ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/JSON-XS/XS.xs

Comparing JSON-XS/XS.xs (file contents):
Revision 1.28 by root, Wed Apr 11 12:23:02 2007 UTC vs.
Revision 1.92 by root, Mon Sep 22 07:29:29 2008 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 <limits.h>
		10	#include <float.h>
9		11
10	#if defined(__BORLANDC__) || defined(_MSC_VER)	12	#if defined(__BORLANDC__) || defined(_MSC_VER)
11	# define snprintf _snprintf // C compilers have this in stdio.h	13	# define snprintf _snprintf // C compilers have this in stdio.h
12	#endif	14	#endif
13		15
		16	// some old perls do not have this, try to make it work, no
		17	// guarentees, though. if it breaks, you get to keep the pieces.
		18	#ifndef UTF8_MAXBYTES
		19	# define UTF8_MAXBYTES 13
		20	#endif
		21
		22	#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
		23
14	#define F_ASCII 0x00000001UL	24	#define F_ASCII 0x00000001UL
		25	#define F_LATIN1 0x00000002UL
15	#define F_UTF8 0x00000002UL	26	#define F_UTF8 0x00000004UL
16	#define F_INDENT 0x00000004UL	27	#define F_INDENT 0x00000008UL
17	#define F_CANONICAL 0x00000008UL	28	#define F_CANONICAL 0x00000010UL
18	#define F_SPACE_BEFORE 0x00000010UL	29	#define F_SPACE_BEFORE 0x00000020UL
19	#define F_SPACE_AFTER 0x00000020UL	30	#define F_SPACE_AFTER 0x00000040UL
20	#define F_ALLOW_NONREF 0x00000080UL	31	#define F_ALLOW_NONREF 0x00000100UL
21	#define F_SHRINK 0x00000100UL	32	#define F_SHRINK 0x00000200UL
		33	#define F_ALLOW_BLESSED 0x00000400UL
		34	#define F_CONV_BLESSED 0x00000800UL
22	#define F_MAXDEPTH 0xf8000000UL	35	#define F_RELAXED 0x00001000UL
23	#define S_MAXDEPTH 27	36	#define F_ALLOW_UNKNOWN 0x00002000UL
24		37	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
25	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
26
27	// F_SELFCONVERT? <=> to_json/toJson
28	// F_BLESSED? <=> { $__class__$ => }
29		38
30	#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER	39	#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
31	#define F_DEFAULT (9UL << S_MAXDEPTH)
32		40
33	#define INIT_SIZE 32 // initial scalar size to be allocated	41	#define INIT_SIZE 32 // initial scalar size to be allocated
34	#define INDENT_STEP 3 // spaces per indentation level	42	#define INDENT_STEP 3 // spaces per indentation level
35		43
36	#define SHORT_STRING_LEN 512 // special-case strings of up to this size	44	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
37		45
38	#define SB do {	46	#define SB do {
39	#define SE } while (0)	47	#define SE } while (0)
40		48
		49	#if __GNUC__ >= 3
		50	# define expect(expr,value) __builtin_expect ((expr), (value))
		51	# define INLINE static inline
		52	#else
		53	# define expect(expr,value) (expr)
		54	# define INLINE static
		55	#endif
		56
		57	#define expect_false(expr) expect ((expr) != 0, 0)
		58	#define expect_true(expr) expect ((expr) != 0, 1)
		59
		60	#define IN_RANGE_INC(type,val,beg,end) \
		61	((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
		62	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
		63
		64	#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
		65
		66	#ifdef USE_ITHREADS
		67	# define JSON_SLOW 1
		68	# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
		69	#else
		70	# define JSON_SLOW 0
		71	# define JSON_STASH json_stash
		72	#endif
		73
41	static HV *json_stash; // JSON::XS::	74	static HV *json_stash, *json_boolean_stash; // JSON::XS::
		75	static SV *json_true, *json_false;
		76
		77	enum {
		78	INCR_M_WS = 0, // initial whitespace skipping, must be 0
		79	INCR_M_STR, // inside string
		80	INCR_M_BS, // inside backslash
		81	INCR_M_JSON // outside anything, count nesting
		82	};
		83
		84	#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
		85
		86	typedef struct {
		87	U32 flags;
		88	U32 max_depth;
		89	STRLEN max_size;
		90
		91	SV *cb_object;
		92	HV *cb_sk_object;
		93
		94	// for the incremental parser
		95	SV *incr_text; // the source text so far
		96	STRLEN incr_pos; // the current offset into the text
		97	int incr_nest; // {[]}-nesting level
		98	unsigned char incr_mode;
		99	} JSON;
		100
		101	INLINE void
		102	json_init (JSON *json)
		103	{
		104	Zero (json, 1, JSON);
		105	json->max_depth = 512;
		106	}
42		107
43	/////////////////////////////////////////////////////////////////////////////	108	/////////////////////////////////////////////////////////////////////////////
44	// utility functions	109	// utility functions
45		110
46	static UV *	111	INLINE SV *
47	SvJSON (SV *sv)	112	get_bool (const char *name)
48	{	113	{
49	if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))	114	SV *sv = get_sv (name, 1);
50	croak ("object is not of type JSON::XS");
51		115
52	return &SvUVX (SvRV (sv));	116	SvREADONLY_on (sv);
53	}	117	SvREADONLY_on (SvRV (sv));
54		118
55	static void	119	return sv;
		120	}
		121
		122	INLINE void
56	shrink (SV *sv)	123	shrink (SV *sv)
57	{	124	{
58	sv_utf8_downgrade (sv, 1);	125	sv_utf8_downgrade (sv, 1);
		126
59	if (SvLEN (sv) > SvCUR (sv) + 1)	127	if (SvLEN (sv) > SvCUR (sv) + 1)
60	{	128	{
61	#ifdef SvPV_shrink_to_cur	129	#ifdef SvPV_shrink_to_cur
62	SvPV_shrink_to_cur (sv);	130	SvPV_shrink_to_cur (sv);
63	#elif defined (SvPV_renew)	131	#elif defined (SvPV_renew)
…		…
69	// decode an utf-8 character and return it, or (UV)-1 in	137	// decode an utf-8 character and return it, or (UV)-1 in
70	// case of an error.	138	// case of an error.
71	// we special-case "safe" characters from U+80 .. U+7FF,	139	// we special-case "safe" characters from U+80 .. U+7FF,
72	// but use the very good perl function to parse anything else.	140	// but use the very good perl function to parse anything else.
73	// note that we never call this function for a ascii codepoints	141	// note that we never call this function for a ascii codepoints
74	static UV	142	INLINE UV
75	decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)	143	decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
76	{	144	{
77	if (s[0] > 0xdf || s[0] < 0xc2)	145	if (expect_true (len >= 2
78	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);	146	&& IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
79	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)	147	&& IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
80	{	148	{
81	*clen = 2;	149	*clen = 2;
82	return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);	150	return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
83	}	151	}
84	else	152	else
85	{	153	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
86	*clen = (STRLEN)-1;	154	}
87	return (UV)-1;	155
88	}	156	// likewise for encoding, also never called for ascii codepoints
		157	// this function takes advantage of this fact, although current gccs
		158	// seem to optimise the check for >= 0x80 away anyways
		159	INLINE unsigned char *
		160	encode_utf8 (unsigned char *s, UV ch)
		161	{
		162	if (expect_false (ch < 0x000080))
		163	*s++ = ch;
		164	else if (expect_true (ch < 0x000800))
		165	*s++ = 0xc0 | ( ch >> 6),
		166	*s++ = 0x80 | ( ch & 0x3f);
		167	else if ( ch < 0x010000)
		168	*s++ = 0xe0 | ( ch >> 12),
		169	*s++ = 0x80 | ((ch >> 6) & 0x3f),
		170	*s++ = 0x80 | ( ch & 0x3f);
		171	else if ( ch < 0x110000)
		172	*s++ = 0xf0 | ( ch >> 18),
		173	*s++ = 0x80 | ((ch >> 12) & 0x3f),
		174	*s++ = 0x80 | ((ch >> 6) & 0x3f),
		175	*s++ = 0x80 | ( ch & 0x3f);
		176
		177	return s;
89	}	178	}
90		179
91	/////////////////////////////////////////////////////////////////////////////	180	/////////////////////////////////////////////////////////////////////////////
92	// encoder	181	// encoder
93		182
…		…
95	typedef struct	184	typedef struct
96	{	185	{
97	char *cur; // SvPVX (sv) + current output position	186	char *cur; // SvPVX (sv) + current output position
98	char *end; // SvEND (sv)	187	char *end; // SvEND (sv)
99	SV *sv; // result scalar	188	SV *sv; // result scalar
100	U32 flags; // F_*	189	JSON json;
101	U32 indent; // indentation level	190	U32 indent; // indentation level
102	U32 maxdepth; // max. indentation/recursion level	191	UV limit; // escape character values >= this value when encoding
103	} enc_t;	192	} enc_t;
104		193
105	static void	194	INLINE void
106	need (enc_t *enc, STRLEN len)	195	need (enc_t *enc, STRLEN len)
107	{	196	{
108	if (enc->cur + len >= enc->end)	197	if (expect_false (enc->cur + len >= enc->end))
109	{	198	{
110	STRLEN cur = enc->cur - SvPVX (enc->sv);	199	STRLEN cur = enc->cur - SvPVX (enc->sv);
111	SvGROW (enc->sv, cur + len + 1);	200	SvGROW (enc->sv, cur + len + 1);
112	enc->cur = SvPVX (enc->sv) + cur;	201	enc->cur = SvPVX (enc->sv) + cur;
113	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	202	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
114	}	203	}
115	}	204	}
116		205
117	static void	206	INLINE void
118	encode_ch (enc_t *enc, char ch)	207	encode_ch (enc_t *enc, char ch)
119	{	208	{
120	need (enc, 1);	209	need (enc, 1);
121	*enc->cur++ = ch;	210	*enc->cur++ = ch;
122	}	211	}
…		…
130		219
131	while (str < end)	220	while (str < end)
132	{	221	{
133	unsigned char ch = *(unsigned char *)str;	222	unsigned char ch = *(unsigned char *)str;
134		223
135	if (ch >= 0x20 && ch < 0x80) // most common case	224	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
136	{	225	{
137	if (ch == '"') // but with slow exceptions	226	if (expect_false (ch == '"')) // but with slow exceptions
138	{	227	{
139	need (enc, len += 1);	228	need (enc, len += 1);
140	*enc->cur++ = '\\';	229	*enc->cur++ = '\\';
141	*enc->cur++ = '"';	230	*enc->cur++ = '"';
142	}	231	}
143	else if (ch == '\\')	232	else if (expect_false (ch == '\\'))
144	{	233	{
145	need (enc, len += 1);	234	need (enc, len += 1);
146	*enc->cur++ = '\\';	235	*enc->cur++ = '\\';
147	*enc->cur++ = '\\';	236	*enc->cur++ = '\\';
148	}	237	}
…		…
166	STRLEN clen;	255	STRLEN clen;
167	UV uch;	256	UV uch;
168		257
169	if (is_utf8)	258	if (is_utf8)
170	{	259	{
171	//uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
172	uch = decode_utf8 (str, end - str, &clen);	260	uch = decode_utf8 (str, end - str, &clen);
173	if (clen == (STRLEN)-1)	261	if (clen == (STRLEN)-1)
174	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);	262	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
175	}	263	}
176	else	264	else
177	{	265	{
178	uch = ch;	266	uch = ch;
179	clen = 1;	267	clen = 1;
180	}	268	}
181		269
182	if (uch > 0x10FFFFUL)	270	if (uch < 0x80/*0x20*/ || uch >= enc->limit)
183	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
184
185	if (uch < 0x80 || enc->flags & F_ASCII)
186	{	271	{
187	if (uch > 0xFFFFUL)	272	if (uch >= 0x10000UL)
188	{	273	{
		274	if (uch >= 0x110000UL)
		275	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
		276
189	need (enc, len += 11);	277	need (enc, len += 11);
190	sprintf (enc->cur, "\\u%04x\\u%04x",	278	sprintf (enc->cur, "\\u%04x\\u%04x",
191	(int)((uch - 0x10000) / 0x400 + 0xD800),	279	(int)((uch - 0x10000) / 0x400 + 0xD800),
192	(int)((uch - 0x10000) % 0x400 + 0xDC00));	280	(int)((uch - 0x10000) % 0x400 + 0xDC00));
193	enc->cur += 12;	281	enc->cur += 12;
194	}	282	}
195	else	283	else
196	{	284	{
197	static char hexdigit [16] = "0123456789abcdef";
198	need (enc, len += 5);	285	need (enc, len += 5);
199	*enc->cur++ = '\\';	286	*enc->cur++ = '\\';
200	*enc->cur++ = 'u';	287	*enc->cur++ = 'u';
201	*enc->cur++ = hexdigit [ uch >> 12 ];	288	*enc->cur++ = PL_hexdigit [ uch >> 12 ];
202	*enc->cur++ = hexdigit [(uch >> 8) & 15];	289	*enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
203	*enc->cur++ = hexdigit [(uch >> 4) & 15];	290	*enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
204	*enc->cur++ = hexdigit [(uch >> 0) & 15];	291	*enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
205	}	292	}
206		293
		294	str += clen;
		295	}
		296	else if (enc->json.flags & F_LATIN1)
		297	{
		298	*enc->cur++ = uch;
207	str += clen;	299	str += clen;
208	}	300	}
209	else if (is_utf8)	301	else if (is_utf8)
210	{	302	{
211	need (enc, len += clen);	303	need (enc, len += clen);
…		…
216	while (--clen);	308	while (--clen);
217	}	309	}
218	else	310	else
219	{	311	{
220	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed	312	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
221	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	313	enc->cur = encode_utf8 (enc->cur, uch);
222	++str;	314	++str;
223	}	315	}
224	}	316	}
225	}	317	}
226	}	318	}
227		319
228	--len;	320	--len;
229	}	321	}
230	}	322	}
231		323
232	static void	324	INLINE void
233	encode_indent (enc_t *enc)	325	encode_indent (enc_t *enc)
234	{	326	{
235	if (enc->flags & F_INDENT)	327	if (enc->json.flags & F_INDENT)
236	{	328	{
237	int spaces = enc->indent * INDENT_STEP;	329	int spaces = enc->indent * INDENT_STEP;
238		330
239	need (enc, spaces);	331	need (enc, spaces);
240	memset (enc->cur, ' ', spaces);	332	memset (enc->cur, ' ', spaces);
241	enc->cur += spaces;	333	enc->cur += spaces;
242	}	334	}
243	}	335	}
244		336
245	static void	337	INLINE void
246	encode_space (enc_t *enc)	338	encode_space (enc_t *enc)
247	{	339	{
248	need (enc, 1);	340	need (enc, 1);
249	encode_ch (enc, ' ');	341	encode_ch (enc, ' ');
250	}	342	}
251		343
252	static void	344	INLINE void
253	encode_nl (enc_t *enc)	345	encode_nl (enc_t *enc)
254	{	346	{
255	if (enc->flags & F_INDENT)	347	if (enc->json.flags & F_INDENT)
256	{	348	{
257	need (enc, 1);	349	need (enc, 1);
258	encode_ch (enc, '\n');	350	encode_ch (enc, '\n');
259	}	351	}
260	}	352	}
261		353
262	static void	354	INLINE void
263	encode_comma (enc_t *enc)	355	encode_comma (enc_t *enc)
264	{	356	{
265	encode_ch (enc, ',');	357	encode_ch (enc, ',');
266		358
267	if (enc->flags & F_INDENT)	359	if (enc->json.flags & F_INDENT)
268	encode_nl (enc);	360	encode_nl (enc);
269	else if (enc->flags & F_SPACE_AFTER)	361	else if (enc->json.flags & F_SPACE_AFTER)
270	encode_space (enc);	362	encode_space (enc);
271	}	363	}
272		364
273	static void encode_sv (enc_t *enc, SV *sv);	365	static void encode_sv (enc_t *enc, SV *sv);
274		366
275	static void	367	static void
276	encode_av (enc_t *enc, AV *av)	368	encode_av (enc_t *enc, AV *av)
277	{	369	{
278	int i, len = av_len (av);	370	int i, len = av_len (av);
279		371
280	if (enc->indent >= enc->maxdepth)	372	if (enc->indent >= enc->json.max_depth)
281	croak ("data structure too deep (hit recursion limit)");	373	croak (ERR_NESTING_EXCEEDED);
282		374
283	encode_ch (enc, '['); encode_nl (enc);	375	encode_ch (enc, '[');
284	++enc->indent;	376
		377	if (len >= 0)
		378	{
		379	encode_nl (enc); ++enc->indent;
285		380
286	for (i = 0; i <= len; ++i)	381	for (i = 0; i <= len; ++i)
287	{	382	{
		383	SV **svp = av_fetch (av, i, 0);
		384
288	encode_indent (enc);	385	encode_indent (enc);
289	encode_sv (enc, *av_fetch (av, i, 0));
290		386
		387	if (svp)
		388	encode_sv (enc, *svp);
		389	else
		390	encode_str (enc, "null", 4, 0);
		391
291	if (i < len)	392	if (i < len)
292	encode_comma (enc);	393	encode_comma (enc);
293	}	394	}
294		395
		396	encode_nl (enc); --enc->indent; encode_indent (enc);
		397	}
		398
295	encode_nl (enc);	399	encode_ch (enc, ']');
296
297	--enc->indent;
298	encode_indent (enc); encode_ch (enc, ']');
299	}	400	}
300		401
301	static void	402	static void
302	encode_he (enc_t *enc, HE *he)	403	encode_hk (enc_t *enc, HE *he)
303	{	404	{
304	encode_ch (enc, '"');	405	encode_ch (enc, '"');
305		406
306	if (HeKLEN (he) == HEf_SVKEY)	407	if (HeKLEN (he) == HEf_SVKEY)
307	{	408	{
…		…
317	else	418	else
318	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));	419	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
319		420
320	encode_ch (enc, '"');	421	encode_ch (enc, '"');
321		422
322	if (enc->flags & F_SPACE_BEFORE) encode_space (enc);	423	if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
323	encode_ch (enc, ':');	424	encode_ch (enc, ':');
324	if (enc->flags & F_SPACE_AFTER ) encode_space (enc);	425	if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
325	encode_sv (enc, HeVAL (he));
326	}	426	}
327		427
328	// compare hash entries, used when all keys are bytestrings	428	// compare hash entries, used when all keys are bytestrings
329	static int	429	static int
330	he_cmp_fast (const void *a_, const void *b_)	430	he_cmp_fast (const void *a_, const void *b_)
…		…
335	HE *b = *(HE **)b_;	435	HE *b = *(HE **)b_;
336		436
337	STRLEN la = HeKLEN (a);	437	STRLEN la = HeKLEN (a);
338	STRLEN lb = HeKLEN (b);	438	STRLEN lb = HeKLEN (b);
339		439
340	if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))	440	if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
341	cmp = la - lb;	441	cmp = lb - la;
342		442
343	return cmp;	443	return cmp;
344	}	444	}
345		445
346	// compare hash entries, used when some keys are sv's or utf-x	446	// compare hash entries, used when some keys are sv's or utf-x
347	static int	447	static int
348	he_cmp_slow (const void *a, const void *b)	448	he_cmp_slow (const void *a, const void *b)
349	{	449	{
350	return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b));	450	return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
351	}	451	}
352		452
353	static void	453	static void
354	encode_hv (enc_t *enc, HV *hv)	454	encode_hv (enc_t *enc, HV *hv)
355	{	455	{
356	int count, i;	456	HE *he;
357		457
358	if (enc->indent >= enc->maxdepth)	458	if (enc->indent >= enc->json.max_depth)
359	croak ("data structure too deep (hit recursion limit)");	459	croak (ERR_NESTING_EXCEEDED);
360		460
361	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;	461	encode_ch (enc, '{');
362		462
363	if ((count = hv_iterinit (hv)))
364	{
365	// for canonical output we have to sort by keys first	463	// for canonical output we have to sort by keys first
366	// actually, this is mostly due to the stupid so-called	464	// actually, this is mostly due to the stupid so-called
367	// security workaround added somewhere in 5.8.x.	465	// security workaround added somewhere in 5.8.x
368	// that randomises hash orderings	466	// that randomises hash orderings
369	if (enc->flags & F_CANONICAL)	467	if (enc->json.flags & F_CANONICAL)
		468	{
		469	int count = hv_iterinit (hv);
		470
		471	if (SvMAGICAL (hv))
370	{	472	{
		473	// need to count by iterating. could improve by dynamically building the vector below
		474	// but I don't care for the speed of this special case.
		475	// note also that we will run into undefined behaviour when the two iterations
		476	// do not result in the same count, something I might care for in some later release.
		477
		478	count = 0;
		479	while (hv_iternext (hv))
		480	++count;
		481
		482	hv_iterinit (hv);
		483	}
		484
		485	if (count)
		486	{
		487	int i, fast = 1;
		488	#if defined(__BORLANDC__) || defined(_MSC_VER)
		489	HE **hes = _alloca (count * sizeof (HE));
		490	#else
371	HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode	491	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
372	int fast = 1;	492	#endif
373		493
374	i = 0;	494	i = 0;
375	while ((he = hv_iternext (hv)))	495	while ((he = hv_iternext (hv)))
376	{	496	{
377	hes [i++] = he;	497	hes [i++] = he;
…		…
399		519
400	FREETMPS;	520	FREETMPS;
401	LEAVE;	521	LEAVE;
402	}	522	}
403		523
404	for (i = 0; i < count; ++i)	524	encode_nl (enc); ++enc->indent;
		525
		526	while (count--)
405	{	527	{
406	encode_indent (enc);	528	encode_indent (enc);
		529	he = hes [count];
407	encode_he (enc, hes [i]);	530	encode_hk (enc, he);
		531	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
408		532
409	if (i < count - 1)	533	if (count)
410	encode_comma (enc);	534	encode_comma (enc);
411	}	535	}
412		536
		537	encode_nl (enc); --enc->indent; encode_indent (enc);
		538	}
		539	}
		540	else
		541	{
		542	if (hv_iterinit (hv) || SvMAGICAL (hv))
		543	if ((he = hv_iternext (hv)))
		544	{
		545	encode_nl (enc); ++enc->indent;
		546
		547	for (;;)
		548	{
413	encode_nl (enc);	549	encode_indent (enc);
		550	encode_hk (enc, he);
		551	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
		552
		553	if (!(he = hv_iternext (hv)))
		554	break;
		555
		556	encode_comma (enc);
		557	}
		558
		559	encode_nl (enc); --enc->indent; encode_indent (enc);
		560	}
		561	}
		562
		563	encode_ch (enc, '}');
		564	}
		565
		566	// encode objects, arrays and special \0=false and \1=true values.
		567	static void
		568	encode_rv (enc_t *enc, SV *sv)
		569	{
		570	svtype svt;
		571
		572	SvGETMAGIC (sv);
		573	svt = SvTYPE (sv);
		574
		575	if (expect_false (SvOBJECT (sv)))
		576	{
		577	HV *stash = !JSON_SLOW || json_boolean_stash
		578	? json_boolean_stash
		579	: gv_stashpv ("JSON::XS::Boolean", 1);
		580
		581	if (SvSTASH (sv) == stash)
		582	{
		583	if (SvIV (sv))
		584	encode_str (enc, "true", 4, 0);
		585	else
		586	encode_str (enc, "false", 5, 0);
414	}	587	}
415	else	588	else
416	{	589	{
417	HE *he = hv_iternext (hv);	590	#if 0
418		591	if (0 && sv_derived_from (rv, "JSON::Literal"))
419	for (;;)
420	{	592	{
421	encode_indent (enc);	593	// not yet
422	encode_he (enc, he);
423
424	if (!(he = hv_iternext (hv)))
425	break;
426
427	encode_comma (enc);
428	}	594	}
		595	#endif
		596	if (enc->json.flags & F_CONV_BLESSED)
		597	{
		598	// we re-bless the reference to get overload and other niceties right
		599	GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
429		600
		601	if (to_json)
		602	{
		603	dSP;
		604
		605	ENTER; SAVETMPS; PUSHMARK (SP);
		606	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
		607
		608	// calling with G_SCALAR ensures that we always get a 1 return value
		609	PUTBACK;
		610	call_sv ((SV *)GvCV (to_json), G_SCALAR);
		611	SPAGAIN;
		612
		613	// catch this surprisingly common error
		614	if (SvROK (TOPs) && SvRV (TOPs) == sv)
		615	croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
		616
		617	sv = POPs;
		618	PUTBACK;
		619
430	encode_nl (enc);	620	encode_sv (enc, sv);
		621
		622	FREETMPS; LEAVE;
		623	}
		624	else if (enc->json.flags & F_ALLOW_BLESSED)
		625	encode_str (enc, "null", 4, 0);
		626	else
		627	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
		628	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		629	}
		630	else if (enc->json.flags & F_ALLOW_BLESSED)
		631	encode_str (enc, "null", 4, 0);
		632	else
		633	croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
		634	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
431	}	635	}
432	}	636	}
433
434	--enc->indent; encode_indent (enc); encode_ch (enc, '}');
435	}
436
437	// encode objects, arrays and special \0=false and \1=true values.
438	static void
439	encode_rv (enc_t *enc, SV *sv)
440	{
441	svtype svt;
442
443	SvGETMAGIC (sv);
444	svt = SvTYPE (sv);
445
446	if (svt == SVt_PVHV)	637	else if (svt == SVt_PVHV)
447	encode_hv (enc, (HV *)sv);	638	encode_hv (enc, (HV *)sv);
448	else if (svt == SVt_PVAV)	639	else if (svt == SVt_PVAV)
449	encode_av (enc, (AV *)sv);	640	encode_av (enc, (AV *)sv);
450	else if (svt < SVt_PVAV)	641	else if (svt < SVt_PVAV)
451	{	642	{
452	if (SvNIOK (sv) && SvIV (sv) == 0)	643	STRLEN len = 0;
		644	char *pv = svt ? SvPV (sv, len) : 0;
		645
		646	if (len == 1 && *pv == '1')
		647	encode_str (enc, "true", 4, 0);
		648	else if (len == 1 && *pv == '0')
453	encode_str (enc, "false", 5, 0);	649	encode_str (enc, "false", 5, 0);
454	else if (SvNIOK (sv) && SvIV (sv) == 1)	650	else if (enc->json.flags & F_ALLOW_UNKNOWN)
455	encode_str (enc, "true", 4, 0);	651	encode_str (enc, "null", 4, 0);
456	else	652	else
457	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",	653	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
458	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	654	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
459	}	655	}
		656	else if (enc->json.flags & F_ALLOW_UNKNOWN)
		657	encode_str (enc, "null", 4, 0);
460	else	658	else
461	croak ("encountered %s, but JSON can only represent references to arrays or hashes",	659	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
462	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	660	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
463	}	661	}
464		662
…		…
475	encode_str (enc, str, len, SvUTF8 (sv));	673	encode_str (enc, str, len, SvUTF8 (sv));
476	encode_ch (enc, '"');	674	encode_ch (enc, '"');
477	}	675	}
478	else if (SvNOKp (sv))	676	else if (SvNOKp (sv))
479	{	677	{
		678	// trust that perl will do the right thing w.r.t. JSON syntax.
480	need (enc, NV_DIG + 32);	679	need (enc, NV_DIG + 32);
481	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	680	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
482	enc->cur += strlen (enc->cur);	681	enc->cur += strlen (enc->cur);
483	}	682	}
484	else if (SvIOKp (sv))	683	else if (SvIOKp (sv))
485	{	684	{
		685	// we assume we can always read an IV as a UV and vice versa
		686	// we assume two's complement
		687	// we assume no aliasing issues in the union
		688	if (SvIsUV (sv) ? SvUVX (sv) <= 59000
		689	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
		690	{
		691	// optimise the "small number case"
		692	// code will likely be branchless and use only a single multiplication
		693	// works for numbers up to 59074
		694	I32 i = SvIVX (sv);
		695	U32 u;
		696	char digit, nz = 0;
		697
486	need (enc, 64);	698	need (enc, 6);
		699
		700	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
		701	u = i < 0 ? -i : i;
		702
		703	// convert to 4.28 fixed-point representation
		704	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
		705
		706	// now output digit by digit, each time masking out the integer part
		707	// and multiplying by 5 while moving the decimal point one to the right,
		708	// resulting in a net multiplication by 10.
		709	// we always write the digit to memory but conditionally increment
		710	// the pointer, to enable the use of conditional move instructions.
		711	digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
		712	digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
		713	digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
		714	digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
		715	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
		716	}
		717	else
		718	{
		719	// large integer, use the (rather slow) snprintf way.
		720	need (enc, IVUV_MAXCHARS);
487	enc->cur +=	721	enc->cur +=
488	SvIsUV(sv)	722	SvIsUV(sv)
489	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	723	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
490	: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));	724	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
		725	}
491	}	726	}
492	else if (SvROK (sv))	727	else if (SvROK (sv))
493	encode_rv (enc, SvRV (sv));	728	encode_rv (enc, SvRV (sv));
494	else if (!SvOK (sv))	729	else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
495	encode_str (enc, "null", 4, 0);	730	encode_str (enc, "null", 4, 0);
496	else	731	else
497	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	732	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
498	SvPV_nolen (sv), SvFLAGS (sv));	733	SvPV_nolen (sv), SvFLAGS (sv));
499	}	734	}
500		735
501	static SV *	736	static SV *
502	encode_json (SV *scalar, U32 flags)	737	encode_json (SV *scalar, JSON *json)
503	{	738	{
504	enc_t enc;	739	enc_t enc;
505		740
506	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	741	if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
507	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");	742	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
508		743
509	enc.flags = flags;	744	enc.json = *json;
510	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	745	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
511	enc.cur = SvPVX (enc.sv);	746	enc.cur = SvPVX (enc.sv);
512	enc.end = SvEND (enc.sv);	747	enc.end = SvEND (enc.sv);
513	enc.indent = 0;	748	enc.indent = 0;
514	enc.maxdepth = DEC_DEPTH (flags);	749	enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
		750	: enc.json.flags & F_LATIN1 ? 0x000100UL
		751	: 0x110000UL;
515		752
516	SvPOK_only (enc.sv);	753	SvPOK_only (enc.sv);
517	encode_sv (&enc, scalar);	754	encode_sv (&enc, scalar);
518		755
519	if (!(flags & (F_ASCII | F_UTF8)))
520	SvUTF8_on (enc.sv);
521
522	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	756	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
523	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	757	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
524		758
		759	if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
		760	SvUTF8_on (enc.sv);
		761
525	if (enc.flags & F_SHRINK)	762	if (enc.json.flags & F_SHRINK)
526	shrink (enc.sv);	763	shrink (enc.sv);
527		764
528	return enc.sv;	765	return enc.sv;
529	}	766	}
530		767
…		…
535	typedef struct	772	typedef struct
536	{	773	{
537	char *cur; // current parser pointer	774	char *cur; // current parser pointer
538	char *end; // end of input string	775	char *end; // end of input string
539	const char *err; // parse error, if != 0	776	const char *err; // parse error, if != 0
540	U32 flags; // F_*	777	JSON json;
541	U32 depth; // recursion depth	778	U32 depth; // recursion depth
542	U32 maxdepth; // recursion depth limit	779	U32 maxdepth; // recursion depth limit
543	} dec_t;	780	} dec_t;
544		781
545	static void	782	INLINE void
		783	decode_comment (dec_t *dec)
		784	{
		785	// only '#'-style comments allowed a.t.m.
		786
		787	while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
		788	++dec->cur;
		789	}
		790
		791	INLINE void
546	decode_ws (dec_t *dec)	792	decode_ws (dec_t *dec)
547	{	793	{
548	for (;;)	794	for (;;)
549	{	795	{
550	char ch = *dec->cur;	796	char ch = *dec->cur;
551		797
552	if (ch > 0x20	798	if (ch > 0x20)
		799	{
		800	if (expect_false (ch == '#'))
		801	{
		802	if (dec->json.flags & F_RELAXED)
		803	decode_comment (dec);
		804	else
		805	break;
		806	}
		807	else
		808	break;
		809	}
553	|| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))	810	else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
554	break;	811	break; // parse error, but let higher level handle it, gives better error messages
555		812
556	++dec->cur;	813	++dec->cur;
557	}	814	}
558	}	815	}
559		816
…		…
563	if (*dec->cur != ch) \	820	if (*dec->cur != ch) \
564	ERR (# ch " expected"); \	821	ERR (# ch " expected"); \
565	++dec->cur; \	822	++dec->cur; \
566	SE	823	SE
567		824
568	#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")	825	#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
569	#define DEC_DEC_DEPTH --dec->depth	826	#define DEC_DEC_DEPTH --dec->depth
570		827
571	static SV *decode_sv (dec_t *dec);	828	static SV *decode_sv (dec_t *dec);
572		829
573	static signed char decode_hexdigit[256];	830	static signed char decode_hexdigit[256];
…		…
576	decode_4hex (dec_t *dec)	833	decode_4hex (dec_t *dec)
577	{	834	{
578	signed char d1, d2, d3, d4;	835	signed char d1, d2, d3, d4;
579	unsigned char *cur = (unsigned char *)dec->cur;	836	unsigned char *cur = (unsigned char *)dec->cur;
580		837
581	d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");	838	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
582	d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");	839	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
583	d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");	840	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
584	d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");	841	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
585		842
586	dec->cur += 4;	843	dec->cur += 4;
587		844
588	return ((UV)d1) << 12	845	return ((UV)d1) << 12
589	| ((UV)d2) << 8	846	| ((UV)d2) << 8
…		…
597	static SV *	854	static SV *
598	decode_str (dec_t *dec)	855	decode_str (dec_t *dec)
599	{	856	{
600	SV *sv = 0;	857	SV *sv = 0;
601	int utf8 = 0;	858	int utf8 = 0;
		859	char *dec_cur = dec->cur;
602		860
603	do	861	do
604	{	862	{
605	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];	863	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
606	char *cur = buf;	864	char *cur = buf;
607		865
608	do	866	do
609	{	867	{
610	unsigned char ch = *(unsigned char *)dec->cur++;	868	unsigned char ch = *(unsigned char *)dec_cur++;
611		869
612	if (ch == '"')	870	if (expect_false (ch == '"'))
613	{	871	{
614	--dec->cur;	872	--dec_cur;
615	break;	873	break;
616	}	874	}
617	else if (ch == '\\')	875	else if (expect_false (ch == '\\'))
618	{	876	{
619	switch (*dec->cur)	877	switch (*dec_cur)
620	{	878	{
621	case '\\':	879	case '\\':
622	case '/':	880	case '/':
623	case '"': *cur++ = *dec->cur++; break;	881	case '"': *cur++ = *dec_cur++; break;
624		882
625	case 'b': ++dec->cur; *cur++ = '\010'; break;	883	case 'b': ++dec_cur; *cur++ = '\010'; break;
626	case 't': ++dec->cur; *cur++ = '\011'; break;	884	case 't': ++dec_cur; *cur++ = '\011'; break;
627	case 'n': ++dec->cur; *cur++ = '\012'; break;	885	case 'n': ++dec_cur; *cur++ = '\012'; break;
628	case 'f': ++dec->cur; *cur++ = '\014'; break;	886	case 'f': ++dec_cur; *cur++ = '\014'; break;
629	case 'r': ++dec->cur; *cur++ = '\015'; break;	887	case 'r': ++dec_cur; *cur++ = '\015'; break;
630		888
631	case 'u':	889	case 'u':
632	{	890	{
633	UV lo, hi;	891	UV lo, hi;
634	++dec->cur;	892	++dec_cur;
635		893
		894	dec->cur = dec_cur;
636	hi = decode_4hex (dec);	895	hi = decode_4hex (dec);
		896	dec_cur = dec->cur;
637	if (hi == (UV)-1)	897	if (hi == (UV)-1)
638	goto fail;	898	goto fail;
639		899
640	// possibly a surrogate pair	900	// possibly a surrogate pair
641	if (hi >= 0xd800)	901	if (hi >= 0xd800)
642	if (hi < 0xdc00)	902	if (hi < 0xdc00)
643	{	903	{
644	if (dec->cur [0] != '\\' || dec->cur [1] != 'u')	904	if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
645	ERR ("missing low surrogate character in surrogate pair");	905	ERR ("missing low surrogate character in surrogate pair");
646		906
647	dec->cur += 2;	907	dec_cur += 2;
648		908
		909	dec->cur = dec_cur;
649	lo = decode_4hex (dec);	910	lo = decode_4hex (dec);
		911	dec_cur = dec->cur;
650	if (lo == (UV)-1)	912	if (lo == (UV)-1)
651	goto fail;	913	goto fail;
652		914
653	if (lo < 0xdc00 || lo >= 0xe000)	915	if (lo < 0xdc00 || lo >= 0xe000)
654	ERR ("surrogate pair expected");	916	ERR ("surrogate pair expected");
…		…
660		922
661	if (hi >= 0x80)	923	if (hi >= 0x80)
662	{	924	{
663	utf8 = 1;	925	utf8 = 1;
664		926
665	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);	927	cur = encode_utf8 (cur, hi);
666	}	928	}
667	else	929	else
668	*cur++ = hi;	930	*cur++ = hi;
669	}	931	}
670	break;	932	break;
671		933
672	default:	934	default:
673	--dec->cur;	935	--dec_cur;
674	ERR ("illegal backslash escape sequence in string");	936	ERR ("illegal backslash escape sequence in string");
675	}	937	}
676	}	938	}
677	else if (ch >= 0x20 && ch <= 0x7f)	939	else if (expect_true (ch >= 0x20 && ch < 0x80))
678	*cur++ = ch;	940	*cur++ = ch;
679	else if (ch >= 0x80)	941	else if (ch >= 0x80)
680	{	942	{
681	STRLEN clen;	943	STRLEN clen;
682	UV uch;	944	UV uch;
683		945
684	--dec->cur;	946	--dec_cur;
685		947
686	uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);	948	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
687	if (clen == (STRLEN)-1)	949	if (clen == (STRLEN)-1)
688	ERR ("malformed UTF-8 character in JSON string");	950	ERR ("malformed UTF-8 character in JSON string");
689		951
690	do	952	do
691	*cur++ = *dec->cur++;	953	*cur++ = *dec_cur++;
692	while (--clen);	954	while (--clen);
693		955
694	utf8 = 1;	956	utf8 = 1;
695	}	957	}
696	else	958	else
697	{	959	{
698	--dec->cur;	960	--dec_cur;
699		961
700	if (!ch)	962	if (!ch)
701	ERR ("unexpected end of string while parsing JSON string");	963	ERR ("unexpected end of string while parsing JSON string");
702	else	964	else
703	ERR ("invalid character encountered while parsing JSON string");	965	ERR ("invalid character encountered while parsing JSON string");
…		…
716	}	978	}
717	else	979	else
718	sv = newSVpvn (buf, len);	980	sv = newSVpvn (buf, len);
719	}	981	}
720	}	982	}
721	while (*dec->cur != '"');	983	while (*dec_cur != '"');
722		984
723	++dec->cur;	985	++dec_cur;
724		986
725	if (sv)	987	if (sv)
726	{	988	{
727	SvPOK_only (sv);	989	SvPOK_only (sv);
728	*SvEND (sv) = 0;	990	*SvEND (sv) = 0;
…		…
731	SvUTF8_on (sv);	993	SvUTF8_on (sv);
732	}	994	}
733	else	995	else
734	sv = newSVpvn ("", 0);	996	sv = newSVpvn ("", 0);
735		997
		998	dec->cur = dec_cur;
736	return sv;	999	return sv;
737		1000
738	fail:	1001	fail:
		1002	dec->cur = dec_cur;
739	return 0;	1003	return 0;
740	}	1004	}
741		1005
742	static SV *	1006	static SV *
743	decode_num (dec_t *dec)	1007	decode_num (dec_t *dec)
…		…
801	is_nv = 1;	1065	is_nv = 1;
802	}	1066	}
803		1067
804	if (!is_nv)	1068	if (!is_nv)
805	{	1069	{
806	UV uv;	1070	int len = dec->cur - start;
807	int numtype = grok_number (start, dec->cur - start, &uv);	1071
808	if (numtype & IS_NUMBER_IN_UV)	1072	// special case the rather common 1..5-digit-int case
809	if (numtype & IS_NUMBER_NEG)	1073	if (*start == '-')
		1074	switch (len)
810	{	1075	{
811	if (uv < (UV)IV_MIN)	1076	case 2: return newSViv (-( start [1] - '0' * 1));
812	return newSViv (-(IV)uv);	1077	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
		1078	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
		1079	case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
		1080	case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
813	}	1081	}
		1082	else
		1083	switch (len)
		1084	{
		1085	case 1: return newSViv ( start [0] - '0' * 1);
		1086	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
		1087	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
		1088	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		1089	case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
		1090	}
		1091
		1092	{
		1093	UV uv;
		1094	int numtype = grok_number (start, len, &uv);
		1095	if (numtype & IS_NUMBER_IN_UV)
		1096	if (numtype & IS_NUMBER_NEG)
		1097	{
		1098	if (uv < (UV)IV_MIN)
		1099	return newSViv (-(IV)uv);
		1100	}
814	else	1101	else
815	return newSVuv (uv);	1102	return newSVuv (uv);
816	}	1103	}
817		1104
		1105	len -= *start == '-' ? 1 : 0;
		1106
		1107	// does not fit into IV or UV, try NV
		1108	if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
		1109	#if defined (LDBL_DIG)
		1110	|| (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
		1111	#endif
		1112	)
		1113	// fits into NV without loss of precision
		1114	return newSVnv (Atof (start));
		1115
		1116	// everything else fails, convert it to a string
		1117	return newSVpvn (start, dec->cur - start);
		1118	}
		1119
		1120	// loss of precision here
818	return newSVnv (Atof (start));	1121	return newSVnv (Atof (start));
819		1122
820	fail:	1123	fail:
821	return 0;	1124	return 0;
822	}	1125	}
…		…
852		1155
853	if (*dec->cur != ',')	1156	if (*dec->cur != ',')
854	ERR (", or ] expected while parsing array");	1157	ERR (", or ] expected while parsing array");
855		1158
856	++dec->cur;	1159	++dec->cur;
		1160
		1161	decode_ws (dec);
		1162
		1163	if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
		1164	{
		1165	++dec->cur;
		1166	break;
		1167	}
857	}	1168	}
858		1169
859	DEC_DEC_DEPTH;	1170	DEC_DEC_DEPTH;
860	return newRV_noinc ((SV *)av);	1171	return newRV_noinc ((SV *)av);
861		1172
…		…
866	}	1177	}
867		1178
868	static SV *	1179	static SV *
869	decode_hv (dec_t *dec)	1180	decode_hv (dec_t *dec)
870	{	1181	{
		1182	SV *sv;
871	HV *hv = newHV ();	1183	HV *hv = newHV ();
872		1184
873	DEC_INC_DEPTH;	1185	DEC_INC_DEPTH;
874	decode_ws (dec);	1186	decode_ws (dec);
875		1187
876	if (*dec->cur == '}')	1188	if (*dec->cur == '}')
877	++dec->cur;	1189	++dec->cur;
878	else	1190	else
879	for (;;)	1191	for (;;)
880	{	1192	{
		1193	EXPECT_CH ('"');
		1194
		1195	// heuristic: assume that
		1196	// a) decode_str + hv_store_ent are abysmally slow.
		1197	// b) most hash keys are short, simple ascii text.
		1198	// => try to "fast-match" such strings to avoid
		1199	// the overhead of decode_str + hv_store_ent.
		1200	{
881	SV *key, *value;	1201	SV *value;
		1202	char *p = dec->cur;
		1203	char *e = p + 24; // only try up to 24 bytes
882		1204
883	decode_ws (dec); EXPECT_CH ('"');	1205	for (;;)
884
885	key = decode_str (dec);
886	if (!key)
887	goto fail;
888
889	decode_ws (dec); EXPECT_CH (':');
890
891	value = decode_sv (dec);
892	if (!value)
893	{	1206	{
		1207	// the >= 0x80 is false on most architectures
		1208	if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
		1209	{
		1210	// slow path, back up and use decode_str
		1211	SV *key = decode_str (dec);
		1212	if (!key)
		1213	goto fail;
		1214
		1215	decode_ws (dec); EXPECT_CH (':');
		1216
		1217	decode_ws (dec);
		1218	value = decode_sv (dec);
		1219	if (!value)
		1220	{
		1221	SvREFCNT_dec (key);
		1222	goto fail;
		1223	}
		1224
		1225	hv_store_ent (hv, key, value, 0);
894	SvREFCNT_dec (key);	1226	SvREFCNT_dec (key);
		1227
		1228	break;
		1229	}
		1230	else if (*p == '"')
		1231	{
		1232	// fast path, got a simple key
		1233	char *key = dec->cur;
		1234	int len = p - key;
		1235	dec->cur = p + 1;
		1236
		1237	decode_ws (dec); EXPECT_CH (':');
		1238
		1239	decode_ws (dec);
		1240	value = decode_sv (dec);
		1241	if (!value)
895	goto fail;	1242	goto fail;
		1243
		1244	hv_store (hv, key, len, value, 0);
		1245
		1246	break;
		1247	}
		1248
		1249	++p;
896	}	1250	}
897		1251	}
898	hv_store_ent (hv, key, value, 0);
899	SvREFCNT_dec (key);
900		1252
901	decode_ws (dec);	1253	decode_ws (dec);
902		1254
903	if (*dec->cur == '}')	1255	if (*dec->cur == '}')
904	{	1256	{
…		…
908		1260
909	if (*dec->cur != ',')	1261	if (*dec->cur != ',')
910	ERR (", or } expected while parsing object/hash");	1262	ERR (", or } expected while parsing object/hash");
911		1263
912	++dec->cur;	1264	++dec->cur;
		1265
		1266	decode_ws (dec);
		1267
		1268	if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
		1269	{
		1270	++dec->cur;
		1271	break;
		1272	}
913	}	1273	}
914		1274
915	DEC_DEC_DEPTH;	1275	DEC_DEC_DEPTH;
916	return newRV_noinc ((SV *)hv);	1276	sv = newRV_noinc ((SV *)hv);
		1277
		1278	// check filter callbacks
		1279	if (dec->json.flags & F_HOOK)
		1280	{
		1281	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
		1282	{
		1283	HE *cb, *he;
		1284
		1285	hv_iterinit (hv);
		1286	he = hv_iternext (hv);
		1287	hv_iterinit (hv);
		1288
		1289	// the next line creates a mortal sv each time its called.
		1290	// might want to optimise this for common cases.
		1291	cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
		1292
		1293	if (cb)
		1294	{
		1295	dSP;
		1296	int count;
		1297
		1298	ENTER; SAVETMPS; PUSHMARK (SP);
		1299	XPUSHs (HeVAL (he));
		1300
		1301	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
		1302
		1303	if (count == 1)
		1304	{
		1305	sv = newSVsv (POPs);
		1306	FREETMPS; LEAVE;
		1307	return sv;
		1308	}
		1309
		1310	FREETMPS; LEAVE;
		1311	}
		1312	}
		1313
		1314	if (dec->json.cb_object)
		1315	{
		1316	dSP;
		1317	int count;
		1318
		1319	ENTER; SAVETMPS; PUSHMARK (SP);
		1320	XPUSHs (sv_2mortal (sv));
		1321
		1322	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
		1323
		1324	if (count == 1)
		1325	{
		1326	sv = newSVsv (POPs);
		1327	FREETMPS; LEAVE;
		1328	return sv;
		1329	}
		1330
		1331	SvREFCNT_inc (sv);
		1332	FREETMPS; LEAVE;
		1333	}
		1334	}
		1335
		1336	return sv;
917		1337
918	fail:	1338	fail:
919	SvREFCNT_dec (hv);	1339	SvREFCNT_dec (hv);
920	DEC_DEC_DEPTH;	1340	DEC_DEC_DEPTH;
921	return 0;	1341	return 0;
922	}	1342	}
923		1343
924	static SV *	1344	static SV *
925	decode_sv (dec_t *dec)	1345	decode_sv (dec_t *dec)
926	{	1346	{
927	decode_ws (dec);	1347	// the beauty of JSON: you need exactly one character lookahead
		1348	// to parse everything.
928	switch (*dec->cur)	1349	switch (*dec->cur)
929	{	1350	{
930	case '"': ++dec->cur; return decode_str (dec);	1351	case '"': ++dec->cur; return decode_str (dec);
931	case '[': ++dec->cur; return decode_av (dec);	1352	case '[': ++dec->cur; return decode_av (dec);
932	case '{': ++dec->cur; return decode_hv (dec);	1353	case '{': ++dec->cur; return decode_hv (dec);
933		1354
934	case '-':	1355	case '-':
935	case '0': case '1': case '2': case '3': case '4':	1356	case '0': case '1': case '2': case '3': case '4':
936	case '5': case '6': case '7': case '8': case '9':	1357	case '5': case '6': case '7': case '8': case '9':
937	return decode_num (dec);	1358	return decode_num (dec);
938		1359
939	case 't':	1360	case 't':
940	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1361	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
941	{	1362	{
942	dec->cur += 4;	1363	dec->cur += 4;
		1364	#if JSON_SLOW
		1365	json_true = get_bool ("JSON::XS::true");
		1366	#endif
943	return newSViv (1);	1367	return newSVsv (json_true);
944	}	1368	}
945	else	1369	else
946	ERR ("'true' expected");	1370	ERR ("'true' expected");
947		1371
948	break;	1372	break;
949		1373
950	case 'f':	1374	case 'f':
951	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1375	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
952	{	1376	{
953	dec->cur += 5;	1377	dec->cur += 5;
		1378	#if JSON_SLOW
		1379	json_false = get_bool ("JSON::XS::false");
		1380	#endif
954	return newSViv (0);	1381	return newSVsv (json_false);
955	}	1382	}
956	else	1383	else
957	ERR ("'false' expected");	1384	ERR ("'false' expected");
958		1385
959	break;	1386	break;
…		…
977	fail:	1404	fail:
978	return 0;	1405	return 0;
979	}	1406	}
980		1407
981	static SV *	1408	static SV *
982	decode_json (SV *string, U32 flags)	1409	decode_json (SV *string, JSON *json, STRLEN *offset_return)
983	{	1410	{
984	dec_t dec;	1411	dec_t dec;
		1412	STRLEN offset;
985	SV *sv;	1413	SV *sv;
986		1414
		1415	SvGETMAGIC (string);
987	SvUPGRADE (string, SVt_PV);	1416	SvUPGRADE (string, SVt_PV);
988		1417
		1418	/* work around a bug in perl 5.10, which causes SvCUR to fail an
		1419	* assertion with -DDEBUGGING, although SvCUR is documented to
		1420	* return the xpv_cur field which certainly exists after upgrading.
		1421	* according to nicholas clark, calling SvPOK fixes this.
		1422	* But it doesn't fix it, so try another workaround, call SvPV_nolen
		1423	* and hope for the best.
		1424	* Damnit, SvPV_nolen still trips over yet another assertion. This
		1425	* assertion business is seriously broken, try yet another workaround
		1426	* for the broken -DDEBUGGING.
		1427	*/
		1428	#ifdef DEBUGGING
		1429	offset = SvOK (string) ? sv_len (string) : 0;
		1430	#else
		1431	offset = SvCUR (string);
		1432	#endif
		1433
		1434	if (offset > json->max_size && json->max_size)
		1435	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1436	(unsigned long)SvCUR (string), (unsigned long)json->max_size);
		1437
989	if (flags & F_UTF8)	1438	if (json->flags & F_UTF8)
990	sv_utf8_downgrade (string, 0);	1439	sv_utf8_downgrade (string, 0);
991	else	1440	else
992	sv_utf8_upgrade (string);	1441	sv_utf8_upgrade (string);
993		1442
994	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1443	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
995		1444
996	dec.flags = flags;	1445	dec.json = *json;
997	dec.cur = SvPVX (string);	1446	dec.cur = SvPVX (string);
998	dec.end = SvEND (string);	1447	dec.end = SvEND (string);
999	dec.err = 0;	1448	dec.err = 0;
1000	dec.depth = 0;	1449	dec.depth = 0;
1001	dec.maxdepth = DEC_DEPTH (dec.flags);
1002		1450
		1451	if (dec.json.cb_object || dec.json.cb_sk_object)
		1452	dec.json.flags |= F_HOOK;
		1453
1003	*dec.end = 0; // this should basically be a nop, too, but make sure its there	1454	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
		1455
		1456	decode_ws (&dec);
1004	sv = decode_sv (&dec);	1457	sv = decode_sv (&dec);
1005		1458
		1459	if (!(offset_return || !sv))
		1460	{
		1461	// check for trailing garbage
		1462	decode_ws (&dec);
		1463
		1464	if (*dec.cur)
		1465	{
		1466	dec.err = "garbage after JSON object";
		1467	SvREFCNT_dec (sv);
		1468	sv = 0;
		1469	}
		1470	}
		1471
		1472	if (offset_return || !sv)
		1473	{
		1474	offset = dec.json.flags & F_UTF8
		1475	? dec.cur - SvPVX (string)
		1476	: utf8_distance (dec.cur, SvPVX (string));
		1477
		1478	if (offset_return)
		1479	*offset_return = offset;
		1480	}
		1481
1006	if (!sv)	1482	if (!sv)
1007	{	1483	{
1008	IV offset = dec.flags & F_UTF8
1009	? dec.cur - SvPVX (string)
1010	: utf8_distance (dec.cur, SvPVX (string));
1011	SV *uni = sv_newmortal ();	1484	SV *uni = sv_newmortal ();
1012		1485
1013	// horrible hack to silence warning inside pv_uni_display	1486	// horrible hack to silence warning inside pv_uni_display
1014	COP cop = *PL_curcop;	1487	COP cop = *PL_curcop;
1015	cop.cop_warnings = pWARN_NONE;	1488	cop.cop_warnings = pWARN_NONE;
…		…
1025	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1498	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1026	}	1499	}
1027		1500
1028	sv = sv_2mortal (sv);	1501	sv = sv_2mortal (sv);
1029		1502
1030	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))	1503	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1031	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");	1504	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1032		1505
1033	return sv;	1506	return sv;
		1507	}
		1508
		1509	/////////////////////////////////////////////////////////////////////////////
		1510	// incremental parser
		1511
		1512	static void
		1513	incr_parse (JSON *self)
		1514	{
		1515	const char *p = SvPVX (self->incr_text) + self->incr_pos;
		1516
		1517	for (;;)
		1518	{
		1519	//printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
		1520	switch (self->incr_mode)
		1521	{
		1522	// only used for intiial whitespace skipping
		1523	case INCR_M_WS:
		1524	for (;;)
		1525	{
		1526	if (*p > 0x20)
		1527	{
		1528	self->incr_mode = INCR_M_JSON;
		1529	goto incr_m_json;
		1530	}
		1531	else if (!*p)
		1532	goto interrupt;
		1533
		1534	++p;
		1535	}
		1536
		1537	// skip a single char inside a string (for \\-processing)
		1538	case INCR_M_BS:
		1539	if (!*p)
		1540	goto interrupt;
		1541
		1542	++p;
		1543	self->incr_mode = INCR_M_STR;
		1544	goto incr_m_str;
		1545
		1546	// inside a string
		1547	case INCR_M_STR:
		1548	incr_m_str:
		1549	for (;;)
		1550	{
		1551	if (*p == '"')
		1552	{
		1553	++p;
		1554	self->incr_mode = INCR_M_JSON;
		1555
		1556	if (!self->incr_nest)
		1557	goto interrupt;
		1558
		1559	goto incr_m_json;
		1560	}
		1561	else if (*p == '\\')
		1562	{
		1563	++p; // "virtually" consumes character after \
		1564
		1565	if (!*p) // if at end of string we have to switch modes
		1566	{
		1567	self->incr_mode = INCR_M_BS;
		1568	goto interrupt;
		1569	}
		1570	}
		1571	else if (!*p)
		1572	goto interrupt;
		1573
		1574	++p;
		1575	}
		1576
		1577	// after initial ws, outside string
		1578	case INCR_M_JSON:
		1579	incr_m_json:
		1580	for (;;)
		1581	{
		1582	switch (*p++)
		1583	{
		1584	case 0:
		1585	--p;
		1586	goto interrupt;
		1587
		1588	case 0x09:
		1589	case 0x0a:
		1590	case 0x0d:
		1591	case 0x20:
		1592	if (!self->incr_nest)
		1593	{
		1594	--p; // do not eat the whitespace, let the next round do it
		1595	goto interrupt;
		1596	}
		1597	break;
		1598
		1599	case '"':
		1600	self->incr_mode = INCR_M_STR;
		1601	goto incr_m_str;
		1602
		1603	case '[':
		1604	case '{':
		1605	if (++self->incr_nest > self->max_depth)
		1606	croak (ERR_NESTING_EXCEEDED);
		1607	break;
		1608
		1609	case ']':
		1610	case '}':
		1611	if (--self->incr_nest <= 0)
		1612	goto interrupt;
		1613	}
		1614	}
		1615	}
		1616
		1617	modechange:
		1618	;
		1619	}
		1620
		1621	interrupt:
		1622	self->incr_pos = p - SvPVX (self->incr_text);
		1623	//printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1034	}	1624	}
1035		1625
1036	/////////////////////////////////////////////////////////////////////////////	1626	/////////////////////////////////////////////////////////////////////////////
1037	// XS interface functions	1627	// XS interface functions
1038		1628
1039	MODULE = JSON::XS PACKAGE = JSON::XS	1629	MODULE = JSON::XS PACKAGE = JSON::XS
1040		1630
1041	BOOT:	1631	BOOT:
1042	{	1632	{
1043	int i;	1633	int i;
1044
1045	memset (decode_hexdigit, 0xff, 256);
1046		1634
1047	for (i = 0; i < 256; ++i)	1635	for (i = 0; i < 256; ++i)
1048	decode_hexdigit [i] =	1636	decode_hexdigit [i] =
1049	i >= '0' && i <= '9' ? i - '0'	1637	i >= '0' && i <= '9' ? i - '0'
1050	: i >= 'a' && i <= 'f' ? i - 'a' + 10	1638	: i >= 'a' && i <= 'f' ? i - 'a' + 10
1051	: i >= 'A' && i <= 'F' ? i - 'A' + 10	1639	: i >= 'A' && i <= 'F' ? i - 'A' + 10
1052	: -1;	1640	: -1;
1053		1641
1054	json_stash = gv_stashpv ("JSON::XS", 1);	1642	json_stash = gv_stashpv ("JSON::XS" , 1);
		1643	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
		1644
		1645	json_true = get_bool ("JSON::XS::true");
		1646	json_false = get_bool ("JSON::XS::false");
1055	}	1647	}
1056		1648
1057	PROTOTYPES: DISABLE	1649	PROTOTYPES: DISABLE
1058		1650
1059	SV *new (char *dummy)	1651	void CLONE (...)
1060	CODE:	1652	CODE:
1061	RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);	1653	json_stash = 0;
		1654	json_boolean_stash = 0;
		1655
		1656	void new (char *klass)
		1657	PPCODE:
		1658	{
		1659	SV *pv = NEWSV (0, sizeof (JSON));
		1660	SvPOK_only (pv);
		1661	json_init ((JSON *)SvPVX (pv));
		1662	XPUSHs (sv_2mortal (sv_bless (
		1663	newRV_noinc (pv),
		1664	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
		1665	)));
		1666	}
		1667
		1668	void ascii (JSON *self, int enable = 1)
		1669	ALIAS:
		1670	ascii = F_ASCII
		1671	latin1 = F_LATIN1
		1672	utf8 = F_UTF8
		1673	indent = F_INDENT
		1674	canonical = F_CANONICAL
		1675	space_before = F_SPACE_BEFORE
		1676	space_after = F_SPACE_AFTER
		1677	pretty = F_PRETTY
		1678	allow_nonref = F_ALLOW_NONREF
		1679	shrink = F_SHRINK
		1680	allow_blessed = F_ALLOW_BLESSED
		1681	convert_blessed = F_CONV_BLESSED
		1682	relaxed = F_RELAXED
		1683	allow_unknown = F_ALLOW_UNKNOWN
		1684	PPCODE:
		1685	{
		1686	if (enable)
		1687	self->flags |= ix;
		1688	else
		1689	self->flags &= ~ix;
		1690
		1691	XPUSHs (ST (0));
		1692	}
		1693
		1694	void get_ascii (JSON *self)
		1695	ALIAS:
		1696	get_ascii = F_ASCII
		1697	get_latin1 = F_LATIN1
		1698	get_utf8 = F_UTF8
		1699	get_indent = F_INDENT
		1700	get_canonical = F_CANONICAL
		1701	get_space_before = F_SPACE_BEFORE
		1702	get_space_after = F_SPACE_AFTER
		1703	get_allow_nonref = F_ALLOW_NONREF
		1704	get_shrink = F_SHRINK
		1705	get_allow_blessed = F_ALLOW_BLESSED
		1706	get_convert_blessed = F_CONV_BLESSED
		1707	get_relaxed = F_RELAXED
		1708	get_allow_unknown = F_ALLOW_UNKNOWN
		1709	PPCODE:
		1710	XPUSHs (boolSV (self->flags & ix));
		1711
		1712	void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
		1713	PPCODE:
		1714	self->max_depth = max_depth;
		1715	XPUSHs (ST (0));
		1716
		1717	U32 get_max_depth (JSON *self)
		1718	CODE:
		1719	RETVAL = self->max_depth;
1062	OUTPUT:	1720	OUTPUT:
1063	RETVAL	1721	RETVAL
1064		1722
1065	SV *ascii (SV *self, int enable = 1)	1723	void max_size (JSON *self, U32 max_size = 0)
1066	ALIAS:	1724	PPCODE:
1067	ascii = F_ASCII	1725	self->max_size = max_size;
1068	utf8 = F_UTF8	1726	XPUSHs (ST (0));
1069	indent = F_INDENT	1727
1070	canonical = F_CANONICAL	1728	int get_max_size (JSON *self)
1071	space_before = F_SPACE_BEFORE
1072	space_after = F_SPACE_AFTER
1073	pretty = F_PRETTY
1074	allow_nonref = F_ALLOW_NONREF
1075	shrink = F_SHRINK
1076	CODE:	1729	CODE:
1077	{	1730	RETVAL = self->max_size;
1078	UV *uv = SvJSON (self);
1079	if (enable)
1080	*uv |= ix;
1081	else
1082	*uv &= ~ix;
1083
1084	RETVAL = newSVsv (self);
1085	}
1086	OUTPUT:	1731	OUTPUT:
1087	RETVAL	1732	RETVAL
1088		1733
1089	SV *max_depth (SV *self, UV max_depth = 0x80000000UL)	1734	void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
		1735	PPCODE:
		1736	{
		1737	SvREFCNT_dec (self->cb_object);
		1738	self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
		1739
		1740	XPUSHs (ST (0));
		1741	}
		1742
		1743	void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
		1744	PPCODE:
		1745	{
		1746	if (!self->cb_sk_object)
		1747	self->cb_sk_object = newHV ();
		1748
		1749	if (SvOK (cb))
		1750	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
		1751	else
		1752	{
		1753	hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
		1754
		1755	if (!HvKEYS (self->cb_sk_object))
		1756	{
		1757	SvREFCNT_dec (self->cb_sk_object);
		1758	self->cb_sk_object = 0;
		1759	}
		1760	}
		1761
		1762	XPUSHs (ST (0));
		1763	}
		1764
		1765	void encode (JSON *self, SV *scalar)
		1766	PPCODE:
		1767	XPUSHs (encode_json (scalar, self));
		1768
		1769	void decode (JSON *self, SV *jsonstr)
		1770	PPCODE:
		1771	XPUSHs (decode_json (jsonstr, self, 0));
		1772
		1773	void decode_prefix (JSON *self, SV *jsonstr)
		1774	PPCODE:
		1775	{
		1776	STRLEN offset;
		1777	EXTEND (SP, 2);
		1778	PUSHs (decode_json (jsonstr, self, &offset));
		1779	PUSHs (sv_2mortal (newSVuv (offset)));
		1780	}
		1781
		1782	void incr_parse (JSON *self, SV *jsonstr = 0)
		1783	PPCODE:
		1784	{
		1785	if (!self->incr_text)
		1786	self->incr_text = newSVpvn ("", 0);
		1787
		1788	// append data, if any
		1789	if (jsonstr)
		1790	{
		1791	if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
		1792	{
		1793	/* utf-8-ness differs, need to upgrade */
		1794	sv_utf8_upgrade (self->incr_text);
		1795
		1796	if (self->incr_pos)
		1797	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
		1798	- (U8 *)SvPVX (self->incr_text);
		1799	}
		1800
		1801	{
		1802	STRLEN len;
		1803	const char *str = SvPV (jsonstr, len);
		1804	SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
		1805	Move (str, SvEND (self->incr_text), len, char);
		1806	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
		1807	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
		1808	}
		1809	}
		1810
		1811	if (GIMME_V != G_VOID)
		1812	do
		1813	{
		1814	STRLEN offset;
		1815
		1816	if (!INCR_DONE (self))
		1817	{
		1818	incr_parse (self);
		1819
		1820	if (self->incr_pos > self->max_size && self->max_size)
		1821	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1822	(unsigned long)self->incr_pos, (unsigned long)self->max_size);
		1823
		1824	if (!INCR_DONE (self))
		1825	break;
		1826	}
		1827
		1828	XPUSHs (decode_json (self->incr_text, self, &offset));
		1829
		1830	sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
		1831	self->incr_pos -= offset;
		1832	self->incr_nest = 0;
		1833	self->incr_mode = 0;
		1834	}
		1835	while (GIMME_V == G_ARRAY);
		1836	}
		1837
		1838	SV *incr_text (JSON *self)
		1839	ATTRS: lvalue
1090	CODE:	1840	CODE:
1091	{	1841	{
1092	UV *uv = SvJSON (self);	1842	if (self->incr_pos)
1093	UV log2 = 0;	1843	croak ("incr_text can not be called when the incremental parser already started parsing");
1094		1844
1095	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;	1845	RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1096
1097	while ((1UL << log2) < max_depth)
1098	++log2;
1099
1100	*uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1101
1102	RETVAL = newSVsv (self);
1103	}	1846	}
1104	OUTPUT:	1847	OUTPUT:
1105	RETVAL	1848	RETVAL
1106		1849
		1850	void incr_skip (JSON *self)
		1851	CODE:
		1852	{
		1853	if (self->incr_pos)
		1854	{
		1855	sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
		1856	self->incr_pos = 0;
		1857	self->incr_nest = 0;
		1858	self->incr_mode = 0;
		1859	}
		1860	}
		1861
		1862	void incr_reset (JSON *self)
		1863	CODE:
		1864	{
		1865	SvREFCNT_dec (self->incr_text);
		1866	self->incr_text = 0;
		1867	self->incr_pos = 0;
		1868	self->incr_nest = 0;
		1869	self->incr_mode = 0;
		1870	}
		1871
		1872	void DESTROY (JSON *self)
		1873	CODE:
		1874	SvREFCNT_dec (self->cb_sk_object);
		1875	SvREFCNT_dec (self->cb_object);
		1876	SvREFCNT_dec (self->incr_text);
		1877
		1878	PROTOTYPES: ENABLE
		1879
1107	void encode (SV *self, SV *scalar)	1880	void encode_json (SV *scalar)
		1881	ALIAS:
		1882	to_json_ = 0
		1883	encode_json = F_UTF8
1108	PPCODE:	1884	PPCODE:
		1885	{
		1886	JSON json;
		1887	json_init (&json);
		1888	json.flags |= ix;
1109	XPUSHs (encode_json (scalar, *SvJSON (self)));	1889	XPUSHs (encode_json (scalar, &json));
		1890	}
1110		1891
1111	void decode (SV *self, SV *jsonstr)	1892	void decode_json (SV *jsonstr)
		1893	ALIAS:
		1894	from_json_ = 0
		1895	decode_json = F_UTF8
1112	PPCODE:	1896	PPCODE:
		1897	{
		1898	JSON json;
		1899	json_init (&json);
		1900	json.flags |= ix;
1113	XPUSHs (decode_json (jsonstr, *SvJSON (self)));	1901	XPUSHs (decode_json (jsonstr, &json, 0));
		1902	}
1114		1903
1115	PROTOTYPES: ENABLE
1116		1904
1117	void to_json (SV *scalar)
1118	ALIAS:
1119	objToJson = 0
1120	PPCODE:
1121	XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1122
1123	void from_json (SV *jsonstr)
1124	ALIAS:
1125	jsonToObj = 0
1126	PPCODE:
1127	XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
1128

Diff Legend

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