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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.16 by root, Sun Mar 25 02:37:00 2007 UTC vs.
Revision 1.46 by root, Mon Jun 25 22:11:39 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_ALLOW_NONREF 0x00000080	27	#define F_ALLOW_NONREF 0x00000100UL
16	#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
17		35
18	// F_SKIPINVALID?	36	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
19	// F_EXECCODEREF?	37	#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
20	// F_SELFCONVERT?
21		38
22	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	39	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
23	#define F_DEFAULT 0	40	#define F_DEFAULT (9UL << S_MAXDEPTH)
24		41
25	#define INIT_SIZE 32 // initial scalar size to be allocated	42	#define INIT_SIZE 32 // initial scalar size to be allocated
26	#define INDENT_STEP 3 // spaces per indentation level	43	#define INDENT_STEP 3 // spaces per indentation level
27		44
28	#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
29	#define SHORT_STRING_LEN 512 // special-case strings of up to this size	45	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
30		46
31	#define SB do {	47	#define SB do {
32	#define SE } while (0)	48	#define SE } while (0)
33		49
		50	#if __GNUC__ >= 3
		51	# define expect(expr,value) __builtin_expect ((expr),(value))
		52	# define inline inline
		53	#else
		54	# define expect(expr,value) (expr)
		55	# define inline static
		56	#endif
		57
		58	#define expect_false(expr) expect ((expr) != 0, 0)
		59	#define expect_true(expr) expect ((expr) != 0, 1)
		60
34	static HV *json_stash; // JSON::XS::	61	static HV json_stash, json_boolean_stash; // JSON::XS::
		62	static SV json_true, json_false;
35		63
36	/////////////////////////////////////////////////////////////////////////////	64	/////////////////////////////////////////////////////////////////////////////
37	// utility functions	65	// utility functions
38		66
39	static UV *	67	static UV *
…		…
62	// decode an utf-8 character and return it, or (UV)-1 in	90	// decode an utf-8 character and return it, or (UV)-1 in
63	// case of an error.	91	// case of an error.
64	// we special-case "safe" characters from U+80 .. U+7FF,	92	// we special-case "safe" characters from U+80 .. U+7FF,
65	// but use the very good perl function to parse anything else.	93	// but use the very good perl function to parse anything else.
66	// note that we never call this function for a ascii codepoints	94	// note that we never call this function for a ascii codepoints
67	static UV	95	inline UV
68	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)	96	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
69	{	97	{
70	if (s[0] > 0xdf \|\| s[0] < 0xc2)	98	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))
71	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);	99	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
72	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)	100	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
73	{	101	{
74	*clen = 2;	102	*clen = 2;
75	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);	103	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
76	}	104	}
77	else	105	else
		106	{
		107	*clen = (STRLEN)-1;
78	return (UV)-1;	108	return (UV)-1;
		109	}
79	}	110	}
80		111
81	/////////////////////////////////////////////////////////////////////////////	112	/////////////////////////////////////////////////////////////////////////////
82	// encoder	113	// encoder
83		114
…		…
85	typedef struct	116	typedef struct
86	{	117	{
87	char *cur; // SvPVX (sv) + current output position	118	char *cur; // SvPVX (sv) + current output position
88	char *end; // SvEND (sv)	119	char *end; // SvEND (sv)
89	SV *sv; // result scalar	120	SV *sv; // result scalar
90	UV flags; // F_*	121	U32 flags; // F_*
91	int indent; // indentation level	122	U32 indent; // indentation level
92	int max_depth; // max. recursion level	123	U32 maxdepth; // max. indentation/recursion level
93	} enc_t;	124	} enc_t;
94		125
95	static void	126	inline void
96	need (enc_t *enc, STRLEN len)	127	need (enc_t *enc, STRLEN len)
97	{	128	{
98	if (enc->cur + len >= enc->end)	129	if (expect_false (enc->cur + len >= enc->end))
99	{	130	{
100	STRLEN cur = enc->cur - SvPVX (enc->sv);	131	STRLEN cur = enc->cur - SvPVX (enc->sv);
101	SvGROW (enc->sv, cur + len + 1);	132	SvGROW (enc->sv, cur + len + 1);
102	enc->cur = SvPVX (enc->sv) + cur;	133	enc->cur = SvPVX (enc->sv) + cur;
103	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);	134	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
104	}	135	}
105	}	136	}
106		137
107	static void	138	inline void
108	encode_ch (enc_t *enc, char ch)	139	encode_ch (enc_t *enc, char ch)
109	{	140	{
110	need (enc, 1);	141	need (enc, 1);
111	*enc->cur++ = ch;	142	*enc->cur++ = ch;
112	}	143	}
…		…
120		151
121	while (str < end)	152	while (str < end)
122	{	153	{
123	unsigned char ch = (unsigned char )str;	154	unsigned char ch = (unsigned char )str;
124		155
125	if (ch >= 0x20 && ch < 0x80) // most common case	156	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
126	{	157	{
127	if (ch == '"') // but with slow exceptions	158	if (expect_false (ch == '"')) // but with slow exceptions
128	{	159	{
129	need (enc, len += 1);	160	need (enc, len += 1);
130	*enc->cur++ = '\\';	161	*enc->cur++ = '\\';
131	*enc->cur++ = '"';	162	*enc->cur++ = '"';
132	}	163	}
133	else if (ch == '\\')	164	else if (expect_false (ch == '\\'))
134	{	165	{
135	need (enc, len += 1);	166	need (enc, len += 1);
136	*enc->cur++ = '\\';	167	*enc->cur++ = '\\';
137	*enc->cur++ = '\\';	168	*enc->cur++ = '\\';
138	}	169	}
…		…
156	STRLEN clen;	187	STRLEN clen;
157	UV uch;	188	UV uch;
158		189
159	if (is_utf8)	190	if (is_utf8)
160	{	191	{
161	//uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
162	uch = decode_utf8 (str, end - str, &clen);	192	uch = decode_utf8 (str, end - str, &clen);
163	if (clen == (STRLEN)-1)	193	if (clen == (STRLEN)-1)
164	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);	194	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
165	}	195	}
166	else	196	else
…		…
170	}	200	}
171		201
172	if (uch > 0x10FFFFUL)	202	if (uch > 0x10FFFFUL)
173	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);	203	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
174		204
175	if (uch < 0x80 \|\| enc->flags & F_ASCII)	205	if (uch < 0x80 \|\| enc->flags & F_ASCII \|\| (enc->flags & F_LATIN1 && uch > 0xFF))
176	{	206	{
177	if (uch > 0xFFFFUL)	207	if (uch > 0xFFFFUL)
178	{	208	{
179	need (enc, len += 11);	209	need (enc, len += 11);
180	sprintf (enc->cur, "\\u%04x\\u%04x",	210	sprintf (enc->cur, "\\u%04x\\u%04x",
…		…
194	*enc->cur++ = hexdigit [(uch >> 0) & 15];	224	*enc->cur++ = hexdigit [(uch >> 0) & 15];
195	}	225	}
196		226
197	str += clen;	227	str += clen;
198	}	228	}
		229	else if (enc->flags & F_LATIN1)
		230	{
		231	*enc->cur++ = uch;
		232	str += clen;
		233	}
199	else if (is_utf8)	234	else if (is_utf8)
200	{	235	{
201	need (enc, len += clen);	236	need (enc, len += clen);
202	do	237	do
203	{	238	{
…		…
205	}	240	}
206	while (--clen);	241	while (--clen);
207	}	242	}
208	else	243	else
209	{	244	{
210	need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed	245	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
211	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	246	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
212	++str;	247	++str;
213	}	248	}
214	}	249	}
215	}	250	}
…		…
217		252
218	--len;	253	--len;
219	}	254	}
220	}	255	}
221		256
222	static void	257	inline void
223	encode_indent (enc_t *enc)	258	encode_indent (enc_t *enc)
224	{	259	{
225	if (enc->flags & F_INDENT)	260	if (enc->flags & F_INDENT)
226	{	261	{
227	int spaces = enc->indent * INDENT_STEP;	262	int spaces = enc->indent * INDENT_STEP;
…		…
230	memset (enc->cur, ' ', spaces);	265	memset (enc->cur, ' ', spaces);
231	enc->cur += spaces;	266	enc->cur += spaces;
232	}	267	}
233	}	268	}
234		269
235	static void	270	inline void
236	encode_space (enc_t *enc)	271	encode_space (enc_t *enc)
237	{	272	{
238	need (enc, 1);	273	need (enc, 1);
239	encode_ch (enc, ' ');	274	encode_ch (enc, ' ');
240	}	275	}
241		276
242	static void	277	inline void
243	encode_nl (enc_t *enc)	278	encode_nl (enc_t *enc)
244	{	279	{
245	if (enc->flags & F_INDENT)	280	if (enc->flags & F_INDENT)
246	{	281	{
247	need (enc, 1);	282	need (enc, 1);
248	encode_ch (enc, '\n');	283	encode_ch (enc, '\n');
249	}	284	}
250	}	285	}
251		286
252	static void	287	inline void
253	encode_comma (enc_t *enc)	288	encode_comma (enc_t *enc)
254	{	289	{
255	encode_ch (enc, ',');	290	encode_ch (enc, ',');
256		291
257	if (enc->flags & F_INDENT)	292	if (enc->flags & F_INDENT)
…		…
264		299
265	static void	300	static void
266	encode_av (enc_t enc, AV av)	301	encode_av (enc_t enc, AV av)
267	{	302	{
268	int i, len = av_len (av);	303	int i, len = av_len (av);
		304
		305	if (enc->indent >= enc->maxdepth)
		306	croak ("data structure too deep (hit recursion limit)");
269		307
270	encode_ch (enc, '['); encode_nl (enc);	308	encode_ch (enc, '['); encode_nl (enc);
271	++enc->indent;	309	++enc->indent;
272		310
273	for (i = 0; i <= len; ++i)	311	for (i = 0; i <= len; ++i)
…		…
340	static void	378	static void
341	encode_hv (enc_t enc, HV hv)	379	encode_hv (enc_t enc, HV hv)
342	{	380	{
343	int count, i;	381	int count, i;
344		382
		383	if (enc->indent >= enc->maxdepth)
		384	croak ("data structure too deep (hit recursion limit)");
		385
345	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;	386	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
346		387
347	if ((count = hv_iterinit (hv)))	388	if ((count = hv_iterinit (hv)))
348	{	389	{
349	// for canonical output we have to sort by keys first	390	// for canonical output we have to sort by keys first
350	// actually, this is mostly due to the stupid so-called	391	// actually, this is mostly due to the stupid so-called
351	// security workaround added somewhere in 5.8.x.	392	// security workaround added somewhere in 5.8.x.
352	// that randomises hash orderings	393	// that randomises hash orderings
353	if (enc->flags & F_CANONICAL)	394	if (enc->flags & F_CANONICAL)
354	{	395	{
355	HE he, hes [count]; // if your compiler dies here, you need to enable C99 mode
356	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
357		403
358	i = 0;	404	i = 0;
359	while ((he = hv_iternext (hv)))	405	while ((he = hv_iternext (hv)))
360	{	406	{
361	hes [i++] = he;	407	hes [i++] = he;
…		…
396		442
397	encode_nl (enc);	443	encode_nl (enc);
398	}	444	}
399	else	445	else
400	{	446	{
401	SV *sv;
402	HE *he = hv_iternext (hv);	447	HE *he = hv_iternext (hv);
403		448
404	for (;;)	449	for (;;)
405	{	450	{
406	encode_indent (enc);	451	encode_indent (enc);
…		…
415	encode_nl (enc);	460	encode_nl (enc);
416	}	461	}
417	}	462	}
418		463
419	--enc->indent; encode_indent (enc); 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))));
420	}	547	}
421		548
422	static void	549	static void
423	encode_sv (enc_t enc, SV sv)	550	encode_sv (enc_t enc, SV sv)
424	{	551	{
…		…
432	encode_str (enc, str, len, SvUTF8 (sv));	559	encode_str (enc, str, len, SvUTF8 (sv));
433	encode_ch (enc, '"');	560	encode_ch (enc, '"');
434	}	561	}
435	else if (SvNOKp (sv))	562	else if (SvNOKp (sv))
436	{	563	{
		564	// trust that perl will do the right thing w.r.t. JSON syntax.
437	need (enc, NV_DIG + 32);	565	need (enc, NV_DIG + 32);
438	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	566	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
439	enc->cur += strlen (enc->cur);	567	enc->cur += strlen (enc->cur);
440	}	568	}
441	else if (SvIOKp (sv))	569	else if (SvIOKp (sv))
442	{	570	{
443	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);
444	enc->cur +=	576	enc->cur +=
445	SvIsUV(sv)	577	SvIsUV(sv)
446	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	578	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
447	: 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	}
448	}	608	}
449	else if (SvROK (sv))	609	else if (SvROK (sv))
450	{	610	encode_rv (enc, SvRV (sv));
451	SV *rv = SvRV (sv);
452
453	if (enc->indent >= enc->max_depth)
454	croak ("data structure too deep (hit recursion limit)");
455
456	switch (SvTYPE (rv))
457	{
458	case SVt_PVAV: encode_av (enc, (AV *)rv); break;
459	case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
460
461	default:
462	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
463	SvPV_nolen (sv));
464	}
465	}
466	else if (!SvOK (sv))	611	else if (!SvOK (sv))
467	encode_str (enc, "null", 4, 0);	612	encode_str (enc, "null", 4, 0);
468	else	613	else
469	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	614	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
470	SvPV_nolen (sv), SvFLAGS (sv));	615	SvPV_nolen (sv), SvFLAGS (sv));
471	}	616	}
472		617
473	static SV *	618	static SV *
474	encode_json (SV *scalar, UV flags)	619	encode_json (SV *scalar, U32 flags)
475	{	620	{
		621	enc_t enc;
		622
476	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	623	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
477	croak ("hash- or arrayref expected (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)");
478		625
479	enc_t enc;
480	enc.flags = flags;	626	enc.flags = flags;
481	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	627	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
482	enc.cur = SvPVX (enc.sv);	628	enc.cur = SvPVX (enc.sv);
483	enc.end = SvEND (enc.sv);	629	enc.end = SvEND (enc.sv);
484	enc.indent = 0;	630	enc.indent = 0;
485	enc.max_depth = 0x7fffffffUL;	631	enc.maxdepth = DEC_DEPTH (flags);
486		632
487	SvPOK_only (enc.sv);	633	SvPOK_only (enc.sv);
488	encode_sv (&enc, scalar);	634	encode_sv (&enc, scalar);
489		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
490	if (!(flags & (F_ASCII \| F_UTF8)))	639	if (!(flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
491	SvUTF8_on (enc.sv);	640	SvUTF8_on (enc.sv);
492
493	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
494		641
495	if (enc.flags & F_SHRINK)	642	if (enc.flags & F_SHRINK)
496	shrink (enc.sv);	643	shrink (enc.sv);
497		644
498	return enc.sv;	645	return enc.sv;
…		…
505	typedef struct	652	typedef struct
506	{	653	{
507	char *cur; // current parser pointer	654	char *cur; // current parser pointer
508	char *end; // end of input string	655	char *end; // end of input string
509	const char *err; // parse error, if != 0	656	const char *err; // parse error, if != 0
510	UV flags; // F_*	657	U32 flags; // F_*
		658	U32 depth; // recursion depth
		659	U32 maxdepth; // recursion depth limit
511	} dec_t;	660	} dec_t;
512		661
513	static void	662	inline void
514	decode_ws (dec_t *dec)	663	decode_ws (dec_t *dec)
515	{	664	{
516	for (;;)	665	for (;;)
517	{	666	{
518	char ch = *dec->cur;	667	char ch = *dec->cur;
…		…
524	++dec->cur;	673	++dec->cur;
525	}	674	}
526	}	675	}
527		676
528	#define ERR(reason) SB dec->err = reason; goto fail; SE	677	#define ERR(reason) SB dec->err = reason; goto fail; SE
		678
529	#define EXPECT_CH(ch) SB \	679	#define EXPECT_CH(ch) SB \
530	if (*dec->cur != ch) \	680	if (*dec->cur != ch) \
531	ERR (# ch " expected"); \	681	ERR (# ch " expected"); \
532	++dec->cur; \	682	++dec->cur; \
533	SE	683	SE
534		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
535	static SV decode_sv (dec_t dec);	688	static SV decode_sv (dec_t dec);
536		689
537	static signed char decode_hexdigit[256];	690	static signed char decode_hexdigit[256];
538		691
539	static UV	692	static UV
540	decode_4hex (dec_t *dec)	693	decode_4hex (dec_t *dec)
541	{	694	{
542	signed char d1, d2, d3, d4;	695	signed char d1, d2, d3, d4;
543	unsigned char cur = (unsigned char )dec->cur;	696	unsigned char cur = (unsigned char )dec->cur;
544		697
545	d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");	698	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
546	d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");	699	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
547	d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");	700	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
548	d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");	701	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
549		702
550	dec->cur += 4;	703	dec->cur += 4;
551		704
552	return ((UV)d1) << 12	705	return ((UV)d1) << 12
553	\| ((UV)d2) << 8	706	\| ((UV)d2) << 8
…		…
561	static SV *	714	static SV *
562	decode_str (dec_t *dec)	715	decode_str (dec_t *dec)
563	{	716	{
564	SV *sv = 0;	717	SV *sv = 0;
565	int utf8 = 0;	718	int utf8 = 0;
		719	char *dec_cur = dec->cur;
566		720
567	do	721	do
568	{	722	{
569	char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];	723	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
570	char *cur = buf;	724	char *cur = buf;
571		725
572	do	726	do
573	{	727	{
574	unsigned char ch = (unsigned char )dec->cur++;	728	unsigned char ch = (unsigned char )dec_cur++;
575		729
576	if (ch == '"')	730	if (expect_false (ch == '"'))
577	{	731	{
578	--dec->cur;	732	--dec_cur;
579	break;	733	break;
580	}	734	}
581	else if (ch == '\\')	735	else if (expect_false (ch == '\\'))
582	{	736	{
583	switch (*dec->cur)	737	switch (*dec_cur)
584	{	738	{
585	case '\\':	739	case '\\':
586	case '/':	740	case '/':
587	case '"': cur++ = dec->cur++; break;	741	case '"': cur++ = dec_cur++; break;
588		742
589	case 'b': ++dec->cur; *cur++ = '\010'; break;	743	case 'b': ++dec_cur; *cur++ = '\010'; break;
590	case 't': ++dec->cur; *cur++ = '\011'; break;	744	case 't': ++dec_cur; *cur++ = '\011'; break;
591	case 'n': ++dec->cur; *cur++ = '\012'; break;	745	case 'n': ++dec_cur; *cur++ = '\012'; break;
592	case 'f': ++dec->cur; *cur++ = '\014'; break;	746	case 'f': ++dec_cur; *cur++ = '\014'; break;
593	case 'r': ++dec->cur; *cur++ = '\015'; break;	747	case 'r': ++dec_cur; *cur++ = '\015'; break;
594		748
595	case 'u':	749	case 'u':
596	{	750	{
597	UV lo, hi;	751	UV lo, hi;
598	++dec->cur;	752	++dec_cur;
599		753
		754	dec->cur = dec_cur;
600	hi = decode_4hex (dec);	755	hi = decode_4hex (dec);
		756	dec_cur = dec->cur;
601	if (hi == (UV)-1)	757	if (hi == (UV)-1)
602	goto fail;	758	goto fail;
603		759
604	// possibly a surrogate pair	760	// possibly a surrogate pair
605	if (hi >= 0xd800)	761	if (hi >= 0xd800)
606	if (hi < 0xdc00)	762	if (hi < 0xdc00)
607	{	763	{
608	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')	764	if (dec_cur [0] != '\\' \|\| dec_cur [1] != 'u')
609	ERR ("missing low surrogate character in surrogate pair");	765	ERR ("missing low surrogate character in surrogate pair");
610		766
611	dec->cur += 2;	767	dec_cur += 2;
612		768
		769	dec->cur = dec_cur;
613	lo = decode_4hex (dec);	770	lo = decode_4hex (dec);
		771	dec_cur = dec->cur;
614	if (lo == (UV)-1)	772	if (lo == (UV)-1)
615	goto fail;	773	goto fail;
616		774
617	if (lo < 0xdc00 \|\| lo >= 0xe000)	775	if (lo < 0xdc00 \|\| lo >= 0xe000)
618	ERR ("surrogate pair expected");	776	ERR ("surrogate pair expected");
…		…
632	*cur++ = hi;	790	*cur++ = hi;
633	}	791	}
634	break;	792	break;
635		793
636	default:	794	default:
637	--dec->cur;	795	--dec_cur;
638	ERR ("illegal backslash escape sequence in string");	796	ERR ("illegal backslash escape sequence in string");
639	}	797	}
640	}	798	}
641	else if (ch >= 0x20 && ch <= 0x7f)	799	else if (expect_true (ch >= 0x20 && ch <= 0x7f))
642	*cur++ = ch;	800	*cur++ = ch;
643	else if (ch >= 0x80)	801	else if (ch >= 0x80)
644	{	802	{
645	--dec->cur;
646
647	STRLEN clen;	803	STRLEN clen;
		804	UV uch;
		805
		806	--dec_cur;
		807
648	UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);	808	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
649	if (clen == (STRLEN)-1)	809	if (clen == (STRLEN)-1)
650	ERR ("malformed UTF-8 character in JSON string");	810	ERR ("malformed UTF-8 character in JSON string");
651		811
652	do	812	do
653	{
654	cur++ = dec->cur++;	813	cur++ = dec_cur++;
655	}
656	while (--clen);	814	while (--clen);
657		815
658	utf8 = 1;	816	utf8 = 1;
659	}	817	}
660	else if (!ch)
661	ERR ("unexpected end of string while parsing json string");
662	else	818	else
		819	{
		820	--dec_cur;
		821
		822	if (!ch)
		823	ERR ("unexpected end of string while parsing JSON string");
		824	else
663	ERR ("invalid character encountered");	825	ERR ("invalid character encountered while parsing JSON string");
664		826	}
665	}	827	}
666	while (cur < buf + SHORT_STRING_LEN);	828	while (cur < buf + SHORT_STRING_LEN);
667		829
		830	{
668	STRLEN len = cur - buf;	831	STRLEN len = cur - buf;
669		832
670	if (sv)	833	if (sv)
671	{	834	{
672	SvGROW (sv, SvCUR (sv) + len + 1);	835	SvGROW (sv, SvCUR (sv) + len + 1);
673	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	836	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
674	SvCUR_set (sv, SvCUR (sv) + len);	837	SvCUR_set (sv, SvCUR (sv) + len);
675	}	838	}
676	else	839	else
677	sv = newSVpvn (buf, len);	840	sv = newSVpvn (buf, len);
678	}	841	}
		842	}
679	while (*dec->cur != '"');	843	while (*dec_cur != '"');
680		844
681	++dec->cur;	845	++dec_cur;
682		846
683	if (sv)	847	if (sv)
684	{	848	{
685	SvPOK_only (sv);	849	SvPOK_only (sv);
686	*SvEND (sv) = 0;	850	*SvEND (sv) = 0;
…		…
689	SvUTF8_on (sv);	853	SvUTF8_on (sv);
690	}	854	}
691	else	855	else
692	sv = newSVpvn ("", 0);	856	sv = newSVpvn ("", 0);
693		857
		858	dec->cur = dec_cur;
694	return sv;	859	return sv;
695		860
696	fail:	861	fail:
		862	dec->cur = dec_cur;
697	return 0;	863	return 0;
698	}	864	}
699		865
700	static SV *	866	static SV *
701	decode_num (dec_t *dec)	867	decode_num (dec_t *dec)
…		…
759	is_nv = 1;	925	is_nv = 1;
760	}	926	}
761		927
762	if (!is_nv)	928	if (!is_nv)
763	{	929	{
764	UV uv;	930	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
765	int numtype = grok_number (start, dec->cur - start, &uv);	931	if (*start == '-')
766	if (numtype & IS_NUMBER_IN_UV)	932	switch (dec->cur - start)
767	if (numtype & IS_NUMBER_NEG)
768	{	933	{
769	if (uv < (UV)IV_MIN)	934	case 2: return newSViv (-( start [1] - '0' * 1));
770	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));
771	}	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	}
772	else	957	else
773	return newSVuv (uv);	958	return newSVuv (uv);
		959
		960	// here would likely be the place for bigint support
774	}	961	}
		962	}
775		963
		964	// if we ever support bigint or bigfloat, this is the place for bigfloat
776	return newSVnv (Atof (start));	965	return newSVnv (Atof (start));
777		966
778	fail:	967	fail:
779	return 0;	968	return 0;
780	}	969	}
…		…
782	static SV *	971	static SV *
783	decode_av (dec_t *dec)	972	decode_av (dec_t *dec)
784	{	973	{
785	AV *av = newAV ();	974	AV *av = newAV ();
786		975
		976	DEC_INC_DEPTH;
787	decode_ws (dec);	977	decode_ws (dec);
		978
788	if (*dec->cur == ']')	979	if (*dec->cur == ']')
789	++dec->cur;	980	++dec->cur;
790	else	981	else
791	for (;;)	982	for (;;)
792	{	983	{
…		…
810	ERR (", or ] expected while parsing array");	1001	ERR (", or ] expected while parsing array");
811		1002
812	++dec->cur;	1003	++dec->cur;
813	}	1004	}
814		1005
		1006	DEC_DEC_DEPTH;
815	return newRV_noinc ((SV *)av);	1007	return newRV_noinc ((SV *)av);
816		1008
817	fail:	1009	fail:
818	SvREFCNT_dec (av);	1010	SvREFCNT_dec (av);
		1011	DEC_DEC_DEPTH;
819	return 0;	1012	return 0;
820	}	1013	}
821		1014
822	static SV *	1015	static SV *
823	decode_hv (dec_t *dec)	1016	decode_hv (dec_t *dec)
824	{	1017	{
825	HV *hv = newHV ();	1018	HV *hv = newHV ();
826		1019
		1020	DEC_INC_DEPTH;
827	decode_ws (dec);	1021	decode_ws (dec);
		1022
828	if (*dec->cur == '}')	1023	if (*dec->cur == '}')
829	++dec->cur;	1024	++dec->cur;
830	else	1025	else
831	for (;;)	1026	for (;;)
832	{	1027	{
833	SV key, value;
834
835	decode_ws (dec); EXPECT_CH ('"');	1028	decode_ws (dec); EXPECT_CH ('"');
836		1029
837	key = decode_str (dec);	1030	// heuristic: assume that
838	if (!key)	1031	// a) decode_str + hv_store_ent are abysmally slow
839	goto fail;	1032	// b) most hash keys are short, simple ascii text
		1033	// so try to "fast-match" such strings to avoid
		1034	// the overhead of hv_store_ent.
		1035	{
		1036	SV *value;
		1037	char *p = dec->cur;
		1038	char *e = p + 24; // only try up to 24 bytes
840		1039
841	decode_ws (dec); EXPECT_CH (':');	1040	for (;;)
842
843	value = decode_sv (dec);
844	if (!value)
845	{	1041	{
		1042	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
		1043	{
		1044	// slow path, back up and use decode_str
		1045	SV *key = decode_str (dec);
		1046	if (!key)
		1047	goto fail;
		1048
		1049	decode_ws (dec); EXPECT_CH (':');
		1050
		1051	value = decode_sv (dec);
		1052	if (!value)
		1053	{
		1054	SvREFCNT_dec (key);
		1055	goto fail;
		1056	}
		1057
		1058	hv_store_ent (hv, key, value, 0);
846	SvREFCNT_dec (key);	1059	SvREFCNT_dec (key);
		1060
		1061	break;
		1062	}
		1063	else if (*p == '"')
		1064	{
		1065	// fast path, got a simple key
		1066	char *key = dec->cur;
		1067	int len = p - key;
		1068	dec->cur = p + 1;
		1069
		1070	decode_ws (dec); EXPECT_CH (':');
		1071
		1072	value = decode_sv (dec);
		1073	if (!value)
847	goto fail;	1074	goto fail;
		1075
		1076	hv_store (hv, key, len, value, 0);
		1077
		1078	break;
		1079	}
		1080
		1081	++p;
848	}	1082	}
849		1083	}
850	//TODO: optimise
851	hv_store_ent (hv, key, value, 0);
852		1084
853	decode_ws (dec);	1085	decode_ws (dec);
854		1086
855	if (*dec->cur == '}')	1087	if (*dec->cur == '}')
856	{	1088	{
…		…
862	ERR (", or } expected while parsing object/hash");	1094	ERR (", or } expected while parsing object/hash");
863		1095
864	++dec->cur;	1096	++dec->cur;
865	}	1097	}
866		1098
		1099	DEC_DEC_DEPTH;
867	return newRV_noinc ((SV *)hv);	1100	return newRV_noinc ((SV *)hv);
868		1101
869	fail:	1102	fail:
870	SvREFCNT_dec (hv);	1103	SvREFCNT_dec (hv);
		1104	DEC_DEC_DEPTH;
871	return 0;	1105	return 0;
872	}	1106	}
873		1107
874	static SV *	1108	static SV *
875	decode_sv (dec_t *dec)	1109	decode_sv (dec_t *dec)
876	{	1110	{
877	decode_ws (dec);	1111	decode_ws (dec);
		1112
		1113	// the beauty of JSON: you need exactly one character lookahead
		1114	// to parse anything.
878	switch (*dec->cur)	1115	switch (*dec->cur)
879	{	1116	{
880	case '"': ++dec->cur; return decode_str (dec);	1117	case '"': ++dec->cur; return decode_str (dec);
881	case '[': ++dec->cur; return decode_av (dec);	1118	case '[': ++dec->cur; return decode_av (dec);
882	case '{': ++dec->cur; return decode_hv (dec);	1119	case '{': ++dec->cur; return decode_hv (dec);
…		…
888		1125
889	case 't':	1126	case 't':
890	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1127	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
891	{	1128	{
892	dec->cur += 4;	1129	dec->cur += 4;
893	return newSViv (1);	1130	return SvREFCNT_inc (json_true);
894	}	1131	}
895	else	1132	else
896	ERR ("'true' expected");	1133	ERR ("'true' expected");
897		1134
898	break;	1135	break;
899		1136
900	case 'f':	1137	case 'f':
901	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1138	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
902	{	1139	{
903	dec->cur += 5;	1140	dec->cur += 5;
904	return newSViv (0);	1141	return SvREFCNT_inc (json_false);
905	}	1142	}
906	else	1143	else
907	ERR ("'false' expected");	1144	ERR ("'false' expected");
908		1145
909	break;	1146	break;
…		…
918	ERR ("'null' expected");	1155	ERR ("'null' expected");
919		1156
920	break;	1157	break;
921		1158
922	default:	1159	default:
923	ERR ("malformed json string, neither array, object, number, string or atom");	1160	ERR ("malformed JSON string, neither array, object, number, string or atom");
924	break;	1161	break;
925	}	1162	}
926		1163
927	fail:	1164	fail:
928	return 0;	1165	return 0;
929	}	1166	}
930		1167
931	static SV *	1168	static SV *
932	decode_json (SV *string, UV flags)	1169	decode_json (SV string, U32 flags, UV offset_return)
933	{	1170	{
		1171	dec_t dec;
		1172	UV offset;
934	SV *sv;	1173	SV *sv;
		1174
		1175	SvGETMAGIC (string);
		1176	SvUPGRADE (string, SVt_PV);
		1177
		1178	if (flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (flags))
		1179	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1180	(unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (flags));
935		1181
936	if (flags & F_UTF8)	1182	if (flags & F_UTF8)
937	sv_utf8_downgrade (string, 0);	1183	sv_utf8_downgrade (string, 0);
938	else	1184	else
939	sv_utf8_upgrade (string);	1185	sv_utf8_upgrade (string);
940		1186
941	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1187	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
942		1188
943	dec_t dec;
944	dec.flags = flags;	1189	dec.flags = flags;
945	dec.cur = SvPVX (string);	1190	dec.cur = SvPVX (string);
946	dec.end = SvEND (string);	1191	dec.end = SvEND (string);
947	dec.err = 0;	1192	dec.err = 0;
		1193	dec.depth = 0;
		1194	dec.maxdepth = DEC_DEPTH (dec.flags);
948		1195
		1196	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
949	sv = decode_sv (&dec);	1197	sv = decode_sv (&dec);
950		1198
		1199	if (!(offset_return \|\| !sv))
		1200	{
		1201	// check for trailing garbage
		1202	decode_ws (&dec);
		1203
		1204	if (*dec.cur)
		1205	{
		1206	dec.err = "garbage after JSON object";
		1207	SvREFCNT_dec (sv);
		1208	sv = 0;
		1209	}
		1210	}
		1211
		1212	if (offset_return \|\| !sv)
		1213	{
		1214	offset = dec.flags & F_UTF8
		1215	? dec.cur - SvPVX (string)
		1216	: utf8_distance (dec.cur, SvPVX (string));
		1217
		1218	if (offset_return)
		1219	*offset_return = offset;
		1220	}
		1221
951	if (!sv)	1222	if (!sv)
952	{	1223	{
953	IV offset = dec.flags & F_UTF8
954	? dec.cur - SvPVX (string)
955	: utf8_distance (dec.cur, SvPVX (string));
956	SV *uni = sv_newmortal ();	1224	SV *uni = sv_newmortal ();
957		1225
958	// horrible hack to silence warning inside pv_uni_display	1226	// horrible hack to silence warning inside pv_uni_display
959	COP cop = *PL_curcop;	1227	COP cop = *PL_curcop;
960	cop.cop_warnings = pWARN_NONE;	1228	cop.cop_warnings = pWARN_NONE;
…		…
962	SAVEVPTR (PL_curcop);	1230	SAVEVPTR (PL_curcop);
963	PL_curcop = &cop;	1231	PL_curcop = &cop;
964	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1232	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
965	LEAVE;	1233	LEAVE;
966		1234
967	croak ("%s, at character offset %d (%s)",	1235	croak ("%s, at character offset %d [\"%s\"]",
968	dec.err,	1236	dec.err,
969	(int)offset,	1237	(int)offset,
970	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1238	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
971	}	1239	}
972		1240
…		…
985		1253
986	BOOT:	1254	BOOT:
987	{	1255	{
988	int i;	1256	int i;
989		1257
990	memset (decode_hexdigit, 0xff, 256);
991	for (i = 10; i--; )	1258	for (i = 0; i < 256; ++i)
992	decode_hexdigit ['0' + i] = i;	1259	decode_hexdigit [i] =
		1260	i >= '0' && i <= '9' ? i - '0'
		1261	: i >= 'a' && i <= 'f' ? i - 'a' + 10
		1262	: i >= 'A' && i <= 'F' ? i - 'A' + 10
		1263	: -1;
993		1264
994	for (i = 7; i--; )
995	{
996	decode_hexdigit ['a' + i] = 10 + i;
997	decode_hexdigit ['A' + i] = 10 + i;
998	}
999
1000	json_stash = gv_stashpv ("JSON::XS", 1);	1265	json_stash = gv_stashpv ("JSON::XS" , 1);
		1266	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
		1267
		1268	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
		1269	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1001	}	1270	}
1002		1271
1003	PROTOTYPES: DISABLE	1272	PROTOTYPES: DISABLE
1004		1273
1005	SV new (char dummy)	1274	SV new (char dummy)
…		…
1008	OUTPUT:	1277	OUTPUT:
1009	RETVAL	1278	RETVAL
1010		1279
1011	SV ascii (SV self, int enable = 1)	1280	SV ascii (SV self, int enable = 1)
1012	ALIAS:	1281	ALIAS:
1013	ascii = F_ASCII	1282	ascii = F_ASCII
		1283	latin1 = F_LATIN1
1014	utf8 = F_UTF8	1284	utf8 = F_UTF8
1015	indent = F_INDENT	1285	indent = F_INDENT
1016	canonical = F_CANONICAL	1286	canonical = F_CANONICAL
1017	space_before = F_SPACE_BEFORE	1287	space_before = F_SPACE_BEFORE
1018	space_after = F_SPACE_AFTER	1288	space_after = F_SPACE_AFTER
1019	pretty = F_PRETTY	1289	pretty = F_PRETTY
1020	allow_nonref = F_ALLOW_NONREF	1290	allow_nonref = F_ALLOW_NONREF
1021	shrink = F_SHRINK	1291	shrink = F_SHRINK
		1292	allow_blessed = F_ALLOW_BLESSED
		1293	convert_blessed = F_CONV_BLESSED
1022	CODE:	1294	CODE:
1023	{	1295	{
1024	UV *uv = SvJSON (self);	1296	UV *uv = SvJSON (self);
1025	if (enable)	1297	if (enable)
1026	*uv \|= ix;	1298	*uv \|= ix;
…		…
1030	RETVAL = newSVsv (self);	1302	RETVAL = newSVsv (self);
1031	}	1303	}
1032	OUTPUT:	1304	OUTPUT:
1033	RETVAL	1305	RETVAL
1034		1306
		1307	SV max_depth (SV self, UV max_depth = 0x80000000UL)
		1308	CODE:
		1309	{
		1310	UV *uv = SvJSON (self);
		1311	UV log2 = 0;
		1312
		1313	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
		1314
		1315	while ((1UL << log2) < max_depth)
		1316	++log2;
		1317
		1318	uv = uv & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
		1319
		1320	RETVAL = newSVsv (self);
		1321	}
		1322	OUTPUT:
		1323	RETVAL
		1324
		1325	SV max_size (SV self, UV max_size = 0)
		1326	CODE:
		1327	{
		1328	UV *uv = SvJSON (self);
		1329	UV log2 = 0;
		1330
		1331	if (max_size > 0x80000000UL) max_size = 0x80000000UL;
		1332	if (max_size == 1) max_size = 2;
		1333
		1334	while ((1UL << log2) < max_size)
		1335	++log2;
		1336
		1337	uv = uv & ~F_MAXSIZE \| (log2 << S_MAXSIZE);
		1338
		1339	RETVAL = newSVsv (self);
		1340	}
		1341	OUTPUT:
		1342	RETVAL
		1343
1035	void encode (SV self, SV scalar)	1344	void encode (SV self, SV scalar)
1036	PPCODE:	1345	PPCODE:
1037	XPUSHs (encode_json (scalar, *SvJSON (self)));	1346	XPUSHs (encode_json (scalar, *SvJSON (self)));
1038		1347
1039	void decode (SV self, SV jsonstr)	1348	void decode (SV self, SV jsonstr)
1040	PPCODE:	1349	PPCODE:
1041	XPUSHs (decode_json (jsonstr, *SvJSON (self)));	1350	XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
		1351
		1352	void decode_prefix (SV self, SV jsonstr)
		1353	PPCODE:
		1354	{
		1355	UV offset;
		1356	EXTEND (SP, 2);
		1357	PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
		1358	PUSHs (sv_2mortal (newSVuv (offset)));
		1359	}
1042		1360
1043	PROTOTYPES: ENABLE	1361	PROTOTYPES: ENABLE
1044		1362
1045	void to_json (SV *scalar)	1363	void to_json (SV *scalar)
1046	ALIAS:	1364	ALIAS:
1047	objToJson = 0	1365	objToJson = 0
1048	PPCODE:	1366	PPCODE:
1049	XPUSHs (encode_json (scalar, F_UTF8));	1367	XPUSHs (encode_json (scalar, F_DEFAULT \| F_UTF8));
1050		1368
1051	void from_json (SV *jsonstr)	1369	void from_json (SV *jsonstr)
1052	ALIAS:	1370	ALIAS:
1053	jsonToObj = 0	1371	jsonToObj = 0
1054	PPCODE:	1372	PPCODE:
1055	XPUSHs (decode_json (jsonstr, F_UTF8));	1373	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8, 0));
1056		1374

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.16 by root, Sun Mar 25 02:37:00 2007 UTC vs. Revision 1.46 by root, Mon Jun 25 22:11:39 2007 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.16 by root, Sun Mar 25 02:37:00 2007 UTC vs.
Revision 1.46 by root, Mon Jun 25 22:11:39 2007 UTC