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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.31 by root, Wed May 9 16:33:53 2007 UTC vs.
Revision 1.58 by root, Mon Aug 13 16:06:25 2007 UTC

…		…
1	#include "EXTERN.h"	1	#include "EXTERN.h"
2	#include "perl.h"	2	#include "perl.h"
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	#include <stdio.h>
		9	#include <float.h>
9		10
10	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	11	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
11	# define snprintf _snprintf // C compilers have this in stdio.h	12	# define snprintf _snprintf // C compilers have this in stdio.h
12	#endif	13	#endif
13		14
		15	// some old perls do not have this, try to make it work, no
		16	// guarentees, though. if it breaks, you get to keep the pieces.
		17	#ifndef UTF8_MAXBYTES
		18	# define UTF8_MAXBYTES 13
		19	#endif
		20
14	#define F_ASCII 0x00000001UL	21	#define F_ASCII 0x00000001UL
15	#define F_LATIN1 0x00000002UL	22	#define F_LATIN1 0x00000002UL
16	#define F_UTF8 0x00000004UL	23	#define F_UTF8 0x00000004UL
17	#define F_INDENT 0x00000008UL	24	#define F_INDENT 0x00000008UL
18	#define F_CANONICAL 0x00000010UL	25	#define F_CANONICAL 0x00000010UL
19	#define F_SPACE_BEFORE 0x00000020UL	26	#define F_SPACE_BEFORE 0x00000020UL
20	#define F_SPACE_AFTER 0x00000040UL	27	#define F_SPACE_AFTER 0x00000040UL
21	#define F_ALLOW_NONREF 0x00000100UL	28	#define F_ALLOW_NONREF 0x00000100UL
22	#define F_SHRINK 0x00000200UL	29	#define F_SHRINK 0x00000200UL
		30	#define F_ALLOW_BLESSED 0x00000400UL
		31	#define F_CONV_BLESSED 0x00000800UL
23	#define F_MAXDEPTH 0xf8000000UL	32	#define F_MAXDEPTH 0xf8000000UL
24	#define S_MAXDEPTH 27	33	#define S_MAXDEPTH 27
		34	#define F_MAXSIZE 0x01f00000UL
		35	#define S_MAXSIZE 20
		36	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
25		37
26	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))	38	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27		39	#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
28	// F_SELFCONVERT? <=> to_json/toJson
29	// F_BLESSED? <=> { $__class__$ => }
30		40
31	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	41	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
32	#define F_DEFAULT (9UL << S_MAXDEPTH)	42	#define F_DEFAULT (9UL << S_MAXDEPTH)
33		43
34	#define INIT_SIZE 32 // initial scalar size to be allocated	44	#define INIT_SIZE 32 // initial scalar size to be allocated
35	#define INDENT_STEP 3 // spaces per indentation level	45	#define INDENT_STEP 3 // spaces per indentation level
36		46
37	#define SHORT_STRING_LEN 512 // special-case strings of up to this size	47	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
38		48
39	#define SB do {	49	#define SB do {
40	#define SE } while (0)	50	#define SE } while (0)
41		51
		52	#if __GNUC__ >= 3
		53	# define expect(expr,value) __builtin_expect ((expr),(value))
		54	# define inline inline
		55	#else
		56	# define expect(expr,value) (expr)
		57	# define inline static
		58	#endif
		59
		60	#define expect_false(expr) expect ((expr) != 0, 0)
		61	#define expect_true(expr) expect ((expr) != 0, 1)
		62
		63	#ifdef USE_ITHREADS
		64	# define JSON_SLOW 1
		65	#else
		66	# define JSON_SLOW 0
		67	#endif
		68
42	static HV *json_stash; // JSON::XS::	69	static HV json_stash, json_boolean_stash; // JSON::XS::
		70	static SV json_true, json_false;
		71
		72	typedef struct {
		73	U32 flags;
		74	SV *cb_object;
		75	HV *cb_sk_object;
		76	} JSON;
43		77
44	/////////////////////////////////////////////////////////////////////////////	78	/////////////////////////////////////////////////////////////////////////////
45	// utility functions	79	// utility functions
46		80
47	static UV *	81	inline void
48	SvJSON (SV *sv)
49	{
50	if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
51	croak ("object is not of type JSON::XS");
52
53	return &SvUVX (SvRV (sv));
54	}
55
56	static void
57	shrink (SV *sv)	82	shrink (SV *sv)
58	{	83	{
59	sv_utf8_downgrade (sv, 1);	84	sv_utf8_downgrade (sv, 1);
60	if (SvLEN (sv) > SvCUR (sv) + 1)	85	if (SvLEN (sv) > SvCUR (sv) + 1)
61	{	86	{
70	// decode an utf-8 character and return it, or (UV)-1 in	95	// decode an utf-8 character and return it, or (UV)-1 in
71	// case of an error.	96	// case of an error.
72	// we special-case "safe" characters from U+80 .. U+7FF,	97	// we special-case "safe" characters from U+80 .. U+7FF,
73	// but use the very good perl function to parse anything else.	98	// but use the very good perl function to parse anything else.
74	// note that we never call this function for a ascii codepoints	99	// note that we never call this function for a ascii codepoints
75	static UV	100	inline UV
76	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)	101	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
77	{	102	{
78	if (s[0] > 0xdf \|\| s[0] < 0xc2)	103	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))
79	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);	104	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
80	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)	105	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
81	{	106	{
82	*clen = 2;	107	*clen = 2;
83	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);	108	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
…		…
96	typedef struct	121	typedef struct
97	{	122	{
98	char *cur; // SvPVX (sv) + current output position	123	char *cur; // SvPVX (sv) + current output position
99	char *end; // SvEND (sv)	124	char *end; // SvEND (sv)
100	SV *sv; // result scalar	125	SV *sv; // result scalar
101	U32 flags; // F_*	126	JSON json;
102	U32 indent; // indentation level	127	U32 indent; // indentation level
103	U32 maxdepth; // max. indentation/recursion level	128	U32 maxdepth; // max. indentation/recursion level
104	} enc_t;	129	} enc_t;
105		130
106	static void	131	inline void
107	need (enc_t *enc, STRLEN len)	132	need (enc_t *enc, STRLEN len)
108	{	133	{
109	if (enc->cur + len >= enc->end)	134	if (expect_false (enc->cur + len >= enc->end))
110	{	135	{
111	STRLEN cur = enc->cur - SvPVX (enc->sv);	136	STRLEN cur = enc->cur - SvPVX (enc->sv);
112	SvGROW (enc->sv, cur + len + 1);	137	SvGROW (enc->sv, cur + len + 1);
113	enc->cur = SvPVX (enc->sv) + cur;	138	enc->cur = SvPVX (enc->sv) + cur;
114	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	139	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
115	}	140	}
116	}	141	}
117		142
118	static void	143	inline void
119	encode_ch (enc_t *enc, char ch)	144	encode_ch (enc_t *enc, char ch)
120	{	145	{
121	need (enc, 1);	146	need (enc, 1);
122	*enc->cur++ = ch;	147	*enc->cur++ = ch;
123	}	148	}
…		…
131		156
132	while (str < end)	157	while (str < end)
133	{	158	{
134	unsigned char ch = (unsigned char )str;	159	unsigned char ch = (unsigned char )str;
135		160
136	if (ch >= 0x20 && ch < 0x80) // most common case	161	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
137	{	162	{
138	if (ch == '"') // but with slow exceptions	163	if (expect_false (ch == '"')) // but with slow exceptions
139	{	164	{
140	need (enc, len += 1);	165	need (enc, len += 1);
141	*enc->cur++ = '\\';	166	*enc->cur++ = '\\';
142	*enc->cur++ = '"';	167	*enc->cur++ = '"';
143	}	168	}
144	else if (ch == '\\')	169	else if (expect_false (ch == '\\'))
145	{	170	{
146	need (enc, len += 1);	171	need (enc, len += 1);
147	*enc->cur++ = '\\';	172	*enc->cur++ = '\\';
148	*enc->cur++ = '\\';	173	*enc->cur++ = '\\';
149	}	174	}
…		…
167	STRLEN clen;	192	STRLEN clen;
168	UV uch;	193	UV uch;
169		194
170	if (is_utf8)	195	if (is_utf8)
171	{	196	{
172	//uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
173	uch = decode_utf8 (str, end - str, &clen);	197	uch = decode_utf8 (str, end - str, &clen);
174	if (clen == (STRLEN)-1)	198	if (clen == (STRLEN)-1)
175	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);	199	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
176	}	200	}
177	else	201	else
…		…
181	}	205	}
182		206
183	if (uch > 0x10FFFFUL)	207	if (uch > 0x10FFFFUL)
184	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);	208	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
185		209
186	if (uch < 0x80 \|\| enc->flags & F_ASCII \|\| (enc->flags & F_LATIN1 && uch > 0xFF))	210	if (uch < 0x80 \|\| enc->json.flags & F_ASCII \|\| (enc->json.flags & F_LATIN1 && uch > 0xFF))
187	{	211	{
188	if (uch > 0xFFFFUL)	212	if (uch > 0xFFFFUL)
189	{	213	{
190	need (enc, len += 11);	214	need (enc, len += 11);
191	sprintf (enc->cur, "\\u%04x\\u%04x",	215	sprintf (enc->cur, "\\u%04x\\u%04x",
…		…
205	*enc->cur++ = hexdigit [(uch >> 0) & 15];	229	*enc->cur++ = hexdigit [(uch >> 0) & 15];
206	}	230	}
207		231
208	str += clen;	232	str += clen;
209	}	233	}
210	else if (enc->flags & F_LATIN1)	234	else if (enc->json.flags & F_LATIN1)
211	{	235	{
212	*enc->cur++ = uch;	236	*enc->cur++ = uch;
213	str += clen;	237	str += clen;
214	}	238	}
215	else if (is_utf8)	239	else if (is_utf8)
…		…
233		257
234	--len;	258	--len;
235	}	259	}
236	}	260	}
237		261
238	static void	262	inline void
239	encode_indent (enc_t *enc)	263	encode_indent (enc_t *enc)
240	{	264	{
241	if (enc->flags & F_INDENT)	265	if (enc->json.flags & F_INDENT)
242	{	266	{
243	int spaces = enc->indent * INDENT_STEP;	267	int spaces = enc->indent * INDENT_STEP;
244		268
245	need (enc, spaces);	269	need (enc, spaces);
246	memset (enc->cur, ' ', spaces);	270	memset (enc->cur, ' ', spaces);
247	enc->cur += spaces;	271	enc->cur += spaces;
248	}	272	}
249	}	273	}
250		274
251	static void	275	inline void
252	encode_space (enc_t *enc)	276	encode_space (enc_t *enc)
253	{	277	{
254	need (enc, 1);	278	need (enc, 1);
255	encode_ch (enc, ' ');	279	encode_ch (enc, ' ');
256	}	280	}
257		281
258	static void	282	inline void
259	encode_nl (enc_t *enc)	283	encode_nl (enc_t *enc)
260	{	284	{
261	if (enc->flags & F_INDENT)	285	if (enc->json.flags & F_INDENT)
262	{	286	{
263	need (enc, 1);	287	need (enc, 1);
264	encode_ch (enc, '\n');	288	encode_ch (enc, '\n');
265	}	289	}
266	}	290	}
267		291
268	static void	292	inline void
269	encode_comma (enc_t *enc)	293	encode_comma (enc_t *enc)
270	{	294	{
271	encode_ch (enc, ',');	295	encode_ch (enc, ',');
272		296
273	if (enc->flags & F_INDENT)	297	if (enc->json.flags & F_INDENT)
274	encode_nl (enc);	298	encode_nl (enc);
275	else if (enc->flags & F_SPACE_AFTER)	299	else if (enc->json.flags & F_SPACE_AFTER)
276	encode_space (enc);	300	encode_space (enc);
277	}	301	}
278		302
279	static void encode_sv (enc_t enc, SV sv);	303	static void encode_sv (enc_t enc, SV sv);
280		304
…		…
289	encode_ch (enc, '['); encode_nl (enc);	313	encode_ch (enc, '['); encode_nl (enc);
290	++enc->indent;	314	++enc->indent;
291		315
292	for (i = 0; i <= len; ++i)	316	for (i = 0; i <= len; ++i)
293	{	317	{
		318	SV **svp = av_fetch (av, i, 0);
		319
294	encode_indent (enc);	320	encode_indent (enc);
295	encode_sv (enc, *av_fetch (av, i, 0));	321
		322	if (svp)
		323	encode_sv (enc, *svp);
		324	else
		325	encode_str (enc, "null", 4, 0);
296		326
297	if (i < len)	327	if (i < len)
298	encode_comma (enc);	328	encode_comma (enc);
299	}	329	}
300		330
…		…
323	else	353	else
324	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));	354	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
325		355
326	encode_ch (enc, '"');	356	encode_ch (enc, '"');
327		357
328	if (enc->flags & F_SPACE_BEFORE) encode_space (enc);	358	if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
329	encode_ch (enc, ':');	359	encode_ch (enc, ':');
330	if (enc->flags & F_SPACE_AFTER ) encode_space (enc);	360	if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
331	encode_sv (enc, HeVAL (he));	361	encode_sv (enc, HeVAL (he));
332	}	362	}
333		363
334	// compare hash entries, used when all keys are bytestrings	364	// compare hash entries, used when all keys are bytestrings
335	static int	365	static int
…		…
370	{	400	{
371	// for canonical output we have to sort by keys first	401	// for canonical output we have to sort by keys first
372	// actually, this is mostly due to the stupid so-called	402	// actually, this is mostly due to the stupid so-called
373	// security workaround added somewhere in 5.8.x.	403	// security workaround added somewhere in 5.8.x.
374	// that randomises hash orderings	404	// that randomises hash orderings
375	if (enc->flags & F_CANONICAL)	405	if (enc->json.flags & F_CANONICAL)
376	{	406	{
377	HE he, hes [count]; // if your compiler dies here, you need to enable C99 mode
378	int fast = 1;	407	int fast = 1;
		408	HE *he;
		409	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		410	HE *hes = _alloca (count sizeof (HE));
		411	#else
		412	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
		413	#endif
379		414
380	i = 0;	415	i = 0;
381	while ((he = hv_iternext (hv)))	416	while ((he = hv_iternext (hv)))
382	{	417	{
383	hes [i++] = he;	418	hes [i++] = he;
…		…
447	svtype svt;	482	svtype svt;
448		483
449	SvGETMAGIC (sv);	484	SvGETMAGIC (sv);
450	svt = SvTYPE (sv);	485	svt = SvTYPE (sv);
451		486
		487	if (expect_false (SvOBJECT (sv)))
		488	{
		489	HV *stash = !JSON_SLOW \|\| json_boolean_stash
		490	? json_boolean_stash
		491	: gv_stashpv ("JSON::XS::Boolean", 1);
		492
		493	if (SvSTASH (sv) == stash)
		494	{
		495	if (SvIV (sv))
		496	encode_str (enc, "true", 4, 0);
		497	else
		498	encode_str (enc, "false", 5, 0);
		499	}
		500	else
		501	{
		502	#if 0
		503	if (0 && sv_derived_from (rv, "JSON::Literal"))
		504	{
		505	// not yet
		506	}
		507	#endif
		508	if (enc->json.flags & F_CONV_BLESSED)
		509	{
		510	// we re-bless the reference to get overload and other niceties right
		511	GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
		512
		513	if (to_json)
		514	{
		515	dSP;
		516
		517	ENTER; SAVETMPS; PUSHMARK (SP);
		518	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
		519
		520	// calling with G_SCALAR ensures that we always get a 1 return value
		521	PUTBACK;
		522	call_sv ((SV *)GvCV (to_json), G_SCALAR);
		523	SPAGAIN;
		524
		525	// catch this surprisingly common error
		526	if (SvROK (TOPs) && SvRV (TOPs) == sv)
		527	croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
		528
		529	sv = POPs;
		530	PUTBACK;
		531
		532	encode_sv (enc, sv);
		533
		534	FREETMPS; LEAVE;
		535	}
		536	else if (enc->json.flags & F_ALLOW_BLESSED)
		537	encode_str (enc, "null", 4, 0);
		538	else
		539	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
		540	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		541	}
		542	else if (enc->json.flags & F_ALLOW_BLESSED)
		543	encode_str (enc, "null", 4, 0);
		544	else
		545	croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
		546	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		547	}
		548	}
452	if (svt == SVt_PVHV)	549	else if (svt == SVt_PVHV)
453	encode_hv (enc, (HV *)sv);	550	encode_hv (enc, (HV *)sv);
454	else if (svt == SVt_PVAV)	551	else if (svt == SVt_PVAV)
455	encode_av (enc, (AV *)sv);	552	encode_av (enc, (AV *)sv);
456	else if (svt < SVt_PVAV)	553	else if (svt < SVt_PVAV)
457	{	554	{
458	if (SvNIOK (sv) && SvIV (sv) == 0)	555	STRLEN len = 0;
		556	char *pv = svt ? SvPV (sv, len) : 0;
		557
		558	if (len == 1 && *pv == '1')
		559	encode_str (enc, "true", 4, 0);
		560	else if (len == 1 && *pv == '0')
459	encode_str (enc, "false", 5, 0);	561	encode_str (enc, "false", 5, 0);
460	else if (SvNIOK (sv) && SvIV (sv) == 1)
461	encode_str (enc, "true", 4, 0);
462	else	562	else
463	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",	563	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
464	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	564	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
465	}	565	}
466	else	566	else
…		…
481	encode_str (enc, str, len, SvUTF8 (sv));	581	encode_str (enc, str, len, SvUTF8 (sv));
482	encode_ch (enc, '"');	582	encode_ch (enc, '"');
483	}	583	}
484	else if (SvNOKp (sv))	584	else if (SvNOKp (sv))
485	{	585	{
		586	// trust that perl will do the right thing w.r.t. JSON syntax.
486	need (enc, NV_DIG + 32);	587	need (enc, NV_DIG + 32);
487	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	588	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
488	enc->cur += strlen (enc->cur);	589	enc->cur += strlen (enc->cur);
489	}	590	}
490	else if (SvIOKp (sv))	591	else if (SvIOKp (sv))
491	{	592	{
492	need (enc, 64);	593	// we assume we can always read an IV as a UV
		594	if (SvUV (sv) & ~(UV)0x7fff)
		595	{
		596	// large integer, use the (rather slow) snprintf way.
		597	need (enc, sizeof (UV) * 3);
493	enc->cur +=	598	enc->cur +=
494	SvIsUV(sv)	599	SvIsUV(sv)
495	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	600	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
496	: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));	601	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
		602	}
		603	else
		604	{
		605	// optimise the "small number case"
		606	// code will likely be branchless and use only a single multiplication
		607	I32 i = SvIV (sv);
		608	U32 u;
		609	char digit, nz = 0;
		610
		611	need (enc, 6);
		612
		613	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
		614	u = i < 0 ? -i : i;
		615
		616	// convert to 4.28 fixed-point representation
		617	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
		618
		619	// now output digit by digit, each time masking out the integer part
		620	// and multiplying by 5 while moving the decimal point one to the right,
		621	// resulting in a net multiplication by 10.
		622	// we always write the digit to memory but conditionally increment
		623	// the pointer, to ease the usage of conditional move instructions.
		624	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffff) 5;
		625	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffff) 5;
		626	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffff) 5;
		627	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffff) 5;
		628	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
		629	}
497	}	630	}
498	else if (SvROK (sv))	631	else if (SvROK (sv))
499	encode_rv (enc, SvRV (sv));	632	encode_rv (enc, SvRV (sv));
500	else if (!SvOK (sv))	633	else if (!SvOK (sv))
501	encode_str (enc, "null", 4, 0);	634	encode_str (enc, "null", 4, 0);
…		…
503	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	636	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
504	SvPV_nolen (sv), SvFLAGS (sv));	637	SvPV_nolen (sv), SvFLAGS (sv));
505	}	638	}
506		639
507	static SV *	640	static SV *
508	encode_json (SV *scalar, U32 flags)	641	encode_json (SV scalar, JSON json)
509	{	642	{
510	enc_t enc;	643	enc_t enc;
511		644
512	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	645	if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
513	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");	646	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
514		647
515	enc.flags = flags;	648	enc.json = *json;
516	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	649	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
517	enc.cur = SvPVX (enc.sv);	650	enc.cur = SvPVX (enc.sv);
518	enc.end = SvEND (enc.sv);	651	enc.end = SvEND (enc.sv);
519	enc.indent = 0;	652	enc.indent = 0;
520	enc.maxdepth = DEC_DEPTH (flags);	653	enc.maxdepth = DEC_DEPTH (enc.json.flags);
521		654
522	SvPOK_only (enc.sv);	655	SvPOK_only (enc.sv);
523	encode_sv (&enc, scalar);	656	encode_sv (&enc, scalar);
524		657
525	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	658	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
526	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	659	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
527		660
528	if (!(flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))	661	if (!(enc.json.flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
529	SvUTF8_on (enc.sv);	662	SvUTF8_on (enc.sv);
530		663
531	if (enc.flags & F_SHRINK)	664	if (enc.json.flags & F_SHRINK)
532	shrink (enc.sv);	665	shrink (enc.sv);
533		666
534	return enc.sv;	667	return enc.sv;
535	}	668	}
536		669
…		…
541	typedef struct	674	typedef struct
542	{	675	{
543	char *cur; // current parser pointer	676	char *cur; // current parser pointer
544	char *end; // end of input string	677	char *end; // end of input string
545	const char *err; // parse error, if != 0	678	const char *err; // parse error, if != 0
546	U32 flags; // F_*	679	JSON json;
547	U32 depth; // recursion depth	680	U32 depth; // recursion depth
548	U32 maxdepth; // recursion depth limit	681	U32 maxdepth; // recursion depth limit
549	} dec_t;	682	} dec_t;
550		683
551	static void	684	inline void
552	decode_ws (dec_t *dec)	685	decode_ws (dec_t *dec)
553	{	686	{
554	for (;;)	687	for (;;)
555	{	688	{
556	char ch = *dec->cur;	689	char ch = *dec->cur;
…		…
582	decode_4hex (dec_t *dec)	715	decode_4hex (dec_t *dec)
583	{	716	{
584	signed char d1, d2, d3, d4;	717	signed char d1, d2, d3, d4;
585	unsigned char cur = (unsigned char )dec->cur;	718	unsigned char cur = (unsigned char )dec->cur;
586		719
587	d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");	720	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
588	d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");	721	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
589	d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");	722	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
590	d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");	723	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
591		724
592	dec->cur += 4;	725	dec->cur += 4;
593		726
594	return ((UV)d1) << 12	727	return ((UV)d1) << 12
595	\| ((UV)d2) << 8	728	\| ((UV)d2) << 8
…		…
603	static SV *	736	static SV *
604	decode_str (dec_t *dec)	737	decode_str (dec_t *dec)
605	{	738	{
606	SV *sv = 0;	739	SV *sv = 0;
607	int utf8 = 0;	740	int utf8 = 0;
		741	char *dec_cur = dec->cur;
608		742
609	do	743	do
610	{	744	{
611	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];	745	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
612	char *cur = buf;	746	char *cur = buf;
613		747
614	do	748	do
615	{	749	{
616	unsigned char ch = (unsigned char )dec->cur++;	750	unsigned char ch = (unsigned char )dec_cur++;
617		751
618	if (ch == '"')	752	if (expect_false (ch == '"'))
619	{	753	{
620	--dec->cur;	754	--dec_cur;
621	break;	755	break;
622	}	756	}
623	else if (ch == '\\')	757	else if (expect_false (ch == '\\'))
624	{	758	{
625	switch (*dec->cur)	759	switch (*dec_cur)
626	{	760	{
627	case '\\':	761	case '\\':
628	case '/':	762	case '/':
629	case '"': cur++ = dec->cur++; break;	763	case '"': cur++ = dec_cur++; break;
630		764
631	case 'b': ++dec->cur; *cur++ = '\010'; break;	765	case 'b': ++dec_cur; *cur++ = '\010'; break;
632	case 't': ++dec->cur; *cur++ = '\011'; break;	766	case 't': ++dec_cur; *cur++ = '\011'; break;
633	case 'n': ++dec->cur; *cur++ = '\012'; break;	767	case 'n': ++dec_cur; *cur++ = '\012'; break;
634	case 'f': ++dec->cur; *cur++ = '\014'; break;	768	case 'f': ++dec_cur; *cur++ = '\014'; break;
635	case 'r': ++dec->cur; *cur++ = '\015'; break;	769	case 'r': ++dec_cur; *cur++ = '\015'; break;
636		770
637	case 'u':	771	case 'u':
638	{	772	{
639	UV lo, hi;	773	UV lo, hi;
640	++dec->cur;	774	++dec_cur;
641		775
		776	dec->cur = dec_cur;
642	hi = decode_4hex (dec);	777	hi = decode_4hex (dec);
		778	dec_cur = dec->cur;
643	if (hi == (UV)-1)	779	if (hi == (UV)-1)
644	goto fail;	780	goto fail;
645		781
646	// possibly a surrogate pair	782	// possibly a surrogate pair
647	if (hi >= 0xd800)	783	if (hi >= 0xd800)
648	if (hi < 0xdc00)	784	if (hi < 0xdc00)
649	{	785	{
650	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')	786	if (dec_cur [0] != '\\' \|\| dec_cur [1] != 'u')
651	ERR ("missing low surrogate character in surrogate pair");	787	ERR ("missing low surrogate character in surrogate pair");
652		788
653	dec->cur += 2;	789	dec_cur += 2;
654		790
		791	dec->cur = dec_cur;
655	lo = decode_4hex (dec);	792	lo = decode_4hex (dec);
		793	dec_cur = dec->cur;
656	if (lo == (UV)-1)	794	if (lo == (UV)-1)
657	goto fail;	795	goto fail;
658		796
659	if (lo < 0xdc00 \|\| lo >= 0xe000)	797	if (lo < 0xdc00 \|\| lo >= 0xe000)
660	ERR ("surrogate pair expected");	798	ERR ("surrogate pair expected");
…		…
674	*cur++ = hi;	812	*cur++ = hi;
675	}	813	}
676	break;	814	break;
677		815
678	default:	816	default:
679	--dec->cur;	817	--dec_cur;
680	ERR ("illegal backslash escape sequence in string");	818	ERR ("illegal backslash escape sequence in string");
681	}	819	}
682	}	820	}
683	else if (ch >= 0x20 && ch <= 0x7f)	821	else if (expect_true (ch >= 0x20 && ch <= 0x7f))
684	*cur++ = ch;	822	*cur++ = ch;
685	else if (ch >= 0x80)	823	else if (ch >= 0x80)
686	{	824	{
687	STRLEN clen;	825	STRLEN clen;
688	UV uch;	826	UV uch;
689		827
690	--dec->cur;	828	--dec_cur;
691		829
692	uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);	830	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
693	if (clen == (STRLEN)-1)	831	if (clen == (STRLEN)-1)
694	ERR ("malformed UTF-8 character in JSON string");	832	ERR ("malformed UTF-8 character in JSON string");
695		833
696	do	834	do
697	cur++ = dec->cur++;	835	cur++ = dec_cur++;
698	while (--clen);	836	while (--clen);
699		837
700	utf8 = 1;	838	utf8 = 1;
701	}	839	}
702	else	840	else
703	{	841	{
704	--dec->cur;	842	--dec_cur;
705		843
706	if (!ch)	844	if (!ch)
707	ERR ("unexpected end of string while parsing JSON string");	845	ERR ("unexpected end of string while parsing JSON string");
708	else	846	else
709	ERR ("invalid character encountered while parsing JSON string");	847	ERR ("invalid character encountered while parsing JSON string");
…		…
722	}	860	}
723	else	861	else
724	sv = newSVpvn (buf, len);	862	sv = newSVpvn (buf, len);
725	}	863	}
726	}	864	}
727	while (*dec->cur != '"');	865	while (*dec_cur != '"');
728		866
729	++dec->cur;	867	++dec_cur;
730		868
731	if (sv)	869	if (sv)
732	{	870	{
733	SvPOK_only (sv);	871	SvPOK_only (sv);
734	*SvEND (sv) = 0;	872	*SvEND (sv) = 0;
…		…
737	SvUTF8_on (sv);	875	SvUTF8_on (sv);
738	}	876	}
739	else	877	else
740	sv = newSVpvn ("", 0);	878	sv = newSVpvn ("", 0);
741		879
		880	dec->cur = dec_cur;
742	return sv;	881	return sv;
743		882
744	fail:	883	fail:
		884	dec->cur = dec_cur;
745	return 0;	885	return 0;
746	}	886	}
747		887
748	static SV *	888	static SV *
749	decode_num (dec_t *dec)	889	decode_num (dec_t *dec)
…		…
807	is_nv = 1;	947	is_nv = 1;
808	}	948	}
809		949
810	if (!is_nv)	950	if (!is_nv)
811	{	951	{
812	UV uv;	952	int len = dec->cur - start;
813	int numtype = grok_number (start, dec->cur - start, &uv);	953
814	if (numtype & IS_NUMBER_IN_UV)	954	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
815	if (numtype & IS_NUMBER_NEG)	955	if (*start == '-')
		956	switch (len)
816	{	957	{
817	if (uv < (UV)IV_MIN)	958	case 2: return newSViv (-( start [1] - '0' * 1));
818	return newSViv (-(IV)uv);	959	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
		960	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
		961	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
819	}	962	}
		963	else
		964	switch (len)
		965	{
		966	case 1: return newSViv ( start [0] - '0' * 1);
		967	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
		968	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
		969	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		970	}
		971
		972	{
		973	UV uv;
		974	int numtype = grok_number (start, len, &uv);
		975	if (numtype & IS_NUMBER_IN_UV)
		976	if (numtype & IS_NUMBER_NEG)
		977	{
		978	if (uv < (UV)IV_MIN)
		979	return newSViv (-(IV)uv);
		980	}
820	else	981	else
821	return newSVuv (uv);	982	return newSVuv (uv);
822	}	983	}
823		984
		985	len -= *start == '-' ? 1 : 0;
		986
		987	// does not fit into IV or UV, try NV
		988	if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
		989	#if defined (LDBL_DIG)
		990	\|\| (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
		991	#endif
		992	)
		993	// fits into NV without loss of precision
		994	return newSVnv (Atof (start));
		995
		996	// everything else fails, convert it to a string
		997	return newSVpvn (start, dec->cur - start);
		998	}
		999
		1000	// loss of precision here
824	return newSVnv (Atof (start));	1001	return newSVnv (Atof (start));
825		1002
826	fail:	1003	fail:
827	return 0;	1004	return 0;
828	}	1005	}
…		…
872	}	1049	}
873		1050
874	static SV *	1051	static SV *
875	decode_hv (dec_t *dec)	1052	decode_hv (dec_t *dec)
876	{	1053	{
		1054	SV *sv;
877	HV *hv = newHV ();	1055	HV *hv = newHV ();
878		1056
879	DEC_INC_DEPTH;	1057	DEC_INC_DEPTH;
880	decode_ws (dec);	1058	decode_ws (dec);
881		1059
882	if (*dec->cur == '}')	1060	if (*dec->cur == '}')
883	++dec->cur;	1061	++dec->cur;
884	else	1062	else
885	for (;;)	1063	for (;;)
886	{	1064	{
887	SV key, value;
888
889	decode_ws (dec); EXPECT_CH ('"');	1065	decode_ws (dec); EXPECT_CH ('"');
890		1066
891	key = decode_str (dec);	1067	// heuristic: assume that
892	if (!key)	1068	// a) decode_str + hv_store_ent are abysmally slow.
893	goto fail;	1069	// b) most hash keys are short, simple ascii text.
		1070	// => try to "fast-match" such strings to avoid
		1071	// the overhead of decode_str + hv_store_ent.
		1072	{
		1073	SV *value;
		1074	char *p = dec->cur;
		1075	char *e = p + 24; // only try up to 24 bytes
894		1076
895	decode_ws (dec); EXPECT_CH (':');	1077	for (;;)
896
897	value = decode_sv (dec);
898	if (!value)
899	{	1078	{
		1079	// the >= 0x80 is true on most architectures
		1080	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
		1081	{
		1082	// slow path, back up and use decode_str
		1083	SV *key = decode_str (dec);
		1084	if (!key)
		1085	goto fail;
		1086
		1087	decode_ws (dec); EXPECT_CH (':');
		1088
		1089	value = decode_sv (dec);
		1090	if (!value)
		1091	{
		1092	SvREFCNT_dec (key);
		1093	goto fail;
		1094	}
		1095
		1096	hv_store_ent (hv, key, value, 0);
900	SvREFCNT_dec (key);	1097	SvREFCNT_dec (key);
		1098
		1099	break;
		1100	}
		1101	else if (*p == '"')
		1102	{
		1103	// fast path, got a simple key
		1104	char *key = dec->cur;
		1105	int len = p - key;
		1106	dec->cur = p + 1;
		1107
		1108	decode_ws (dec); EXPECT_CH (':');
		1109
		1110	value = decode_sv (dec);
		1111	if (!value)
901	goto fail;	1112	goto fail;
		1113
		1114	hv_store (hv, key, len, value, 0);
		1115
		1116	break;
		1117	}
		1118
		1119	++p;
902	}	1120	}
903		1121	}
904	hv_store_ent (hv, key, value, 0);
905	SvREFCNT_dec (key);
906		1122
907	decode_ws (dec);	1123	decode_ws (dec);
908		1124
909	if (*dec->cur == '}')	1125	if (*dec->cur == '}')
910	{	1126	{
…		…
917		1133
918	++dec->cur;	1134	++dec->cur;
919	}	1135	}
920		1136
921	DEC_DEC_DEPTH;	1137	DEC_DEC_DEPTH;
922	return newRV_noinc ((SV *)hv);	1138	sv = newRV_noinc ((SV *)hv);
		1139
		1140	// check filter callbacks
		1141	if (dec->json.flags & F_HOOK)
		1142	{
		1143	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
		1144	{
		1145	HE cb, he;
		1146
		1147	hv_iterinit (hv);
		1148	he = hv_iternext (hv);
		1149	hv_iterinit (hv);
		1150
		1151	// the next line creates a mortal sv each time its called.
		1152	// might want to optimise this for common cases.
		1153	cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
		1154
		1155	if (cb)
		1156	{
		1157	dSP;
		1158	int count;
		1159
		1160	ENTER; SAVETMPS; PUSHMARK (SP);
		1161	XPUSHs (HeVAL (he));
		1162
		1163	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
		1164
		1165	if (count == 1)
		1166	{
		1167	sv = newSVsv (POPs);
		1168	FREETMPS; LEAVE;
		1169	return sv;
		1170	}
		1171
		1172	FREETMPS; LEAVE;
		1173	}
		1174	}
		1175
		1176	if (dec->json.cb_object)
		1177	{
		1178	dSP;
		1179	int count;
		1180
		1181	ENTER; SAVETMPS; PUSHMARK (SP);
		1182	XPUSHs (sv_2mortal (sv));
		1183
		1184	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
		1185
		1186	if (count == 1)
		1187	{
		1188	sv = newSVsv (POPs);
		1189	FREETMPS; LEAVE;
		1190	return sv;
		1191	}
		1192
		1193	SvREFCNT_inc (sv);
		1194	FREETMPS; LEAVE;
		1195	}
		1196	}
		1197
		1198	return sv;
923		1199
924	fail:	1200	fail:
925	SvREFCNT_dec (hv);	1201	SvREFCNT_dec (hv);
926	DEC_DEC_DEPTH;	1202	DEC_DEC_DEPTH;
927	return 0;	1203	return 0;
…		…
929		1205
930	static SV *	1206	static SV *
931	decode_sv (dec_t *dec)	1207	decode_sv (dec_t *dec)
932	{	1208	{
933	decode_ws (dec);	1209	decode_ws (dec);
		1210
		1211	// the beauty of JSON: you need exactly one character lookahead
		1212	// to parse anything.
934	switch (*dec->cur)	1213	switch (*dec->cur)
935	{	1214	{
936	case '"': ++dec->cur; return decode_str (dec);	1215	case '"': ++dec->cur; return decode_str (dec);
937	case '[': ++dec->cur; return decode_av (dec);	1216	case '[': ++dec->cur; return decode_av (dec);
938	case '{': ++dec->cur; return decode_hv (dec);	1217	case '{': ++dec->cur; return decode_hv (dec);
…		…
944		1223
945	case 't':	1224	case 't':
946	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1225	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
947	{	1226	{
948	dec->cur += 4;	1227	dec->cur += 4;
949	return newSViv (1);	1228	return SvREFCNT_inc (json_true);
950	}	1229	}
951	else	1230	else
952	ERR ("'true' expected");	1231	ERR ("'true' expected");
953		1232
954	break;	1233	break;
955		1234
956	case 'f':	1235	case 'f':
957	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1236	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
958	{	1237	{
959	dec->cur += 5;	1238	dec->cur += 5;
960	return newSViv (0);	1239	return SvREFCNT_inc (json_false);
961	}	1240	}
962	else	1241	else
963	ERR ("'false' expected");	1242	ERR ("'false' expected");
964		1243
965	break;	1244	break;
…		…
983	fail:	1262	fail:
984	return 0;	1263	return 0;
985	}	1264	}
986		1265
987	static SV *	1266	static SV *
988	decode_json (SV string, U32 flags, UV offset_return)	1267	decode_json (SV string, JSON json, UV *offset_return)
989	{	1268	{
990	dec_t dec;	1269	dec_t dec;
991	UV offset;	1270	UV offset;
992	SV *sv;	1271	SV *sv;
993		1272
994	SvGETMAGIC (string);	1273	SvGETMAGIC (string);
995	SvUPGRADE (string, SVt_PV);	1274	SvUPGRADE (string, SVt_PV);
996		1275
		1276	if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
		1277	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1278	(unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
		1279
997	if (flags & F_UTF8)	1280	if (json->flags & F_UTF8)
998	sv_utf8_downgrade (string, 0);	1281	sv_utf8_downgrade (string, 0);
999	else	1282	else
1000	sv_utf8_upgrade (string);	1283	sv_utf8_upgrade (string);
1001		1284
1002	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1285	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1003		1286
1004	dec.flags = flags;	1287	dec.json = *json;
1005	dec.cur = SvPVX (string);	1288	dec.cur = SvPVX (string);
1006	dec.end = SvEND (string);	1289	dec.end = SvEND (string);
1007	dec.err = 0;	1290	dec.err = 0;
1008	dec.depth = 0;	1291	dec.depth = 0;
1009	dec.maxdepth = DEC_DEPTH (dec.flags);	1292	dec.maxdepth = DEC_DEPTH (dec.json.flags);
		1293
		1294	if (dec.json.cb_object \|\| dec.json.cb_sk_object)
		1295	dec.json.flags \|= F_HOOK;
1010		1296
1011	*dec.end = 0; // this should basically be a nop, too, but make sure it's there	1297	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
1012	sv = decode_sv (&dec);	1298	sv = decode_sv (&dec);
1013		1299
1014	if (offset_return \|\| !sv)	1300	if (!(offset_return \|\| !sv))
1015	{
1016	offset = dec.flags & F_UTF8
1017	? dec.cur - SvPVX (string)
1018	: utf8_distance (dec.cur, SvPVX (string));
1019
1020	if (offset_return)
1021	*offset_return = offset;
1022	}
1023	else
1024	{	1301	{
1025	// check for trailing garbage	1302	// check for trailing garbage
1026	decode_ws (&dec);	1303	decode_ws (&dec);
1027		1304
1028	if (*dec.cur)	1305	if (*dec.cur)
1029	{	1306	{
1030	dec.err = "garbage after JSON object";	1307	dec.err = "garbage after JSON object";
1031	SvREFCNT_dec (sv);	1308	SvREFCNT_dec (sv);
1032	sv = 0;	1309	sv = 0;
1033	}	1310	}
		1311	}
		1312
		1313	if (offset_return \|\| !sv)
		1314	{
		1315	offset = dec.json.flags & F_UTF8
		1316	? dec.cur - SvPVX (string)
		1317	: utf8_distance (dec.cur, SvPVX (string));
		1318
		1319	if (offset_return)
		1320	*offset_return = offset;
1034	}	1321	}
1035		1322
1036	if (!sv)	1323	if (!sv)
1037	{	1324	{
1038	SV *uni = sv_newmortal ();	1325	SV *uni = sv_newmortal ();
…		…
1052	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1339	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1053	}	1340	}
1054		1341
1055	sv = sv_2mortal (sv);	1342	sv = sv_2mortal (sv);
1056		1343
1057	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))	1344	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1058	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");	1345	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1059		1346
1060	return sv;	1347	return sv;
1061	}	1348	}
1062		1349
…		…
1066	MODULE = JSON::XS PACKAGE = JSON::XS	1353	MODULE = JSON::XS PACKAGE = JSON::XS
1067		1354
1068	BOOT:	1355	BOOT:
1069	{	1356	{
1070	int i;	1357	int i;
1071
1072	memset (decode_hexdigit, 0xff, 256);
1073		1358
1074	for (i = 0; i < 256; ++i)	1359	for (i = 0; i < 256; ++i)
1075	decode_hexdigit [i] =	1360	decode_hexdigit [i] =
1076	i >= '0' && i <= '9' ? i - '0'	1361	i >= '0' && i <= '9' ? i - '0'
1077	: i >= 'a' && i <= 'f' ? i - 'a' + 10	1362	: i >= 'a' && i <= 'f' ? i - 'a' + 10
1078	: i >= 'A' && i <= 'F' ? i - 'A' + 10	1363	: i >= 'A' && i <= 'F' ? i - 'A' + 10
1079	: -1;	1364	: -1;
1080		1365
1081	json_stash = gv_stashpv ("JSON::XS", 1);	1366	json_stash = gv_stashpv ("JSON::XS" , 1);
		1367	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
		1368
		1369	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
		1370	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1082	}	1371	}
1083		1372
1084	PROTOTYPES: DISABLE	1373	PROTOTYPES: DISABLE
1085		1374
1086	SV new (char dummy)	1375	void CLONE (...)
1087	CODE:	1376	CODE:
1088	RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);	1377	json_stash = 0;
1089	OUTPUT:	1378	json_boolean_stash = 0;
1090	RETVAL
1091		1379
		1380	void new (char *klass)
		1381	PPCODE:
		1382	{
		1383	HV *stash = !JSON_SLOW \|\| json_stash
		1384	? json_stash
		1385	: gv_stashpv ("JSON::XS", 1);
		1386	SV *pv = NEWSV (0, sizeof (JSON));
		1387	SvPOK_only (pv);
		1388	Zero (SvPVX (pv), 1, JSON);
		1389	((JSON *)SvPVX (pv))->flags = F_DEFAULT;
		1390	XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), stash)));
		1391	}
		1392
1092	SV ascii (SV self, int enable = 1)	1393	void ascii (JSON *self, int enable = 1)
1093	ALIAS:	1394	ALIAS:
1094	ascii = F_ASCII	1395	ascii = F_ASCII
1095	latin1 = F_LATIN1	1396	latin1 = F_LATIN1
1096	utf8 = F_UTF8	1397	utf8 = F_UTF8
1097	indent = F_INDENT	1398	indent = F_INDENT
1098	canonical = F_CANONICAL	1399	canonical = F_CANONICAL
1099	space_before = F_SPACE_BEFORE	1400	space_before = F_SPACE_BEFORE
1100	space_after = F_SPACE_AFTER	1401	space_after = F_SPACE_AFTER
1101	pretty = F_PRETTY	1402	pretty = F_PRETTY
1102	allow_nonref = F_ALLOW_NONREF	1403	allow_nonref = F_ALLOW_NONREF
1103	shrink = F_SHRINK	1404	shrink = F_SHRINK
		1405	allow_blessed = F_ALLOW_BLESSED
		1406	convert_blessed = F_CONV_BLESSED
1104	CODE:	1407	PPCODE:
1105	{	1408	{
1106	UV *uv = SvJSON (self);
1107	if (enable)	1409	if (enable)
1108	*uv \|= ix;	1410	self->flags \|= ix;
1109	else	1411	else
1110	*uv &= ~ix;	1412	self->flags &= ~ix;
1111		1413
1112	RETVAL = newSVsv (self);	1414	XPUSHs (ST (0));
1113	}	1415	}
1114	OUTPUT:
1115	RETVAL
1116		1416
1117	SV max_depth (SV self, UV max_depth = 0x80000000UL)	1417	void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1118	CODE:	1418	PPCODE:
1119	{	1419	{
1120	UV *uv = SvJSON (self);
1121	UV log2 = 0;	1420	UV log2 = 0;
1122		1421
1123	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;	1422	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1124		1423
1125	while ((1UL << log2) < max_depth)	1424	while ((1UL << log2) < max_depth)
1126	++log2;	1425	++log2;
1127		1426
1128	uv = uv & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);	1427	self->flags = self->flags & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
1129		1428
1130	RETVAL = newSVsv (self);	1429	XPUSHs (ST (0));
1131	}	1430	}
1132	OUTPUT:
1133	RETVAL
1134		1431
1135	void encode (SV self, SV scalar)	1432	void max_size (JSON *self, UV max_size = 0)
1136	PPCODE:	1433	PPCODE:
1137	XPUSHs (encode_json (scalar, *SvJSON (self)));	1434	{
		1435	UV log2 = 0;
1138		1436
1139	void decode (SV self, SV jsonstr)	1437	if (max_size > 0x80000000UL) max_size = 0x80000000UL;
		1438	if (max_size == 1) max_size = 2;
		1439
		1440	while ((1UL << log2) < max_size)
		1441	++log2;
		1442
		1443	self->flags = self->flags & ~F_MAXSIZE \| (log2 << S_MAXSIZE);
		1444
		1445	XPUSHs (ST (0));
		1446	}
		1447
		1448	void filter_json_object (JSON self, SV cb = &PL_sv_undef)
1140	PPCODE:	1449	PPCODE:
		1450	{
		1451	SvREFCNT_dec (self->cb_object);
		1452	self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
		1453
		1454	XPUSHs (ST (0));
		1455	}
		1456
		1457	void filter_json_single_key_object (JSON self, SV key, SV *cb = &PL_sv_undef)
		1458	PPCODE:
		1459	{
		1460	if (!self->cb_sk_object)
		1461	self->cb_sk_object = newHV ();
		1462
		1463	if (SvOK (cb))
		1464	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
		1465	else
		1466	{
		1467	hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
		1468
		1469	if (!HvKEYS (self->cb_sk_object))
		1470	{
		1471	SvREFCNT_dec (self->cb_sk_object);
		1472	self->cb_sk_object = 0;
		1473	}
		1474	}
		1475
		1476	XPUSHs (ST (0));
		1477	}
		1478
		1479	void encode (JSON self, SV scalar)
		1480	PPCODE:
		1481	XPUSHs (encode_json (scalar, self));
		1482
		1483	void decode (JSON self, SV jsonstr)
		1484	PPCODE:
1141	XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));	1485	XPUSHs (decode_json (jsonstr, self, 0));
1142		1486
1143	void decode_prefix (SV self, SV jsonstr)	1487	void decode_prefix (JSON self, SV jsonstr)
1144	PPCODE:	1488	PPCODE:
1145	{	1489	{
1146	UV offset;	1490	UV offset;
1147	EXTEND (SP, 2);	1491	EXTEND (SP, 2);
1148	PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));	1492	PUSHs (decode_json (jsonstr, self, &offset));
1149	PUSHs (sv_2mortal (newSVuv (offset)));	1493	PUSHs (sv_2mortal (newSVuv (offset)));
1150	}	1494	}
1151		1495
		1496	void DESTROY (JSON *self)
		1497	CODE:
		1498	SvREFCNT_dec (self->cb_sk_object);
		1499	SvREFCNT_dec (self->cb_object);
		1500
1152	PROTOTYPES: ENABLE	1501	PROTOTYPES: ENABLE
1153		1502
1154	void to_json (SV *scalar)	1503	void to_json (SV *scalar)
1155	ALIAS:
1156	objToJson = 0
1157	PPCODE:	1504	PPCODE:
		1505	{
		1506	JSON json = { F_DEFAULT \| F_UTF8 };
1158	XPUSHs (encode_json (scalar, F_DEFAULT \| F_UTF8));	1507	XPUSHs (encode_json (scalar, &json));
		1508	}
1159		1509
1160	void from_json (SV *jsonstr)	1510	void from_json (SV *jsonstr)
1161	ALIAS:
1162	jsonToObj = 0
1163	PPCODE:	1511	PPCODE:
		1512	{
		1513	JSON json = { F_DEFAULT \| F_UTF8 };
1164	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8, 0));	1514	XPUSHs (decode_json (jsonstr, &json, 0));
		1515	}
1165		1516

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.31 by root, Wed May 9 16:33:53 2007 UTC vs. Revision 1.58 by root, Mon Aug 13 16:06:25 2007 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.31 by root, Wed May 9 16:33:53 2007 UTC vs.
Revision 1.58 by root, Mon Aug 13 16:06:25 2007 UTC