ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/JSON-XS/XS.xs

Comparing JSON-XS/XS.xs (file contents):
Revision 1.11 by root, Sat Mar 24 19:42:14 2007 UTC vs.
Revision 1.63 by root, Mon Aug 27 01:49:01 2007 UTC

1	#include "EXTERN.h"	1	#include "EXTERN.h"
2	#include "perl.h"	2	#include "perl.h"
3	#include "XSUB.h"	3	#include "XSUB.h"
4		4
5	#include "assert.h"	5	#include <assert.h>
6	#include "string.h"	6	#include <string.h>
7	#include "stdlib.h"	7	#include <stdlib.h>
		8	#include <stdio.h>
		9	#include <float.h>
8		10
		11	#if defined(__BORLANDC__) || defined(_MSC_VER)
		12	# define snprintf _snprintf // C compilers have this in stdio.h
		13	#endif
		14
		15	// some old perls do not have this, try to make it work, no
		16	// guarentees, though. if it breaks, you get to keep the pieces.
		17	#ifndef UTF8_MAXBYTES
		18	# define UTF8_MAXBYTES 13
		19	#endif
		20
9	#define F_ASCII 0x00000001	21	#define F_ASCII 0x00000001UL
		22	#define F_LATIN1 0x00000002UL
10	#define F_UTF8 0x00000002	23	#define F_UTF8 0x00000004UL
11	#define F_INDENT 0x00000004	24	#define F_INDENT 0x00000008UL
12	#define F_CANONICAL 0x00000008	25	#define F_CANONICAL 0x00000010UL
13	#define F_SPACE_BEFORE 0x00000010	26	#define F_SPACE_BEFORE 0x00000020UL
14	#define F_SPACE_AFTER 0x00000020	27	#define F_SPACE_AFTER 0x00000040UL
15	#define F_ALLOW_NONREF 0x00000080	28	#define F_ALLOW_NONREF 0x00000100UL
16	#define F_SHRINK 0x00000100	29	#define F_SHRINK 0x00000200UL
		30	#define F_ALLOW_BLESSED 0x00000400UL
		31	#define F_CONV_BLESSED 0x00000800UL
		32	#define F_RELAXED 0x00001000UL
		33
		34	#define F_MAXDEPTH 0xf8000000UL
		35	#define S_MAXDEPTH 27
		36	#define F_MAXSIZE 0x01f00000UL
		37	#define S_MAXSIZE 20
		38	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
		39
		40	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
		41	#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
17		42
18	#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER	43	#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19	#define F_DEFAULT 0	44	#define F_DEFAULT (9UL << S_MAXDEPTH)
20		45
21	#define INIT_SIZE 32 // initial scalar size to be allocated	46	#define INIT_SIZE 32 // initial scalar size to be allocated
		47	#define INDENT_STEP 3 // spaces per indentation level
		48
		49	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
22		50
23	#define SB do {	51	#define SB do {
24	#define SE } while (0)	52	#define SE } while (0)
25		53
26	static HV *json_stash;	54	#if __GNUC__ >= 3
		55	# define expect(expr,value) __builtin_expect ((expr),(value))
		56	# define inline inline
		57	#else
		58	# define expect(expr,value) (expr)
		59	# define inline static
		60	#endif
		61
		62	#define expect_false(expr) expect ((expr) != 0, 0)
		63	#define expect_true(expr) expect ((expr) != 0, 1)
		64
		65	#ifdef USE_ITHREADS
		66	# define JSON_SLOW 1
		67	# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
		68	#else
		69	# define JSON_SLOW 0
		70	# define JSON_STASH json_stash
		71	#endif
		72
		73	static HV *json_stash, *json_boolean_stash; // JSON::XS::
		74	static SV *json_true, *json_false;
		75
		76	typedef struct {
		77	U32 flags;
		78	SV *cb_object;
		79	HV *cb_sk_object;
		80	} JSON;
		81
		82	/////////////////////////////////////////////////////////////////////////////
		83	// utility functions
		84
		85	inline void
		86	shrink (SV *sv)
		87	{
		88	sv_utf8_downgrade (sv, 1);
		89	if (SvLEN (sv) > SvCUR (sv) + 1)
		90	{
		91	#ifdef SvPV_shrink_to_cur
		92	SvPV_shrink_to_cur (sv);
		93	#elif defined (SvPV_renew)
		94	SvPV_renew (sv, SvCUR (sv) + 1);
		95	#endif
		96	}
		97	}
		98
		99	// decode an utf-8 character and return it, or (UV)-1 in
		100	// case of an error.
		101	// we special-case "safe" characters from U+80 .. U+7FF,
		102	// but use the very good perl function to parse anything else.
		103	// note that we never call this function for a ascii codepoints
		104	inline UV
		105	decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
		106	{
		107	if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
		108	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
		109	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
		110	{
		111	*clen = 2;
		112	return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
		113	}
		114	else
		115	{
		116	*clen = (STRLEN)-1;
		117	return (UV)-1;
		118	}
		119	}
		120
		121	/////////////////////////////////////////////////////////////////////////////
		122	// encoder
27		123
28	// structure used for encoding JSON	124	// structure used for encoding JSON
29	typedef struct	125	typedef struct
30	{	126	{
31	char *cur;	127	char *cur; // SvPVX (sv) + current output position
32	STRLEN len; // SvLEN (sv)
33	char *end; // SvEND (sv)	128	char *end; // SvEND (sv)
34	SV *sv;	129	SV *sv; // result scalar
35	UV flags;	130	JSON json;
36	int max_recurse;	131	U32 indent; // indentation level
37	int indent;	132	U32 maxdepth; // max. indentation/recursion level
38	} enc_t;	133	} enc_t;
39		134
40	// structure used for decoding JSON	135	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	static void
59	shrink (SV *sv)
60	{
61	sv_utf8_downgrade (sv, 1);
62	#ifdef SvPV_shrink_to_cur
63	SvPV_shrink_to_cur (sv);
64	#endif
65	}
66
67	/////////////////////////////////////////////////////////////////////////////
68
69	static void
70	need (enc_t *enc, STRLEN len)	136	need (enc_t *enc, STRLEN len)
71	{	137	{
72	if (enc->cur + len >= enc->end)	138	if (expect_false (enc->cur + len >= enc->end))
73	{	139	{
74	STRLEN cur = enc->cur - SvPVX (enc->sv);	140	STRLEN cur = enc->cur - SvPVX (enc->sv);
75	SvGROW (enc->sv, cur + len + 1);	141	SvGROW (enc->sv, cur + len + 1);
76	enc->cur = SvPVX (enc->sv) + cur;	142	enc->cur = SvPVX (enc->sv) + cur;
77	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);	143	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
78	}	144	}
79	}	145	}
80		146
81	static void	147	inline void
82	encode_ch (enc_t *enc, char ch)	148	encode_ch (enc_t *enc, char ch)
83	{	149	{
84	need (enc, 1);	150	need (enc, 1);
85	*enc->cur++ = ch;	151	*enc->cur++ = ch;
86	}	152	}
…		…
94		160
95	while (str < end)	161	while (str < end)
96	{	162	{
97	unsigned char ch = *(unsigned char *)str;	163	unsigned char ch = *(unsigned char *)str;
98		164
99	if (ch >= 0x20 && ch < 0x80) // most common case	165	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
100	{	166	{
101	if (ch == '"') // but with slow exceptions	167	if (expect_false (ch == '"')) // but with slow exceptions
102	{	168	{
103	need (enc, len += 1);	169	need (enc, len += 1);
104	*enc->cur++ = '\\';	170	*enc->cur++ = '\\';
105	*enc->cur++ = '"';	171	*enc->cur++ = '"';
106	}	172	}
107	else if (ch == '\\')	173	else if (expect_false (ch == '\\'))
108	{	174	{
109	need (enc, len += 1);	175	need (enc, len += 1);
110	*enc->cur++ = '\\';	176	*enc->cur++ = '\\';
111	*enc->cur++ = '\\';	177	*enc->cur++ = '\\';
112	}	178	}
…		…
130	STRLEN clen;	196	STRLEN clen;
131	UV uch;	197	UV uch;
132		198
133	if (is_utf8)	199	if (is_utf8)
134	{	200	{
135	uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);	201	uch = decode_utf8 (str, end - str, &clen);
136	if (clen == (STRLEN)-1)	202	if (clen == (STRLEN)-1)
137	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);	203	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
138	}	204	}
139	else	205	else
140	{	206	{
…		…
143	}	209	}
144		210
145	if (uch > 0x10FFFFUL)	211	if (uch > 0x10FFFFUL)
146	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);	212	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
147		213
148	if (uch < 0x80 || enc->flags & F_ASCII)	214	if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
149	{	215	{
150	if (uch > 0xFFFFUL)	216	if (uch > 0xFFFFUL)
151	{	217	{
152	need (enc, len += 11);	218	need (enc, len += 11);
153	sprintf (enc->cur, "\\u%04x\\u%04x",	219	sprintf (enc->cur, "\\u%04x\\u%04x",
…		…
167	*enc->cur++ = hexdigit [(uch >> 0) & 15];	233	*enc->cur++ = hexdigit [(uch >> 0) & 15];
168	}	234	}
169		235
170	str += clen;	236	str += clen;
171	}	237	}
		238	else if (enc->json.flags & F_LATIN1)
		239	{
		240	*enc->cur++ = uch;
		241	str += clen;
		242	}
172	else if (is_utf8)	243	else if (is_utf8)
173	{	244	{
174	need (enc, len += clen);	245	need (enc, len += clen);
175	do	246	do
176	{	247	{
…		…
178	}	249	}
179	while (--clen);	250	while (--clen);
180	}	251	}
181	else	252	else
182	{	253	{
183	need (enc, len += 10); // never more than 11 bytes needed	254	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
184	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	255	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
185	++str;	256	++str;
186	}	257	}
187	}	258	}
188	}	259	}
…		…
190		261
191	--len;	262	--len;
192	}	263	}
193	}	264	}
194		265
195	#define INDENT SB \	266	inline void
		267	encode_indent (enc_t *enc)
		268	{
196	if (enc->flags & F_INDENT) \	269	if (enc->json.flags & F_INDENT)
197	{ \	270	{
198	int i_; \	271	int spaces = enc->indent * INDENT_STEP;
199	need (enc, enc->indent); \	272
200	for (i_ = enc->indent * 3; i_--; )\	273	need (enc, spaces);
		274	memset (enc->cur, ' ', spaces);
		275	enc->cur += spaces;
		276	}
		277	}
		278
		279	inline void
		280	encode_space (enc_t *enc)
		281	{
		282	need (enc, 1);
		283	encode_ch (enc, ' ');
		284	}
		285
		286	inline void
		287	encode_nl (enc_t *enc)
		288	{
		289	if (enc->json.flags & F_INDENT)
		290	{
		291	need (enc, 1);
201	encode_ch (enc, ' '); \	292	encode_ch (enc, '\n');
202	} \	293	}
203	SE	294	}
204		295
205	#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE	296	inline void
206	#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE	297	encode_comma (enc_t *enc)
207	#define COMMA SB \	298	{
208	encode_ch (enc, ','); \	299	encode_ch (enc, ',');
		300
209	if (enc->flags & F_INDENT) \	301	if (enc->json.flags & F_INDENT)
210	NL; \	302	encode_nl (enc);
211	else if (enc->flags & F_SPACE_AFTER) \	303	else if (enc->json.flags & F_SPACE_AFTER)
212	SPACE; \	304	encode_space (enc);
213	SE	305	}
214		306
215	static void encode_sv (enc_t *enc, SV *sv);	307	static void encode_sv (enc_t *enc, SV *sv);
216		308
217	static void	309	static void
218	encode_av (enc_t *enc, AV *av)	310	encode_av (enc_t *enc, AV *av)
219	{	311	{
220	int i, len = av_len (av);	312	int i, len = av_len (av);
221		313
222	encode_ch (enc, '['); NL;	314	if (enc->indent >= enc->maxdepth)
		315	croak ("data structure too deep (hit recursion limit)");
		316
		317	encode_ch (enc, '['); encode_nl (enc);
223	++enc->indent;	318	++enc->indent;
224		319
225	for (i = 0; i <= len; ++i)	320	for (i = 0; i <= len; ++i)
226	{	321	{
227	INDENT;	322	SV **svp = av_fetch (av, i, 0);
228	encode_sv (enc, *av_fetch (av, i, 0));	323
		324	encode_indent (enc);
		325
		326	if (svp)
		327	encode_sv (enc, *svp);
		328	else
		329	encode_str (enc, "null", 4, 0);
229		330
230	if (i < len)	331	if (i < len)
231	COMMA;	332	encode_comma (enc);
232	}	333	}
233		334
234	NL;	335	encode_nl (enc);
235		336
236	--enc->indent;	337	--enc->indent;
237	INDENT; encode_ch (enc, ']');	338	encode_indent (enc); encode_ch (enc, ']');
238	}	339	}
239		340
240	static void	341	static void
241	encode_he (enc_t *enc, HE *he)	342	encode_hk (enc_t *enc, HE *he)
242	{	343	{
243	encode_ch (enc, '"');	344	encode_ch (enc, '"');
244		345
245	if (HeKLEN (he) == HEf_SVKEY)	346	if (HeKLEN (he) == HEf_SVKEY)
246	{	347	{
…		…
256	else	357	else
257	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));	358	encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
258		359
259	encode_ch (enc, '"');	360	encode_ch (enc, '"');
260		361
261	if (enc->flags & F_SPACE_BEFORE) SPACE;	362	if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
262	encode_ch (enc, ':');	363	encode_ch (enc, ':');
263	if (enc->flags & F_SPACE_AFTER ) SPACE;	364	if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
264	encode_sv (enc, HeVAL (he));
265	}	365	}
266		366
267	// compare hash entries, used when all keys are bytestrings	367	// compare hash entries, used when all keys are bytestrings
268	static int	368	static int
269	he_cmp_fast (const void *a_, const void *b_)	369	he_cmp_fast (const void *a_, const void *b_)
…		…
274	HE *b = *(HE **)b_;	374	HE *b = *(HE **)b_;
275		375
276	STRLEN la = HeKLEN (a);	376	STRLEN la = HeKLEN (a);
277	STRLEN lb = HeKLEN (b);	377	STRLEN lb = HeKLEN (b);
278		378
279	if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))	379	if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
280	cmp = la - lb;	380	cmp = lb - la;
281		381
282	return cmp;	382	return cmp;
283	}	383	}
284		384
285	// compare hash entries, used when some keys are sv's or utf-x	385	// compare hash entries, used when some keys are sv's or utf-x
286	static int	386	static int
287	he_cmp_slow (const void *a, const void *b)	387	he_cmp_slow (const void *a, const void *b)
288	{	388	{
289	return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b));	389	return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
290	}	390	}
291		391
292	static void	392	static void
293	encode_hv (enc_t *enc, HV *hv)	393	encode_hv (enc_t *enc, HV *hv)
294	{	394	{
		395	HE *he;
295	int count, i;	396	int count;
296		397
		398	if (enc->indent >= enc->maxdepth)
		399	croak ("data structure too deep (hit recursion limit)");
		400
297	encode_ch (enc, '{'); NL; ++enc->indent;	401	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
298		402
299	if ((count = hv_iterinit (hv)))
300	{
301	// for canonical output we have to sort by keys first	403	// for canonical output we have to sort by keys first
302	// actually, this is mostly due to the stupid so-called	404	// actually, this is mostly due to the stupid so-called
303	// security workaround added somewhere in 5.8.x.	405	// security workaround added somewhere in 5.8.x.
304	// that randomises hash orderings	406	// that randomises hash orderings
305	if (enc->flags & F_CANONICAL)	407	if (enc->json.flags & F_CANONICAL)
		408	{
		409	int count = hv_iterinit (hv);
		410
		411	if (SvMAGICAL (hv))
306	{	412	{
307	HE *he, *hes [count];	413	// need to count by iterating. could improve by dynamically building the vector below
		414	// but I don't care for the speed of this special case.
		415	// note also that we will run into undefined behaviour when the two iterations
		416	// do not result in the same count, something I might care for in some later release.
		417
		418	count = 0;
		419	while (hv_iternext (hv))
		420	++count;
		421
		422	hv_iterinit (hv);
		423	}
		424
		425	if (count)
		426	{
308	int fast = 1;	427	int i, fast = 1;
		428	#if defined(__BORLANDC__) || defined(_MSC_VER)
		429	HE **hes = _alloca (count * sizeof (HE));
		430	#else
		431	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
		432	#endif
309		433
310	i = 0;	434	i = 0;
311	while ((he = hv_iternext (hv)))	435	while ((he = hv_iternext (hv)))
312	{	436	{
313	hes [i++] = he;	437	hes [i++] = he;
…		…
335		459
336	FREETMPS;	460	FREETMPS;
337	LEAVE;	461	LEAVE;
338	}	462	}
339		463
340	for (i = 0; i < count; ++i)	464	while (count--)
341	{
342	INDENT;
343	encode_he (enc, hes [i]);
344
345	if (i < count - 1)
346	COMMA;
347	}	465	{
		466	encode_indent (enc);
		467	he = hes [count];
		468	encode_hk (enc, he);
		469	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
348		470
		471	if (count)
		472	encode_comma (enc);
349	NL;	473	}
		474	}
		475	}
		476	else
		477	{
		478	if (hv_iterinit (hv) || SvMAGICAL (hv))
		479	if ((he = hv_iternext (hv)))
		480	for (;;)
		481	{
		482	encode_indent (enc);
		483	encode_hk (enc, he);
		484	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
		485
		486	if (!(he = hv_iternext (hv)))
		487	break;
		488
		489	encode_comma (enc);
		490	}
		491	}
		492
		493	encode_nl (enc);
		494
		495	--enc->indent; encode_indent (enc); encode_ch (enc, '}');
		496	}
		497
		498	// encode objects, arrays and special \0=false and \1=true values.
		499	static void
		500	encode_rv (enc_t *enc, SV *sv)
		501	{
		502	svtype svt;
		503
		504	SvGETMAGIC (sv);
		505	svt = SvTYPE (sv);
		506
		507	if (expect_false (SvOBJECT (sv)))
		508	{
		509	HV *stash = !JSON_SLOW || json_boolean_stash
		510	? json_boolean_stash
		511	: gv_stashpv ("JSON::XS::Boolean", 1);
		512
		513	if (SvSTASH (sv) == stash)
		514	{
		515	if (SvIV (sv))
		516	encode_str (enc, "true", 4, 0);
		517	else
		518	encode_str (enc, "false", 5, 0);
350	}	519	}
351	else	520	else
352	{	521	{
353	SV *sv;	522	#if 0
354	HE *he = hv_iternext (hv);	523	if (0 && sv_derived_from (rv, "JSON::Literal"))
355
356	for (;;)
357	{
358	INDENT;
359	encode_he (enc, he);
360
361	if (!(he = hv_iternext (hv)))
362	break;
363
364	COMMA;
365	}	524	{
366		525	// not yet
367	NL;	526	}
		527	#endif
		528	if (enc->json.flags & F_CONV_BLESSED)
		529	{
		530	// we re-bless the reference to get overload and other niceties right
		531	GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
		532
		533	if (to_json)
		534	{
		535	dSP;
		536
		537	ENTER; SAVETMPS; PUSHMARK (SP);
		538	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
		539
		540	// calling with G_SCALAR ensures that we always get a 1 return value
		541	PUTBACK;
		542	call_sv ((SV *)GvCV (to_json), G_SCALAR);
		543	SPAGAIN;
		544
		545	// catch this surprisingly common error
		546	if (SvROK (TOPs) && SvRV (TOPs) == sv)
		547	croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
		548
		549	sv = POPs;
		550	PUTBACK;
		551
		552	encode_sv (enc, sv);
		553
		554	FREETMPS; LEAVE;
		555	}
		556	else if (enc->json.flags & F_ALLOW_BLESSED)
		557	encode_str (enc, "null", 4, 0);
		558	else
		559	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
		560	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		561	}
		562	else if (enc->json.flags & F_ALLOW_BLESSED)
		563	encode_str (enc, "null", 4, 0);
		564	else
		565	croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
		566	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
368	}	567	}
369	}	568	}
		569	else if (svt == SVt_PVHV)
		570	encode_hv (enc, (HV *)sv);
		571	else if (svt == SVt_PVAV)
		572	encode_av (enc, (AV *)sv);
		573	else if (svt < SVt_PVAV)
		574	{
		575	STRLEN len = 0;
		576	char *pv = svt ? SvPV (sv, len) : 0;
370		577
371	--enc->indent; INDENT; encode_ch (enc, '}');	578	if (len == 1 && *pv == '1')
		579	encode_str (enc, "true", 4, 0);
		580	else if (len == 1 && *pv == '0')
		581	encode_str (enc, "false", 5, 0);
		582	else
		583	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
		584	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		585	}
		586	else
		587	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
		588	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
372	}	589	}
373		590
374	static void	591	static void
375	encode_sv (enc_t *enc, SV *sv)	592	encode_sv (enc_t *enc, SV *sv)
376	{	593	{
…		…
384	encode_str (enc, str, len, SvUTF8 (sv));	601	encode_str (enc, str, len, SvUTF8 (sv));
385	encode_ch (enc, '"');	602	encode_ch (enc, '"');
386	}	603	}
387	else if (SvNOKp (sv))	604	else if (SvNOKp (sv))
388	{	605	{
		606	// trust that perl will do the right thing w.r.t. JSON syntax.
389	need (enc, NV_DIG + 32);	607	need (enc, NV_DIG + 32);
390	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	608	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
391	enc->cur += strlen (enc->cur);	609	enc->cur += strlen (enc->cur);
392	}	610	}
393	else if (SvIOKp (sv))	611	else if (SvIOKp (sv))
394	{	612	{
395	need (enc, 64);	613	// we assume we can always read an IV as a UV
		614	if (SvUV (sv) & ~(UV)0x7fff)
		615	{
		616	// large integer, use the (rather slow) snprintf way.
		617	need (enc, sizeof (UV) * 3);
396	enc->cur +=	618	enc->cur +=
397	SvIsUV(sv)	619	SvIsUV(sv)
398	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	620	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
399	: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));	621	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
		622	}
		623	else
		624	{
		625	// optimise the "small number case"
		626	// code will likely be branchless and use only a single multiplication
		627	I32 i = SvIV (sv);
		628	U32 u;
		629	char digit, nz = 0;
		630
		631	need (enc, 6);
		632
		633	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
		634	u = i < 0 ? -i : i;
		635
		636	// convert to 4.28 fixed-point representation
		637	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
		638
		639	// now output digit by digit, each time masking out the integer part
		640	// and multiplying by 5 while moving the decimal point one to the right,
		641	// resulting in a net multiplication by 10.
		642	// we always write the digit to memory but conditionally increment
		643	// the pointer, to ease the usage of conditional move instructions.
		644	digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
		645	digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
		646	digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
		647	digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
		648	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
		649	}
400	}	650	}
401	else if (SvROK (sv))	651	else if (SvROK (sv))
402	{	652	encode_rv (enc, SvRV (sv));
403	SV *rv = SvRV (sv);
404
405	if (!--enc->max_recurse)
406	croak ("data structure too deep (hit recursion limit)");
407
408	switch (SvTYPE (rv))
409	{
410	case SVt_PVAV: encode_av (enc, (AV *)rv); break;
411	case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
412
413	default:
414	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
415	SvPV_nolen (sv));
416	}
417	}
418	else if (!SvOK (sv))	653	else if (!SvOK (sv))
419	encode_str (enc, "null", 4, 0);	654	encode_str (enc, "null", 4, 0);
420	else	655	else
421	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	656	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
422	SvPV_nolen (sv), SvFLAGS (sv));	657	SvPV_nolen (sv), SvFLAGS (sv));
423	}	658	}
424		659
425	static SV *	660	static SV *
426	encode_json (SV *scalar, UV flags)	661	encode_json (SV *scalar, JSON *json)
427	{	662	{
		663	enc_t enc;
		664
428	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	665	if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
429	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");	666	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
430		667
431	enc_t enc;	668	enc.json = *json;
432	enc.flags = flags;
433	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	669	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
434	enc.cur = SvPVX (enc.sv);	670	enc.cur = SvPVX (enc.sv);
435	enc.end = SvEND (enc.sv);	671	enc.end = SvEND (enc.sv);
436	enc.max_recurse = 0;
437	enc.indent = 0;	672	enc.indent = 0;
		673	enc.maxdepth = DEC_DEPTH (enc.json.flags);
438		674
439	SvPOK_only (enc.sv);	675	SvPOK_only (enc.sv);
440	encode_sv (&enc, scalar);	676	encode_sv (&enc, scalar);
441		677
		678	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
		679	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
		680
442	if (!(flags & (F_ASCII | F_UTF8)))	681	if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
443	SvUTF8_on (enc.sv);	682	SvUTF8_on (enc.sv);
444		683
445	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
446
447	if (enc.flags & F_SHRINK)	684	if (enc.json.flags & F_SHRINK)
448	shrink (enc.sv);	685	shrink (enc.sv);
449		686
450	return enc.sv;	687	return enc.sv;
451	}	688	}
452		689
453	/////////////////////////////////////////////////////////////////////////////	690	/////////////////////////////////////////////////////////////////////////////
		691	// decoder
454		692
455	#define WS \	693	// structure used for decoding JSON
		694	typedef struct
		695	{
		696	char *cur; // current parser pointer
		697	char *end; // end of input string
		698	const char *err; // parse error, if != 0
		699	JSON json;
		700	U32 depth; // recursion depth
		701	U32 maxdepth; // recursion depth limit
		702	} dec_t;
		703
		704	inline void
		705	decode_ws (dec_t *dec)
		706	{
456	for (;;) \	707	for (;;)
457	{ \	708	{
458	char ch = *dec->cur; \	709	char ch = *dec->cur;
		710
459	if (ch > 0x20 \	711	if (ch > 0x20
460	|| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \	712	|| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
461	break; \	713	break;
		714
		715	if (ch == '#' && dec->json.flags & F_RELAXED)
		716	++dec->cur;
		717
462	++dec->cur; \	718	++dec->cur;
463	}	719	}
		720	}
464		721
465	#define ERR(reason) SB dec->err = reason; goto fail; SE	722	#define ERR(reason) SB dec->err = reason; goto fail; SE
		723
466	#define EXPECT_CH(ch) SB \	724	#define EXPECT_CH(ch) SB \
467	if (*dec->cur != ch) \	725	if (*dec->cur != ch) \
468	ERR (# ch " expected"); \	726	ERR (# ch " expected"); \
469	++dec->cur; \	727	++dec->cur; \
470	SE	728	SE
471		729
		730	#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
		731	#define DEC_DEC_DEPTH --dec->depth
		732
472	static SV *decode_sv (dec_t *dec);	733	static SV *decode_sv (dec_t *dec);
473		734
474	static signed char decode_hexdigit[256];	735	static signed char decode_hexdigit[256];
475		736
476	static UV	737	static UV
477	decode_4hex (dec_t *dec)	738	decode_4hex (dec_t *dec)
478	{	739	{
479	signed char d1, d2, d3, d4;	740	signed char d1, d2, d3, d4;
		741	unsigned char *cur = (unsigned char *)dec->cur;
480		742
481	d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];	743	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
482	if (d1 < 0) ERR ("four hexadecimal digits expected");	744	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
483	d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];	745	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
484	if (d2 < 0) ERR ("four hexadecimal digits expected");	746	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
485	d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
486	if (d3 < 0) ERR ("four hexadecimal digits expected");
487	d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
488	if (d4 < 0) ERR ("four hexadecimal digits expected");
489		747
490	dec->cur += 4;	748	dec->cur += 4;
491		749
492	return ((UV)d1) << 12	750	return ((UV)d1) << 12
493	| ((UV)d2) << 8	751	| ((UV)d2) << 8
…		…
496		754
497	fail:	755	fail:
498	return (UV)-1;	756	return (UV)-1;
499	}	757	}
500		758
501	#define APPEND_GROW(n) SB \
502	if (cur + (n) >= end) \
503	{ \
504	STRLEN ofs = cur - SvPVX (sv); \
505	SvGROW (sv, ofs + (n) + 1); \
506	cur = SvPVX (sv) + ofs; \
507	end = SvEND (sv); \
508	} \
509	SE
510
511	#define APPEND_CH(ch) SB \
512	APPEND_GROW (1); \
513	*cur++ = (ch); \
514	SE
515
516	static SV *	759	static SV *
517	decode_str (dec_t *dec)	760	decode_str (dec_t *dec)
518	{	761	{
519	SV *sv = NEWSV (0,2);	762	SV *sv = 0;
520	int utf8 = 0;	763	int utf8 = 0;
521	char *cur = SvPVX (sv);	764	char *dec_cur = dec->cur;
522	char *end = SvEND (sv);
523		765
524	for (;;)	766	do
525	{	767	{
526	unsigned char ch = *(unsigned char *)dec->cur;	768	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
		769	char *cur = buf;
527		770
528	if (ch == '"')	771	do
529	break;
530	else if (ch == '\\')
531	{	772	{
532	switch (*++dec->cur)	773	unsigned char ch = *(unsigned char *)dec_cur++;
		774
		775	if (expect_false (ch == '"'))
		776	{
		777	--dec_cur;
		778	break;
533	{	779	}
534	case '\\':	780	else if (expect_false (ch == '\\'))
535	case '/':	781	{
536	case '"': APPEND_CH (*dec->cur++); break;	782	switch (*dec_cur)
537
538	case 'b': APPEND_CH ('\010'); ++dec->cur; break;
539	case 't': APPEND_CH ('\011'); ++dec->cur; break;
540	case 'n': APPEND_CH ('\012'); ++dec->cur; break;
541	case 'f': APPEND_CH ('\014'); ++dec->cur; break;
542	case 'r': APPEND_CH ('\015'); ++dec->cur; break;
543
544	case 'u':
545	{	783	{
546	UV lo, hi;	784	case '\\':
547	++dec->cur;	785	case '/':
		786	case '"': *cur++ = *dec_cur++; break;
548		787
549	hi = decode_4hex (dec);	788	case 'b': ++dec_cur; *cur++ = '\010'; break;
550	if (hi == (UV)-1)	789	case 't': ++dec_cur; *cur++ = '\011'; break;
551	goto fail;	790	case 'n': ++dec_cur; *cur++ = '\012'; break;
		791	case 'f': ++dec_cur; *cur++ = '\014'; break;
		792	case 'r': ++dec_cur; *cur++ = '\015'; break;
552		793
553	// possibly a surrogate pair	794	case 'u':
554	if (hi >= 0xd800 && hi < 0xdc00)
555	{	795	{
556	if (dec->cur [0] != '\\' || dec->cur [1] != 'u')	796	UV lo, hi;
557	ERR ("missing low surrogate character in surrogate pair");	797	++dec_cur;
558		798
559	dec->cur += 2;	799	dec->cur = dec_cur;
560
561	lo = decode_4hex (dec);	800	hi = decode_4hex (dec);
		801	dec_cur = dec->cur;
562	if (lo == (UV)-1)	802	if (hi == (UV)-1)
563	goto fail;	803	goto fail;
564		804
		805	// possibly a surrogate pair
		806	if (hi >= 0xd800)
		807	if (hi < 0xdc00)
		808	{
		809	if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
		810	ERR ("missing low surrogate character in surrogate pair");
		811
		812	dec_cur += 2;
		813
		814	dec->cur = dec_cur;
		815	lo = decode_4hex (dec);
		816	dec_cur = dec->cur;
		817	if (lo == (UV)-1)
		818	goto fail;
		819
565	if (lo < 0xdc00 || lo >= 0xe000)	820	if (lo < 0xdc00 || lo >= 0xe000)
566	ERR ("surrogate pair expected");	821	ERR ("surrogate pair expected");
567		822
568	hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;	823	hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
		824	}
		825	else if (hi < 0xe000)
		826	ERR ("missing high surrogate character in surrogate pair");
		827
		828	if (hi >= 0x80)
		829	{
		830	utf8 = 1;
		831
		832	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
		833	}
		834	else
		835	*cur++ = hi;
569	}	836	}
570	else if (hi >= 0xdc00 && hi < 0xe000)
571	ERR ("missing high surrogate character in surrogate pair");
572
573	if (hi >= 0x80)
574	{	837	break;
575	utf8 = 1;
576		838
577	APPEND_GROW (4); // at most 4 bytes for 21 bits
578	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
579	}
580	else	839	default:
581	APPEND_CH (hi);	840	--dec_cur;
		841	ERR ("illegal backslash escape sequence in string");
582	}	842	}
583	break;	843	}
		844	else if (expect_true (ch >= 0x20 && ch <= 0x7f))
		845	*cur++ = ch;
		846	else if (ch >= 0x80)
		847	{
		848	STRLEN clen;
		849	UV uch;
584		850
585	default:
586	--dec->cur;	851	--dec_cur;
587	ERR ("illegal backslash escape sequence in string");	852
		853	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
		854	if (clen == (STRLEN)-1)
		855	ERR ("malformed UTF-8 character in JSON string");
		856
		857	do
		858	*cur++ = *dec_cur++;
		859	while (--clen);
		860
		861	utf8 = 1;
		862	}
		863	else
		864	{
		865	--dec_cur;
		866
		867	if (!ch)
		868	ERR ("unexpected end of string while parsing JSON string");
		869	else
		870	ERR ("invalid character encountered while parsing JSON string");
588	}	871	}
589	}	872	}
590	else if (ch >= 0x20 && ch <= 0x7f)	873	while (cur < buf + SHORT_STRING_LEN);
591	APPEND_CH (*dec->cur++);	874
592	else if (ch >= 0x80)
593	{	875	{
594	STRLEN clen;	876	STRLEN len = cur - buf;
595	UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
596	if (clen == (STRLEN)-1)
597	ERR ("malformed UTF-8 character in JSON string");
598		877
599	APPEND_GROW (clen);	878	if (sv)
600	do
601	{	879	{
602	*cur++ = *dec->cur++;	880	SvGROW (sv, SvCUR (sv) + len + 1);
		881	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
		882	SvCUR_set (sv, SvCUR (sv) + len);
603	}	883	}
604	while (--clen);
605
606	utf8 = 1;
607	}
608	else if (dec->cur == dec->end)
609	ERR ("unexpected end of string while parsing json string");
610	else	884	else
611	ERR ("invalid character encountered");	885	sv = newSVpvn (buf, len);
612	}	886	}
		887	}
		888	while (*dec_cur != '"');
613		889
614	++dec->cur;	890	++dec_cur;
615		891
616	SvCUR_set (sv, cur - SvPVX (sv));	892	if (sv)
617		893	{
618	SvPOK_only (sv);	894	SvPOK_only (sv);
619	*SvEND (sv) = 0;	895	*SvEND (sv) = 0;
620		896
621	if (utf8)	897	if (utf8)
622	SvUTF8_on (sv);	898	SvUTF8_on (sv);
		899	}
		900	else
		901	sv = newSVpvn ("", 0);
623		902
624	if (dec->flags & F_SHRINK)	903	dec->cur = dec_cur;
625	shrink (sv);
626
627	return sv;	904	return sv;
628		905
629	fail:	906	fail:
630	SvREFCNT_dec (sv);	907	dec->cur = dec_cur;
631	return 0;	908	return 0;
632	}	909	}
633		910
634	static SV *	911	static SV *
635	decode_num (dec_t *dec)	912	decode_num (dec_t *dec)
…		…
693	is_nv = 1;	970	is_nv = 1;
694	}	971	}
695		972
696	if (!is_nv)	973	if (!is_nv)
697	{	974	{
698	UV uv;	975	int len = dec->cur - start;
699	int numtype = grok_number (start, dec->cur - start, &uv);	976
700	if (numtype & IS_NUMBER_IN_UV)	977	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
701	if (numtype & IS_NUMBER_NEG)	978	if (*start == '-')
		979	switch (len)
702	{	980	{
703	if (uv < (UV)IV_MIN)	981	case 2: return newSViv (-( start [1] - '0' * 1));
704	return newSViv (-(IV)uv);	982	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
		983	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
		984	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
705	}	985	}
		986	else
		987	switch (len)
		988	{
		989	case 1: return newSViv ( start [0] - '0' * 1);
		990	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
		991	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
		992	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		993	}
		994
		995	{
		996	UV uv;
		997	int numtype = grok_number (start, len, &uv);
		998	if (numtype & IS_NUMBER_IN_UV)
		999	if (numtype & IS_NUMBER_NEG)
		1000	{
		1001	if (uv < (UV)IV_MIN)
		1002	return newSViv (-(IV)uv);
		1003	}
706	else	1004	else
707	return newSVuv (uv);	1005	return newSVuv (uv);
708	}	1006	}
709		1007
		1008	len -= *start == '-' ? 1 : 0;
		1009
		1010	// does not fit into IV or UV, try NV
		1011	if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
		1012	#if defined (LDBL_DIG)
		1013	|| (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
		1014	#endif
		1015	)
		1016	// fits into NV without loss of precision
		1017	return newSVnv (Atof (start));
		1018
		1019	// everything else fails, convert it to a string
		1020	return newSVpvn (start, dec->cur - start);
		1021	}
		1022
		1023	// loss of precision here
710	return newSVnv (Atof (start));	1024	return newSVnv (Atof (start));
711		1025
712	fail:	1026	fail:
713	return 0;	1027	return 0;
714	}	1028	}
…		…
716	static SV *	1030	static SV *
717	decode_av (dec_t *dec)	1031	decode_av (dec_t *dec)
718	{	1032	{
719	AV *av = newAV ();	1033	AV *av = newAV ();
720		1034
721	WS;	1035	DEC_INC_DEPTH;
		1036	decode_ws (dec);
		1037
722	if (*dec->cur == ']')	1038	if (*dec->cur == ']')
723	++dec->cur;	1039	++dec->cur;
724	else	1040	else
725	for (;;)	1041	for (;;)
726	{	1042	{
…		…
730	if (!value)	1046	if (!value)
731	goto fail;	1047	goto fail;
732		1048
733	av_push (av, value);	1049	av_push (av, value);
734		1050
735	WS;	1051	decode_ws (dec);
736		1052
737	if (*dec->cur == ']')	1053	if (*dec->cur == ']')
738	{	1054	{
739	++dec->cur;	1055	++dec->cur;
740	break;	1056	break;
…		…
742		1058
743	if (*dec->cur != ',')	1059	if (*dec->cur != ',')
744	ERR (", or ] expected while parsing array");	1060	ERR (", or ] expected while parsing array");
745		1061
746	++dec->cur;	1062	++dec->cur;
		1063
		1064	decode_ws (dec);
		1065
		1066	if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
		1067	{
		1068	++dec->cur;
		1069	break;
		1070	}
747	}	1071	}
748		1072
		1073	DEC_DEC_DEPTH;
749	return newRV_noinc ((SV *)av);	1074	return newRV_noinc ((SV *)av);
750		1075
751	fail:	1076	fail:
752	SvREFCNT_dec (av);	1077	SvREFCNT_dec (av);
		1078	DEC_DEC_DEPTH;
753	return 0;	1079	return 0;
754	}	1080	}
755		1081
756	static SV *	1082	static SV *
757	decode_hv (dec_t *dec)	1083	decode_hv (dec_t *dec)
758	{	1084	{
		1085	SV *sv;
759	HV *hv = newHV ();	1086	HV *hv = newHV ();
760		1087
761	WS;	1088	DEC_INC_DEPTH;
		1089	decode_ws (dec);
		1090
762	if (*dec->cur == '}')	1091	if (*dec->cur == '}')
763	++dec->cur;	1092	++dec->cur;
764	else	1093	else
765	for (;;)	1094	for (;;)
766	{	1095	{
767	SV *key, *value;
768
769	WS; EXPECT_CH ('"');	1096	EXPECT_CH ('"');
770		1097
771	key = decode_str (dec);	1098	// heuristic: assume that
772	if (!key)	1099	// a) decode_str + hv_store_ent are abysmally slow.
773	goto fail;	1100	// b) most hash keys are short, simple ascii text.
		1101	// => try to "fast-match" such strings to avoid
		1102	// the overhead of decode_str + hv_store_ent.
		1103	{
		1104	SV *value;
		1105	char *p = dec->cur;
		1106	char *e = p + 24; // only try up to 24 bytes
774		1107
775	WS; EXPECT_CH (':');	1108	for (;;)
776
777	value = decode_sv (dec);
778	if (!value)
779	{	1109	{
		1110	// the >= 0x80 is true on most architectures
		1111	if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
		1112	{
		1113	// slow path, back up and use decode_str
		1114	SV *key = decode_str (dec);
		1115	if (!key)
		1116	goto fail;
		1117
		1118	decode_ws (dec); EXPECT_CH (':');
		1119
		1120	decode_ws (dec);
		1121	value = decode_sv (dec);
		1122	if (!value)
		1123	{
		1124	SvREFCNT_dec (key);
		1125	goto fail;
		1126	}
		1127
		1128	hv_store_ent (hv, key, value, 0);
780	SvREFCNT_dec (key);	1129	SvREFCNT_dec (key);
		1130
		1131	break;
		1132	}
		1133	else if (*p == '"')
		1134	{
		1135	// fast path, got a simple key
		1136	char *key = dec->cur;
		1137	int len = p - key;
		1138	dec->cur = p + 1;
		1139
		1140	decode_ws (dec); EXPECT_CH (':');
		1141
		1142	decode_ws (dec);
		1143	value = decode_sv (dec);
		1144	if (!value)
781	goto fail;	1145	goto fail;
		1146
		1147	hv_store (hv, key, len, value, 0);
		1148
		1149	break;
		1150	}
		1151
		1152	++p;
782	}	1153	}
783
784	//TODO: optimise
785	hv_store_ent (hv, key, value, 0);
786
787	WS;	1154	}
		1155
		1156	decode_ws (dec);
788		1157
789	if (*dec->cur == '}')	1158	if (*dec->cur == '}')
790	{	1159	{
791	++dec->cur;	1160	++dec->cur;
792	break;	1161	break;
…		…
794		1163
795	if (*dec->cur != ',')	1164	if (*dec->cur != ',')
796	ERR (", or } expected while parsing object/hash");	1165	ERR (", or } expected while parsing object/hash");
797		1166
798	++dec->cur;	1167	++dec->cur;
		1168
		1169	decode_ws (dec);
		1170
		1171	if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
		1172	{
		1173	++dec->cur;
		1174	break;
		1175	}
799	}	1176	}
800		1177
		1178	DEC_DEC_DEPTH;
801	return newRV_noinc ((SV *)hv);	1179	sv = newRV_noinc ((SV *)hv);
		1180
		1181	// check filter callbacks
		1182	if (dec->json.flags & F_HOOK)
		1183	{
		1184	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
		1185	{
		1186	HE *cb, *he;
		1187
		1188	hv_iterinit (hv);
		1189	he = hv_iternext (hv);
		1190	hv_iterinit (hv);
		1191
		1192	// the next line creates a mortal sv each time its called.
		1193	// might want to optimise this for common cases.
		1194	cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
		1195
		1196	if (cb)
		1197	{
		1198	dSP;
		1199	int count;
		1200
		1201	ENTER; SAVETMPS; PUSHMARK (SP);
		1202	XPUSHs (HeVAL (he));
		1203
		1204	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
		1205
		1206	if (count == 1)
		1207	{
		1208	sv = newSVsv (POPs);
		1209	FREETMPS; LEAVE;
		1210	return sv;
		1211	}
		1212
		1213	FREETMPS; LEAVE;
		1214	}
		1215	}
		1216
		1217	if (dec->json.cb_object)
		1218	{
		1219	dSP;
		1220	int count;
		1221
		1222	ENTER; SAVETMPS; PUSHMARK (SP);
		1223	XPUSHs (sv_2mortal (sv));
		1224
		1225	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
		1226
		1227	if (count == 1)
		1228	{
		1229	sv = newSVsv (POPs);
		1230	FREETMPS; LEAVE;
		1231	return sv;
		1232	}
		1233
		1234	SvREFCNT_inc (sv);
		1235	FREETMPS; LEAVE;
		1236	}
		1237	}
		1238
		1239	return sv;
802		1240
803	fail:	1241	fail:
804	SvREFCNT_dec (hv);	1242	SvREFCNT_dec (hv);
		1243	DEC_DEC_DEPTH;
805	return 0;	1244	return 0;
806	}	1245	}
807		1246
808	static SV *	1247	static SV *
809	decode_sv (dec_t *dec)	1248	decode_sv (dec_t *dec)
810	{	1249	{
811	WS;	1250	// the beauty of JSON: you need exactly one character lookahead
		1251	// to parse anything.
812	switch (*dec->cur)	1252	switch (*dec->cur)
813	{	1253	{
814	case '"': ++dec->cur; return decode_str (dec);	1254	case '"': ++dec->cur; return decode_str (dec);
815	case '[': ++dec->cur; return decode_av (dec);	1255	case '[': ++dec->cur; return decode_av (dec);
816	case '{': ++dec->cur; return decode_hv (dec);	1256	case '{': ++dec->cur; return decode_hv (dec);
…		…
822		1262
823	case 't':	1263	case 't':
824	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1264	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
825	{	1265	{
826	dec->cur += 4;	1266	dec->cur += 4;
827	return newSViv (1);	1267	#if JSON_SLOW
		1268	json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
		1269	#endif
		1270	return SvREFCNT_inc (json_true);
828	}	1271	}
829	else	1272	else
830	ERR ("'true' expected");	1273	ERR ("'true' expected");
831		1274
832	break;	1275	break;
833		1276
834	case 'f':	1277	case 'f':
835	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1278	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
836	{	1279	{
837	dec->cur += 5;	1280	dec->cur += 5;
838	return newSViv (0);	1281	#if JSON_SLOW
		1282	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
		1283	#endif
		1284	return SvREFCNT_inc (json_false);
839	}	1285	}
840	else	1286	else
841	ERR ("'false' expected");	1287	ERR ("'false' expected");
842		1288
843	break;	1289	break;
…		…
852	ERR ("'null' expected");	1298	ERR ("'null' expected");
853		1299
854	break;	1300	break;
855		1301
856	default:	1302	default:
857	ERR ("malformed json string, neither array, object, number, string or atom");	1303	ERR ("malformed JSON string, neither array, object, number, string or atom");
858	break;	1304	break;
859	}	1305	}
860		1306
861	fail:	1307	fail:
862	return 0;	1308	return 0;
863	}	1309	}
864		1310
865	static SV *	1311	static SV *
866	decode_json (SV *string, UV flags)	1312	decode_json (SV *string, JSON *json, UV *offset_return)
867	{	1313	{
		1314	dec_t dec;
		1315	UV offset;
868	SV *sv;	1316	SV *sv;
869		1317
		1318	SvGETMAGIC (string);
		1319	SvUPGRADE (string, SVt_PV);
		1320
		1321	if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
		1322	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1323	(unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
		1324
870	if (flags & F_UTF8)	1325	if (json->flags & F_UTF8)
871	sv_utf8_downgrade (string, 0);	1326	sv_utf8_downgrade (string, 0);
872	else	1327	else
873	sv_utf8_upgrade (string);	1328	sv_utf8_upgrade (string);
874		1329
875	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1330	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
876		1331
877	dec_t dec;	1332	dec.json = *json;
878	dec.flags = flags;
879	dec.cur = SvPVX (string);	1333	dec.cur = SvPVX (string);
880	dec.end = SvEND (string);	1334	dec.end = SvEND (string);
881	dec.err = 0;	1335	dec.err = 0;
		1336	dec.depth = 0;
		1337	dec.maxdepth = DEC_DEPTH (dec.json.flags);
882		1338
		1339	if (dec.json.cb_object || dec.json.cb_sk_object)
		1340	dec.json.flags |= F_HOOK;
		1341
		1342	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
		1343
		1344	decode_ws (&dec);
883	sv = decode_sv (&dec);	1345	sv = decode_sv (&dec);
884		1346
		1347	if (!(offset_return || !sv))
		1348	{
		1349	// check for trailing garbage
		1350	decode_ws (&dec);
		1351
		1352	if (*dec.cur)
		1353	{
		1354	dec.err = "garbage after JSON object";
		1355	SvREFCNT_dec (sv);
		1356	sv = 0;
		1357	}
		1358	}
		1359
		1360	if (offset_return || !sv)
		1361	{
		1362	offset = dec.json.flags & F_UTF8
		1363	? dec.cur - SvPVX (string)
		1364	: utf8_distance (dec.cur, SvPVX (string));
		1365
		1366	if (offset_return)
		1367	*offset_return = offset;
		1368	}
		1369
885	if (!sv)	1370	if (!sv)
886	{	1371	{
887	IV offset = dec.flags & F_UTF8
888	? dec.cur - SvPVX (string)
889	: utf8_distance (dec.cur, SvPVX (string));
890	SV *uni = sv_newmortal ();	1372	SV *uni = sv_newmortal ();
891		1373
892	// horrible hack to silence warning inside pv_uni_display	1374	// horrible hack to silence warning inside pv_uni_display
893	COP cop = *PL_curcop;	1375	COP cop = *PL_curcop;
894	cop.cop_warnings = pWARN_NONE;	1376	cop.cop_warnings = pWARN_NONE;
…		…
896	SAVEVPTR (PL_curcop);	1378	SAVEVPTR (PL_curcop);
897	PL_curcop = &cop;	1379	PL_curcop = &cop;
898	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1380	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
899	LEAVE;	1381	LEAVE;
900		1382
901	croak ("%s, at character offset %d (%s)",	1383	croak ("%s, at character offset %d [\"%s\"]",
902	dec.err,	1384	dec.err,
903	(int)offset,	1385	(int)offset,
904	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1386	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
905	}	1387	}
906		1388
907	sv = sv_2mortal (sv);	1389	sv = sv_2mortal (sv);
908		1390
909	if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))	1391	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
910	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");	1392	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
911		1393
912	return sv;	1394	return sv;
913	}	1395	}
914		1396
		1397	/////////////////////////////////////////////////////////////////////////////
		1398	// XS interface functions
		1399
915	MODULE = JSON::XS PACKAGE = JSON::XS	1400	MODULE = JSON::XS PACKAGE = JSON::XS
916		1401
917	BOOT:	1402	BOOT:
918	{	1403	{
919	int i;	1404	int i;
920		1405
921	memset (decode_hexdigit, 0xff, 256);
922	for (i = 10; i--; )	1406	for (i = 0; i < 256; ++i)
923	decode_hexdigit ['0' + i] = i;	1407	decode_hexdigit [i] =
		1408	i >= '0' && i <= '9' ? i - '0'
		1409	: i >= 'a' && i <= 'f' ? i - 'a' + 10
		1410	: i >= 'A' && i <= 'F' ? i - 'A' + 10
		1411	: -1;
924		1412
925	for (i = 7; i--; )	1413	json_stash = gv_stashpv ("JSON::XS" , 1);
		1414	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
		1415
		1416	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
		1417	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
		1418	}
		1419
		1420	PROTOTYPES: DISABLE
		1421
		1422	void CLONE (...)
		1423	CODE:
		1424	json_stash = 0;
		1425	json_boolean_stash = 0;
		1426
		1427	void new (char *klass)
		1428	PPCODE:
		1429	{
		1430	SV *pv = NEWSV (0, sizeof (JSON));
		1431	SvPOK_only (pv);
		1432	Zero (SvPVX (pv), 1, JSON);
		1433	((JSON *)SvPVX (pv))->flags = F_DEFAULT;
		1434	XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));
		1435	}
		1436
		1437	void ascii (JSON *self, int enable = 1)
		1438	ALIAS:
		1439	ascii = F_ASCII
		1440	latin1 = F_LATIN1
		1441	utf8 = F_UTF8
		1442	indent = F_INDENT
		1443	canonical = F_CANONICAL
		1444	space_before = F_SPACE_BEFORE
		1445	space_after = F_SPACE_AFTER
		1446	pretty = F_PRETTY
		1447	allow_nonref = F_ALLOW_NONREF
		1448	shrink = F_SHRINK
		1449	allow_blessed = F_ALLOW_BLESSED
		1450	convert_blessed = F_CONV_BLESSED
		1451	relaxed = F_RELAXED
		1452	PPCODE:
		1453	{
		1454	if (enable)
		1455	self->flags |= ix;
		1456	else
		1457	self->flags &= ~ix;
		1458
		1459	XPUSHs (ST (0));
		1460	}
		1461
		1462	void max_depth (JSON *self, UV max_depth = 0x80000000UL)
		1463	PPCODE:
		1464	{
		1465	UV log2 = 0;
		1466
		1467	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
		1468
		1469	while ((1UL << log2) < max_depth)
		1470	++log2;
		1471
		1472	self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
		1473
		1474	XPUSHs (ST (0));
		1475	}
		1476
		1477	void max_size (JSON *self, UV max_size = 0)
		1478	PPCODE:
		1479	{
		1480	UV log2 = 0;
		1481
		1482	if (max_size > 0x80000000UL) max_size = 0x80000000UL;
		1483	if (max_size == 1) max_size = 2;
		1484
		1485	while ((1UL << log2) < max_size)
		1486	++log2;
		1487
		1488	self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
		1489
		1490	XPUSHs (ST (0));
		1491	}
		1492
		1493	void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
		1494	PPCODE:
		1495	{
		1496	SvREFCNT_dec (self->cb_object);
		1497	self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
		1498
		1499	XPUSHs (ST (0));
		1500	}
		1501
		1502	void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
		1503	PPCODE:
		1504	{
		1505	if (!self->cb_sk_object)
		1506	self->cb_sk_object = newHV ();
		1507
		1508	if (SvOK (cb))
		1509	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
		1510	else
926	{	1511	{
927	decode_hexdigit ['a' + i] = 10 + i;	1512	hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
928	decode_hexdigit ['A' + i] = 10 + i;	1513
		1514	if (!HvKEYS (self->cb_sk_object))
		1515	{
		1516	SvREFCNT_dec (self->cb_sk_object);
		1517	self->cb_sk_object = 0;
		1518	}
929	}	1519	}
930		1520
931	json_stash = gv_stashpv ("JSON::XS", 1);	1521	XPUSHs (ST (0));
932	}	1522	}
933		1523
934	PROTOTYPES: DISABLE	1524	void encode (JSON *self, SV *scalar)
		1525	PPCODE:
		1526	XPUSHs (encode_json (scalar, self));
935		1527
936	SV *new (char *dummy)	1528	void decode (JSON *self, SV *jsonstr)
		1529	PPCODE:
		1530	XPUSHs (decode_json (jsonstr, self, 0));
		1531
		1532	void decode_prefix (JSON *self, SV *jsonstr)
		1533	PPCODE:
		1534	{
		1535	UV offset;
		1536	EXTEND (SP, 2);
		1537	PUSHs (decode_json (jsonstr, self, &offset));
		1538	PUSHs (sv_2mortal (newSVuv (offset)));
		1539	}
		1540
		1541	void DESTROY (JSON *self)
937	CODE:	1542	CODE:
938	RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);	1543	SvREFCNT_dec (self->cb_sk_object);
939	OUTPUT:	1544	SvREFCNT_dec (self->cb_object);
940	RETVAL
941
942	SV *ascii (SV *self, int enable = 1)
943	ALIAS:
944	ascii = F_ASCII
945	utf8 = F_UTF8
946	indent = F_INDENT
947	canonical = F_CANONICAL
948	space_before = F_SPACE_BEFORE
949	space_after = F_SPACE_AFTER
950	pretty = F_PRETTY
951	allow_nonref = F_ALLOW_NONREF
952	shrink = F_SHRINK
953	CODE:
954	{
955	UV *uv = SvJSON (self);
956	if (enable)
957	*uv |= ix;
958	else
959	*uv &= ~ix;
960
961	RETVAL = newSVsv (self);
962	}
963	OUTPUT:
964	RETVAL
965
966	void encode (SV *self, SV *scalar)
967	PPCODE:
968	XPUSHs (encode_json (scalar, *SvJSON (self)));
969
970	void decode (SV *self, SV *jsonstr)
971	PPCODE:
972	XPUSHs (decode_json (jsonstr, *SvJSON (self)));
973		1545
974	PROTOTYPES: ENABLE	1546	PROTOTYPES: ENABLE
975		1547
976	void to_json (SV *scalar)	1548	void to_json (SV *scalar)
977	PPCODE:	1549	PPCODE:
		1550	{
		1551	JSON json = { F_DEFAULT | F_UTF8 };
978	XPUSHs (encode_json (scalar, F_UTF8));	1552	XPUSHs (encode_json (scalar, &json));
		1553	}
979		1554
980	void from_json (SV *jsonstr)	1555	void from_json (SV *jsonstr)
981	PPCODE:	1556	PPCODE:
		1557	{
		1558	JSON json = { F_DEFAULT | F_UTF8 };
982	XPUSHs (decode_json (jsonstr, F_UTF8));	1559	XPUSHs (decode_json (jsonstr, &json, 0));
		1560	}
983		1561

Diff Legend

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