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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.58 by root, Mon Aug 13 16:06:25 2007 UTC vs.
Revision 1.84 by root, Thu Mar 27 06:37:35 2008 UTC

…		…
4		4
5	#include <assert.h>	5	#include <assert.h>
6	#include <string.h>	6	#include <string.h>
7	#include <stdlib.h>	7	#include <stdlib.h>
8	#include <stdio.h>	8	#include <stdio.h>
		9	#include <limits.h>
9	#include <float.h>	10	#include <float.h>
10		11
11	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	12	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
12	# define snprintf _snprintf // C compilers have this in stdio.h	13	# define snprintf _snprintf // C compilers have this in stdio.h
13	#endif	14	#endif
…		…
15	// some old perls do not have this, try to make it work, no	16	// 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	// guarentees, though. if it breaks, you get to keep the pieces.
17	#ifndef UTF8_MAXBYTES	18	#ifndef UTF8_MAXBYTES
18	# define UTF8_MAXBYTES 13	19	# define UTF8_MAXBYTES 13
19	#endif	20	#endif
		21
		22	#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
20		23
21	#define F_ASCII 0x00000001UL	24	#define F_ASCII 0x00000001UL
22	#define F_LATIN1 0x00000002UL	25	#define F_LATIN1 0x00000002UL
23	#define F_UTF8 0x00000004UL	26	#define F_UTF8 0x00000004UL
24	#define F_INDENT 0x00000008UL	27	#define F_INDENT 0x00000008UL
…		…
27	#define F_SPACE_AFTER 0x00000040UL	30	#define F_SPACE_AFTER 0x00000040UL
28	#define F_ALLOW_NONREF 0x00000100UL	31	#define F_ALLOW_NONREF 0x00000100UL
29	#define F_SHRINK 0x00000200UL	32	#define F_SHRINK 0x00000200UL
30	#define F_ALLOW_BLESSED 0x00000400UL	33	#define F_ALLOW_BLESSED 0x00000400UL
31	#define F_CONV_BLESSED 0x00000800UL	34	#define F_CONV_BLESSED 0x00000800UL
		35	#define F_RELAXED 0x00001000UL
		36	#define F_ALLOW_UNKNOWN 0x00002000UL
		37
32	#define F_MAXDEPTH 0xf8000000UL	38	#define F_MAXDEPTH 0xf8000000UL
33	#define S_MAXDEPTH 27	39	#define S_MAXDEPTH 27
34	#define F_MAXSIZE 0x01f00000UL	40	#define F_MAXSIZE 0x01f00000UL
35	#define S_MAXSIZE 20	41	#define S_MAXSIZE 20
36	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing	42	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
…		…
48		54
49	#define SB do {	55	#define SB do {
50	#define SE } while (0)	56	#define SE } while (0)
51		57
52	#if __GNUC__ >= 3	58	#if __GNUC__ >= 3
53	# define expect(expr,value) __builtin_expect ((expr),(value))	59	# define expect(expr,value) __builtin_expect ((expr), (value))
54	# define inline inline	60	# define INLINE static inline
55	#else	61	#else
56	# define expect(expr,value) (expr)	62	# define expect(expr,value) (expr)
57	# define inline static	63	# define INLINE static
58	#endif	64	#endif
59		65
60	#define expect_false(expr) expect ((expr) != 0, 0)	66	#define expect_false(expr) expect ((expr) != 0, 0)
61	#define expect_true(expr) expect ((expr) != 0, 1)	67	#define expect_true(expr) expect ((expr) != 0, 1)
62		68
		69	#define IN_RANGE_INC(type,val,beg,end) \
		70	((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
		71	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
		72
63	#ifdef USE_ITHREADS	73	#ifdef USE_ITHREADS
64	# define JSON_SLOW 1	74	# define JSON_SLOW 1
		75	# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
65	#else	76	#else
66	# define JSON_SLOW 0	77	# define JSON_SLOW 0
		78	# define JSON_STASH json_stash
67	#endif	79	#endif
68		80
69	static HV json_stash, json_boolean_stash; // JSON::XS::	81	static HV json_stash, json_boolean_stash; // JSON::XS::
70	static SV json_true, json_false;	82	static SV json_true, json_false;
		83
		84	enum {
		85	INCR_M_WS = 0, // initial whitespace skipping, must be 0
		86	INCR_M_STR, // inside string
		87	INCR_M_BS, // inside backslash
		88	INCR_M_JSON // outside anything, count nesting
		89	};
		90
		91	#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON)
71		92
72	typedef struct {	93	typedef struct {
73	U32 flags;	94	U32 flags;
74	SV *cb_object;	95	SV *cb_object;
75	HV *cb_sk_object;	96	HV *cb_sk_object;
		97
		98	// for the incremental parser
		99	SV *incr_text; // the source text so far
		100	STRLEN incr_pos; // the current offset into the text
		101	int incr_nest; // {[]}-nesting level
		102	int incr_mode;
76	} JSON;	103	} JSON;
77		104
78	/////////////////////////////////////////////////////////////////////////////	105	/////////////////////////////////////////////////////////////////////////////
79	// utility functions	106	// utility functions
80		107
81	inline void	108	INLINE SV *
		109	get_bool (const char *name)
		110	{
		111	SV *sv = get_sv (name, 1);
		112
		113	SvREADONLY_on (sv);
		114	SvREADONLY_on (SvRV (sv));
		115
		116	return sv;
		117	}
		118
		119	INLINE void
82	shrink (SV *sv)	120	shrink (SV *sv)
83	{	121	{
84	sv_utf8_downgrade (sv, 1);	122	sv_utf8_downgrade (sv, 1);
85	if (SvLEN (sv) > SvCUR (sv) + 1)	123	if (SvLEN (sv) > SvCUR (sv) + 1)
86	{	124	{
…		…
95	// decode an utf-8 character and return it, or (UV)-1 in	133	// decode an utf-8 character and return it, or (UV)-1 in
96	// case of an error.	134	// case of an error.
97	// we special-case "safe" characters from U+80 .. U+7FF,	135	// we special-case "safe" characters from U+80 .. U+7FF,
98	// but use the very good perl function to parse anything else.	136	// but use the very good perl function to parse anything else.
99	// note that we never call this function for a ascii codepoints	137	// note that we never call this function for a ascii codepoints
100	inline UV	138	INLINE UV
101	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)	139	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
102	{	140	{
103	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))	141	if (expect_true (len >= 2
104	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);	142	&& IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
105	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)	143	&& IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
106	{	144	{
107	*clen = 2;	145	*clen = 2;
108	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);	146	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
109	}	147	}
110	else	148	else
111	{	149	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
112	*clen = (STRLEN)-1;	150	}
113	return (UV)-1;	151
114	}	152	// likewise for encoding, also never called for ascii codepoints
		153	// this function takes advantage of this fact, although current gccs
		154	// seem to optimise the check for >= 0x80 away anyways
		155	INLINE unsigned char *
		156	encode_utf8 (unsigned char *s, UV ch)
		157	{
		158	if (expect_false (ch < 0x000080))
		159	*s++ = ch;
		160	else if (expect_true (ch < 0x000800))
		161	*s++ = 0xc0 \| ( ch >> 6),
		162	*s++ = 0x80 \| ( ch & 0x3f);
		163	else if ( ch < 0x010000)
		164	*s++ = 0xe0 \| ( ch >> 12),
		165	*s++ = 0x80 \| ((ch >> 6) & 0x3f),
		166	*s++ = 0x80 \| ( ch & 0x3f);
		167	else if ( ch < 0x110000)
		168	*s++ = 0xf0 \| ( ch >> 18),
		169	*s++ = 0x80 \| ((ch >> 12) & 0x3f),
		170	*s++ = 0x80 \| ((ch >> 6) & 0x3f),
		171	*s++ = 0x80 \| ( ch & 0x3f);
		172
		173	return s;
115	}	174	}
116		175
117	/////////////////////////////////////////////////////////////////////////////	176	/////////////////////////////////////////////////////////////////////////////
118	// encoder	177	// encoder
119		178
…		…
124	char *end; // SvEND (sv)	183	char *end; // SvEND (sv)
125	SV *sv; // result scalar	184	SV *sv; // result scalar
126	JSON json;	185	JSON json;
127	U32 indent; // indentation level	186	U32 indent; // indentation level
128	U32 maxdepth; // max. indentation/recursion level	187	U32 maxdepth; // max. indentation/recursion level
		188	UV limit; // escape character values >= this value when encoding
129	} enc_t;	189	} enc_t;
130		190
131	inline void	191	INLINE void
132	need (enc_t *enc, STRLEN len)	192	need (enc_t *enc, STRLEN len)
133	{	193	{
134	if (expect_false (enc->cur + len >= enc->end))	194	if (expect_false (enc->cur + len >= enc->end))
135	{	195	{
136	STRLEN cur = enc->cur - SvPVX (enc->sv);	196	STRLEN cur = enc->cur - SvPVX (enc->sv);
…		…
138	enc->cur = SvPVX (enc->sv) + cur;	198	enc->cur = SvPVX (enc->sv) + cur;
139	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	199	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
140	}	200	}
141	}	201	}
142		202
143	inline void	203	INLINE void
144	encode_ch (enc_t *enc, char ch)	204	encode_ch (enc_t *enc, char ch)
145	{	205	{
146	need (enc, 1);	206	need (enc, 1);
147	*enc->cur++ = ch;	207	*enc->cur++ = ch;
148	}	208	}
…		…
202	{	262	{
203	uch = ch;	263	uch = ch;
204	clen = 1;	264	clen = 1;
205	}	265	}
206		266
207	if (uch > 0x10FFFFUL)	267	if (uch < 0x80/0x20/ \|\| uch >= enc->limit)
208	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
209
210	if (uch < 0x80 \|\| enc->json.flags & F_ASCII \|\| (enc->json.flags & F_LATIN1 && uch > 0xFF))
211	{	268	{
212	if (uch > 0xFFFFUL)	269	if (uch >= 0x10000UL)
213	{	270	{
		271	if (uch >= 0x110000UL)
		272	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
		273
214	need (enc, len += 11);	274	need (enc, len += 11);
215	sprintf (enc->cur, "\\u%04x\\u%04x",	275	sprintf (enc->cur, "\\u%04x\\u%04x",
216	(int)((uch - 0x10000) / 0x400 + 0xD800),	276	(int)((uch - 0x10000) / 0x400 + 0xD800),
217	(int)((uch - 0x10000) % 0x400 + 0xDC00));	277	(int)((uch - 0x10000) % 0x400 + 0xDC00));
218	enc->cur += 12;	278	enc->cur += 12;
…		…
246	while (--clen);	306	while (--clen);
247	}	307	}
248	else	308	else
249	{	309	{
250	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed	310	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
251	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	311	enc->cur = encode_utf8 (enc->cur, uch);
252	++str;	312	++str;
253	}	313	}
254	}	314	}
255	}	315	}
256	}	316	}
257		317
258	--len;	318	--len;
259	}	319	}
260	}	320	}
261		321
262	inline void	322	INLINE void
263	encode_indent (enc_t *enc)	323	encode_indent (enc_t *enc)
264	{	324	{
265	if (enc->json.flags & F_INDENT)	325	if (enc->json.flags & F_INDENT)
266	{	326	{
267	int spaces = enc->indent * INDENT_STEP;	327	int spaces = enc->indent * INDENT_STEP;
…		…
270	memset (enc->cur, ' ', spaces);	330	memset (enc->cur, ' ', spaces);
271	enc->cur += spaces;	331	enc->cur += spaces;
272	}	332	}
273	}	333	}
274		334
275	inline void	335	INLINE void
276	encode_space (enc_t *enc)	336	encode_space (enc_t *enc)
277	{	337	{
278	need (enc, 1);	338	need (enc, 1);
279	encode_ch (enc, ' ');	339	encode_ch (enc, ' ');
280	}	340	}
281		341
282	inline void	342	INLINE void
283	encode_nl (enc_t *enc)	343	encode_nl (enc_t *enc)
284	{	344	{
285	if (enc->json.flags & F_INDENT)	345	if (enc->json.flags & F_INDENT)
286	{	346	{
287	need (enc, 1);	347	need (enc, 1);
288	encode_ch (enc, '\n');	348	encode_ch (enc, '\n');
289	}	349	}
290	}	350	}
291		351
292	inline void	352	INLINE void
293	encode_comma (enc_t *enc)	353	encode_comma (enc_t *enc)
294	{	354	{
295	encode_ch (enc, ',');	355	encode_ch (enc, ',');
296		356
297	if (enc->json.flags & F_INDENT)	357	if (enc->json.flags & F_INDENT)
…		…
308	int i, len = av_len (av);	368	int i, len = av_len (av);
309		369
310	if (enc->indent >= enc->maxdepth)	370	if (enc->indent >= enc->maxdepth)
311	croak ("data structure too deep (hit recursion limit)");	371	croak ("data structure too deep (hit recursion limit)");
312		372
313	encode_ch (enc, '['); encode_nl (enc);	373	encode_ch (enc, '[');
314	++enc->indent;	374
		375	if (len >= 0)
		376	{
		377	encode_nl (enc); ++enc->indent;
315		378
316	for (i = 0; i <= len; ++i)	379	for (i = 0; i <= len; ++i)
317	{	380	{
318	SV **svp = av_fetch (av, i, 0);	381	SV **svp = av_fetch (av, i, 0);
319		382
320	encode_indent (enc);	383	encode_indent (enc);
321		384
322	if (svp)	385	if (svp)
323	encode_sv (enc, *svp);	386	encode_sv (enc, *svp);
324	else	387	else
325	encode_str (enc, "null", 4, 0);	388	encode_str (enc, "null", 4, 0);
326		389
327	if (i < len)	390	if (i < len)
328	encode_comma (enc);	391	encode_comma (enc);
329	}	392	}
330		393
		394	encode_nl (enc); --enc->indent; encode_indent (enc);
		395	}
		396
331	encode_nl (enc);	397	encode_ch (enc, ']');
332
333	--enc->indent;
334	encode_indent (enc); encode_ch (enc, ']');
335	}	398	}
336		399
337	static void	400	static void
338	encode_he (enc_t enc, HE he)	401	encode_hk (enc_t enc, HE he)
339	{	402	{
340	encode_ch (enc, '"');	403	encode_ch (enc, '"');
341		404
342	if (HeKLEN (he) == HEf_SVKEY)	405	if (HeKLEN (he) == HEf_SVKEY)
343	{	406	{
…		…
356	encode_ch (enc, '"');	419	encode_ch (enc, '"');
357		420
358	if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);	421	if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
359	encode_ch (enc, ':');	422	encode_ch (enc, ':');
360	if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);	423	if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
361	encode_sv (enc, HeVAL (he));
362	}	424	}
363		425
364	// compare hash entries, used when all keys are bytestrings	426	// compare hash entries, used when all keys are bytestrings
365	static int	427	static int
366	he_cmp_fast (const void a_, const void b_)	428	he_cmp_fast (const void a_, const void b_)
…		…
371	HE b = (HE **)b_;	433	HE b = (HE **)b_;
372		434
373	STRLEN la = HeKLEN (a);	435	STRLEN la = HeKLEN (a);
374	STRLEN lb = HeKLEN (b);	436	STRLEN lb = HeKLEN (b);
375		437
376	if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))	438	if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
377	cmp = la - lb;	439	cmp = lb - la;
378		440
379	return cmp;	441	return cmp;
380	}	442	}
381		443
382	// compare hash entries, used when some keys are sv's or utf-x	444	// compare hash entries, used when some keys are sv's or utf-x
383	static int	445	static int
384	he_cmp_slow (const void a, const void b)	446	he_cmp_slow (const void a, const void b)
385	{	447	{
386	return sv_cmp (HeSVKEY_force ((HE )a), HeSVKEY_force ((HE **)b));	448	return sv_cmp (HeSVKEY_force ((HE )b), HeSVKEY_force ((HE **)a));
387	}	449	}
388		450
389	static void	451	static void
390	encode_hv (enc_t enc, HV hv)	452	encode_hv (enc_t enc, HV hv)
391	{	453	{
392	int count, i;	454	HE *he;
393		455
394	if (enc->indent >= enc->maxdepth)	456	if (enc->indent >= enc->maxdepth)
395	croak ("data structure too deep (hit recursion limit)");	457	croak ("data structure too deep (hit recursion limit)");
396		458
397	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;	459	encode_ch (enc, '{');
398		460
399	if ((count = hv_iterinit (hv)))
400	{
401	// for canonical output we have to sort by keys first	461	// for canonical output we have to sort by keys first
402	// actually, this is mostly due to the stupid so-called	462	// actually, this is mostly due to the stupid so-called
403	// security workaround added somewhere in 5.8.x.	463	// security workaround added somewhere in 5.8.x.
404	// that randomises hash orderings	464	// that randomises hash orderings
405	if (enc->json.flags & F_CANONICAL)	465	if (enc->json.flags & F_CANONICAL)
		466	{
		467	int count = hv_iterinit (hv);
		468
		469	if (SvMAGICAL (hv))
406	{	470	{
		471	// need to count by iterating. could improve by dynamically building the vector below
		472	// but I don't care for the speed of this special case.
		473	// note also that we will run into undefined behaviour when the two iterations
		474	// do not result in the same count, something I might care for in some later release.
		475
		476	count = 0;
		477	while (hv_iternext (hv))
		478	++count;
		479
		480	hv_iterinit (hv);
		481	}
		482
		483	if (count)
		484	{
407	int fast = 1;	485	int i, fast = 1;
408	HE *he;
409	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	486	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
410	HE *hes = _alloca (count sizeof (HE));	487	HE *hes = _alloca (count sizeof (HE));
411	#else	488	#else
412	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode	489	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
413	#endif	490	#endif
…		…
440		517
441	FREETMPS;	518	FREETMPS;
442	LEAVE;	519	LEAVE;
443	}	520	}
444		521
445	for (i = 0; i < count; ++i)	522	encode_nl (enc); ++enc->indent;
		523
		524	while (count--)
446	{	525	{
447	encode_indent (enc);	526	encode_indent (enc);
		527	he = hes [count];
448	encode_he (enc, hes [i]);	528	encode_hk (enc, he);
		529	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
449		530
450	if (i < count - 1)	531	if (count)
451	encode_comma (enc);	532	encode_comma (enc);
452	}	533	}
453		534
454	encode_nl (enc);	535	encode_nl (enc); --enc->indent; encode_indent (enc);
455	}	536	}
		537	}
456	else	538	else
		539	{
		540	if (hv_iterinit (hv) \|\| SvMAGICAL (hv))
		541	if ((he = hv_iternext (hv)))
457	{	542	{
458	HE *he = hv_iternext (hv);	543	encode_nl (enc); ++enc->indent;
459		544
460	for (;;)	545	for (;;)
461	{	546	{
462	encode_indent (enc);	547	encode_indent (enc);
463	encode_he (enc, he);	548	encode_hk (enc, he);
		549	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
464		550
465	if (!(he = hv_iternext (hv)))	551	if (!(he = hv_iternext (hv)))
466	break;	552	break;
467		553
468	encode_comma (enc);	554	encode_comma (enc);
469	}	555	}
470		556
471	encode_nl (enc);	557	encode_nl (enc); --enc->indent; encode_indent (enc);
472	}	558	}
473	}	559	}
474		560
475	--enc->indent; encode_indent (enc); encode_ch (enc, '}');	561	encode_ch (enc, '}');
476	}	562	}
477		563
478	// encode objects, arrays and special \0=false and \1=true values.	564	// encode objects, arrays and special \0=false and \1=true values.
479	static void	565	static void
480	encode_rv (enc_t enc, SV sv)	566	encode_rv (enc_t enc, SV sv)
…		…
557		643
558	if (len == 1 && *pv == '1')	644	if (len == 1 && *pv == '1')
559	encode_str (enc, "true", 4, 0);	645	encode_str (enc, "true", 4, 0);
560	else if (len == 1 && *pv == '0')	646	else if (len == 1 && *pv == '0')
561	encode_str (enc, "false", 5, 0);	647	encode_str (enc, "false", 5, 0);
		648	else if (enc->json.flags & F_ALLOW_UNKNOWN)
		649	encode_str (enc, "null", 4, 0);
562	else	650	else
563	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",	651	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
564	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	652	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
565	}	653	}
		654	else if (enc->json.flags & F_ALLOW_UNKNOWN)
		655	encode_str (enc, "null", 4, 0);
566	else	656	else
567	croak ("encountered %s, but JSON can only represent references to arrays or hashes",	657	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
568	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	658	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
569	}	659	}
570		660
…		…
588	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	678	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
589	enc->cur += strlen (enc->cur);	679	enc->cur += strlen (enc->cur);
590	}	680	}
591	else if (SvIOKp (sv))	681	else if (SvIOKp (sv))
592	{	682	{
593	// we assume we can always read an IV as a UV	683	// we assume we can always read an IV as a UV and vice versa
594	if (SvUV (sv) & ~(UV)0x7fff)	684	// we assume two's complement
595	{	685	// we assume no aliasing issues in the union
596	// large integer, use the (rather slow) snprintf way.	686	if (SvIsUV (sv) ? SvUVX (sv) <= 59000
597	need (enc, sizeof (UV) * 3);	687	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
598	enc->cur +=
599	SvIsUV(sv)
600	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
601	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
602	}
603	else
604	{	688	{
605	// optimise the "small number case"	689	// optimise the "small number case"
606	// code will likely be branchless and use only a single multiplication	690	// code will likely be branchless and use only a single multiplication
		691	// works for numbers up to 59074
607	I32 i = SvIV (sv);	692	I32 i = SvIVX (sv);
608	U32 u;	693	U32 u;
609	char digit, nz = 0;	694	char digit, nz = 0;
610		695
611	need (enc, 6);	696	need (enc, 6);
612		697
…		…
618		703
619	// now output digit by digit, each time masking out the integer part	704	// now output digit by digit, each time masking out the integer part
620	// and multiplying by 5 while moving the decimal point one to the right,	705	// and multiplying by 5 while moving the decimal point one to the right,
621	// resulting in a net multiplication by 10.	706	// resulting in a net multiplication by 10.
622	// we always write the digit to memory but conditionally increment	707	// we always write the digit to memory but conditionally increment
623	// the pointer, to ease the usage of conditional move instructions.	708	// the pointer, to enable the use of conditional move instructions.
624	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffff) 5;	709	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffffUL) 5;
625	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffff) 5;	710	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffffUL) 5;
626	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffff) 5;	711	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffffUL) 5;
627	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffff) 5;	712	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffffUL) 5;
628	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'	713	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
629	}	714	}
		715	else
		716	{
		717	// large integer, use the (rather slow) snprintf way.
		718	need (enc, IVUV_MAXCHARS);
		719	enc->cur +=
		720	SvIsUV(sv)
		721	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
		722	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
		723	}
630	}	724	}
631	else if (SvROK (sv))	725	else if (SvROK (sv))
632	encode_rv (enc, SvRV (sv));	726	encode_rv (enc, SvRV (sv));
633	else if (!SvOK (sv))	727	else if (!SvOK (sv) \|\| enc->json.flags & F_ALLOW_UNKNOWN)
634	encode_str (enc, "null", 4, 0);	728	encode_str (enc, "null", 4, 0);
635	else	729	else
636	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	730	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
637	SvPV_nolen (sv), SvFLAGS (sv));	731	SvPV_nolen (sv), SvFLAGS (sv));
638	}	732	}
…		…
649	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	743	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
650	enc.cur = SvPVX (enc.sv);	744	enc.cur = SvPVX (enc.sv);
651	enc.end = SvEND (enc.sv);	745	enc.end = SvEND (enc.sv);
652	enc.indent = 0;	746	enc.indent = 0;
653	enc.maxdepth = DEC_DEPTH (enc.json.flags);	747	enc.maxdepth = DEC_DEPTH (enc.json.flags);
		748	enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
		749	: enc.json.flags & F_LATIN1 ? 0x000100UL
		750	: 0x110000UL;
654		751
655	SvPOK_only (enc.sv);	752	SvPOK_only (enc.sv);
656	encode_sv (&enc, scalar);	753	encode_sv (&enc, scalar);
657		754
658	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	755	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
…		…
679	JSON json;	776	JSON json;
680	U32 depth; // recursion depth	777	U32 depth; // recursion depth
681	U32 maxdepth; // recursion depth limit	778	U32 maxdepth; // recursion depth limit
682	} dec_t;	779	} dec_t;
683		780
684	inline void	781	INLINE void
		782	decode_comment (dec_t *dec)
		783	{
		784	// only '#'-style comments allowed a.t.m.
		785
		786	while (dec->cur && dec->cur != 0x0a && *dec->cur != 0x0d)
		787	++dec->cur;
		788	}
		789
		790	INLINE void
685	decode_ws (dec_t *dec)	791	decode_ws (dec_t *dec)
686	{	792	{
687	for (;;)	793	for (;;)
688	{	794	{
689	char ch = *dec->cur;	795	char ch = *dec->cur;
690		796
691	if (ch > 0x20	797	if (ch > 0x20)
		798	{
		799	if (expect_false (ch == '#'))
		800	{
		801	if (dec->json.flags & F_RELAXED)
		802	decode_comment (dec);
		803	else
		804	break;
		805	}
		806	else
		807	break;
		808	}
692	\|\| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))	809	else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
693	break;	810	break; // parse error, but let higher level handle it, gives better error messages
694		811
695	++dec->cur;	812	++dec->cur;
696	}	813	}
697	}	814	}
698		815
…		…
804		921
805	if (hi >= 0x80)	922	if (hi >= 0x80)
806	{	923	{
807	utf8 = 1;	924	utf8 = 1;
808		925
809	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);	926	cur = encode_utf8 (cur, hi);
810	}	927	}
811	else	928	else
812	*cur++ = hi;	929	*cur++ = hi;
813	}	930	}
814	break;	931	break;
…		…
816	default:	933	default:
817	--dec_cur;	934	--dec_cur;
818	ERR ("illegal backslash escape sequence in string");	935	ERR ("illegal backslash escape sequence in string");
819	}	936	}
820	}	937	}
821	else if (expect_true (ch >= 0x20 && ch <= 0x7f))	938	else if (expect_true (ch >= 0x20 && ch < 0x80))
822	*cur++ = ch;	939	*cur++ = ch;
823	else if (ch >= 0x80)	940	else if (ch >= 0x80)
824	{	941	{
825	STRLEN clen;	942	STRLEN clen;
826	UV uch;	943	UV uch;
…		…
949		1066
950	if (!is_nv)	1067	if (!is_nv)
951	{	1068	{
952	int len = dec->cur - start;	1069	int len = dec->cur - start;
953		1070
954	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so	1071	// special case the rather common 1..5-digit-int case
955	if (*start == '-')	1072	if (*start == '-')
956	switch (len)	1073	switch (len)
957	{	1074	{
958	case 2: return newSViv (-( start [1] - '0' * 1));	1075	case 2: return newSViv (-( start [1] - '0' * 1));
959	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));	1076	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
960	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));	1077	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
961	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));	1078	case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
		1079	case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
962	}	1080	}
963	else	1081	else
964	switch (len)	1082	switch (len)
965	{	1083	{
966	case 1: return newSViv ( start [0] - '0' * 1);	1084	case 1: return newSViv ( start [0] - '0' * 1);
967	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);	1085	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
968	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);	1086	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
969	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);	1087	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		1088	case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
970	}	1089	}
971		1090
972	{	1091	{
973	UV uv;	1092	UV uv;
974	int numtype = grok_number (start, len, &uv);	1093	int numtype = grok_number (start, len, &uv);
…		…
1035		1154
1036	if (*dec->cur != ',')	1155	if (*dec->cur != ',')
1037	ERR (", or ] expected while parsing array");	1156	ERR (", or ] expected while parsing array");
1038		1157
1039	++dec->cur;	1158	++dec->cur;
		1159
		1160	decode_ws (dec);
		1161
		1162	if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
		1163	{
		1164	++dec->cur;
		1165	break;
		1166	}
1040	}	1167	}
1041		1168
1042	DEC_DEC_DEPTH;	1169	DEC_DEC_DEPTH;
1043	return newRV_noinc ((SV *)av);	1170	return newRV_noinc ((SV *)av);
1044		1171
…		…
1060	if (*dec->cur == '}')	1187	if (*dec->cur == '}')
1061	++dec->cur;	1188	++dec->cur;
1062	else	1189	else
1063	for (;;)	1190	for (;;)
1064	{	1191	{
1065	decode_ws (dec); EXPECT_CH ('"');	1192	EXPECT_CH ('"');
1066		1193
1067	// heuristic: assume that	1194	// heuristic: assume that
1068	// a) decode_str + hv_store_ent are abysmally slow.	1195	// a) decode_str + hv_store_ent are abysmally slow.
1069	// b) most hash keys are short, simple ascii text.	1196	// b) most hash keys are short, simple ascii text.
1070	// => try to "fast-match" such strings to avoid	1197	// => try to "fast-match" such strings to avoid
…		…
1074	char *p = dec->cur;	1201	char *p = dec->cur;
1075	char *e = p + 24; // only try up to 24 bytes	1202	char *e = p + 24; // only try up to 24 bytes
1076		1203
1077	for (;;)	1204	for (;;)
1078	{	1205	{
1079	// the >= 0x80 is true on most architectures	1206	// the >= 0x80 is false on most architectures
1080	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')	1207	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
1081	{	1208	{
1082	// slow path, back up and use decode_str	1209	// slow path, back up and use decode_str
1083	SV *key = decode_str (dec);	1210	SV *key = decode_str (dec);
1084	if (!key)	1211	if (!key)
1085	goto fail;	1212	goto fail;
1086		1213
1087	decode_ws (dec); EXPECT_CH (':');	1214	decode_ws (dec); EXPECT_CH (':');
1088		1215
		1216	decode_ws (dec);
1089	value = decode_sv (dec);	1217	value = decode_sv (dec);
1090	if (!value)	1218	if (!value)
1091	{	1219	{
1092	SvREFCNT_dec (key);	1220	SvREFCNT_dec (key);
1093	goto fail;	1221	goto fail;
…		…
1105	int len = p - key;	1233	int len = p - key;
1106	dec->cur = p + 1;	1234	dec->cur = p + 1;
1107		1235
1108	decode_ws (dec); EXPECT_CH (':');	1236	decode_ws (dec); EXPECT_CH (':');
1109		1237
		1238	decode_ws (dec);
1110	value = decode_sv (dec);	1239	value = decode_sv (dec);
1111	if (!value)	1240	if (!value)
1112	goto fail;	1241	goto fail;
1113		1242
1114	hv_store (hv, key, len, value, 0);	1243	hv_store (hv, key, len, value, 0);
…		…
1130		1259
1131	if (*dec->cur != ',')	1260	if (*dec->cur != ',')
1132	ERR (", or } expected while parsing object/hash");	1261	ERR (", or } expected while parsing object/hash");
1133		1262
1134	++dec->cur;	1263	++dec->cur;
		1264
		1265	decode_ws (dec);
		1266
		1267	if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
		1268	{
		1269	++dec->cur;
		1270	break;
		1271	}
1135	}	1272	}
1136		1273
1137	DEC_DEC_DEPTH;	1274	DEC_DEC_DEPTH;
1138	sv = newRV_noinc ((SV *)hv);	1275	sv = newRV_noinc ((SV *)hv);
1139		1276
…		…
1204	}	1341	}
1205		1342
1206	static SV *	1343	static SV *
1207	decode_sv (dec_t *dec)	1344	decode_sv (dec_t *dec)
1208	{	1345	{
1209	decode_ws (dec);
1210
1211	// the beauty of JSON: you need exactly one character lookahead	1346	// the beauty of JSON: you need exactly one character lookahead
1212	// to parse anything.	1347	// to parse everything.
1213	switch (*dec->cur)	1348	switch (*dec->cur)
1214	{	1349	{
1215	case '"': ++dec->cur; return decode_str (dec);	1350	case '"': ++dec->cur; return decode_str (dec);
1216	case '[': ++dec->cur; return decode_av (dec);	1351	case '[': ++dec->cur; return decode_av (dec);
1217	case '{': ++dec->cur; return decode_hv (dec);	1352	case '{': ++dec->cur; return decode_hv (dec);
1218		1353
1219	case '-':	1354	case '-':
1220	case '0': case '1': case '2': case '3': case '4':	1355	case '0': case '1': case '2': case '3': case '4':
1221	case '5': case '6': case '7': case '8': case '9':	1356	case '5': case '6': case '7': case '8': case '9':
1222	return decode_num (dec);	1357	return decode_num (dec);
1223		1358
1224	case 't':	1359	case 't':
1225	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1360	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1226	{	1361	{
1227	dec->cur += 4;	1362	dec->cur += 4;
		1363	#if JSON_SLOW
		1364	json_true = get_bool ("JSON::XS::true");
		1365	#endif
1228	return SvREFCNT_inc (json_true);	1366	return newSVsv (json_true);
1229	}	1367	}
1230	else	1368	else
1231	ERR ("'true' expected");	1369	ERR ("'true' expected");
1232		1370
1233	break;	1371	break;
1234		1372
1235	case 'f':	1373	case 'f':
1236	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1374	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1237	{	1375	{
1238	dec->cur += 5;	1376	dec->cur += 5;
		1377	#if JSON_SLOW
		1378	json_false = get_bool ("JSON::XS::false");
		1379	#endif
1239	return SvREFCNT_inc (json_false);	1380	return newSVsv (json_false);
1240	}	1381	}
1241	else	1382	else
1242	ERR ("'false' expected");	1383	ERR ("'false' expected");
1243		1384
1244	break;	1385	break;
…		…
1262	fail:	1403	fail:
1263	return 0;	1404	return 0;
1264	}	1405	}
1265		1406
1266	static SV *	1407	static SV *
1267	decode_json (SV string, JSON json, UV *offset_return)	1408	decode_json (SV string, JSON json, STRLEN *offset_return)
1268	{	1409	{
1269	dec_t dec;	1410	dec_t dec;
1270	UV offset;	1411	STRLEN offset;
1271	SV *sv;	1412	SV *sv;
1272		1413
1273	SvGETMAGIC (string);	1414	SvGETMAGIC (string);
1274	SvUPGRADE (string, SVt_PV);	1415	SvUPGRADE (string, SVt_PV);
1275		1416
…		…
1293		1434
1294	if (dec.json.cb_object \|\| dec.json.cb_sk_object)	1435	if (dec.json.cb_object \|\| dec.json.cb_sk_object)
1295	dec.json.flags \|= F_HOOK;	1436	dec.json.flags \|= F_HOOK;
1296		1437
1297	*dec.end = 0; // this should basically be a nop, too, but make sure it's there	1438	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
		1439
		1440	decode_ws (&dec);
1298	sv = decode_sv (&dec);	1441	sv = decode_sv (&dec);
1299		1442
1300	if (!(offset_return \|\| !sv))	1443	if (!(offset_return \|\| !sv))
1301	{	1444	{
1302	// check for trailing garbage	1445	// check for trailing garbage
…		…
1346		1489
1347	return sv;	1490	return sv;
1348	}	1491	}
1349		1492
1350	/////////////////////////////////////////////////////////////////////////////	1493	/////////////////////////////////////////////////////////////////////////////
		1494	// incremental parser
		1495
		1496	static void
		1497	incr_parse (JSON *self)
		1498	{
		1499	const char *p = SvPVX (self->incr_text) + self->incr_pos;
		1500
		1501	for (;;)
		1502	{
		1503	//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
		1504	switch (self->incr_mode)
		1505	{
		1506	// only used for intiial whitespace skipping
		1507	case INCR_M_WS:
		1508	for (;;)
		1509	{
		1510	if (*p > 0x20)
		1511	{
		1512	self->incr_mode = INCR_M_JSON;
		1513	goto incr_m_json;
		1514	}
		1515	else if (!*p)
		1516	goto interrupt;
		1517
		1518	++p;
		1519	}
		1520
		1521	// skip a single char inside a string (for \\-processing)
		1522	case INCR_M_BS:
		1523	if (!*p)
		1524	goto interrupt;
		1525
		1526	++p;
		1527	self->incr_mode = INCR_M_STR;
		1528	goto incr_m_str;
		1529
		1530	// inside a string
		1531	case INCR_M_STR:
		1532	incr_m_str:
		1533	for (;;)
		1534	{
		1535	if (*p == '"')
		1536	{
		1537	++p;
		1538	self->incr_mode = INCR_M_JSON;
		1539
		1540	if (!self->incr_nest)
		1541	goto interrupt;
		1542
		1543	goto incr_m_json;
		1544	}
		1545	else if (*p == '\\')
		1546	{
		1547	++p; // "virtually" consumes character after \
		1548
		1549	if (!*p) // if at end of string we have to switch modes
		1550	{
		1551	self->incr_mode = INCR_M_BS;
		1552	goto interrupt;
		1553	}
		1554	}
		1555	else if (!*p)
		1556	goto interrupt;
		1557
		1558	++p;
		1559	}
		1560
		1561	// after initial ws, outside string
		1562	case INCR_M_JSON:
		1563	incr_m_json:
		1564	for (;;)
		1565	{
		1566	switch (*p++)
		1567	{
		1568	case 0:
		1569	--p;
		1570	goto interrupt;
		1571
		1572	case 0x09:
		1573	case 0x0a:
		1574	case 0x0d:
		1575	case 0x20:
		1576	if (!self->incr_nest)
		1577	{
		1578	--p; // do not eat the whitespace, let the next round do it
		1579	goto interrupt;
		1580	}
		1581	break;
		1582
		1583	case '"':
		1584	self->incr_mode = INCR_M_STR;
		1585	goto incr_m_str;
		1586
		1587	case '[':
		1588	case '{':
		1589	++self->incr_nest;
		1590	break;
		1591
		1592	case ']':
		1593	case '}':
		1594	if (!--self->incr_nest)
		1595	goto interrupt;
		1596	}
		1597	}
		1598	}
		1599
		1600	modechange:
		1601	;
		1602	}
		1603
		1604	interrupt:
		1605	self->incr_pos = p - SvPVX (self->incr_text);
		1606	//printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
		1607	}
		1608
		1609	/////////////////////////////////////////////////////////////////////////////
1351	// XS interface functions	1610	// XS interface functions
1352		1611
1353	MODULE = JSON::XS PACKAGE = JSON::XS	1612	MODULE = JSON::XS PACKAGE = JSON::XS
1354		1613
1355	BOOT:	1614	BOOT:
…		…
1364	: -1;	1623	: -1;
1365		1624
1366	json_stash = gv_stashpv ("JSON::XS" , 1);	1625	json_stash = gv_stashpv ("JSON::XS" , 1);
1367	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);	1626	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1368		1627
1369	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );	1628	json_true = get_bool ("JSON::XS::true");
1370	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);	1629	json_false = get_bool ("JSON::XS::false");
1371	}	1630	}
1372		1631
1373	PROTOTYPES: DISABLE	1632	PROTOTYPES: DISABLE
1374		1633
1375	void CLONE (...)	1634	void CLONE (...)
…		…
1378	json_boolean_stash = 0;	1637	json_boolean_stash = 0;
1379		1638
1380	void new (char *klass)	1639	void new (char *klass)
1381	PPCODE:	1640	PPCODE:
1382	{	1641	{
1383	HV *stash = !JSON_SLOW \|\| json_stash
1384	? json_stash
1385	: gv_stashpv ("JSON::XS", 1);
1386	SV *pv = NEWSV (0, sizeof (JSON));	1642	SV *pv = NEWSV (0, sizeof (JSON));
1387	SvPOK_only (pv);	1643	SvPOK_only (pv);
1388	Zero (SvPVX (pv), 1, JSON);	1644	Zero (SvPVX (pv), 1, JSON);
1389	((JSON *)SvPVX (pv))->flags = F_DEFAULT;	1645	((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1390	XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), stash)));	1646	XPUSHs (sv_2mortal (sv_bless (
		1647	newRV_noinc (pv),
		1648	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
		1649	)));
1391	}	1650	}
1392		1651
1393	void ascii (JSON *self, int enable = 1)	1652	void ascii (JSON *self, int enable = 1)
1394	ALIAS:	1653	ALIAS:
1395	ascii = F_ASCII	1654	ascii = F_ASCII
…		…
1402	pretty = F_PRETTY	1661	pretty = F_PRETTY
1403	allow_nonref = F_ALLOW_NONREF	1662	allow_nonref = F_ALLOW_NONREF
1404	shrink = F_SHRINK	1663	shrink = F_SHRINK
1405	allow_blessed = F_ALLOW_BLESSED	1664	allow_blessed = F_ALLOW_BLESSED
1406	convert_blessed = F_CONV_BLESSED	1665	convert_blessed = F_CONV_BLESSED
		1666	relaxed = F_RELAXED
		1667	allow_unknown = F_ALLOW_UNKNOWN
1407	PPCODE:	1668	PPCODE:
1408	{	1669	{
1409	if (enable)	1670	if (enable)
1410	self->flags \|= ix;	1671	self->flags \|= ix;
1411	else	1672	else
1412	self->flags &= ~ix;	1673	self->flags &= ~ix;
1413		1674
1414	XPUSHs (ST (0));	1675	XPUSHs (ST (0));
1415	}	1676	}
1416		1677
		1678	void get_ascii (JSON *self)
		1679	ALIAS:
		1680	get_ascii = F_ASCII
		1681	get_latin1 = F_LATIN1
		1682	get_utf8 = F_UTF8
		1683	get_indent = F_INDENT
		1684	get_canonical = F_CANONICAL
		1685	get_space_before = F_SPACE_BEFORE
		1686	get_space_after = F_SPACE_AFTER
		1687	get_allow_nonref = F_ALLOW_NONREF
		1688	get_shrink = F_SHRINK
		1689	get_allow_blessed = F_ALLOW_BLESSED
		1690	get_convert_blessed = F_CONV_BLESSED
		1691	get_relaxed = F_RELAXED
		1692	get_allow_unknown = F_ALLOW_UNKNOWN
		1693	PPCODE:
		1694	XPUSHs (boolSV (self->flags & ix));
		1695
1417	void max_depth (JSON *self, UV max_depth = 0x80000000UL)	1696	void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1418	PPCODE:	1697	PPCODE:
1419	{	1698	{
1420	UV log2 = 0;	1699	UV log2 = 0;
1421		1700
…		…
1427	self->flags = self->flags & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);	1706	self->flags = self->flags & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
1428		1707
1429	XPUSHs (ST (0));	1708	XPUSHs (ST (0));
1430	}	1709	}
1431		1710
		1711	U32 get_max_depth (JSON *self)
		1712	CODE:
		1713	RETVAL = DEC_DEPTH (self->flags);
		1714	OUTPUT:
		1715	RETVAL
		1716
1432	void max_size (JSON *self, UV max_size = 0)	1717	void max_size (JSON *self, UV max_size = 0)
1433	PPCODE:	1718	PPCODE:
1434	{	1719	{
1435	UV log2 = 0;	1720	UV log2 = 0;
1436		1721
…		…
1442		1727
1443	self->flags = self->flags & ~F_MAXSIZE \| (log2 << S_MAXSIZE);	1728	self->flags = self->flags & ~F_MAXSIZE \| (log2 << S_MAXSIZE);
1444		1729
1445	XPUSHs (ST (0));	1730	XPUSHs (ST (0));
1446	}	1731	}
		1732
		1733	int get_max_size (JSON *self)
		1734	CODE:
		1735	RETVAL = DEC_SIZE (self->flags);
		1736	OUTPUT:
		1737	RETVAL
1447		1738
1448	void filter_json_object (JSON self, SV cb = &PL_sv_undef)	1739	void filter_json_object (JSON self, SV cb = &PL_sv_undef)
1449	PPCODE:	1740	PPCODE:
1450	{	1741	{
1451	SvREFCNT_dec (self->cb_object);	1742	SvREFCNT_dec (self->cb_object);
…		…
1485	XPUSHs (decode_json (jsonstr, self, 0));	1776	XPUSHs (decode_json (jsonstr, self, 0));
1486		1777
1487	void decode_prefix (JSON self, SV jsonstr)	1778	void decode_prefix (JSON self, SV jsonstr)
1488	PPCODE:	1779	PPCODE:
1489	{	1780	{
1490	UV offset;	1781	STRLEN offset;
1491	EXTEND (SP, 2);	1782	EXTEND (SP, 2);
1492	PUSHs (decode_json (jsonstr, self, &offset));	1783	PUSHs (decode_json (jsonstr, self, &offset));
1493	PUSHs (sv_2mortal (newSVuv (offset)));	1784	PUSHs (sv_2mortal (newSVuv (offset)));
		1785	}
		1786
		1787	void incr_parse (JSON self, SV jsonstr = 0)
		1788	PPCODE:
		1789	{
		1790	if (!self->incr_text)
		1791	self->incr_text = newSVpvn ("", 0);
		1792
		1793	// append data, if any
		1794	if (jsonstr)
		1795	{
		1796	if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
		1797	{
		1798	/* utf-8-ness differs, need to upgrade */
		1799	sv_utf8_upgrade (self->incr_text);
		1800
		1801	if (self->incr_pos)
		1802	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
		1803	- (U8 *)SvPVX (self->incr_text);
		1804	}
		1805
		1806	{
		1807	STRLEN len;
		1808	const char *str = SvPV (jsonstr, len);
		1809	SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
		1810	Move (str, SvEND (self->incr_text), len, char);
		1811	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
		1812	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
		1813	}
		1814	}
		1815
		1816	if (GIMME_V != G_VOID)
		1817	do
		1818	{
		1819	STRLEN offset;
		1820
		1821	if (!INCR_DONE (self))
		1822	{
		1823	incr_parse (self);
		1824	if (!INCR_DONE (self))
		1825	break;
		1826	}
		1827
		1828	XPUSHs (decode_json (self->incr_text, self, &offset));
		1829
		1830	sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
		1831	self->incr_pos -= offset;
		1832	self->incr_nest = 0;
		1833	self->incr_mode = 0;
		1834	}
		1835	while (GIMME_V == G_ARRAY);
		1836	}
		1837
		1838	SV incr_text (JSON self)
		1839	ATTRS: lvalue
		1840	CODE:
		1841	{
		1842	if (self->incr_pos)
		1843	croak ("incr_text can not be called when the incremental parser already started parsing");
		1844
		1845	RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
		1846	}
		1847	OUTPUT:
		1848	RETVAL
		1849
		1850	void incr_skip (JSON *self)
		1851	CODE:
		1852	{
		1853	if (self->incr_pos)
		1854	{
		1855	sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
		1856	self->incr_pos = 0;
		1857	self->incr_nest = 0;
		1858	self->incr_mode = 0;
		1859	}
1494	}	1860	}
1495		1861
1496	void DESTROY (JSON *self)	1862	void DESTROY (JSON *self)
1497	CODE:	1863	CODE:
1498	SvREFCNT_dec (self->cb_sk_object);	1864	SvREFCNT_dec (self->cb_sk_object);
1499	SvREFCNT_dec (self->cb_object);	1865	SvREFCNT_dec (self->cb_object);
		1866	SvREFCNT_dec (self->incr_text);
1500		1867
1501	PROTOTYPES: ENABLE	1868	PROTOTYPES: ENABLE
1502		1869
1503	void to_json (SV *scalar)	1870	void encode_json (SV *scalar)
		1871	ALIAS:
		1872	to_json_ = 0
		1873	encode_json = F_UTF8
1504	PPCODE:	1874	PPCODE:
1505	{	1875	{
1506	JSON json = { F_DEFAULT \| F_UTF8 };	1876	JSON json = { F_DEFAULT \| ix };
1507	XPUSHs (encode_json (scalar, &json));	1877	XPUSHs (encode_json (scalar, &json));
1508	}	1878	}
1509		1879
1510	void from_json (SV *jsonstr)	1880	void decode_json (SV *jsonstr)
		1881	ALIAS:
		1882	from_json_ = 0
		1883	decode_json = F_UTF8
1511	PPCODE:	1884	PPCODE:
1512	{	1885	{
1513	JSON json = { F_DEFAULT \| F_UTF8 };	1886	JSON json = { F_DEFAULT \| ix };
1514	XPUSHs (decode_json (jsonstr, &json, 0));	1887	XPUSHs (decode_json (jsonstr, &json, 0));
1515	}	1888	}
1516		1889
		1890

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.58 by root, Mon Aug 13 16:06:25 2007 UTC vs. Revision 1.84 by root, Thu Mar 27 06:37:35 2008 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.58 by root, Mon Aug 13 16:06:25 2007 UTC vs.
Revision 1.84 by root, Thu Mar 27 06:37:35 2008 UTC