigned char )dec->cur++;
		660
		661	if (expect_false (ch == '"'))
513	{	662	{
514	case '\\':	663	--dec->cur;
515	case '/':	664	break;
516	case '"': APPEND_CH (*dec->cur++); break;	665	}
517		666	else if (expect_false (ch == '\\'))
518	case 'b': APPEND_CH ('\010'); ++dec->cur; break;	667	{
519	case 't': APPEND_CH ('\011'); ++dec->cur; break;	668	switch (*dec->cur)
520	case 'n': APPEND_CH ('\012'); ++dec->cur; break;
521	case 'f': APPEND_CH ('\014'); ++dec->cur; break;
522	case 'r': APPEND_CH ('\015'); ++dec->cur; break;
523
524	case 'u':
525	{	669	{
526	UV lo, hi;	670	case '\\':
527	++dec->cur;	671	case '/':
		672	case '"': cur++ = dec->cur++; break;
528		673
529	hi = decode_4hex (dec);	674	case 'b': ++dec->cur; *cur++ = '\010'; break;
530	if (hi == (UV)-1)	675	case 't': ++dec->cur; *cur++ = '\011'; break;
531	goto fail;	676	case 'n': ++dec->cur; *cur++ = '\012'; break;
		677	case 'f': ++dec->cur; *cur++ = '\014'; break;
		678	case 'r': ++dec->cur; *cur++ = '\015'; break;
532		679
533	// possibly a surrogate pair	680	case 'u':
534	if (hi >= 0xd800 && hi < 0xdc00)
535	{	681	{
536	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')	682	UV lo, hi;
537	ERR ("missing low surrogate character in surrogate pair");
538
539	dec->cur += 2;	683	++dec->cur;
540		684
541	lo = decode_4hex (dec);	685	hi = decode_4hex (dec);
542	if (lo == (UV)-1)	686	if (hi == (UV)-1)
543	goto fail;	687	goto fail;
544		688
		689	// possibly a surrogate pair
		690	if (hi >= 0xd800)
		691	if (hi < 0xdc00)
		692	{
		693	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')
		694	ERR ("missing low surrogate character in surrogate pair");
		695
		696	dec->cur += 2;
		697
		698	lo = decode_4hex (dec);
		699	if (lo == (UV)-1)
		700	goto fail;
		701
545	if (lo < 0xdc00 \|\| lo >= 0xe000)	702	if (lo < 0xdc00 \|\| lo >= 0xe000)
546	ERR ("surrogate pair expected");	703	ERR ("surrogate pair expected");
547		704
548	hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;	705	hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
		706	}
		707	else if (hi < 0xe000)
		708	ERR ("missing high surrogate character in surrogate pair");
		709
		710	if (hi >= 0x80)
		711	{
		712	utf8 = 1;
		713
		714	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
		715	}
		716	else
		717	*cur++ = hi;
549	}	718	}
550	else if (hi >= 0xdc00 && hi < 0xe000)
551	ERR ("missing high surrogate character in surrogate pair");
552
553	if (hi >= 0x80)
554	{	719	break;
555	utf8 = 1;
556		720
557	APPEND_GROW (4); // at most 4 bytes for 21 bits
558	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
559	}
560	else	721	default:
561	APPEND_CH (hi);	722	--dec->cur;
		723	ERR ("illegal backslash escape sequence in string");
562	}	724	}
563	break;	725	}
		726	else if (expect_true (ch >= 0x20 && ch <= 0x7f))
		727	*cur++ = ch;
		728	else if (ch >= 0x80)
		729	{
		730	STRLEN clen;
		731	UV uch;
564		732
565	default:
566	--dec->cur;	733	--dec->cur;
567	ERR ("illegal backslash escape sequence in string");	734
		735	uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
		736	if (clen == (STRLEN)-1)
		737	ERR ("malformed UTF-8 character in JSON string");
		738
		739	do
		740	cur++ = dec->cur++;
		741	while (--clen);
		742
		743	utf8 = 1;
		744	}
		745	else
		746	{
		747	--dec->cur;
		748
		749	if (!ch)
		750	ERR ("unexpected end of string while parsing JSON string");
		751	else
		752	ERR ("invalid character encountered while parsing JSON string");
568	}	753	}
569	}	754	}
570	else if (ch >= 0x20 && ch <= 0x7f)	755	while (cur < buf + SHORT_STRING_LEN);
571	APPEND_CH (*dec->cur++);	756
572	else if (ch >= 0x80)
573	{	757	{
574	STRLEN clen;	758	STRLEN len = cur - buf;
575	UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
576	if (clen == (STRLEN)-1)
577	ERR ("malformed UTF-8 character in string, cannot convert to JSON");
578		759
579	APPEND_GROW (clen);	760	if (sv)
580	do
581	{	761	{
582	cur++ = dec->cur++;	762	SvGROW (sv, SvCUR (sv) + len + 1);
		763	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
		764	SvCUR_set (sv, SvCUR (sv) + len);
583	}	765	}
584	while (--clen);
585
586	utf8 = 1;
587	}
588	else if (dec->cur == dec->end)
589	ERR ("unexpected end of string while parsing json string");
590	else	766	else
591	ERR ("invalid character encountered");	767	sv = newSVpvn (buf, len);
592	}	768	}
		769	}
		770	while (*dec->cur != '"');
593		771
594	++dec->cur;	772	++dec->cur;
595		773
596	SvCUR_set (sv, cur - SvPVX (sv));	774	if (sv)
597		775	{
598	SvPOK_only (sv);	776	SvPOK_only (sv);
599	*SvEND (sv) = 0;	777	*SvEND (sv) = 0;
600		778
601	if (utf8)	779	if (utf8)
602	SvUTF8_on (sv);	780	SvUTF8_on (sv);
		781	}
		782	else
		783	sv = newSVpvn ("", 0);
603		784
604	return sv;	785	return sv;
605		786
606	fail:	787	fail:
607	SvREFCNT_dec (sv);
608	return 0;	788	return 0;
609	}	789	}
610		790
611	static SV *	791	static SV *
612	decode_num (dec_t *dec)	792	decode_num (dec_t *dec)
…		…
670	is_nv = 1;	850	is_nv = 1;
671	}	851	}
672		852
673	if (!is_nv)	853	if (!is_nv)
674	{	854	{
675	UV uv;	855	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
676	int numtype = grok_number (start, dec->cur - start, &uv);	856	if (*start == '-')
677	if (numtype & IS_NUMBER_IN_UV)	857	switch (dec->cur - start)
678	if (numtype & IS_NUMBER_NEG)
679	{	858	{
680	if (uv < (UV)IV_MIN)	859	case 2: return newSViv (-( start [1] - '0' ));
681	return newSViv (-(IV)uv);	860	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
		861	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
		862	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
682	}	863	}
		864	else
		865	switch (dec->cur - start)
		866	{
		867	case 1: return newSViv ( start [0] - '0' );
		868	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
		869	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
		870	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		871	}
		872
		873	{
		874	UV uv;
		875	int numtype = grok_number (start, dec->cur - start, &uv);
		876	if (numtype & IS_NUMBER_IN_UV)
		877	if (numtype & IS_NUMBER_NEG)
		878	{
		879	if (uv < (UV)IV_MIN)
		880	return newSViv (-(IV)uv);
		881	}
683	else	882	else
684	return newSVuv (uv);	883	return newSVuv (uv);
		884	}
685	}	885	}
686		886
687	return newSVnv (Atof (start));	887	return newSVnv (Atof (start));
688		888
689	fail:	889	fail:
…		…
693	static SV *	893	static SV *
694	decode_av (dec_t *dec)	894	decode_av (dec_t *dec)
695	{	895	{
696	AV *av = newAV ();	896	AV *av = newAV ();
697		897
698	WS;	898	DEC_INC_DEPTH;
		899	decode_ws (dec);
		900
699	if (*dec->cur == ']')	901	if (*dec->cur == ']')
700	++dec->cur;	902	++dec->cur;
701	else	903	else
702	for (;;)	904	for (;;)
703	{	905	{
…		…
707	if (!value)	909	if (!value)
708	goto fail;	910	goto fail;
709		911
710	av_push (av, value);	912	av_push (av, value);
711		913
712	WS;	914	decode_ws (dec);
713		915
714	if (*dec->cur == ']')	916	if (*dec->cur == ']')
715	{	917	{
716	++dec->cur;	918	++dec->cur;
717	break;	919	break;
…		…
721	ERR (", or ] expected while parsing array");	923	ERR (", or ] expected while parsing array");
722		924
723	++dec->cur;	925	++dec->cur;
724	}	926	}
725		927
		928	DEC_DEC_DEPTH;
726	return newRV_noinc ((SV *)av);	929	return newRV_noinc ((SV *)av);
727		930
728	fail:	931	fail:
729	SvREFCNT_dec (av);	932	SvREFCNT_dec (av);
		933	DEC_DEC_DEPTH;
730	return 0;	934	return 0;
731	}	935	}
732		936
733	static SV *	937	static SV *
734	decode_hv (dec_t *dec)	938	decode_hv (dec_t *dec)
735	{	939	{
736	HV *hv = newHV ();	940	HV *hv = newHV ();
737		941
738	WS;	942	DEC_INC_DEPTH;
		943	decode_ws (dec);
		944
739	if (*dec->cur == '}')	945	if (*dec->cur == '}')
740	++dec->cur;	946	++dec->cur;
741	else	947	else
742	for (;;)	948	for (;;)
743	{	949	{
744	SV key, value;	950	SV key, value;
745		951
746	WS; EXPECT_CH ('"');	952	decode_ws (dec); EXPECT_CH ('"');
747		953
748	key = decode_str (dec);	954	key = decode_str (dec);
749	if (!key)	955	if (!key)
750	goto fail;	956	goto fail;
751		957
752	WS; EXPECT_CH (':');	958	decode_ws (dec); EXPECT_CH (':');
753		959
754	value = decode_sv (dec);	960	value = decode_sv (dec);
755	if (!value)	961	if (!value)
756	{	962	{
757	SvREFCNT_dec (key);	963	SvREFCNT_dec (key);
758	goto fail;	964	goto fail;
759	}	965	}
760		966
761	//TODO: optimise
762	hv_store_ent (hv, key, value, 0);	967	hv_store_ent (hv, key, value, 0);
		968	SvREFCNT_dec (key);
763		969
764	WS;	970	decode_ws (dec);
765		971
766	if (*dec->cur == '}')	972	if (*dec->cur == '}')
767	{	973	{
768	++dec->cur;	974	++dec->cur;
769	break;	975	break;
…		…
773	ERR (", or } expected while parsing object/hash");	979	ERR (", or } expected while parsing object/hash");
774		980
775	++dec->cur;	981	++dec->cur;
776	}	982	}
777		983
		984	DEC_DEC_DEPTH;
778	return newRV_noinc ((SV *)hv);	985	return newRV_noinc ((SV *)hv);
779		986
780	fail:	987	fail:
781	SvREFCNT_dec (hv);	988	SvREFCNT_dec (hv);
		989	DEC_DEC_DEPTH;
782	return 0;	990	return 0;
783	}	991	}
784		992
785	static SV *	993	static SV *
786	decode_sv (dec_t *dec)	994	decode_sv (dec_t *dec)
787	{	995	{
788	WS;	996	decode_ws (dec);
789	switch (*dec->cur)	997	switch (*dec->cur)
790	{	998	{
791	case '"': ++dec->cur; return decode_str (dec);	999	case '"': ++dec->cur; return decode_str (dec);
792	case '[': ++dec->cur; return decode_av (dec);	1000	case '[': ++dec->cur; return decode_av (dec);
793	case '{': ++dec->cur; return decode_hv (dec);	1001	case '{': ++dec->cur; return decode_hv (dec);
…		…
829	ERR ("'null' expected");	1037	ERR ("'null' expected");
830		1038
831	break;	1039	break;
832		1040
833	default:	1041	default:
834	ERR ("malformed json string");	1042	ERR ("malformed JSON string, neither array, object, number, string or atom");
835	break;	1043	break;
836	}	1044	}
837		1045
838	fail:	1046	fail:
839	return 0;	1047	return 0;
840	}	1048	}
841		1049
842	static SV *	1050	static SV *
843	decode_json (SV *string, UV flags)	1051	decode_json (SV string, U32 flags, UV offset_return)
844	{	1052	{
		1053	dec_t dec;
		1054	UV offset;
845	SV *sv;	1055	SV *sv;
		1056
		1057	SvGETMAGIC (string);
		1058	SvUPGRADE (string, SVt_PV);
846		1059
847	if (flags & F_UTF8)	1060	if (flags & F_UTF8)
848	sv_utf8_downgrade (string, 0);	1061	sv_utf8_downgrade (string, 0);
849	else	1062	else
850	sv_utf8_upgrade (string);	1063	sv_utf8_upgrade (string);
851		1064
852	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1065	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
853		1066
854	dec_t dec;
855	dec.flags = flags;	1067	dec.flags = flags;
856	dec.cur = SvPVX (string);	1068	dec.cur = SvPVX (string);
857	dec.end = SvEND (string);	1069	dec.end = SvEND (string);
858	dec.err = 0;	1070	dec.err = 0;
		1071	dec.depth = 0;
		1072	dec.maxdepth = DEC_DEPTH (dec.flags);
859		1073
		1074	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
860	sv = decode_sv (&dec);	1075	sv = decode_sv (&dec);
861		1076
		1077	if (!(offset_return \|\| !sv))
		1078	{
		1079	// check for trailing garbage
		1080	decode_ws (&dec);
		1081
		1082	if (*dec.cur)
		1083	{
		1084	dec.err = "garbage after JSON object";
		1085	SvREFCNT_dec (sv);
		1086	sv = 0;
		1087	}
		1088	}
		1089
		1090	if (offset_return \|\| !sv)
		1091	{
		1092	offset = dec.flags & F_UTF8
		1093	? dec.cur - SvPVX (string)
		1094	: utf8_distance (dec.cur, SvPVX (string));
		1095
		1096	if (offset_return)
		1097	*offset_return = offset;
		1098	}
		1099
862	if (!sv)	1100	if (!sv)
863	{	1101	{
864	IV offset = utf8_distance (dec.cur, SvPVX (string));
865	SV *uni = sv_newmortal ();	1102	SV *uni = sv_newmortal ();
		1103
866	// horrible hack to silence warning inside pv_uni_display	1104	// horrible hack to silence warning inside pv_uni_display
867	COP cop;	1105	COP cop = *PL_curcop;
868	memset (&cop, 0, sizeof (cop));
869	cop.cop_warnings = pWARN_NONE;	1106	cop.cop_warnings = pWARN_NONE;
		1107	ENTER;
870	SAVEVPTR (PL_curcop);	1108	SAVEVPTR (PL_curcop);
871	PL_curcop = &cop;	1109	PL_curcop = &cop;
872
873	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1110	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
		1111	LEAVE;
		1112
874	croak ("%s, at character offset %d (%s)",	1113	croak ("%s, at character offset %d [\"%s\"]",
875	dec.err,	1114	dec.err,
876	(int)offset,	1115	(int)offset,
877	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1116	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
878	}	1117	}
879		1118
880	sv = sv_2mortal (sv);	1119	sv = sv_2mortal (sv);
881		1120
882	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))	1121	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
883	croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)");	1122	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
884		1123
885	return sv;	1124	return sv;
886	}	1125	}
887		1126
		1127	/////////////////////////////////////////////////////////////////////////////
		1128	// XS interface functions
		1129
888	MODULE = JSON::XS PACKAGE = JSON::XS	1130	MODULE = JSON::XS PACKAGE = JSON::XS
889		1131
890	BOOT:	1132	BOOT:
891	{	1133	{
892	int i;	1134	int i;
893		1135
894	memset (decode_hexdigit, 0xff, 256);	1136	memset (decode_hexdigit, 0xff, 256);
		1137
895	for (i = 10; i--; )	1138	for (i = 0; i < 256; ++i)
896	decode_hexdigit ['0' + i] = i;	1139	decode_hexdigit [i] =
897		1140	i >= '0' && i <= '9' ? i - '0'
898	for (i = 7; i--; )	1141	: i >= 'a' && i <= 'f' ? i - 'a' + 10
899	{	1142	: i >= 'A' && i <= 'F' ? i - 'A' + 10
900	decode_hexdigit ['a' + i] = 10 + i;	1143	: -1;
901	decode_hexdigit ['A' + i] = 10 + i;
902	}
903		1144
904	json_stash = gv_stashpv ("JSON::XS", 1);	1145	json_stash = gv_stashpv ("JSON::XS", 1);
905	}	1146	}
906		1147
907	PROTOTYPES: DISABLE	1148	PROTOTYPES: DISABLE
…		…
910	CODE:	1151	CODE:
911	RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);	1152	RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
912	OUTPUT:	1153	OUTPUT:
913	RETVAL	1154	RETVAL
914		1155
915	SV ascii (SV self, int enable)	1156	SV ascii (SV self, int enable = 1)
916	ALIAS:	1157	ALIAS:
917	ascii = F_ASCII	1158	ascii = F_ASCII
		1159	latin1 = F_LATIN1
918	utf8 = F_UTF8	1160	utf8 = F_UTF8
919	indent = F_INDENT	1161	indent = F_INDENT
920	canonical = F_CANONICAL	1162	canonical = F_CANONICAL
921	space_before = F_SPACE_BEFORE	1163	space_before = F_SPACE_BEFORE
922	space_after = F_SPACE_AFTER	1164	space_after = F_SPACE_AFTER
923	json_rpc = F_JSON_RPC
924	pretty = F_PRETTY	1165	pretty = F_PRETTY
925	allow_nonref = F_ALLOW_NONREF	1166	allow_nonref = F_ALLOW_NONREF
		1167	shrink = F_SHRINK
926	CODE:	1168	CODE:
927	{	1169	{
928	UV *uv = SvJSON (self);	1170	UV *uv = SvJSON (self);
929	if (enable)	1171	if (enable)
930	*uv \|= ix;	1172	*uv \|= ix;
…		…
934	RETVAL = newSVsv (self);	1176	RETVAL = newSVsv (self);
935	}	1177	}
936	OUTPUT:	1178	OUTPUT:
937	RETVAL	1179	RETVAL
938		1180
		1181	SV max_depth (SV self, UV max_depth = 0x80000000UL)
		1182	CODE:
		1183	{
		1184	UV *uv = SvJSON (self);
		1185	UV log2 = 0;
		1186
		1187	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
		1188
		1189	while ((1UL << log2) < max_depth)
		1190	++log2;
		1191
		1192	uv = uv & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
		1193
		1194	RETVAL = newSVsv (self);
		1195	}
		1196	OUTPUT:
		1197	RETVAL
		1198
939	void encode (SV self, SV scalar)	1199	void encode (SV self, SV scalar)
940	PPCODE:	1200	PPCODE:
941	XPUSHs (encode_json (scalar, *SvJSON (self)));	1201	XPUSHs (encode_json (scalar, *SvJSON (self)));
942		1202
943	void decode (SV self, SV jsonstr)	1203	void decode (SV self, SV jsonstr)
944	PPCODE:	1204	PPCODE:
945	XPUSHs (decode_json (jsonstr, *SvJSON (self)));	1205	XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
		1206
		1207	void decode_prefix (SV self, SV jsonstr)
		1208	PPCODE:
		1209	{
		1210	UV offset;
		1211	EXTEND (SP, 2);
		1212	PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
		1213	PUSHs (sv_2mortal (newSVuv (offset)));
		1214	}
946		1215
947	PROTOTYPES: ENABLE	1216	PROTOTYPES: ENABLE
948		1217
949	void to_json (SV *scalar)	1218	void to_json (SV *scalar)
		1219	ALIAS:
		1220	objToJson = 0
950	PPCODE:	1221	PPCODE:
951	XPUSHs (encode_json (scalar, F_UTF8));	1222	XPUSHs (encode_json (scalar, F_DEFAULT \| F_UTF8));
952		1223
953	void from_json (SV *jsonstr)	1224	void from_json (SV *jsonstr)
		1225	ALIAS:
		1226	jsonToObj = 0
954	PPCODE:	1227	PPCODE:
955	XPUSHs (decode_json (jsonstr, F_UTF8));	1228	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8, 0));
956		1229

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.5 by root, Thu Mar 22 23:24:18 2007 UTC vs. Revision 1.35 by root, Wed Jun 6 14:52:49 2007 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.5 by root, Thu Mar 22 23:24:18 2007 UTC vs.
Revision 1.35 by root, Wed Jun 6 14:52:49 2007 UTC