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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.5 by root, Thu Mar 22 23:24:18 2007 UTC vs.
Revision 1.100 by root, Sat Sep 5 23:00:56 2009 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>
		9	#include <limits.h>
		10	#include <float.h>
8		11
		12	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		13	# define snprintf _snprintf // C compilers have this in stdio.h
		14	#endif
		15
		16	// some old perls do not have this, try to make it work, no
		17	// guarantees, 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
9	#define F_ASCII 0x00000001	24	#define F_ASCII 0x00000001UL
		25	#define F_LATIN1 0x00000002UL
10	#define F_UTF8 0x00000002	26	#define F_UTF8 0x00000004UL
11	#define F_INDENT 0x00000004	27	#define F_INDENT 0x00000008UL
12	#define F_CANONICAL 0x00000008	28	#define F_CANONICAL 0x00000010UL
13	#define F_SPACE_BEFORE 0x00000010	29	#define F_SPACE_BEFORE 0x00000020UL
14	#define F_SPACE_AFTER 0x00000020	30	#define F_SPACE_AFTER 0x00000040UL
15	#define F_JSON_RPC 0x00000040
16	#define F_ALLOW_NONREF 0x00000080	31	#define F_ALLOW_NONREF 0x00000100UL
		32	#define F_SHRINK 0x00000200UL
		33	#define F_ALLOW_BLESSED 0x00000400UL
		34	#define F_CONV_BLESSED 0x00000800UL
		35	#define F_RELAXED 0x00001000UL
		36	#define F_ALLOW_UNKNOWN 0x00002000UL
		37	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
17		38
18	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	39	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
19	#define F_DEFAULT 0
20		40
21	#define INIT_SIZE 32 // initial scalar size to be allocated	41	#define INIT_SIZE 32 // initial scalar size to be allocated
		42	#define INDENT_STEP 3 // spaces per indentation level
		43
		44	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
22		45
23	#define SB do {	46	#define SB do {
24	#define SE } while (0)	47	#define SE } while (0)
25		48
26	static HV *json_stash;	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
		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	}
		107
		108	/////////////////////////////////////////////////////////////////////////////
		109	// utility functions
		110
		111	INLINE SV *
		112	get_bool (const char *name)
		113	{
		114	SV *sv = get_sv (name, 1);
		115
		116	SvREADONLY_on (sv);
		117	SvREADONLY_on (SvRV (sv));
		118
		119	return sv;
		120	}
		121
		122	INLINE void
		123	shrink (SV *sv)
		124	{
		125	sv_utf8_downgrade (sv, 1);
		126
		127	if (SvLEN (sv) > SvCUR (sv) + 1)
		128	{
		129	#ifdef SvPV_shrink_to_cur
		130	SvPV_shrink_to_cur (sv);
		131	#elif defined (SvPV_renew)
		132	SvPV_renew (sv, SvCUR (sv) + 1);
		133	#endif
		134	}
		135	}
		136
		137	// decode an utf-8 character and return it, or (UV)-1 in
		138	// case of an error.
		139	// we special-case "safe" characters from U+80 .. U+7FF,
		140	// but use the very good perl function to parse anything else.
		141	// note that we never call this function for a ascii codepoints
		142	INLINE UV
		143	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
		144	{
		145	if (expect_true (len >= 2
		146	&& IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
		147	&& IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
		148	{
		149	*clen = 2;
		150	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
		151	}
		152	else
		153	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
		154	}
		155
		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;
		178	}
		179
		180	// convert offset pointer to character index, sv must be string
		181	static STRLEN
		182	ptr_to_index (SV sv, char offset)
		183	{
		184	return SvUTF8 (sv)
		185	? utf8_distance (offset, SvPVX (sv))
		186	: offset - SvPVX (sv);
		187	}
		188
		189	/////////////////////////////////////////////////////////////////////////////
		190	// encoder
27		191
28	// structure used for encoding JSON	192	// structure used for encoding JSON
29	typedef struct	193	typedef struct
30	{	194	{
31	char *cur;	195	char *cur; // SvPVX (sv) + current output position
32	STRLEN len; // SvLEN (sv)
33	char *end; // SvEND (sv)	196	char *end; // SvEND (sv)
34	SV *sv;	197	SV *sv; // result scalar
35	UV flags;	198	JSON json;
36	int max_recurse;	199	U32 indent; // indentation level
37	int indent;	200	UV limit; // escape character values >= this value when encoding
38	} enc_t;	201	} enc_t;
39		202
40	// structure used for decoding JSON	203	INLINE void
41	typedef struct
42	{
43	char *cur;
44	char *end;
45	const char *err;
46	UV flags;
47	} dec_t;
48
49	static UV *
50	SvJSON (SV *sv)
51	{
52	if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
53	croak ("object is not of type JSON::XS");
54
55	return &SvUVX (SvRV (sv));
56	}
57
58	/////////////////////////////////////////////////////////////////////////////
59
60	static void
61	need (enc_t *enc, STRLEN len)	204	need (enc_t *enc, STRLEN len)
62	{	205	{
63	if (enc->cur + len >= enc->end)	206	if (expect_false (enc->cur + len >= enc->end))
64	{	207	{
65	STRLEN cur = enc->cur - SvPVX (enc->sv);	208	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
66	SvGROW (enc->sv, cur + len + 1);	209	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
67	enc->cur = SvPVX (enc->sv) + cur;	210	enc->cur = SvPVX (enc->sv) + cur;
68	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);	211	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
69	}	212	}
70	}	213	}
71		214
72	static void	215	INLINE void
73	encode_ch (enc_t *enc, char ch)	216	encode_ch (enc_t *enc, char ch)
74	{	217	{
75	need (enc, 1);	218	need (enc, 1);
76	*enc->cur++ = ch;	219	*enc->cur++ = ch;
77	}	220	}
85		228
86	while (str < end)	229	while (str < end)
87	{	230	{
88	unsigned char ch = (unsigned char )str;	231	unsigned char ch = (unsigned char )str;
89		232
90	if (ch == '"')	233	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
91	{	234	{
		235	if (expect_false (ch == '"')) // but with slow exceptions
		236	{
92	need (enc, len += 1);	237	need (enc, len += 1);
93	*enc->cur++ = '\\';	238	*enc->cur++ = '\\';
94	*enc->cur++ = '"';	239	*enc->cur++ = '"';
95	++str;
96	}	240	}
97	else if (ch == '\\')	241	else if (expect_false (ch == '\\'))
98	{	242	{
99	need (enc, len += 1);	243	need (enc, len += 1);
100	*enc->cur++ = '\\';	244	*enc->cur++ = '\\';
101	*enc->cur++ = '\\';	245	*enc->cur++ = '\\';
102	++str;
103	}	246	}
104	else if (ch >= 0x20 && ch < 0x80) // most common case	247	else
105	{
106	*enc->cur++ = ch;	248	*enc->cur++ = ch;
107	++str;	249
108	}
109	else if (ch == '\015')
110	{
111	need (enc, len += 1);
112	*enc->cur++ = '\\';
113	*enc->cur++ = 'r';
114	++str;
115	}
116	else if (ch == '\012')
117	{
118	need (enc, len += 1);
119	*enc->cur++ = '\\';
120	*enc->cur++ = 'n';
121	++str;	250	++str;
122	}	251	}
123	else	252	else
124	{	253	{
125	STRLEN clen;	254	switch (ch)
126	UV uch;
127
128	if (is_utf8)
129	{	255	{
130	uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);	256	case '\010': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 'b'; ++str; break;
131	if (clen == (STRLEN)-1)	257	case '\011': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 't'; ++str; break;
132	croak ("malformed UTF-8 character in string, cannot convert to JSON");	258	case '\012': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 'n'; ++str; break;
133	}	259	case '\014': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 'f'; ++str; break;
134	else	260	case '\015': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 'r'; ++str; break;
135	{
136	uch = ch;
137	clen = 1;
138	}
139		261
140	if (uch < 0x80 \|\| enc->flags & F_ASCII)	262	default:
141	{
142	if (uch > 0xFFFFUL)
143	{	263	{
		264	STRLEN clen;
		265	UV uch;
		266
		267	if (is_utf8)
		268	{
		269	uch = decode_utf8 (str, end - str, &clen);
		270	if (clen == (STRLEN)-1)
		271	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
		272	}
		273	else
		274	{
		275	uch = ch;
		276	clen = 1;
		277	}
		278
		279	if (uch < 0x80/0x20/ \|\| uch >= enc->limit)
		280	{
		281	if (uch >= 0x10000UL)
		282	{
		283	if (uch >= 0x110000UL)
		284	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
		285
144	need (enc, len += 11);	286	need (enc, len += 11);
145	sprintf (enc->cur, "\\u%04x\\u%04x",	287	sprintf (enc->cur, "\\u%04x\\u%04x",
146	(uch - 0x10000) / 0x400 + 0xD800,	288	(int)((uch - 0x10000) / 0x400 + 0xD800),
147	(uch - 0x10000) % 0x400 + 0xDC00);	289	(int)((uch - 0x10000) % 0x400 + 0xDC00));
148	enc->cur += 12;	290	enc->cur += 12;
		291	}
		292	else
		293	{
		294	need (enc, len += 5);
		295	*enc->cur++ = '\\';
		296	*enc->cur++ = 'u';
		297	*enc->cur++ = PL_hexdigit [ uch >> 12 ];
		298	*enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
		299	*enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
		300	*enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
		301	}
		302
		303	str += clen;
		304	}
		305	else if (enc->json.flags & F_LATIN1)
		306	{
		307	*enc->cur++ = uch;
		308	str += clen;
		309	}
		310	else if (is_utf8)
		311	{
		312	need (enc, len += clen);
		313	do
		314	{
		315	enc->cur++ = str++;
		316	}
		317	while (--clen);
		318	}
		319	else
		320	{
		321	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
		322	enc->cur = encode_utf8 (enc->cur, uch);
		323	++str;
		324	}
149	}	325	}
150	else
151	{
152	static char hexdigit [16] = "0123456789abcdef";
153	need (enc, len += 5);
154	*enc->cur++ = '\\';
155	*enc->cur++ = 'u';
156	*enc->cur++ = hexdigit [ uch >> 12 ];
157	*enc->cur++ = hexdigit [(uch >> 8) & 15];
158	*enc->cur++ = hexdigit [(uch >> 4) & 15];
159	*enc->cur++ = hexdigit [(uch >> 0) & 15];
160	}
161
162	str += clen;
163	}
164	else if (is_utf8)
165	{
166	need (enc, len += clen);
167	do
168	{
169	enc->cur++ = str++;
170	}
171	while (--clen);
172	}
173	else
174	{
175	need (enc, 10); // never more than 11 bytes needed
176	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
177	++str;
178	}	326	}
179	}	327	}
180		328
181	--len;	329	--len;
182	}	330	}
183	}	331	}
184		332
185	#define INDENT SB \	333	INLINE void
		334	encode_indent (enc_t *enc)
		335	{
186	if (enc->flags & F_INDENT) \	336	if (enc->json.flags & F_INDENT)
187	{ \	337	{
188	int i_; \	338	int spaces = enc->indent * INDENT_STEP;
189	need (enc, enc->indent); \	339
190	for (i_ = enc->indent * 3; i_--; )\	340	need (enc, spaces);
		341	memset (enc->cur, ' ', spaces);
		342	enc->cur += spaces;
		343	}
		344	}
		345
		346	INLINE void
		347	encode_space (enc_t *enc)
		348	{
		349	need (enc, 1);
		350	encode_ch (enc, ' ');
		351	}
		352
		353	INLINE void
		354	encode_nl (enc_t *enc)
		355	{
		356	if (enc->json.flags & F_INDENT)
		357	{
		358	need (enc, 1);
191	encode_ch (enc, ' '); \	359	encode_ch (enc, '\n');
192	} \	360	}
193	SE	361	}
194		362
195	#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE	363	INLINE void
196	#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE	364	encode_comma (enc_t *enc)
197	#define COMMA SB \	365	{
198	encode_ch (enc, ','); \	366	encode_ch (enc, ',');
		367
199	if (enc->flags & F_INDENT) \	368	if (enc->json.flags & F_INDENT)
200	NL; \	369	encode_nl (enc);
201	else if (enc->flags & F_SPACE_AFTER) \	370	else if (enc->json.flags & F_SPACE_AFTER)
202	SPACE; \	371	encode_space (enc);
203	SE	372	}
204		373
205	static void encode_sv (enc_t enc, SV sv);	374	static void encode_sv (enc_t enc, SV sv);
206		375
207	static void	376	static void
208	encode_av (enc_t enc, AV av)	377	encode_av (enc_t enc, AV av)
209	{	378	{
210	int i, len = av_len (av);	379	int i, len = av_len (av);
211		380
		381	if (enc->indent >= enc->json.max_depth)
		382	croak (ERR_NESTING_EXCEEDED);
		383
212	encode_ch (enc, '['); NL;	384	encode_ch (enc, '[');
213	++enc->indent;	385
		386	if (len >= 0)
		387	{
		388	encode_nl (enc); ++enc->indent;
214		389
215	for (i = 0; i <= len; ++i)	390	for (i = 0; i <= len; ++i)
216	{	391	{
217	INDENT;	392	SV **svp = av_fetch (av, i, 0);
218	encode_sv (enc, *av_fetch (av, i, 0));
219		393
		394	encode_indent (enc);
		395
		396	if (svp)
		397	encode_sv (enc, *svp);
		398	else
		399	encode_str (enc, "null", 4, 0);
		400
220	if (i < len)	401	if (i < len)
221	COMMA;	402	encode_comma (enc);
222	}	403	}
223		404
224	NL;	405	encode_nl (enc); --enc->indent; encode_indent (enc);
225		406	}
226	--enc->indent;	407
227	INDENT; encode_ch (enc, ']');	408	encode_ch (enc, ']');
228	}	409	}
229		410
230	static void	411	static void
231	encode_he (enc_t enc, HE he)	412	encode_hk (enc_t enc, HE he)
232	{	413	{
233	encode_ch (enc, '"');	414	encode_ch (enc, '"');
234		415
235	if (HeKLEN (he) == HEf_SVKEY)	416	if (HeKLEN (he) == HEf_SVKEY)
236	{	417	{
…		…
246	else	427	else
247	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));	428	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
248		429
249	encode_ch (enc, '"');	430	encode_ch (enc, '"');
250		431
251	if (enc->flags & F_SPACE_BEFORE) SPACE;	432	if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
252	encode_ch (enc, ':');	433	encode_ch (enc, ':');
253	if (enc->flags & F_SPACE_AFTER ) SPACE;	434	if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
254	encode_sv (enc, HeVAL (he));
255	}	435	}
256		436
257	// compare hash entries, used when all keys are bytestrings	437	// compare hash entries, used when all keys are bytestrings
258	static int	438	static int
259	he_cmp_fast (const void a_, const void b_)	439	he_cmp_fast (const void a_, const void b_)
…		…
264	HE b = (HE **)b_;	444	HE b = (HE **)b_;
265		445
266	STRLEN la = HeKLEN (a);	446	STRLEN la = HeKLEN (a);
267	STRLEN lb = HeKLEN (b);	447	STRLEN lb = HeKLEN (b);
268		448
269	if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))	449	if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
270	cmp = la < lb ? -1 : la == lb ? 0 : 1;	450	cmp = lb - la;
271		451
272	return cmp;	452	return cmp;
273	}	453	}
274		454
275	// compare hash entries, used when some keys are sv's or utf-x	455	// compare hash entries, used when some keys are sv's or utf-x
276	static int	456	static int
277	he_cmp_slow (const void a, const void b)	457	he_cmp_slow (const void a, const void b)
278	{	458	{
279	return sv_cmp (HeSVKEY_force ((HE )a), HeSVKEY_force ((HE **)b));	459	return sv_cmp (HeSVKEY_force ((HE )b), HeSVKEY_force ((HE **)a));
280	}	460	}
281		461
282	static void	462	static void
283	encode_hv (enc_t enc, HV hv)	463	encode_hv (enc_t enc, HV hv)
284	{	464	{
285	int count, i;	465	HE *he;
286		466
287	encode_ch (enc, '{'); NL; ++enc->indent;	467	if (enc->indent >= enc->json.max_depth)
		468	croak (ERR_NESTING_EXCEEDED);
288		469
289	if ((count = hv_iterinit (hv)))	470	encode_ch (enc, '{');
290	{	471
291	// for canonical output we have to sort by keys first	472	// for canonical output we have to sort by keys first
292	// actually, this is mostly due to the stupid so-called	473	// actually, this is mostly due to the stupid so-called
293	// security workaround added somewhere in 5.8.x.	474	// security workaround added somewhere in 5.8.x
294	// that randomises hash orderings	475	// that randomises hash orderings
295	if (enc->flags & F_CANONICAL)	476	if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
		477	{
		478	int count = hv_iterinit (hv);
		479
		480	if (SvMAGICAL (hv))
296	{	481	{
297	HE he, hes [count];	482	// need to count by iterating. could improve by dynamically building the vector below
		483	// but I don't care for the speed of this special case.
		484	// note also that we will run into undefined behaviour when the two iterations
		485	// do not result in the same count, something I might care for in some later release.
		486
		487	count = 0;
		488	while (hv_iternext (hv))
		489	++count;
		490
		491	hv_iterinit (hv);
		492	}
		493
		494	if (count)
		495	{
298	int fast = 1;	496	int i, fast = 1;
		497	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		498	HE *hes = _alloca (count sizeof (HE));
		499	#else
		500	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
		501	#endif
299		502
300	i = 0;	503	i = 0;
301	while ((he = hv_iternext (hv)))	504	while ((he = hv_iternext (hv)))
302	{	505	{
303	hes [i++] = he;	506	hes [i++] = he;
…		…
309		512
310	if (fast)	513	if (fast)
311	qsort (hes, count, sizeof (HE *), he_cmp_fast);	514	qsort (hes, count, sizeof (HE *), he_cmp_fast);
312	else	515	else
313	{	516	{
314	// hack to disable "use bytes"	517	// hack to forcefully disable "use bytes"
315	COP *oldcop = PL_curcop, cop;	518	COP cop = *PL_curcop;
316	cop.op_private = 0;	519	cop.op_private = 0;
		520
		521	ENTER;
		522	SAVETMPS;
		523
		524	SAVEVPTR (PL_curcop);
317	PL_curcop = &cop;	525	PL_curcop = &cop;
318		526
319	SAVETMPS;
320	qsort (hes, count, sizeof (HE *), he_cmp_slow);	527	qsort (hes, count, sizeof (HE *), he_cmp_slow);
		528
321	FREETMPS;	529	FREETMPS;
322		530	LEAVE;
323	PL_curcop = oldcop;
324	}	531	}
325		532
326	for (i = 0; i < count; ++i)	533	encode_nl (enc); ++enc->indent;
		534
		535	while (count--)
327	{	536	{
328	INDENT;	537	encode_indent (enc);
		538	he = hes [count];
329	encode_he (enc, hes [i]);	539	encode_hk (enc, he);
		540	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
330		541
331	if (i < count - 1)	542	if (count)
332	COMMA;	543	encode_comma (enc);
333	}	544	}
334		545
		546	encode_nl (enc); --enc->indent; encode_indent (enc);
		547	}
		548	}
		549	else
		550	{
		551	if (hv_iterinit (hv) \|\| SvMAGICAL (hv))
		552	if ((he = hv_iternext (hv)))
335	NL;	553	{
		554	encode_nl (enc); ++enc->indent;
		555
		556	for (;;)
		557	{
		558	encode_indent (enc);
		559	encode_hk (enc, he);
		560	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
		561
		562	if (!(he = hv_iternext (hv)))
		563	break;
		564
		565	encode_comma (enc);
		566	}
		567
		568	encode_nl (enc); --enc->indent; encode_indent (enc);
		569	}
		570	}
		571
		572	encode_ch (enc, '}');
		573	}
		574
		575	// encode objects, arrays and special \0=false and \1=true values.
		576	static void
		577	encode_rv (enc_t enc, SV sv)
		578	{
		579	svtype svt;
		580
		581	SvGETMAGIC (sv);
		582	svt = SvTYPE (sv);
		583
		584	if (expect_false (SvOBJECT (sv)))
		585	{
		586	HV *stash = !JSON_SLOW \|\| json_boolean_stash
		587	? json_boolean_stash
		588	: gv_stashpv ("JSON::XS::Boolean", 1);
		589
		590	if (SvSTASH (sv) == stash)
		591	{
		592	if (SvIV (sv))
		593	encode_str (enc, "true", 4, 0);
		594	else
		595	encode_str (enc, "false", 5, 0);
336	}	596	}
337	else	597	else
338	{	598	{
339	SV *sv;	599	#if 0
340	HE *he = hv_iternext (hv);	600	if (0 && sv_derived_from (rv, "JSON::Literal"))
341
342	for (;;)
343	{	601	{
344	INDENT;	602	// not yet
345	encode_he (enc, he);
346
347	if (!(he = hv_iternext (hv)))
348	break;
349
350	COMMA;
351	}	603	}
352		604	#endif
		605	if (enc->json.flags & F_CONV_BLESSED)
353	NL;	606	{
		607	// we re-bless the reference to get overload and other niceties right
		608	GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
		609
		610	if (to_json)
		611	{
		612	dSP;
		613
		614	ENTER; SAVETMPS; PUSHMARK (SP);
		615	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
		616
		617	// calling with G_SCALAR ensures that we always get a 1 return value
		618	PUTBACK;
		619	call_sv ((SV *)GvCV (to_json), G_SCALAR);
		620	SPAGAIN;
		621
		622	// catch this surprisingly common error
		623	if (SvROK (TOPs) && SvRV (TOPs) == sv)
		624	croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
		625
		626	sv = POPs;
		627	PUTBACK;
		628
		629	encode_sv (enc, sv);
		630
		631	FREETMPS; LEAVE;
		632	}
		633	else if (enc->json.flags & F_ALLOW_BLESSED)
		634	encode_str (enc, "null", 4, 0);
		635	else
		636	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
		637	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		638	}
		639	else if (enc->json.flags & F_ALLOW_BLESSED)
		640	encode_str (enc, "null", 4, 0);
		641	else
		642	croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
		643	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
354	}	644	}
355	}	645	}
		646	else if (svt == SVt_PVHV)
		647	encode_hv (enc, (HV *)sv);
		648	else if (svt == SVt_PVAV)
		649	encode_av (enc, (AV *)sv);
		650	else if (svt < SVt_PVAV)
		651	{
		652	STRLEN len = 0;
		653	char *pv = svt ? SvPV (sv, len) : 0;
356		654
357	--enc->indent; INDENT; encode_ch (enc, '}');	655	if (len == 1 && *pv == '1')
		656	encode_str (enc, "true", 4, 0);
		657	else if (len == 1 && *pv == '0')
		658	encode_str (enc, "false", 5, 0);
		659	else if (enc->json.flags & F_ALLOW_UNKNOWN)
		660	encode_str (enc, "null", 4, 0);
		661	else
		662	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
		663	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		664	}
		665	else if (enc->json.flags & F_ALLOW_UNKNOWN)
		666	encode_str (enc, "null", 4, 0);
		667	else
		668	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
		669	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
358	}	670	}
359		671
360	static void	672	static void
361	encode_sv (enc_t enc, SV sv)	673	encode_sv (enc_t enc, SV sv)
362	{	674	{
…		…
370	encode_str (enc, str, len, SvUTF8 (sv));	682	encode_str (enc, str, len, SvUTF8 (sv));
371	encode_ch (enc, '"');	683	encode_ch (enc, '"');
372	}	684	}
373	else if (SvNOKp (sv))	685	else if (SvNOKp (sv))
374	{	686	{
		687	// trust that perl will do the right thing w.r.t. JSON syntax.
375	need (enc, NV_DIG + 32);	688	need (enc, NV_DIG + 32);
376	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	689	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
377	enc->cur += strlen (enc->cur);	690	enc->cur += strlen (enc->cur);
378	}	691	}
379	else if (SvIOKp (sv))	692	else if (SvIOKp (sv))
380	{	693	{
		694	// we assume we can always read an IV as a UV and vice versa
		695	// we assume two's complement
		696	// we assume no aliasing issues in the union
		697	if (SvIsUV (sv) ? SvUVX (sv) <= 59000
		698	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
		699	{
		700	// optimise the "small number case"
		701	// code will likely be branchless and use only a single multiplication
		702	// works for numbers up to 59074
		703	I32 i = SvIVX (sv);
		704	U32 u;
		705	char digit, nz = 0;
		706
381	need (enc, 64);	707	need (enc, 6);
		708
		709	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
		710	u = i < 0 ? -i : i;
		711
		712	// convert to 4.28 fixed-point representation
		713	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
		714
		715	// now output digit by digit, each time masking out the integer part
		716	// and multiplying by 5 while moving the decimal point one to the right,
		717	// resulting in a net multiplication by 10.
		718	// we always write the digit to memory but conditionally increment
		719	// the pointer, to enable the use of conditional move instructions.
		720	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffffUL) 5;
		721	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffffUL) 5;
		722	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffffUL) 5;
		723	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffffUL) 5;
		724	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
		725	}
		726	else
		727	{
		728	// large integer, use the (rather slow) snprintf way.
		729	need (enc, IVUV_MAXCHARS);
382	enc->cur +=	730	enc->cur +=
383	SvIsUV(sv)	731	SvIsUV(sv)
384	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	732	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
385	: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));	733	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
		734	}
386	}	735	}
387	else if (SvROK (sv))	736	else if (SvROK (sv))
388	{	737	encode_rv (enc, SvRV (sv));
389	if (!--enc->max_recurse)	738	else if (!SvOK (sv) \|\| enc->json.flags & F_ALLOW_UNKNOWN)
390	croak ("data structure too deep (hit recursion limit)");
391
392	sv = SvRV (sv);
393
394	switch (SvTYPE (sv))
395	{
396	case SVt_PVAV: encode_av (enc, (AV *)sv); break;
397	case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
398
399	default:
400	croak ("JSON can only represent references to arrays or hashes");
401	}
402	}
403	else if (!SvOK (sv))
404	encode_str (enc, "null", 4, 0);	739	encode_str (enc, "null", 4, 0);
405	else	740	else
406	croak ("encountered perl type that JSON cannot handle");	741	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
		742	SvPV_nolen (sv), SvFLAGS (sv));
407	}	743	}
408		744
409	static SV *	745	static SV *
410	encode_json (SV *scalar, UV flags)	746	encode_json (SV scalar, JSON json)
411	{	747	{
412	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
413	croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)");
414
415	enc_t enc;	748	enc_t enc;
416	enc.flags = flags;	749
		750	if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
		751	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
		752
		753	enc.json = *json;
417	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	754	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
418	enc.cur = SvPVX (enc.sv);	755	enc.cur = SvPVX (enc.sv);
419	enc.end = SvEND (enc.sv);	756	enc.end = SvEND (enc.sv);
420	enc.max_recurse = 0;
421	enc.indent = 0;	757	enc.indent = 0;
		758	enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
		759	: enc.json.flags & F_LATIN1 ? 0x000100UL
		760	: 0x110000UL;
422		761
423	SvPOK_only (enc.sv);	762	SvPOK_only (enc.sv);
424	encode_sv (&enc, scalar);	763	encode_sv (&enc, scalar);
		764	encode_nl (&enc);
425		765
		766	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
		767	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
		768
426	if (!(flags & (F_ASCII \| F_UTF8)))	769	if (!(enc.json.flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
427	SvUTF8_on (enc.sv);	770	SvUTF8_on (enc.sv);
428		771
429	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	772	if (enc.json.flags & F_SHRINK)
		773	shrink (enc.sv);
		774
430	return enc.sv;	775	return enc.sv;
431	}	776	}
432		777
433	/////////////////////////////////////////////////////////////////////////////	778	/////////////////////////////////////////////////////////////////////////////
		779	// decoder
434		780
435	#define WS \	781	// structure used for decoding JSON
		782	typedef struct
		783	{
		784	char *cur; // current parser pointer
		785	char *end; // end of input string
		786	const char *err; // parse error, if != 0
		787	JSON json;
		788	U32 depth; // recursion depth
		789	U32 maxdepth; // recursion depth limit
		790	} dec_t;
		791
		792	INLINE void
		793	decode_comment (dec_t *dec)
		794	{
		795	// only '#'-style comments allowed a.t.m.
		796
		797	while (dec->cur && dec->cur != 0x0a && *dec->cur != 0x0d)
		798	++dec->cur;
		799	}
		800
		801	INLINE void
		802	decode_ws (dec_t *dec)
		803	{
436	for (;;) \	804	for (;;)
437	{ \	805	{
438	char ch = *dec->cur; \	806	char ch = *dec->cur;
		807
439	if (ch > 0x20 \	808	if (ch > 0x20)
		809	{
		810	if (expect_false (ch == '#'))
		811	{
		812	if (dec->json.flags & F_RELAXED)
		813	decode_comment (dec);
		814	else
		815	break;
		816	}
		817	else
		818	break;
		819	}
440	\|\| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \	820	else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
441	break; \	821	break; // parse error, but let higher level handle it, gives better error messages
		822
442	++dec->cur; \	823	++dec->cur;
443	}	824	}
		825	}
444		826
445	#define ERR(reason) SB dec->err = reason; goto fail; SE	827	#define ERR(reason) SB dec->err = reason; goto fail; SE
		828
446	#define EXPECT_CH(ch) SB \	829	#define EXPECT_CH(ch) SB \
447	if (*dec->cur != ch) \	830	if (*dec->cur != ch) \
448	ERR (# ch " expected"); \	831	ERR (# ch " expected"); \
449	++dec->cur; \	832	++dec->cur; \
450	SE	833	SE
451		834
		835	#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
		836	#define DEC_DEC_DEPTH --dec->depth
		837
452	static SV decode_sv (dec_t dec);	838	static SV decode_sv (dec_t dec);
453		839
454	static signed char decode_hexdigit[256];	840	static signed char decode_hexdigit[256];
455		841
456	static UV	842	static UV
457	decode_4hex (dec_t *dec)	843	decode_4hex (dec_t *dec)
458	{	844	{
459	signed char d1, d2, d3, d4;	845	signed char d1, d2, d3, d4;
		846	unsigned char cur = (unsigned char )dec->cur;
460		847
461	d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];	848	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
462	if (d1 < 0) ERR ("four hexadecimal digits expected");	849	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
463	d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];	850	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
464	if (d2 < 0) ERR ("four hexadecimal digits expected");	851	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
465	d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
466	if (d3 < 0) ERR ("four hexadecimal digits expected");
467	d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
468	if (d4 < 0) ERR ("four hexadecimal digits expected");
469		852
470	dec->cur += 4;	853	dec->cur += 4;
471		854
472	return ((UV)d1) << 12	855	return ((UV)d1) << 12
473	\| ((UV)d2) << 8	856	\| ((UV)d2) << 8
…		…
476		859
477	fail:	860	fail:
478	return (UV)-1;	861	return (UV)-1;
479	}	862	}
480		863
481	#define APPEND_GROW(n) SB \
482	if (cur + (n) >= end) \
483	{ \
484	STRLEN ofs = cur - SvPVX (sv); \
485	SvGROW (sv, ofs + (n) + 1); \
486	cur = SvPVX (sv) + ofs; \
487	end = SvEND (sv); \
488	} \
489	SE
490
491	#define APPEND_CH(ch) SB \
492	APPEND_GROW (1); \
493	*cur++ = (ch); \
494	SE
495
496	static SV *	864	static SV *
497	decode_str (dec_t *dec)	865	decode_str (dec_t *dec)
498	{	866	{
499	SV *sv = NEWSV (0,2);	867	SV *sv = 0;
500	int utf8 = 0;	868	int utf8 = 0;
501	char *cur = SvPVX (sv);	869	char *dec_cur = dec->cur;
502	char *end = SvEND (sv);
503		870
504	for (;;)	871	do
505	{	872	{
506	unsigned char ch = (unsigned char )dec->cur;	873	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
		874	char *cur = buf;
507		875
508	if (ch == '"')	876	do
509	break;
510	else if (ch == '\\')
511	{	877	{
512	switch (*++dec->cur)	878	unsigned char ch = (unsigned char )dec_cur++;
		879
		880	if (expect_false (ch == '"'))
513	{	881	{
514	case '\\':	882	--dec_cur;
515	case '/':	883	break;
516	case '"': APPEND_CH (*dec->cur++); break;	884	}
517		885	else if (expect_false (ch == '\\'))
518	case 'b': APPEND_CH ('\010'); ++dec->cur; break;	886	{
519	case 't': APPEND_CH ('\011'); ++dec->cur; break;	887	switch (*dec_cur)
520	case 'n': APPEND_CH ('\012'); ++dec->cur; break;
521	case 'f': APPEND_CH ('\014'); ++dec->cur; break;
522	case 'r': APPEND_CH ('\015'); ++dec->cur; break;
523
524	case 'u':
525	{	888	{
526	UV lo, hi;	889	case '\\':
527	++dec->cur;	890	case '/':
		891	case '"': cur++ = dec_cur++; break;
528		892
529	hi = decode_4hex (dec);	893	case 'b': ++dec_cur; *cur++ = '\010'; break;
530	if (hi == (UV)-1)	894	case 't': ++dec_cur; *cur++ = '\011'; break;
531	goto fail;	895	case 'n': ++dec_cur; *cur++ = '\012'; break;
		896	case 'f': ++dec_cur; *cur++ = '\014'; break;
		897	case 'r': ++dec_cur; *cur++ = '\015'; break;
532		898
533	// possibly a surrogate pair	899	case 'u':
534	if (hi >= 0xd800 && hi < 0xdc00)
535	{	900	{
536	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')	901	UV lo, hi;
537	ERR ("missing low surrogate character in surrogate pair");	902	++dec_cur;
538		903
539	dec->cur += 2;	904	dec->cur = dec_cur;
540
541	lo = decode_4hex (dec);	905	hi = decode_4hex (dec);
		906	dec_cur = dec->cur;
542	if (lo == (UV)-1)	907	if (hi == (UV)-1)
543	goto fail;	908	goto fail;
544		909
		910	// possibly a surrogate pair
		911	if (hi >= 0xd800)
		912	if (hi < 0xdc00)
		913	{
		914	if (dec_cur [0] != '\\' \|\| dec_cur [1] != 'u')
		915	ERR ("missing low surrogate character in surrogate pair");
		916
		917	dec_cur += 2;
		918
		919	dec->cur = dec_cur;
		920	lo = decode_4hex (dec);
		921	dec_cur = dec->cur;
		922	if (lo == (UV)-1)
		923	goto fail;
		924
545	if (lo < 0xdc00 \|\| lo >= 0xe000)	925	if (lo < 0xdc00 \|\| lo >= 0xe000)
546	ERR ("surrogate pair expected");	926	ERR ("surrogate pair expected");
547		927
548	hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;	928	hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
		929	}
		930	else if (hi < 0xe000)
		931	ERR ("missing high surrogate character in surrogate pair");
		932
		933	if (hi >= 0x80)
		934	{
		935	utf8 = 1;
		936
		937	cur = encode_utf8 (cur, hi);
		938	}
		939	else
		940	*cur++ = hi;
549	}	941	}
550	else if (hi >= 0xdc00 && hi < 0xe000)
551	ERR ("missing high surrogate character in surrogate pair");
552
553	if (hi >= 0x80)
554	{	942	break;
555	utf8 = 1;
556		943
557	APPEND_GROW (4); // at most 4 bytes for 21 bits
558	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
559	}
560	else	944	default:
561	APPEND_CH (hi);	945	--dec_cur;
		946	ERR ("illegal backslash escape sequence in string");
562	}	947	}
563	break;	948	}
		949	else if (expect_true (ch >= 0x20 && ch < 0x80))
		950	*cur++ = ch;
		951	else if (ch >= 0x80)
		952	{
		953	STRLEN clen;
		954	UV uch;
564		955
565	default:
566	--dec->cur;	956	--dec_cur;
567	ERR ("illegal backslash escape sequence in string");	957
		958	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
		959	if (clen == (STRLEN)-1)
		960	ERR ("malformed UTF-8 character in JSON string");
		961
		962	do
		963	cur++ = dec_cur++;
		964	while (--clen);
		965
		966	utf8 = 1;
		967	}
		968	else
		969	{
		970	--dec_cur;
		971
		972	if (!ch)
		973	ERR ("unexpected end of string while parsing JSON string");
		974	else
		975	ERR ("invalid character encountered while parsing JSON string");
568	}	976	}
569	}	977	}
570	else if (ch >= 0x20 && ch <= 0x7f)	978	while (cur < buf + SHORT_STRING_LEN);
571	APPEND_CH (*dec->cur++);	979
572	else if (ch >= 0x80)
573	{	980	{
574	STRLEN clen;	981	STRLEN len = cur - buf;
575	UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
576	if (clen == (STRLEN)-1)
577	ERR ("malformed UTF-8 character in string, cannot convert to JSON");
578		982
579	APPEND_GROW (clen);	983	if (sv)
580	do
581	{	984	{
582	cur++ = dec->cur++;	985	STRLEN cur = SvCUR (sv);
		986
		987	if (SvLEN (sv) <= cur + len)
		988	SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
		989
		990	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
		991	SvCUR_set (sv, SvCUR (sv) + len);
583	}	992	}
584	while (--clen);
585
586	utf8 = 1;
587	}
588	else if (dec->cur == dec->end)
589	ERR ("unexpected end of string while parsing json string");
590	else	993	else
591	ERR ("invalid character encountered");	994	sv = newSVpvn (buf, len);
592	}	995	}
		996	}
		997	while (*dec_cur != '"');
593		998
594	++dec->cur;	999	++dec_cur;
595		1000
596	SvCUR_set (sv, cur - SvPVX (sv));	1001	if (sv)
597		1002	{
598	SvPOK_only (sv);	1003	SvPOK_only (sv);
599	*SvEND (sv) = 0;	1004	*SvEND (sv) = 0;
600		1005
601	if (utf8)	1006	if (utf8)
602	SvUTF8_on (sv);	1007	SvUTF8_on (sv);
		1008	}
		1009	else
		1010	sv = newSVpvn ("", 0);
603		1011
		1012	dec->cur = dec_cur;
604	return sv;	1013	return sv;
605		1014
606	fail:	1015	fail:
607	SvREFCNT_dec (sv);	1016	dec->cur = dec_cur;
608	return 0;	1017	return 0;
609	}	1018	}
610		1019
611	static SV *	1020	static SV *
612	decode_num (dec_t *dec)	1021	decode_num (dec_t *dec)
…		…
670	is_nv = 1;	1079	is_nv = 1;
671	}	1080	}
672		1081
673	if (!is_nv)	1082	if (!is_nv)
674	{	1083	{
675	UV uv;	1084	int len = dec->cur - start;
676	int numtype = grok_number (start, dec->cur - start, &uv);	1085
677	if (numtype & IS_NUMBER_IN_UV)	1086	// special case the rather common 1..5-digit-int case
678	if (numtype & IS_NUMBER_NEG)	1087	if (*start == '-')
		1088	switch (len)
679	{	1089	{
680	if (uv < (UV)IV_MIN)	1090	case 2: return newSViv (-(IV)( start [1] - '0' * 1));
681	return newSViv (-(IV)uv);	1091	case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
		1092	case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
		1093	case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
		1094	case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
682	}	1095	}
		1096	else
		1097	switch (len)
		1098	{
		1099	case 1: return newSViv ( start [0] - '0' * 1);
		1100	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
		1101	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
		1102	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		1103	case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
		1104	}
		1105
		1106	{
		1107	UV uv;
		1108	int numtype = grok_number (start, len, &uv);
		1109	if (numtype & IS_NUMBER_IN_UV)
		1110	if (numtype & IS_NUMBER_NEG)
		1111	{
		1112	if (uv < (UV)IV_MIN)
		1113	return newSViv (-(IV)uv);
		1114	}
683	else	1115	else
684	return newSVuv (uv);	1116	return newSVuv (uv);
685	}	1117	}
686		1118
		1119	len -= *start == '-' ? 1 : 0;
		1120
		1121	// does not fit into IV or UV, try NV
		1122	if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
		1123	#if defined (LDBL_DIG)
		1124	\|\| (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
		1125	#endif
		1126	)
		1127	// fits into NV without loss of precision
		1128	return newSVnv (Atof (start));
		1129
		1130	// everything else fails, convert it to a string
		1131	return newSVpvn (start, dec->cur - start);
		1132	}
		1133
		1134	// loss of precision here
687	return newSVnv (Atof (start));	1135	return newSVnv (Atof (start));
688		1136
689	fail:	1137	fail:
690	return 0;	1138	return 0;
691	}	1139	}
…		…
693	static SV *	1141	static SV *
694	decode_av (dec_t *dec)	1142	decode_av (dec_t *dec)
695	{	1143	{
696	AV *av = newAV ();	1144	AV *av = newAV ();
697		1145
698	WS;	1146	DEC_INC_DEPTH;
		1147	decode_ws (dec);
		1148
699	if (*dec->cur == ']')	1149	if (*dec->cur == ']')
700	++dec->cur;	1150	++dec->cur;
701	else	1151	else
702	for (;;)	1152	for (;;)
703	{	1153	{
…		…
707	if (!value)	1157	if (!value)
708	goto fail;	1158	goto fail;
709		1159
710	av_push (av, value);	1160	av_push (av, value);
711		1161
712	WS;	1162	decode_ws (dec);
713		1163
714	if (*dec->cur == ']')	1164	if (*dec->cur == ']')
715	{	1165	{
716	++dec->cur;	1166	++dec->cur;
717	break;	1167	break;
…		…
719		1169
720	if (*dec->cur != ',')	1170	if (*dec->cur != ',')
721	ERR (", or ] expected while parsing array");	1171	ERR (", or ] expected while parsing array");
722		1172
723	++dec->cur;	1173	++dec->cur;
		1174
		1175	decode_ws (dec);
		1176
		1177	if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
		1178	{
		1179	++dec->cur;
		1180	break;
		1181	}
724	}	1182	}
725		1183
		1184	DEC_DEC_DEPTH;
726	return newRV_noinc ((SV *)av);	1185	return newRV_noinc ((SV *)av);
727		1186
728	fail:	1187	fail:
729	SvREFCNT_dec (av);	1188	SvREFCNT_dec (av);
		1189	DEC_DEC_DEPTH;
730	return 0;	1190	return 0;
731	}	1191	}
732		1192
733	static SV *	1193	static SV *
734	decode_hv (dec_t *dec)	1194	decode_hv (dec_t *dec)
735	{	1195	{
		1196	SV *sv;
736	HV *hv = newHV ();	1197	HV *hv = newHV ();
737		1198
738	WS;	1199	DEC_INC_DEPTH;
		1200	decode_ws (dec);
		1201
739	if (*dec->cur == '}')	1202	if (*dec->cur == '}')
740	++dec->cur;	1203	++dec->cur;
741	else	1204	else
742	for (;;)	1205	for (;;)
743	{	1206	{
744	SV key, value;
745
746	WS; EXPECT_CH ('"');	1207	EXPECT_CH ('"');
747		1208
748	key = decode_str (dec);	1209	// heuristic: assume that
749	if (!key)	1210	// a) decode_str + hv_store_ent are abysmally slow.
750	goto fail;	1211	// b) most hash keys are short, simple ascii text.
		1212	// => try to "fast-match" such strings to avoid
		1213	// the overhead of decode_str + hv_store_ent.
		1214	{
		1215	SV *value;
		1216	char *p = dec->cur;
		1217	char *e = p + 24; // only try up to 24 bytes
751		1218
752	WS; EXPECT_CH (':');	1219	for (;;)
753
754	value = decode_sv (dec);
755	if (!value)
756	{	1220	{
		1221	// the >= 0x80 is false on most architectures
		1222	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
		1223	{
		1224	// slow path, back up and use decode_str
		1225	SV *key = decode_str (dec);
		1226	if (!key)
		1227	goto fail;
		1228
		1229	decode_ws (dec); EXPECT_CH (':');
		1230
		1231	decode_ws (dec);
		1232	value = decode_sv (dec);
		1233	if (!value)
		1234	{
		1235	SvREFCNT_dec (key);
		1236	goto fail;
		1237	}
		1238
		1239	hv_store_ent (hv, key, value, 0);
757	SvREFCNT_dec (key);	1240	SvREFCNT_dec (key);
		1241
		1242	break;
		1243	}
		1244	else if (*p == '"')
		1245	{
		1246	// fast path, got a simple key
		1247	char *key = dec->cur;
		1248	int len = p - key;
		1249	dec->cur = p + 1;
		1250
		1251	decode_ws (dec); EXPECT_CH (':');
		1252
		1253	decode_ws (dec);
		1254	value = decode_sv (dec);
		1255	if (!value)
758	goto fail;	1256	goto fail;
		1257
		1258	hv_store (hv, key, len, value, 0);
		1259
		1260	break;
		1261	}
		1262
		1263	++p;
759	}	1264	}
760
761	//TODO: optimise
762	hv_store_ent (hv, key, value, 0);
763
764	WS;	1265	}
		1266
		1267	decode_ws (dec);
765		1268
766	if (*dec->cur == '}')	1269	if (*dec->cur == '}')
767	{	1270	{
768	++dec->cur;	1271	++dec->cur;
769	break;	1272	break;
…		…
771		1274
772	if (*dec->cur != ',')	1275	if (*dec->cur != ',')
773	ERR (", or } expected while parsing object/hash");	1276	ERR (", or } expected while parsing object/hash");
774		1277
775	++dec->cur;	1278	++dec->cur;
		1279
		1280	decode_ws (dec);
		1281
		1282	if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
		1283	{
		1284	++dec->cur;
		1285	break;
		1286	}
776	}	1287	}
777		1288
		1289	DEC_DEC_DEPTH;
778	return newRV_noinc ((SV *)hv);	1290	sv = newRV_noinc ((SV *)hv);
		1291
		1292	// check filter callbacks
		1293	if (dec->json.flags & F_HOOK)
		1294	{
		1295	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
		1296	{
		1297	HE cb, he;
		1298
		1299	hv_iterinit (hv);
		1300	he = hv_iternext (hv);
		1301	hv_iterinit (hv);
		1302
		1303	// the next line creates a mortal sv each time its called.
		1304	// might want to optimise this for common cases.
		1305	cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
		1306
		1307	if (cb)
		1308	{
		1309	dSP;
		1310	int count;
		1311
		1312	ENTER; SAVETMPS; PUSHMARK (SP);
		1313	XPUSHs (HeVAL (he));
		1314
		1315	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
		1316
		1317	if (count == 1)
		1318	{
		1319	sv = newSVsv (POPs);
		1320	FREETMPS; LEAVE;
		1321	return sv;
		1322	}
		1323
		1324	FREETMPS; LEAVE;
		1325	}
		1326	}
		1327
		1328	if (dec->json.cb_object)
		1329	{
		1330	dSP;
		1331	int count;
		1332
		1333	ENTER; SAVETMPS; PUSHMARK (SP);
		1334	XPUSHs (sv_2mortal (sv));
		1335
		1336	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
		1337
		1338	if (count == 1)
		1339	{
		1340	sv = newSVsv (POPs);
		1341	FREETMPS; LEAVE;
		1342	return sv;
		1343	}
		1344
		1345	SvREFCNT_inc (sv);
		1346	FREETMPS; LEAVE;
		1347	}
		1348	}
		1349
		1350	return sv;
779		1351
780	fail:	1352	fail:
781	SvREFCNT_dec (hv);	1353	SvREFCNT_dec (hv);
		1354	DEC_DEC_DEPTH;
782	return 0;	1355	return 0;
783	}	1356	}
784		1357
785	static SV *	1358	static SV *
786	decode_sv (dec_t *dec)	1359	decode_sv (dec_t *dec)
787	{	1360	{
788	WS;	1361	// the beauty of JSON: you need exactly one character lookahead
		1362	// to parse everything.
789	switch (*dec->cur)	1363	switch (*dec->cur)
790	{	1364	{
791	case '"': ++dec->cur; return decode_str (dec);	1365	case '"': ++dec->cur; return decode_str (dec);
792	case '[': ++dec->cur; return decode_av (dec);	1366	case '[': ++dec->cur; return decode_av (dec);
793	case '{': ++dec->cur; return decode_hv (dec);	1367	case '{': ++dec->cur; return decode_hv (dec);
794		1368
795	case '-':	1369	case '-':
796	case '0': case '1': case '2': case '3': case '4':	1370	case '0': case '1': case '2': case '3': case '4':
797	case '5': case '6': case '7': case '8': case '9':	1371	case '5': case '6': case '7': case '8': case '9':
798	return decode_num (dec);	1372	return decode_num (dec);
799		1373
800	case 't':	1374	case 't':
801	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1375	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
802	{	1376	{
803	dec->cur += 4;	1377	dec->cur += 4;
		1378	#if JSON_SLOW
		1379	json_true = get_bool ("JSON::XS::true");
		1380	#endif
804	return newSViv (1);	1381	return newSVsv (json_true);
805	}	1382	}
806	else	1383	else
807	ERR ("'true' expected");	1384	ERR ("'true' expected");
808		1385
809	break;	1386	break;
810		1387
811	case 'f':	1388	case 'f':
812	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1389	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
813	{	1390	{
814	dec->cur += 5;	1391	dec->cur += 5;
		1392	#if JSON_SLOW
		1393	json_false = get_bool ("JSON::XS::false");
		1394	#endif
815	return newSViv (0);	1395	return newSVsv (json_false);
816	}	1396	}
817	else	1397	else
818	ERR ("'false' expected");	1398	ERR ("'false' expected");
819		1399
820	break;	1400	break;
…		…
829	ERR ("'null' expected");	1409	ERR ("'null' expected");
830		1410
831	break;	1411	break;
832		1412
833	default:	1413	default:
834	ERR ("malformed json string");	1414	ERR ("malformed JSON string, neither array, object, number, string or atom");
835	break;	1415	break;
836	}	1416	}
837		1417
838	fail:	1418	fail:
839	return 0;	1419	return 0;
840	}	1420	}
841		1421
842	static SV *	1422	static SV *
843	decode_json (SV *string, UV flags)	1423	decode_json (SV string, JSON json, char **offset_return)
844	{	1424	{
		1425	dec_t dec;
845	SV *sv;	1426	SV *sv;
846		1427
		1428	/* work around bugs in 5.10 where manipulating magic values
		1429	* will perl ignore the magic in subsequent accesses
		1430	*/
		1431	/SvGETMAGIC (string);/
		1432	if (SvMAGICAL (string))
		1433	string = sv_2mortal (newSVsv (string));
		1434
		1435	SvUPGRADE (string, SVt_PV);
		1436
		1437	/* work around a bug in perl 5.10, which causes SvCUR to fail an
		1438	* assertion with -DDEBUGGING, although SvCUR is documented to
		1439	* return the xpv_cur field which certainly exists after upgrading.
		1440	* according to nicholas clark, calling SvPOK fixes this.
		1441	* But it doesn't fix it, so try another workaround, call SvPV_nolen
		1442	* and hope for the best.
		1443	* Damnit, SvPV_nolen still trips over yet another assertion. This
		1444	* assertion business is seriously broken, try yet another workaround
		1445	* for the broken -DDEBUGGING.
		1446	*/
		1447	{
		1448	#ifdef DEBUGGING
		1449	STRLEN offset = SvOK (string) ? sv_len (string) : 0;
		1450	#else
		1451	STRLEN offset = SvCUR (string);
		1452	#endif
		1453
		1454	if (offset > json->max_size && json->max_size)
		1455	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1456	(unsigned long)SvCUR (string), (unsigned long)json->max_size);
		1457	}
		1458
847	if (flags & F_UTF8)	1459	if (json->flags & F_UTF8)
848	sv_utf8_downgrade (string, 0);	1460	sv_utf8_downgrade (string, 0);
849	else	1461	else
850	sv_utf8_upgrade (string);	1462	sv_utf8_upgrade (string);
851		1463
852	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1464	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
853		1465
854	dec_t dec;	1466	dec.json = *json;
855	dec.flags = flags;
856	dec.cur = SvPVX (string);	1467	dec.cur = SvPVX (string);
857	dec.end = SvEND (string);	1468	dec.end = SvEND (string);
858	dec.err = 0;	1469	dec.err = 0;
		1470	dec.depth = 0;
859		1471
		1472	if (dec.json.cb_object \|\| dec.json.cb_sk_object)
		1473	dec.json.flags \|= F_HOOK;
		1474
		1475	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
		1476
		1477	decode_ws (&dec);
860	sv = decode_sv (&dec);	1478	sv = decode_sv (&dec);
861		1479
		1480	if (offset_return)
		1481	*offset_return = dec.cur;
		1482
		1483	if (!(offset_return \|\| !sv))
		1484	{
		1485	// check for trailing garbage
		1486	decode_ws (&dec);
		1487
		1488	if (*dec.cur)
		1489	{
		1490	dec.err = "garbage after JSON object";
		1491	SvREFCNT_dec (sv);
		1492	sv = 0;
		1493	}
		1494	}
		1495
862	if (!sv)	1496	if (!sv)
863	{	1497	{
864	IV offset = utf8_distance (dec.cur, SvPVX (string));
865	SV *uni = sv_newmortal ();	1498	SV *uni = sv_newmortal ();
		1499
866	// horrible hack to silence warning inside pv_uni_display	1500	// horrible hack to silence warning inside pv_uni_display
867	COP cop;	1501	COP cop = *PL_curcop;
868	memset (&cop, 0, sizeof (cop));
869	cop.cop_warnings = pWARN_NONE;	1502	cop.cop_warnings = pWARN_NONE;
		1503	ENTER;
870	SAVEVPTR (PL_curcop);	1504	SAVEVPTR (PL_curcop);
871	PL_curcop = &cop;	1505	PL_curcop = &cop;
872
873	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1506	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
		1507	LEAVE;
		1508
874	croak ("%s, at character offset %d (%s)",	1509	croak ("%s, at character offset %d (before \"%s\")",
875	dec.err,	1510	dec.err,
876	(int)offset,	1511	ptr_to_index (string, dec.cur),
877	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1512	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
878	}	1513	}
879		1514
880	sv = sv_2mortal (sv);	1515	sv = sv_2mortal (sv);
881		1516
882	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))	1517	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
883	croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)");	1518	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
884		1519
885	return sv;	1520	return sv;
886	}	1521	}
887		1522
		1523	/////////////////////////////////////////////////////////////////////////////
		1524	// incremental parser
		1525
		1526	static void
		1527	incr_parse (JSON *self)
		1528	{
		1529	const char *p = SvPVX (self->incr_text) + self->incr_pos;
		1530
		1531	for (;;)
		1532	{
		1533	//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
		1534	switch (self->incr_mode)
		1535	{
		1536	// only used for intiial whitespace skipping
		1537	case INCR_M_WS:
		1538	for (;;)
		1539	{
		1540	if (*p > 0x20)
		1541	{
		1542	self->incr_mode = INCR_M_JSON;
		1543	goto incr_m_json;
		1544	}
		1545	else if (!*p)
		1546	goto interrupt;
		1547
		1548	++p;
		1549	}
		1550
		1551	// skip a single char inside a string (for \\-processing)
		1552	case INCR_M_BS:
		1553	if (!*p)
		1554	goto interrupt;
		1555
		1556	++p;
		1557	self->incr_mode = INCR_M_STR;
		1558	goto incr_m_str;
		1559
		1560	// inside a string
		1561	case INCR_M_STR:
		1562	incr_m_str:
		1563	for (;;)
		1564	{
		1565	if (*p == '"')
		1566	{
		1567	++p;
		1568	self->incr_mode = INCR_M_JSON;
		1569
		1570	if (!self->incr_nest)
		1571	goto interrupt;
		1572
		1573	goto incr_m_json;
		1574	}
		1575	else if (*p == '\\')
		1576	{
		1577	++p; // "virtually" consumes character after \
		1578
		1579	if (!*p) // if at end of string we have to switch modes
		1580	{
		1581	self->incr_mode = INCR_M_BS;
		1582	goto interrupt;
		1583	}
		1584	}
		1585	else if (!*p)
		1586	goto interrupt;
		1587
		1588	++p;
		1589	}
		1590
		1591	// after initial ws, outside string
		1592	case INCR_M_JSON:
		1593	incr_m_json:
		1594	for (;;)
		1595	{
		1596	switch (*p++)
		1597	{
		1598	case 0:
		1599	--p;
		1600	goto interrupt;
		1601
		1602	case 0x09:
		1603	case 0x0a:
		1604	case 0x0d:
		1605	case 0x20:
		1606	if (!self->incr_nest)
		1607	{
		1608	--p; // do not eat the whitespace, let the next round do it
		1609	goto interrupt;
		1610	}
		1611	break;
		1612
		1613	case '"':
		1614	self->incr_mode = INCR_M_STR;
		1615	goto incr_m_str;
		1616
		1617	case '[':
		1618	case '{':
		1619	if (++self->incr_nest > self->max_depth)
		1620	croak (ERR_NESTING_EXCEEDED);
		1621	break;
		1622
		1623	case ']':
		1624	case '}':
		1625	if (--self->incr_nest <= 0)
		1626	goto interrupt;
		1627	}
		1628	}
		1629	}
		1630
		1631	modechange:
		1632	;
		1633	}
		1634
		1635	interrupt:
		1636	self->incr_pos = p - SvPVX (self->incr_text);
		1637	//printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
		1638	}
		1639
		1640	/////////////////////////////////////////////////////////////////////////////
		1641	// XS interface functions
		1642
888	MODULE = JSON::XS PACKAGE = JSON::XS	1643	MODULE = JSON::XS PACKAGE = JSON::XS
889		1644
890	BOOT:	1645	BOOT:
891	{	1646	{
892	int i;	1647	int i;
893		1648
894	memset (decode_hexdigit, 0xff, 256);
895	for (i = 10; i--; )	1649	for (i = 0; i < 256; ++i)
896	decode_hexdigit ['0' + i] = i;	1650	decode_hexdigit [i] =
		1651	i >= '0' && i <= '9' ? i - '0'
		1652	: i >= 'a' && i <= 'f' ? i - 'a' + 10
		1653	: i >= 'A' && i <= 'F' ? i - 'A' + 10
		1654	: -1;
897		1655
898	for (i = 7; i--; )
899	{
900	decode_hexdigit ['a' + i] = 10 + i;
901	decode_hexdigit ['A' + i] = 10 + i;
902	}
903
904	json_stash = gv_stashpv ("JSON::XS", 1);	1656	json_stash = gv_stashpv ("JSON::XS" , 1);
		1657	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
		1658
		1659	json_true = get_bool ("JSON::XS::true");
		1660	json_false = get_bool ("JSON::XS::false");
		1661
		1662	CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
905	}	1663	}
906		1664
907	PROTOTYPES: DISABLE	1665	PROTOTYPES: DISABLE
908		1666
909	SV new (char dummy)	1667	void CLONE (...)
910	CODE:	1668	CODE:
911	RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);	1669	json_stash = 0;
		1670	json_boolean_stash = 0;
		1671
		1672	void new (char *klass)
		1673	PPCODE:
		1674	{
		1675	SV *pv = NEWSV (0, sizeof (JSON));
		1676	SvPOK_only (pv);
		1677	json_init ((JSON *)SvPVX (pv));
		1678	XPUSHs (sv_2mortal (sv_bless (
		1679	newRV_noinc (pv),
		1680	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
		1681	)));
		1682	}
		1683
		1684	void ascii (JSON *self, int enable = 1)
		1685	ALIAS:
		1686	ascii = F_ASCII
		1687	latin1 = F_LATIN1
		1688	utf8 = F_UTF8
		1689	indent = F_INDENT
		1690	canonical = F_CANONICAL
		1691	space_before = F_SPACE_BEFORE
		1692	space_after = F_SPACE_AFTER
		1693	pretty = F_PRETTY
		1694	allow_nonref = F_ALLOW_NONREF
		1695	shrink = F_SHRINK
		1696	allow_blessed = F_ALLOW_BLESSED
		1697	convert_blessed = F_CONV_BLESSED
		1698	relaxed = F_RELAXED
		1699	allow_unknown = F_ALLOW_UNKNOWN
		1700	PPCODE:
		1701	{
		1702	if (enable)
		1703	self->flags \|= ix;
		1704	else
		1705	self->flags &= ~ix;
		1706
		1707	XPUSHs (ST (0));
		1708	}
		1709
		1710	void get_ascii (JSON *self)
		1711	ALIAS:
		1712	get_ascii = F_ASCII
		1713	get_latin1 = F_LATIN1
		1714	get_utf8 = F_UTF8
		1715	get_indent = F_INDENT
		1716	get_canonical = F_CANONICAL
		1717	get_space_before = F_SPACE_BEFORE
		1718	get_space_after = F_SPACE_AFTER
		1719	get_allow_nonref = F_ALLOW_NONREF
		1720	get_shrink = F_SHRINK
		1721	get_allow_blessed = F_ALLOW_BLESSED
		1722	get_convert_blessed = F_CONV_BLESSED
		1723	get_relaxed = F_RELAXED
		1724	get_allow_unknown = F_ALLOW_UNKNOWN
		1725	PPCODE:
		1726	XPUSHs (boolSV (self->flags & ix));
		1727
		1728	void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
		1729	PPCODE:
		1730	self->max_depth = max_depth;
		1731	XPUSHs (ST (0));
		1732
		1733	U32 get_max_depth (JSON *self)
		1734	CODE:
		1735	RETVAL = self->max_depth;
912	OUTPUT:	1736	OUTPUT:
913	RETVAL	1737	RETVAL
914		1738
915	SV ascii (SV self, int enable)	1739	void max_size (JSON *self, U32 max_size = 0)
916	ALIAS:	1740	PPCODE:
917	ascii = F_ASCII	1741	self->max_size = max_size;
918	utf8 = F_UTF8	1742	XPUSHs (ST (0));
919	indent = F_INDENT	1743
920	canonical = F_CANONICAL	1744	int get_max_size (JSON *self)
921	space_before = F_SPACE_BEFORE
922	space_after = F_SPACE_AFTER
923	json_rpc = F_JSON_RPC
924	pretty = F_PRETTY
925	allow_nonref = F_ALLOW_NONREF
926	CODE:	1745	CODE:
927	{	1746	RETVAL = self->max_size;
928	UV *uv = SvJSON (self);
929	if (enable)
930	*uv \|= ix;
931	else
932	*uv &= ~ix;
933
934	RETVAL = newSVsv (self);
935	}
936	OUTPUT:	1747	OUTPUT:
937	RETVAL	1748	RETVAL
938		1749
939	void encode (SV self, SV scalar)	1750	void filter_json_object (JSON self, SV cb = &PL_sv_undef)
940	PPCODE:	1751	PPCODE:
941	XPUSHs (encode_json (scalar, *SvJSON (self)));	1752	{
		1753	SvREFCNT_dec (self->cb_object);
		1754	self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
942		1755
943	void decode (SV self, SV jsonstr)	1756	XPUSHs (ST (0));
		1757	}
		1758
		1759	void filter_json_single_key_object (JSON self, SV key, SV *cb = &PL_sv_undef)
944	PPCODE:	1760	PPCODE:
		1761	{
		1762	if (!self->cb_sk_object)
		1763	self->cb_sk_object = newHV ();
		1764
		1765	if (SvOK (cb))
		1766	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
		1767	else
		1768	{
		1769	hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
		1770
		1771	if (!HvKEYS (self->cb_sk_object))
		1772	{
		1773	SvREFCNT_dec (self->cb_sk_object);
		1774	self->cb_sk_object = 0;
		1775	}
		1776	}
		1777
		1778	XPUSHs (ST (0));
		1779	}
		1780
		1781	void encode (JSON self, SV scalar)
		1782	PPCODE:
		1783	XPUSHs (encode_json (scalar, self));
		1784
		1785	void decode (JSON self, SV jsonstr)
		1786	PPCODE:
945	XPUSHs (decode_json (jsonstr, *SvJSON (self)));	1787	XPUSHs (decode_json (jsonstr, self, 0));
		1788
		1789	void decode_prefix (JSON self, SV jsonstr)
		1790	PPCODE:
		1791	{
		1792	char *offset;
		1793	EXTEND (SP, 2);
		1794	PUSHs (decode_json (jsonstr, self, &offset));
		1795	PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
		1796	}
		1797
		1798	void incr_parse (JSON self, SV jsonstr = 0)
		1799	PPCODE:
		1800	{
		1801	if (!self->incr_text)
		1802	self->incr_text = newSVpvn ("", 0);
		1803
		1804	// append data, if any
		1805	if (jsonstr)
		1806	{
		1807	if (SvUTF8 (jsonstr))
		1808	{
		1809	if (!SvUTF8 (self->incr_text))
		1810	{
		1811	/* utf-8-ness differs, need to upgrade */
		1812	sv_utf8_upgrade (self->incr_text);
		1813
		1814	if (self->incr_pos)
		1815	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
		1816	- (U8 *)SvPVX (self->incr_text);
		1817	}
		1818	}
		1819	else if (SvUTF8 (self->incr_text))
		1820	sv_utf8_upgrade (jsonstr);
		1821
		1822	{
		1823	STRLEN len;
		1824	const char *str = SvPV (jsonstr, len);
		1825	STRLEN cur = SvCUR (self->incr_text);
		1826
		1827	if (SvLEN (self->incr_text) <= cur + len)
		1828	SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
		1829
		1830	Move (str, SvEND (self->incr_text), len, char);
		1831	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
		1832	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
		1833	}
		1834	}
		1835
		1836	if (GIMME_V != G_VOID)
		1837	do
		1838	{
		1839	char *offset;
		1840
		1841	if (!INCR_DONE (self))
		1842	{
		1843	incr_parse (self);
		1844
		1845	if (self->incr_pos > self->max_size && self->max_size)
		1846	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1847	(unsigned long)self->incr_pos, (unsigned long)self->max_size);
		1848
		1849	if (!INCR_DONE (self))
		1850	break;
		1851	}
		1852
		1853	XPUSHs (decode_json (self->incr_text, self, &offset));
		1854
		1855	self->incr_pos -= offset - SvPVX (self->incr_text);
		1856	self->incr_nest = 0;
		1857	self->incr_mode = 0;
		1858
		1859	sv_chop (self->incr_text, offset);
		1860	}
		1861	while (GIMME_V == G_ARRAY);
		1862	}
		1863
		1864	SV incr_text (JSON self)
		1865	ATTRS: lvalue
		1866	CODE:
		1867	{
		1868	if (self->incr_pos)
		1869	croak ("incr_text can not be called when the incremental parser already started parsing");
		1870
		1871	RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
		1872	}
		1873	OUTPUT:
		1874	RETVAL
		1875
		1876	void incr_skip (JSON *self)
		1877	CODE:
		1878	{
		1879	if (self->incr_pos)
		1880	{
		1881	sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
		1882	self->incr_pos = 0;
		1883	self->incr_nest = 0;
		1884	self->incr_mode = 0;
		1885	}
		1886	}
		1887
		1888	void incr_reset (JSON *self)
		1889	CODE:
		1890	{
		1891	SvREFCNT_dec (self->incr_text);
		1892	self->incr_text = 0;
		1893	self->incr_pos = 0;
		1894	self->incr_nest = 0;
		1895	self->incr_mode = 0;
		1896	}
		1897
		1898	void DESTROY (JSON *self)
		1899	CODE:
		1900	SvREFCNT_dec (self->cb_sk_object);
		1901	SvREFCNT_dec (self->cb_object);
		1902	SvREFCNT_dec (self->incr_text);
946		1903
947	PROTOTYPES: ENABLE	1904	PROTOTYPES: ENABLE
948		1905
949	void to_json (SV *scalar)	1906	void encode_json (SV *scalar)
		1907	ALIAS:
		1908	to_json_ = 0
		1909	encode_json = F_UTF8
950	PPCODE:	1910	PPCODE:
		1911	{
		1912	JSON json;
		1913	json_init (&json);
		1914	json.flags \|= ix;
951	XPUSHs (encode_json (scalar, F_UTF8));	1915	XPUSHs (encode_json (scalar, &json));
		1916	}
952		1917
953	void from_json (SV *jsonstr)	1918	void decode_json (SV *jsonstr)
		1919	ALIAS:
		1920	from_json_ = 0
		1921	decode_json = F_UTF8
954	PPCODE:	1922	PPCODE:
		1923	{
		1924	JSON json;
		1925	json_init (&json);
		1926	json.flags \|= ix;
955	XPUSHs (decode_json (jsonstr, F_UTF8));	1927	XPUSHs (decode_json (jsonstr, &json, 0));
		1928	}
956		1929
		1930

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.5 by root, Thu Mar 22 23:24:18 2007 UTC vs. Revision 1.100 by root, Sat Sep 5 23:00:56 2009 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.5 by root, Thu Mar 22 23:24:18 2007 UTC vs.
Revision 1.100 by root, Sat Sep 5 23:00:56 2009 UTC