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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.24 by root, Tue Apr 3 23:59:04 2007 UTC vs.
Revision 1.50 by root, Mon Jul 2 00:29:38 2007 UTC

…		…
3	#include "XSUB.h"	3	#include "XSUB.h"
4		4
5	#include "assert.h"	5	#include "assert.h"
6	#include "string.h"	6	#include "string.h"
7	#include "stdlib.h"	7	#include "stdlib.h"
		8	#include "stdio.h"
8		9
		10	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		11	# define snprintf _snprintf // C compilers have this in stdio.h
		12	#endif
		13
		14	// some old perls do not have this, try to make it work, no
		15	// guarentees, though. if it breaks, you get to keep the pieces.
		16	#ifndef UTF8_MAXBYTES
		17	# define UTF8_MAXBYTES 13
		18	#endif
		19
9	#define F_ASCII 0x00000001UL	20	#define F_ASCII 0x00000001UL
		21	#define F_LATIN1 0x00000002UL
10	#define F_UTF8 0x00000002UL	22	#define F_UTF8 0x00000004UL
11	#define F_INDENT 0x00000004UL	23	#define F_INDENT 0x00000008UL
12	#define F_CANONICAL 0x00000008UL	24	#define F_CANONICAL 0x00000010UL
13	#define F_SPACE_BEFORE 0x00000010UL	25	#define F_SPACE_BEFORE 0x00000020UL
14	#define F_SPACE_AFTER 0x00000020UL	26	#define F_SPACE_AFTER 0x00000040UL
15	#define F_ALLOW_NONREF 0x00000080UL	27	#define F_ALLOW_NONREF 0x00000100UL
16	#define F_SHRINK 0x00000100UL	28	#define F_SHRINK 0x00000200UL
		29	#define F_ALLOW_BLESSED 0x00000400UL
		30	#define F_CONV_BLESSED 0x00000800UL
17	#define F_MAXDEPTH 0xf8000000UL	31	#define F_MAXDEPTH 0xf8000000UL
18	#define S_MAXDEPTH 27	32	#define S_MAXDEPTH 27
		33	#define F_MAXSIZE 0x01f00000UL
		34	#define S_MAXSIZE 20
		35	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
19		36
20	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))	37	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21		38	#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
22	// F_SELFCONVERT? <=> to_json/toJson
23	// F_BLESSED? <=> { $__class__$ => }
24		39
25	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	40	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
26	#define F_DEFAULT (9UL << S_MAXDEPTH)	41	#define F_DEFAULT (9UL << S_MAXDEPTH)
27		42
28	#define INIT_SIZE 32 // initial scalar size to be allocated	43	#define INIT_SIZE 32 // initial scalar size to be allocated
29	#define INDENT_STEP 3 // spaces per indentation level	44	#define INDENT_STEP 3 // spaces per indentation level
30		45
31	#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32	#define SHORT_STRING_LEN 512 // special-case strings of up to this size	46	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
33		47
34	#define SB do {	48	#define SB do {
35	#define SE } while (0)	49	#define SE } while (0)
36		50
		51	#if __GNUC__ >= 3
		52	# define expect(expr,value) __builtin_expect ((expr),(value))
		53	# define inline inline
		54	#else
		55	# define expect(expr,value) (expr)
		56	# define inline static
		57	#endif
		58
		59	#define expect_false(expr) expect ((expr) != 0, 0)
		60	#define expect_true(expr) expect ((expr) != 0, 1)
		61
37	static HV *json_stash; // JSON::XS::	62	static HV json_stash, json_boolean_stash; // JSON::XS::
		63	static SV json_true, json_false;
		64
		65	typedef struct {
		66	U32 flags;
		67	SV cb_object, cb_sk_object;
		68	} JSON;
38		69
39	/////////////////////////////////////////////////////////////////////////////	70	/////////////////////////////////////////////////////////////////////////////
40	// utility functions	71	// utility functions
41		72
42	static UV *	73	inline void
43	SvJSON (SV *sv)
44	{
45	if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
46	croak ("object is not of type JSON::XS");
47
48	return &SvUVX (SvRV (sv));
49	}
50
51	static void
52	shrink (SV *sv)	74	shrink (SV *sv)
53	{	75	{
54	sv_utf8_downgrade (sv, 1);	76	sv_utf8_downgrade (sv, 1);
55	if (SvLEN (sv) > SvCUR (sv) + 1)	77	if (SvLEN (sv) > SvCUR (sv) + 1)
56	{	78	{
…		…
65	// decode an utf-8 character and return it, or (UV)-1 in	87	// decode an utf-8 character and return it, or (UV)-1 in
66	// case of an error.	88	// case of an error.
67	// we special-case "safe" characters from U+80 .. U+7FF,	89	// we special-case "safe" characters from U+80 .. U+7FF,
68	// but use the very good perl function to parse anything else.	90	// but use the very good perl function to parse anything else.
69	// note that we never call this function for a ascii codepoints	91	// note that we never call this function for a ascii codepoints
70	static UV	92	inline UV
71	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)	93	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
72	{	94	{
73	if (s[0] > 0xdf \|\| s[0] < 0xc2)	95	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))
74	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);	96	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)	97	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76	{	98	{
77	*clen = 2;	99	*clen = 2;
78	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);	100	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
…		…
91	typedef struct	113	typedef struct
92	{	114	{
93	char *cur; // SvPVX (sv) + current output position	115	char *cur; // SvPVX (sv) + current output position
94	char *end; // SvEND (sv)	116	char *end; // SvEND (sv)
95	SV *sv; // result scalar	117	SV *sv; // result scalar
96	U32 flags; // F_*	118	JSON json;
97	U32 indent; // indentation level	119	U32 indent; // indentation level
98	U32 maxdepth; // max. indentation/recursion level	120	U32 maxdepth; // max. indentation/recursion level
99	} enc_t;	121	} enc_t;
100		122
101	static void	123	inline void
102	need (enc_t *enc, STRLEN len)	124	need (enc_t *enc, STRLEN len)
103	{	125	{
104	if (enc->cur + len >= enc->end)	126	if (expect_false (enc->cur + len >= enc->end))
105	{	127	{
106	STRLEN cur = enc->cur - SvPVX (enc->sv);	128	STRLEN cur = enc->cur - SvPVX (enc->sv);
107	SvGROW (enc->sv, cur + len + 1);	129	SvGROW (enc->sv, cur + len + 1);
108	enc->cur = SvPVX (enc->sv) + cur;	130	enc->cur = SvPVX (enc->sv) + cur;
109	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);	131	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
110	}	132	}
111	}	133	}
112		134
113	static void	135	inline void
114	encode_ch (enc_t *enc, char ch)	136	encode_ch (enc_t *enc, char ch)
115	{	137	{
116	need (enc, 1);	138	need (enc, 1);
117	*enc->cur++ = ch;	139	*enc->cur++ = ch;
118	}	140	}
…		…
126		148
127	while (str < end)	149	while (str < end)
128	{	150	{
129	unsigned char ch = (unsigned char )str;	151	unsigned char ch = (unsigned char )str;
130		152
131	if (ch >= 0x20 && ch < 0x80) // most common case	153	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
132	{	154	{
133	if (ch == '"') // but with slow exceptions	155	if (expect_false (ch == '"')) // but with slow exceptions
134	{	156	{
135	need (enc, len += 1);	157	need (enc, len += 1);
136	*enc->cur++ = '\\';	158	*enc->cur++ = '\\';
137	*enc->cur++ = '"';	159	*enc->cur++ = '"';
138	}	160	}
139	else if (ch == '\\')	161	else if (expect_false (ch == '\\'))
140	{	162	{
141	need (enc, len += 1);	163	need (enc, len += 1);
142	*enc->cur++ = '\\';	164	*enc->cur++ = '\\';
143	*enc->cur++ = '\\';	165	*enc->cur++ = '\\';
144	}	166	}
…		…
162	STRLEN clen;	184	STRLEN clen;
163	UV uch;	185	UV uch;
164		186
165	if (is_utf8)	187	if (is_utf8)
166	{	188	{
167	//uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
168	uch = decode_utf8 (str, end - str, &clen);	189	uch = decode_utf8 (str, end - str, &clen);
169	if (clen == (STRLEN)-1)	190	if (clen == (STRLEN)-1)
170	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);	191	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
171	}	192	}
172	else	193	else
…		…
176	}	197	}
177		198
178	if (uch > 0x10FFFFUL)	199	if (uch > 0x10FFFFUL)
179	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);	200	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
180		201
181	if (uch < 0x80 \|\| enc->flags & F_ASCII)	202	if (uch < 0x80 \|\| enc->json.flags & F_ASCII \|\| (enc->json.flags & F_LATIN1 && uch > 0xFF))
182	{	203	{
183	if (uch > 0xFFFFUL)	204	if (uch > 0xFFFFUL)
184	{	205	{
185	need (enc, len += 11);	206	need (enc, len += 11);
186	sprintf (enc->cur, "\\u%04x\\u%04x",	207	sprintf (enc->cur, "\\u%04x\\u%04x",
…		…
200	*enc->cur++ = hexdigit [(uch >> 0) & 15];	221	*enc->cur++ = hexdigit [(uch >> 0) & 15];
201	}	222	}
202		223
203	str += clen;	224	str += clen;
204	}	225	}
		226	else if (enc->json.flags & F_LATIN1)
		227	{
		228	*enc->cur++ = uch;
		229	str += clen;
		230	}
205	else if (is_utf8)	231	else if (is_utf8)
206	{	232	{
207	need (enc, len += clen);	233	need (enc, len += clen);
208	do	234	do
209	{	235	{
…		…
211	}	237	}
212	while (--clen);	238	while (--clen);
213	}	239	}
214	else	240	else
215	{	241	{
216	need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed	242	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
217	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	243	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
218	++str;	244	++str;
219	}	245	}
220	}	246	}
221	}	247	}
…		…
223		249
224	--len;	250	--len;
225	}	251	}
226	}	252	}
227		253
228	static void	254	inline void
229	encode_indent (enc_t *enc)	255	encode_indent (enc_t *enc)
230	{	256	{
231	if (enc->flags & F_INDENT)	257	if (enc->json.flags & F_INDENT)
232	{	258	{
233	int spaces = enc->indent * INDENT_STEP;	259	int spaces = enc->indent * INDENT_STEP;
234		260
235	need (enc, spaces);	261	need (enc, spaces);
236	memset (enc->cur, ' ', spaces);	262	memset (enc->cur, ' ', spaces);
237	enc->cur += spaces;	263	enc->cur += spaces;
238	}	264	}
239	}	265	}
240		266
241	static void	267	inline void
242	encode_space (enc_t *enc)	268	encode_space (enc_t *enc)
243	{	269	{
244	need (enc, 1);	270	need (enc, 1);
245	encode_ch (enc, ' ');	271	encode_ch (enc, ' ');
246	}	272	}
247		273
248	static void	274	inline void
249	encode_nl (enc_t *enc)	275	encode_nl (enc_t *enc)
250	{	276	{
251	if (enc->flags & F_INDENT)	277	if (enc->json.flags & F_INDENT)
252	{	278	{
253	need (enc, 1);	279	need (enc, 1);
254	encode_ch (enc, '\n');	280	encode_ch (enc, '\n');
255	}	281	}
256	}	282	}
257		283
258	static void	284	inline void
259	encode_comma (enc_t *enc)	285	encode_comma (enc_t *enc)
260	{	286	{
261	encode_ch (enc, ',');	287	encode_ch (enc, ',');
262		288
263	if (enc->flags & F_INDENT)	289	if (enc->json.flags & F_INDENT)
264	encode_nl (enc);	290	encode_nl (enc);
265	else if (enc->flags & F_SPACE_AFTER)	291	else if (enc->json.flags & F_SPACE_AFTER)
266	encode_space (enc);	292	encode_space (enc);
267	}	293	}
268		294
269	static void encode_sv (enc_t enc, SV sv);	295	static void encode_sv (enc_t enc, SV sv);
270		296
…		…
313	else	339	else
314	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));	340	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
315		341
316	encode_ch (enc, '"');	342	encode_ch (enc, '"');
317		343
318	if (enc->flags & F_SPACE_BEFORE) encode_space (enc);	344	if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
319	encode_ch (enc, ':');	345	encode_ch (enc, ':');
320	if (enc->flags & F_SPACE_AFTER ) encode_space (enc);	346	if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
321	encode_sv (enc, HeVAL (he));	347	encode_sv (enc, HeVAL (he));
322	}	348	}
323		349
324	// compare hash entries, used when all keys are bytestrings	350	// compare hash entries, used when all keys are bytestrings
325	static int	351	static int
…		…
360	{	386	{
361	// for canonical output we have to sort by keys first	387	// for canonical output we have to sort by keys first
362	// actually, this is mostly due to the stupid so-called	388	// actually, this is mostly due to the stupid so-called
363	// security workaround added somewhere in 5.8.x.	389	// security workaround added somewhere in 5.8.x.
364	// that randomises hash orderings	390	// that randomises hash orderings
365	if (enc->flags & F_CANONICAL)	391	if (enc->json.flags & F_CANONICAL)
366	{	392	{
367	HE he, hes [count]; // if your compiler dies here, you need to enable C99 mode
368	int fast = 1;	393	int fast = 1;
		394	HE *he;
		395	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		396	HE *hes = _alloca (count sizeof (HE));
		397	#else
		398	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
		399	#endif
369		400
370	i = 0;	401	i = 0;
371	while ((he = hv_iternext (hv)))	402	while ((he = hv_iternext (hv)))
372	{	403	{
373	hes [i++] = he;	404	hes [i++] = he;
…		…
408		439
409	encode_nl (enc);	440	encode_nl (enc);
410	}	441	}
411	else	442	else
412	{	443	{
413	SV *sv;
414	HE *he = hv_iternext (hv);	444	HE *he = hv_iternext (hv);
415		445
416	for (;;)	446	for (;;)
417	{	447	{
418	encode_indent (enc);	448	encode_indent (enc);
…		…
433		463
434	// encode objects, arrays and special \0=false and \1=true values.	464	// encode objects, arrays and special \0=false and \1=true values.
435	static void	465	static void
436	encode_rv (enc_t enc, SV sv)	466	encode_rv (enc_t enc, SV sv)
437	{	467	{
		468	svtype svt;
		469
438	SvGETMAGIC (sv);	470	SvGETMAGIC (sv);
439
440	svtype svt = SvTYPE (sv);	471	svt = SvTYPE (sv);
441		472
		473	if (expect_false (SvOBJECT (sv)))
		474	{
		475	if (SvSTASH (sv) == json_boolean_stash)
		476	{
		477	if (SvIV (sv) == 0)
		478	encode_str (enc, "false", 5, 0);
		479	else
		480	encode_str (enc, "true", 4, 0);
		481	}
		482	else
		483	{
		484	#if 0
		485	if (0 && sv_derived_from (rv, "JSON::Literal"))
		486	{
		487	// not yet
		488	}
		489	#endif
		490	if (enc->json.flags & F_CONV_BLESSED)
		491	{
		492	// we re-bless the reference to get overload and other niceties right
		493	GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
		494
		495	if (to_json)
		496	{
		497	dSP; ENTER; SAVETMPS; PUSHMARK (SP);
		498	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
		499
		500	// calling with G_SCALAR ensures that we always get a 1 reutrn value
		501	// check anyways.
		502	PUTBACK;
		503	assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
		504	SPAGAIN;
		505
		506	encode_sv (enc, POPs);
		507
		508	FREETMPS; LEAVE;
		509	}
		510	else if (enc->json.flags & F_ALLOW_BLESSED)
		511	encode_str (enc, "null", 4, 0);
		512	else
		513	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
		514	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		515	}
		516	else if (enc->json.flags & F_ALLOW_BLESSED)
		517	encode_str (enc, "null", 4, 0);
		518	else
		519	croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
		520	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		521	}
		522	}
442	if (svt == SVt_PVHV)	523	else if (svt == SVt_PVHV)
443	encode_hv (enc, (HV *)sv);	524	encode_hv (enc, (HV *)sv);
444	else if (svt == SVt_PVAV)	525	else if (svt == SVt_PVAV)
445	encode_av (enc, (AV *)sv);	526	encode_av (enc, (AV *)sv);
446	else if (svt < SVt_PVAV)	527	else if (svt < SVt_PVAV)
447	{	528	{
448	if (SvNIOK (sv) && SvIV (sv) == 0)	529	STRLEN len = 0;
		530	char *pv = svt ? SvPV (sv, len) : 0;
		531
		532	if (len == 1 && *pv == '0')
449	encode_str (enc, "false", 5, 0);	533	encode_str (enc, "false", 5, 0);
450	else if (SvNIOK (sv) && SvIV (sv) == 1)	534	else if (len == 1 && *pv == '1')
451	encode_str (enc, "true", 4, 0);	535	encode_str (enc, "true", 4, 0);
452	else	536	else
453	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",	537	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
454	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	538	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
455	}	539	}
…		…
471	encode_str (enc, str, len, SvUTF8 (sv));	555	encode_str (enc, str, len, SvUTF8 (sv));
472	encode_ch (enc, '"');	556	encode_ch (enc, '"');
473	}	557	}
474	else if (SvNOKp (sv))	558	else if (SvNOKp (sv))
475	{	559	{
		560	// trust that perl will do the right thing w.r.t. JSON syntax.
476	need (enc, NV_DIG + 32);	561	need (enc, NV_DIG + 32);
477	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	562	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
478	enc->cur += strlen (enc->cur);	563	enc->cur += strlen (enc->cur);
479	}	564	}
480	else if (SvIOKp (sv))	565	else if (SvIOKp (sv))
481	{	566	{
482	need (enc, 64);	567	// we assume we can always read an IV as a UV
		568	if (SvUV (sv) & ~(UV)0x7fff)
		569	{
		570	// large integer, use the (rather slow) snprintf way.
		571	need (enc, sizeof (UV) * 3);
483	enc->cur +=	572	enc->cur +=
484	SvIsUV(sv)	573	SvIsUV(sv)
485	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	574	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
486	: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));	575	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
		576	}
		577	else
		578	{
		579	// optimise the "small number case"
		580	// code will likely be branchless and use only a single multiplication
		581	I32 i = SvIV (sv);
		582	U32 u;
		583	char digit, nz = 0;
		584
		585	need (enc, 6);
		586
		587	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
		588	u = i < 0 ? -i : i;
		589
		590	// convert to 4.28 fixed-point representation
		591	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
		592
		593	// now output digit by digit, each time masking out the integer part
		594	// and multiplying by 5 while moving the decimal point one to the right,
		595	// resulting in a net multiplication by 10.
		596	// we always write the digit to memory but conditionally increment
		597	// the pointer, to ease the usage of conditional move instructions.
		598	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffff) 5;
		599	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffff) 5;
		600	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffff) 5;
		601	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffff) 5;
		602	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
		603	}
487	}	604	}
488	else if (SvROK (sv))	605	else if (SvROK (sv))
489	encode_rv (enc, SvRV (sv));	606	encode_rv (enc, SvRV (sv));
490	else if (!SvOK (sv))	607	else if (!SvOK (sv))
491	encode_str (enc, "null", 4, 0);	608	encode_str (enc, "null", 4, 0);
…		…
493	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	610	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
494	SvPV_nolen (sv), SvFLAGS (sv));	611	SvPV_nolen (sv), SvFLAGS (sv));
495	}	612	}
496		613
497	static SV *	614	static SV *
498	encode_json (SV *scalar, U32 flags)	615	encode_json (SV scalar, JSON json)
499	{	616	{
		617	enc_t enc;
		618
500	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	619	if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
501	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");	620	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
502		621
503	enc_t enc;	622	enc.json = *json;
504	enc.flags = flags;
505	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	623	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
506	enc.cur = SvPVX (enc.sv);	624	enc.cur = SvPVX (enc.sv);
507	enc.end = SvEND (enc.sv);	625	enc.end = SvEND (enc.sv);
508	enc.indent = 0;	626	enc.indent = 0;
509	enc.maxdepth = DEC_DEPTH (flags);	627	enc.maxdepth = DEC_DEPTH (enc.json.flags);
510		628
511	SvPOK_only (enc.sv);	629	SvPOK_only (enc.sv);
512	encode_sv (&enc, scalar);	630	encode_sv (&enc, scalar);
513		631
		632	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
		633	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
		634
514	if (!(flags & (F_ASCII \| F_UTF8)))	635	if (!(enc.json.flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
515	SvUTF8_on (enc.sv);	636	SvUTF8_on (enc.sv);
516		637
517	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
518
519	if (enc.flags & F_SHRINK)	638	if (enc.json.flags & F_SHRINK)
520	shrink (enc.sv);	639	shrink (enc.sv);
521		640
522	return enc.sv;	641	return enc.sv;
523	}	642	}
524		643
…		…
529	typedef struct	648	typedef struct
530	{	649	{
531	char *cur; // current parser pointer	650	char *cur; // current parser pointer
532	char *end; // end of input string	651	char *end; // end of input string
533	const char *err; // parse error, if != 0	652	const char *err; // parse error, if != 0
534	U32 flags; // F_*	653	JSON json;
535	U32 depth; // recursion depth	654	U32 depth; // recursion depth
536	U32 maxdepth; // recursion depth limit	655	U32 maxdepth; // recursion depth limit
537	} dec_t;	656	} dec_t;
538		657
539	static void	658	inline void
540	decode_ws (dec_t *dec)	659	decode_ws (dec_t *dec)
541	{	660	{
542	for (;;)	661	for (;;)
543	{	662	{
544	char ch = *dec->cur;	663	char ch = *dec->cur;
…		…
570	decode_4hex (dec_t *dec)	689	decode_4hex (dec_t *dec)
571	{	690	{
572	signed char d1, d2, d3, d4;	691	signed char d1, d2, d3, d4;
573	unsigned char cur = (unsigned char )dec->cur;	692	unsigned char cur = (unsigned char )dec->cur;
574		693
575	d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");	694	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
576	d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");	695	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
577	d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");	696	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
578	d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");	697	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
579		698
580	dec->cur += 4;	699	dec->cur += 4;
581		700
582	return ((UV)d1) << 12	701	return ((UV)d1) << 12
583	\| ((UV)d2) << 8	702	\| ((UV)d2) << 8
…		…
591	static SV *	710	static SV *
592	decode_str (dec_t *dec)	711	decode_str (dec_t *dec)
593	{	712	{
594	SV *sv = 0;	713	SV *sv = 0;
595	int utf8 = 0;	714	int utf8 = 0;
		715	char *dec_cur = dec->cur;
596		716
597	do	717	do
598	{	718	{
599	char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];	719	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
600	char *cur = buf;	720	char *cur = buf;
601		721
602	do	722	do
603	{	723	{
604	unsigned char ch = (unsigned char )dec->cur++;	724	unsigned char ch = (unsigned char )dec_cur++;
605		725
606	if (ch == '"')	726	if (expect_false (ch == '"'))
607	{	727	{
608	--dec->cur;	728	--dec_cur;
609	break;	729	break;
610	}	730	}
611	else if (ch == '\\')	731	else if (expect_false (ch == '\\'))
612	{	732	{
613	switch (*dec->cur)	733	switch (*dec_cur)
614	{	734	{
615	case '\\':	735	case '\\':
616	case '/':	736	case '/':
617	case '"': cur++ = dec->cur++; break;	737	case '"': cur++ = dec_cur++; break;
618		738
619	case 'b': ++dec->cur; *cur++ = '\010'; break;	739	case 'b': ++dec_cur; *cur++ = '\010'; break;
620	case 't': ++dec->cur; *cur++ = '\011'; break;	740	case 't': ++dec_cur; *cur++ = '\011'; break;
621	case 'n': ++dec->cur; *cur++ = '\012'; break;	741	case 'n': ++dec_cur; *cur++ = '\012'; break;
622	case 'f': ++dec->cur; *cur++ = '\014'; break;	742	case 'f': ++dec_cur; *cur++ = '\014'; break;
623	case 'r': ++dec->cur; *cur++ = '\015'; break;	743	case 'r': ++dec_cur; *cur++ = '\015'; break;
624		744
625	case 'u':	745	case 'u':
626	{	746	{
627	UV lo, hi;	747	UV lo, hi;
628	++dec->cur;	748	++dec_cur;
629		749
		750	dec->cur = dec_cur;
630	hi = decode_4hex (dec);	751	hi = decode_4hex (dec);
		752	dec_cur = dec->cur;
631	if (hi == (UV)-1)	753	if (hi == (UV)-1)
632	goto fail;	754	goto fail;
633		755
634	// possibly a surrogate pair	756	// possibly a surrogate pair
635	if (hi >= 0xd800)	757	if (hi >= 0xd800)
636	if (hi < 0xdc00)	758	if (hi < 0xdc00)
637	{	759	{
638	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')	760	if (dec_cur [0] != '\\' \|\| dec_cur [1] != 'u')
639	ERR ("missing low surrogate character in surrogate pair");	761	ERR ("missing low surrogate character in surrogate pair");
640		762
641	dec->cur += 2;	763	dec_cur += 2;
642		764
		765	dec->cur = dec_cur;
643	lo = decode_4hex (dec);	766	lo = decode_4hex (dec);
		767	dec_cur = dec->cur;
644	if (lo == (UV)-1)	768	if (lo == (UV)-1)
645	goto fail;	769	goto fail;
646		770
647	if (lo < 0xdc00 \|\| lo >= 0xe000)	771	if (lo < 0xdc00 \|\| lo >= 0xe000)
648	ERR ("surrogate pair expected");	772	ERR ("surrogate pair expected");
…		…
662	*cur++ = hi;	786	*cur++ = hi;
663	}	787	}
664	break;	788	break;
665		789
666	default:	790	default:
667	--dec->cur;	791	--dec_cur;
668	ERR ("illegal backslash escape sequence in string");	792	ERR ("illegal backslash escape sequence in string");
669	}	793	}
670	}	794	}
671	else if (ch >= 0x20 && ch <= 0x7f)	795	else if (expect_true (ch >= 0x20 && ch <= 0x7f))
672	*cur++ = ch;	796	*cur++ = ch;
673	else if (ch >= 0x80)	797	else if (ch >= 0x80)
674	{	798	{
675	--dec->cur;
676
677	STRLEN clen;	799	STRLEN clen;
		800	UV uch;
		801
		802	--dec_cur;
		803
678	UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);	804	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
679	if (clen == (STRLEN)-1)	805	if (clen == (STRLEN)-1)
680	ERR ("malformed UTF-8 character in JSON string");	806	ERR ("malformed UTF-8 character in JSON string");
681		807
682	do	808	do
683	cur++ = dec->cur++;	809	cur++ = dec_cur++;
684	while (--clen);	810	while (--clen);
685		811
686	utf8 = 1;	812	utf8 = 1;
687	}	813	}
688	else	814	else
689	{	815	{
690	--dec->cur;	816	--dec_cur;
691		817
692	if (!ch)	818	if (!ch)
693	ERR ("unexpected end of string while parsing JSON string");	819	ERR ("unexpected end of string while parsing JSON string");
694	else	820	else
695	ERR ("invalid character encountered while parsing JSON string");	821	ERR ("invalid character encountered while parsing JSON string");
696	}	822	}
697	}	823	}
698	while (cur < buf + SHORT_STRING_LEN);	824	while (cur < buf + SHORT_STRING_LEN);
699		825
		826	{
700	STRLEN len = cur - buf;	827	STRLEN len = cur - buf;
701		828
702	if (sv)	829	if (sv)
703	{	830	{
704	SvGROW (sv, SvCUR (sv) + len + 1);	831	SvGROW (sv, SvCUR (sv) + len + 1);
705	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	832	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
706	SvCUR_set (sv, SvCUR (sv) + len);	833	SvCUR_set (sv, SvCUR (sv) + len);
707	}	834	}
708	else	835	else
709	sv = newSVpvn (buf, len);	836	sv = newSVpvn (buf, len);
710	}	837	}
		838	}
711	while (*dec->cur != '"');	839	while (*dec_cur != '"');
712		840
713	++dec->cur;	841	++dec_cur;
714		842
715	if (sv)	843	if (sv)
716	{	844	{
717	SvPOK_only (sv);	845	SvPOK_only (sv);
718	*SvEND (sv) = 0;	846	*SvEND (sv) = 0;
…		…
721	SvUTF8_on (sv);	849	SvUTF8_on (sv);
722	}	850	}
723	else	851	else
724	sv = newSVpvn ("", 0);	852	sv = newSVpvn ("", 0);
725		853
		854	dec->cur = dec_cur;
726	return sv;	855	return sv;
727		856
728	fail:	857	fail:
		858	dec->cur = dec_cur;
729	return 0;	859	return 0;
730	}	860	}
731		861
732	static SV *	862	static SV *
733	decode_num (dec_t *dec)	863	decode_num (dec_t *dec)
…		…
791	is_nv = 1;	921	is_nv = 1;
792	}	922	}
793		923
794	if (!is_nv)	924	if (!is_nv)
795	{	925	{
796	UV uv;	926	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
797	int numtype = grok_number (start, dec->cur - start, &uv);	927	if (*start == '-')
798	if (numtype & IS_NUMBER_IN_UV)	928	switch (dec->cur - start)
799	if (numtype & IS_NUMBER_NEG)
800	{	929	{
801	if (uv < (UV)IV_MIN)	930	case 2: return newSViv (-( start [1] - '0' * 1));
802	return newSViv (-(IV)uv);	931	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
		932	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
		933	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
803	}	934	}
		935	else
		936	switch (dec->cur - start)
		937	{
		938	case 1: return newSViv ( start [0] - '0' * 1);
		939	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
		940	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
		941	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		942	}
		943
		944	{
		945	UV uv;
		946	int numtype = grok_number (start, dec->cur - start, &uv);
		947	if (numtype & IS_NUMBER_IN_UV)
		948	if (numtype & IS_NUMBER_NEG)
		949	{
		950	if (uv < (UV)IV_MIN)
		951	return newSViv (-(IV)uv);
		952	}
804	else	953	else
805	return newSVuv (uv);	954	return newSVuv (uv);
		955
		956	// here would likely be the place for bigint support
806	}	957	}
		958	}
807		959
		960	// if we ever support bigint or bigfloat, this is the place for bigfloat
808	return newSVnv (Atof (start));	961	return newSVnv (Atof (start));
809		962
810	fail:	963	fail:
811	return 0;	964	return 0;
812	}	965	}
…		…
856	}	1009	}
857		1010
858	static SV *	1011	static SV *
859	decode_hv (dec_t *dec)	1012	decode_hv (dec_t *dec)
860	{	1013	{
		1014	SV *sv;
861	HV *hv = newHV ();	1015	HV *hv = newHV ();
862		1016
863	DEC_INC_DEPTH;	1017	DEC_INC_DEPTH;
864	decode_ws (dec);	1018	decode_ws (dec);
865		1019
866	if (*dec->cur == '}')	1020	if (*dec->cur == '}')
867	++dec->cur;	1021	++dec->cur;
868	else	1022	else
869	for (;;)	1023	for (;;)
870	{	1024	{
871	SV key, value;
872
873	decode_ws (dec); EXPECT_CH ('"');	1025	decode_ws (dec); EXPECT_CH ('"');
874		1026
875	key = decode_str (dec);	1027	// heuristic: assume that
876	if (!key)	1028	// a) decode_str + hv_store_ent are abysmally slow.
877	goto fail;	1029	// b) most hash keys are short, simple ascii text.
		1030	// => try to "fast-match" such strings to avoid
		1031	// the overhead of decode_str + hv_store_ent.
		1032	{
		1033	SV *value;
		1034	char *p = dec->cur;
		1035	char *e = p + 24; // only try up to 24 bytes
878		1036
879	decode_ws (dec); EXPECT_CH (':');	1037	for (;;)
880
881	value = decode_sv (dec);
882	if (!value)
883	{	1038	{
		1039	// the >= 0x80 is true on most architectures
		1040	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
		1041	{
		1042	// slow path, back up and use decode_str
		1043	SV *key = decode_str (dec);
		1044	if (!key)
		1045	goto fail;
		1046
		1047	decode_ws (dec); EXPECT_CH (':');
		1048
		1049	value = decode_sv (dec);
		1050	if (!value)
		1051	{
		1052	SvREFCNT_dec (key);
		1053	goto fail;
		1054	}
		1055
		1056	hv_store_ent (hv, key, value, 0);
884	SvREFCNT_dec (key);	1057	SvREFCNT_dec (key);
		1058
		1059	break;
		1060	}
		1061	else if (*p == '"')
		1062	{
		1063	// fast path, got a simple key
		1064	char *key = dec->cur;
		1065	int len = p - key;
		1066	dec->cur = p + 1;
		1067
		1068	decode_ws (dec); EXPECT_CH (':');
		1069
		1070	value = decode_sv (dec);
		1071	if (!value)
885	goto fail;	1072	goto fail;
		1073
		1074	hv_store (hv, key, len, value, 0);
		1075
		1076	break;
		1077	}
		1078
		1079	++p;
886	}	1080	}
887		1081	}
888	hv_store_ent (hv, key, value, 0);
889	SvREFCNT_dec (key);
890		1082
891	decode_ws (dec);	1083	decode_ws (dec);
892		1084
893	if (*dec->cur == '}')	1085	if (*dec->cur == '}')
894	{	1086	{
…		…
901		1093
902	++dec->cur;	1094	++dec->cur;
903	}	1095	}
904		1096
905	DEC_DEC_DEPTH;	1097	DEC_DEC_DEPTH;
		1098	sv = newRV_noinc ((SV *)hv);
		1099
		1100	// check filter callbacks
		1101	if (dec->json.flags & F_HOOK)
		1102	{
		1103	ENTER; SAVETMPS;
		1104
		1105	if (HvKEYS (hv) == 1 && dec->json.cb_sk_object)
		1106	{
		1107	int count;
		1108
		1109	dSP; PUSHMARK (SP);
		1110	XPUSHs (sv_2mortal (sv));
		1111
		1112	PUTBACK; count = call_sv (dec->json.cb_sk_object, G_ARRAY); SPAGAIN;
		1113
		1114	if (count == 1)
		1115	sv = newSVsv (POPs);
		1116	else
		1117	SvREFCNT_inc (sv);
		1118	}
		1119
		1120	if (dec->json.cb_object)
		1121	{
		1122	int count;
		1123
		1124	dSP; ENTER; SAVETMPS; PUSHMARK (SP);
		1125	XPUSHs (sv_2mortal (sv));
		1126
		1127	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
		1128
		1129	if (count == 1)
		1130	sv = newSVsv (POPs);
		1131	else
		1132	SvREFCNT_inc (sv);
		1133	}
		1134
		1135	FREETMPS; LEAVE;
		1136	}
		1137
906	return newRV_noinc ((SV *)hv);	1138	return newRV_noinc ((SV *)hv);
907		1139
908	fail:	1140	fail:
909	SvREFCNT_dec (hv);	1141	SvREFCNT_dec (hv);
910	DEC_DEC_DEPTH;	1142	DEC_DEC_DEPTH;
…		…
913		1145
914	static SV *	1146	static SV *
915	decode_sv (dec_t *dec)	1147	decode_sv (dec_t *dec)
916	{	1148	{
917	decode_ws (dec);	1149	decode_ws (dec);
		1150
		1151	// the beauty of JSON: you need exactly one character lookahead
		1152	// to parse anything.
918	switch (*dec->cur)	1153	switch (*dec->cur)
919	{	1154	{
920	case '"': ++dec->cur; return decode_str (dec);	1155	case '"': ++dec->cur; return decode_str (dec);
921	case '[': ++dec->cur; return decode_av (dec);	1156	case '[': ++dec->cur; return decode_av (dec);
922	case '{': ++dec->cur; return decode_hv (dec);	1157	case '{': ++dec->cur; return decode_hv (dec);
…		…
928		1163
929	case 't':	1164	case 't':
930	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1165	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
931	{	1166	{
932	dec->cur += 4;	1167	dec->cur += 4;
933	return newSViv (1);	1168	return SvREFCNT_inc (json_true);
934	}	1169	}
935	else	1170	else
936	ERR ("'true' expected");	1171	ERR ("'true' expected");
937		1172
938	break;	1173	break;
939		1174
940	case 'f':	1175	case 'f':
941	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1176	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
942	{	1177	{
943	dec->cur += 5;	1178	dec->cur += 5;
944	return newSViv (0);	1179	return SvREFCNT_inc (json_false);
945	}	1180	}
946	else	1181	else
947	ERR ("'false' expected");	1182	ERR ("'false' expected");
948		1183
949	break;	1184	break;
…		…
967	fail:	1202	fail:
968	return 0;	1203	return 0;
969	}	1204	}
970		1205
971	static SV *	1206	static SV *
972	decode_json (SV *string, U32 flags)	1207	decode_json (SV string, JSON json, UV *offset_return)
973	{	1208	{
		1209	dec_t dec;
		1210	UV offset;
974	SV *sv;	1211	SV *sv;
975		1212
		1213	SvGETMAGIC (string);
976	SvUPGRADE (string, SVt_PV);	1214	SvUPGRADE (string, SVt_PV);
977		1215
		1216	if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
		1217	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1218	(unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
		1219
978	if (flags & F_UTF8)	1220	if (json->flags & F_UTF8)
979	sv_utf8_downgrade (string, 0);	1221	sv_utf8_downgrade (string, 0);
980	else	1222	else
981	sv_utf8_upgrade (string);	1223	sv_utf8_upgrade (string);
982		1224
983	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1225	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
984		1226
985	dec_t dec;	1227	dec.json = *json;
986	dec.flags = flags;
987	dec.cur = SvPVX (string);	1228	dec.cur = SvPVX (string);
988	dec.end = SvEND (string);	1229	dec.end = SvEND (string);
989	dec.err = 0;	1230	dec.err = 0;
990	dec.depth = 0;	1231	dec.depth = 0;
991	dec.maxdepth = DEC_DEPTH (dec.flags);	1232	dec.maxdepth = DEC_DEPTH (dec.json.flags);
992		1233
		1234	if (dec.json.cb_object \|\| dec.json.cb_sk_object)
		1235	dec.json.flags \|= F_HOOK;
		1236
993	*dec.end = 0; // this should basically be a nop, too, but make sure its there	1237	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
994	sv = decode_sv (&dec);	1238	sv = decode_sv (&dec);
995		1239
		1240	if (!(offset_return \|\| !sv))
		1241	{
		1242	// check for trailing garbage
		1243	decode_ws (&dec);
		1244
		1245	if (*dec.cur)
		1246	{
		1247	dec.err = "garbage after JSON object";
		1248	SvREFCNT_dec (sv);
		1249	sv = 0;
		1250	}
		1251	}
		1252
		1253	if (offset_return \|\| !sv)
		1254	{
		1255	offset = dec.json.flags & F_UTF8
		1256	? dec.cur - SvPVX (string)
		1257	: utf8_distance (dec.cur, SvPVX (string));
		1258
		1259	if (offset_return)
		1260	*offset_return = offset;
		1261	}
		1262
996	if (!sv)	1263	if (!sv)
997	{	1264	{
998	IV offset = dec.flags & F_UTF8
999	? dec.cur - SvPVX (string)
1000	: utf8_distance (dec.cur, SvPVX (string));
1001	SV *uni = sv_newmortal ();	1265	SV *uni = sv_newmortal ();
1002		1266
1003	// horrible hack to silence warning inside pv_uni_display	1267	// horrible hack to silence warning inside pv_uni_display
1004	COP cop = *PL_curcop;	1268	COP cop = *PL_curcop;
1005	cop.cop_warnings = pWARN_NONE;	1269	cop.cop_warnings = pWARN_NONE;
…		…
1015	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1279	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1016	}	1280	}
1017		1281
1018	sv = sv_2mortal (sv);	1282	sv = sv_2mortal (sv);
1019		1283
1020	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))	1284	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1021	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");	1285	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1022		1286
1023	return sv;	1287	return sv;
1024	}	1288	}
1025		1289
…		…
1029	MODULE = JSON::XS PACKAGE = JSON::XS	1293	MODULE = JSON::XS PACKAGE = JSON::XS
1030		1294
1031	BOOT:	1295	BOOT:
1032	{	1296	{
1033	int i;	1297	int i;
1034
1035	memset (decode_hexdigit, 0xff, 256);
1036		1298
1037	for (i = 0; i < 256; ++i)	1299	for (i = 0; i < 256; ++i)
1038	decode_hexdigit [i] =	1300	decode_hexdigit [i] =
1039	i >= '0' && i <= '9' ? i - '0'	1301	i >= '0' && i <= '9' ? i - '0'
1040	: i >= 'a' && i <= 'f' ? i - 'a' + 10	1302	: i >= 'a' && i <= 'f' ? i - 'a' + 10
1041	: i >= 'A' && i <= 'F' ? i - 'A' + 10	1303	: i >= 'A' && i <= 'F' ? i - 'A' + 10
1042	: -1;	1304	: -1;
1043		1305
1044	json_stash = gv_stashpv ("JSON::XS", 1);	1306	json_stash = gv_stashpv ("JSON::XS" , 1);
		1307	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
		1308
		1309	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
		1310	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1045	}	1311	}
1046		1312
1047	PROTOTYPES: DISABLE	1313	PROTOTYPES: DISABLE
1048		1314
1049	SV new (char dummy)	1315	void new (char *klass)
1050	CODE:	1316	PPCODE:
1051	RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);	1317	{
1052	OUTPUT:	1318	SV *pv = NEWSV (0, sizeof (JSON));
1053	RETVAL	1319	SvPOK_only (pv);
		1320	Zero (SvPVX (pv), 1, JSON);
		1321	((JSON *)SvPVX (pv))->flags = F_DEFAULT;
		1322	XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
		1323	}
1054		1324
1055	SV ascii (SV self, int enable = 1)	1325	void ascii (JSON *self, int enable = 1)
1056	ALIAS:	1326	ALIAS:
1057	ascii = F_ASCII	1327	ascii = F_ASCII
		1328	latin1 = F_LATIN1
1058	utf8 = F_UTF8	1329	utf8 = F_UTF8
1059	indent = F_INDENT	1330	indent = F_INDENT
1060	canonical = F_CANONICAL	1331	canonical = F_CANONICAL
1061	space_before = F_SPACE_BEFORE	1332	space_before = F_SPACE_BEFORE
1062	space_after = F_SPACE_AFTER	1333	space_after = F_SPACE_AFTER
1063	pretty = F_PRETTY	1334	pretty = F_PRETTY
1064	allow_nonref = F_ALLOW_NONREF	1335	allow_nonref = F_ALLOW_NONREF
1065	shrink = F_SHRINK	1336	shrink = F_SHRINK
		1337	allow_blessed = F_ALLOW_BLESSED
		1338	convert_blessed = F_CONV_BLESSED
1066	CODE:	1339	PPCODE:
1067	{	1340	{
1068	UV *uv = SvJSON (self);
1069	if (enable)	1341	if (enable)
1070	*uv \|= ix;	1342	self->flags \|= ix;
1071	else	1343	else
1072	*uv &= ~ix;	1344	self->flags &= ~ix;
1073		1345
1074	RETVAL = newSVsv (self);	1346	XPUSHs (ST (0));
1075	}	1347	}
1076	OUTPUT:
1077	RETVAL
1078		1348
1079	SV max_depth (SV self, int max_depth = 0x80000000UL)	1349	void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1080	CODE:	1350	PPCODE:
1081	{	1351	{
1082	UV *uv = SvJSON (self);
1083	UV log2 = 0;	1352	UV log2 = 0;
1084		1353
1085	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;	1354	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1086		1355
1087	while ((1UL << log2) < max_depth)	1356	while ((1UL << log2) < max_depth)
1088	++log2;	1357	++log2;
1089		1358
1090	uv = uv & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);	1359	self->flags = self->flags & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
1091		1360
1092	RETVAL = newSVsv (self);	1361	XPUSHs (ST (0));
1093	}	1362	}
1094	OUTPUT:
1095	RETVAL
1096		1363
1097	void encode (SV self, SV scalar)	1364	void max_size (JSON *self, UV max_size = 0)
1098	PPCODE:	1365	PPCODE:
1099	XPUSHs (encode_json (scalar, *SvJSON (self)));	1366	{
		1367	UV log2 = 0;
1100		1368
1101	void decode (SV self, SV jsonstr)	1369	if (max_size > 0x80000000UL) max_size = 0x80000000UL;
		1370	if (max_size == 1) max_size = 2;
		1371
		1372	while ((1UL << log2) < max_size)
		1373	++log2;
		1374
		1375	self->flags = self->flags & ~F_MAXSIZE \| (log2 << S_MAXSIZE);
		1376
		1377	XPUSHs (ST (0));
		1378	}
		1379
		1380	void filter_json_objects (JSON self, SV cb = &PL_sv_undef)
		1381	ALIAS:
		1382	filter_sk_json_objects = 1
1102	PPCODE:	1383	PPCODE:
		1384	{
		1385	if (!SvOK (cb))
		1386	cb = 0;
		1387
		1388	switch (ix)
		1389	{
		1390	case 0: self->cb_object = cb; break;
		1391	case 1: self->cb_sk_object = cb; break;
		1392	}
		1393
		1394	XPUSHs (ST (0));
		1395	}
		1396
		1397	void encode (JSON self, SV scalar)
		1398	PPCODE:
		1399	XPUSHs (encode_json (scalar, self));
		1400
		1401	void decode (JSON self, SV jsonstr)
		1402	PPCODE:
1103	XPUSHs (decode_json (jsonstr, *SvJSON (self)));	1403	XPUSHs (decode_json (jsonstr, self, 0));
		1404
		1405	void decode_prefix (JSON self, SV jsonstr)
		1406	PPCODE:
		1407	{
		1408	UV offset;
		1409	EXTEND (SP, 2);
		1410	PUSHs (decode_json (jsonstr, self, &offset));
		1411	PUSHs (sv_2mortal (newSVuv (offset)));
		1412	}
1104		1413
1105	PROTOTYPES: ENABLE	1414	PROTOTYPES: ENABLE
1106		1415
1107	void to_json (SV *scalar)	1416	void to_json (SV *scalar)
1108	ALIAS:
1109	objToJson = 0
1110	PPCODE:	1417	PPCODE:
		1418	{
		1419	JSON json = { F_DEFAULT \| F_UTF8 };
1111	XPUSHs (encode_json (scalar, F_DEFAULT \| F_UTF8));	1420	XPUSHs (encode_json (scalar, &json));
		1421	}
1112		1422
1113	void from_json (SV *jsonstr)	1423	void from_json (SV *jsonstr)
1114	ALIAS:
1115	jsonToObj = 0
1116	PPCODE:	1424	PPCODE:
		1425	{
		1426	JSON json = { F_DEFAULT \| F_UTF8 };
1117	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8));	1427	XPUSHs (decode_json (jsonstr, &json, 0));
		1428	}
1118		1429

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.24 by root, Tue Apr 3 23:59:04 2007 UTC vs. Revision 1.50 by root, Mon Jul 2 00:29:38 2007 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.24 by root, Tue Apr 3 23:59:04 2007 UTC vs.
Revision 1.50 by root, Mon Jul 2 00:29:38 2007 UTC