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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.60 by root, Mon Aug 13 16:19:13 2007 UTC vs.
Revision 1.78 by root, Tue Mar 25 19:31:02 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
32	#define F_MAXDEPTH 0xf8000000UL	37	#define F_MAXDEPTH 0xf8000000UL
33	#define S_MAXDEPTH 27	38	#define S_MAXDEPTH 27
34	#define F_MAXSIZE 0x01f00000UL	39	#define F_MAXSIZE 0x01f00000UL
35	#define S_MAXSIZE 20	40	#define S_MAXSIZE 20
36	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing	41	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
…		…
48		53
49	#define SB do {	54	#define SB do {
50	#define SE } while (0)	55	#define SE } while (0)
51		56
52	#if __GNUC__ >= 3	57	#if __GNUC__ >= 3
53	# define expect(expr,value) __builtin_expect ((expr),(value))	58	# define expect(expr,value) __builtin_expect ((expr), (value))
54	# define inline inline	59	# define INLINE static inline
55	#else	60	#else
56	# define expect(expr,value) (expr)	61	# define expect(expr,value) (expr)
57	# define inline static	62	# define INLINE static
58	#endif	63	#endif
59		64
60	#define expect_false(expr) expect ((expr) != 0, 0)	65	#define expect_false(expr) expect ((expr) != 0, 0)
61	#define expect_true(expr) expect ((expr) != 0, 1)	66	#define expect_true(expr) expect ((expr) != 0, 1)
		67
		68	#define IN_RANGE_INC(type,val,beg,end) \
		69	((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
		70	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
62		71
63	#ifdef USE_ITHREADS	72	#ifdef USE_ITHREADS
64	# define JSON_SLOW 1	73	# define JSON_SLOW 1
65	# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))	74	# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
66	#else	75	#else
…		…
69	#endif	78	#endif
70		79
71	static HV json_stash, json_boolean_stash; // JSON::XS::	80	static HV json_stash, json_boolean_stash; // JSON::XS::
72	static SV json_true, json_false;	81	static SV json_true, json_false;
73		82
		83	enum {
		84	INCR_M_WS = 0, // initial whitespace skipping, must be 0
		85	INCR_M_STR, // inside string
		86	INCR_M_BS, // inside backslash
		87	INCR_M_JSON // outside anything, count nesting
		88	};
		89
		90	#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON)
		91
74	typedef struct {	92	typedef struct {
75	U32 flags;	93	U32 flags;
76	SV *cb_object;	94	SV *cb_object;
77	HV *cb_sk_object;	95	HV *cb_sk_object;
		96
		97	// for the incremental parser
		98	SV *incr_text; // the source text so far
		99	STRLEN incr_pos; // the current offset into the text
		100	int incr_nest; // {[]}-nesting level
		101	int incr_mode;
78	} JSON;	102	} JSON;
79		103
80	/////////////////////////////////////////////////////////////////////////////	104	/////////////////////////////////////////////////////////////////////////////
81	// utility functions	105	// utility functions
82		106
83	inline void	107	INLINE void
84	shrink (SV *sv)	108	shrink (SV *sv)
85	{	109	{
86	sv_utf8_downgrade (sv, 1);	110	sv_utf8_downgrade (sv, 1);
87	if (SvLEN (sv) > SvCUR (sv) + 1)	111	if (SvLEN (sv) > SvCUR (sv) + 1)
88	{	112	{
…		…
97	// decode an utf-8 character and return it, or (UV)-1 in	121	// decode an utf-8 character and return it, or (UV)-1 in
98	// case of an error.	122	// case of an error.
99	// we special-case "safe" characters from U+80 .. U+7FF,	123	// we special-case "safe" characters from U+80 .. U+7FF,
100	// but use the very good perl function to parse anything else.	124	// but use the very good perl function to parse anything else.
101	// note that we never call this function for a ascii codepoints	125	// note that we never call this function for a ascii codepoints
102	inline UV	126	INLINE UV
103	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)	127	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
104	{	128	{
105	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))	129	if (expect_true (len >= 2
106	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);	130	&& IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
107	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)	131	&& IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
108	{	132	{
109	*clen = 2;	133	*clen = 2;
110	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);	134	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
111	}	135	}
112	else	136	else
113	{	137	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
114	*clen = (STRLEN)-1;	138	}
115	return (UV)-1;	139
116	}	140	// likewise for encoding, also never called for ascii codepoints
		141	// this function takes advantage of this fact, although current gccs
		142	// seem to optimise the check for >= 0x80 away anyways
		143	INLINE unsigned char *
		144	encode_utf8 (unsigned char *s, UV ch)
		145	{
		146	if (expect_false (ch < 0x000080))
		147	*s++ = ch;
		148	else if (expect_true (ch < 0x000800))
		149	*s++ = 0xc0 \| ( ch >> 6),
		150	*s++ = 0x80 \| ( ch & 0x3f);
		151	else if ( ch < 0x010000)
		152	*s++ = 0xe0 \| ( ch >> 12),
		153	*s++ = 0x80 \| ((ch >> 6) & 0x3f),
		154	*s++ = 0x80 \| ( ch & 0x3f);
		155	else if ( ch < 0x110000)
		156	*s++ = 0xf0 \| ( ch >> 18),
		157	*s++ = 0x80 \| ((ch >> 12) & 0x3f),
		158	*s++ = 0x80 \| ((ch >> 6) & 0x3f),
		159	*s++ = 0x80 \| ( ch & 0x3f);
		160
		161	return s;
117	}	162	}
118		163
119	/////////////////////////////////////////////////////////////////////////////	164	/////////////////////////////////////////////////////////////////////////////
120	// encoder	165	// encoder
121		166
…		…
126	char *end; // SvEND (sv)	171	char *end; // SvEND (sv)
127	SV *sv; // result scalar	172	SV *sv; // result scalar
128	JSON json;	173	JSON json;
129	U32 indent; // indentation level	174	U32 indent; // indentation level
130	U32 maxdepth; // max. indentation/recursion level	175	U32 maxdepth; // max. indentation/recursion level
		176	UV limit; // escape character values >= this value when encoding
131	} enc_t;	177	} enc_t;
132		178
133	inline void	179	INLINE void
134	need (enc_t *enc, STRLEN len)	180	need (enc_t *enc, STRLEN len)
135	{	181	{
136	if (expect_false (enc->cur + len >= enc->end))	182	if (expect_false (enc->cur + len >= enc->end))
137	{	183	{
138	STRLEN cur = enc->cur - SvPVX (enc->sv);	184	STRLEN cur = enc->cur - SvPVX (enc->sv);
…		…
140	enc->cur = SvPVX (enc->sv) + cur;	186	enc->cur = SvPVX (enc->sv) + cur;
141	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	187	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
142	}	188	}
143	}	189	}
144		190
145	inline void	191	INLINE void
146	encode_ch (enc_t *enc, char ch)	192	encode_ch (enc_t *enc, char ch)
147	{	193	{
148	need (enc, 1);	194	need (enc, 1);
149	*enc->cur++ = ch;	195	*enc->cur++ = ch;
150	}	196	}
…		…
204	{	250	{
205	uch = ch;	251	uch = ch;
206	clen = 1;	252	clen = 1;
207	}	253	}
208		254
209	if (uch > 0x10FFFFUL)	255	if (uch < 0x80/0x20/ \|\| uch >= enc->limit)
210	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
211
212	if (uch < 0x80 \|\| enc->json.flags & F_ASCII \|\| (enc->json.flags & F_LATIN1 && uch > 0xFF))
213	{	256	{
214	if (uch > 0xFFFFUL)	257	if (uch >= 0x10000UL)
215	{	258	{
		259	if (uch >= 0x110000UL)
		260	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
		261
216	need (enc, len += 11);	262	need (enc, len += 11);
217	sprintf (enc->cur, "\\u%04x\\u%04x",	263	sprintf (enc->cur, "\\u%04x\\u%04x",
218	(int)((uch - 0x10000) / 0x400 + 0xD800),	264	(int)((uch - 0x10000) / 0x400 + 0xD800),
219	(int)((uch - 0x10000) % 0x400 + 0xDC00));	265	(int)((uch - 0x10000) % 0x400 + 0xDC00));
220	enc->cur += 12;	266	enc->cur += 12;
…		…
248	while (--clen);	294	while (--clen);
249	}	295	}
250	else	296	else
251	{	297	{
252	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed	298	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
253	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	299	enc->cur = encode_utf8 (enc->cur, uch);
254	++str;	300	++str;
255	}	301	}
256	}	302	}
257	}	303	}
258	}	304	}
259		305
260	--len;	306	--len;
261	}	307	}
262	}	308	}
263		309
264	inline void	310	INLINE void
265	encode_indent (enc_t *enc)	311	encode_indent (enc_t *enc)
266	{	312	{
267	if (enc->json.flags & F_INDENT)	313	if (enc->json.flags & F_INDENT)
268	{	314	{
269	int spaces = enc->indent * INDENT_STEP;	315	int spaces = enc->indent * INDENT_STEP;
…		…
272	memset (enc->cur, ' ', spaces);	318	memset (enc->cur, ' ', spaces);
273	enc->cur += spaces;	319	enc->cur += spaces;
274	}	320	}
275	}	321	}
276		322
277	inline void	323	INLINE void
278	encode_space (enc_t *enc)	324	encode_space (enc_t *enc)
279	{	325	{
280	need (enc, 1);	326	need (enc, 1);
281	encode_ch (enc, ' ');	327	encode_ch (enc, ' ');
282	}	328	}
283		329
284	inline void	330	INLINE void
285	encode_nl (enc_t *enc)	331	encode_nl (enc_t *enc)
286	{	332	{
287	if (enc->json.flags & F_INDENT)	333	if (enc->json.flags & F_INDENT)
288	{	334	{
289	need (enc, 1);	335	need (enc, 1);
290	encode_ch (enc, '\n');	336	encode_ch (enc, '\n');
291	}	337	}
292	}	338	}
293		339
294	inline void	340	INLINE void
295	encode_comma (enc_t *enc)	341	encode_comma (enc_t *enc)
296	{	342	{
297	encode_ch (enc, ',');	343	encode_ch (enc, ',');
298		344
299	if (enc->json.flags & F_INDENT)	345	if (enc->json.flags & F_INDENT)
…		…
310	int i, len = av_len (av);	356	int i, len = av_len (av);
311		357
312	if (enc->indent >= enc->maxdepth)	358	if (enc->indent >= enc->maxdepth)
313	croak ("data structure too deep (hit recursion limit)");	359	croak ("data structure too deep (hit recursion limit)");
314		360
315	encode_ch (enc, '['); encode_nl (enc);	361	encode_ch (enc, '[');
316	++enc->indent;	362
		363	if (len >= 0)
		364	{
		365	encode_nl (enc); ++enc->indent;
317		366
318	for (i = 0; i <= len; ++i)	367	for (i = 0; i <= len; ++i)
319	{	368	{
320	SV **svp = av_fetch (av, i, 0);	369	SV **svp = av_fetch (av, i, 0);
321		370
322	encode_indent (enc);	371	encode_indent (enc);
323		372
324	if (svp)	373	if (svp)
325	encode_sv (enc, *svp);	374	encode_sv (enc, *svp);
326	else	375	else
327	encode_str (enc, "null", 4, 0);	376	encode_str (enc, "null", 4, 0);
328		377
329	if (i < len)	378	if (i < len)
330	encode_comma (enc);	379	encode_comma (enc);
331	}	380	}
332		381
		382	encode_nl (enc); --enc->indent; encode_indent (enc);
		383	}
		384
333	encode_nl (enc);	385	encode_ch (enc, ']');
334
335	--enc->indent;
336	encode_indent (enc); encode_ch (enc, ']');
337	}	386	}
338		387
339	static void	388	static void
340	encode_he (enc_t enc, HE he)	389	encode_hk (enc_t enc, HE he)
341	{	390	{
342	encode_ch (enc, '"');	391	encode_ch (enc, '"');
343		392
344	if (HeKLEN (he) == HEf_SVKEY)	393	if (HeKLEN (he) == HEf_SVKEY)
345	{	394	{
…		…
358	encode_ch (enc, '"');	407	encode_ch (enc, '"');
359		408
360	if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);	409	if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
361	encode_ch (enc, ':');	410	encode_ch (enc, ':');
362	if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);	411	if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
363	encode_sv (enc, HeVAL (he));
364	}	412	}
365		413
366	// compare hash entries, used when all keys are bytestrings	414	// compare hash entries, used when all keys are bytestrings
367	static int	415	static int
368	he_cmp_fast (const void a_, const void b_)	416	he_cmp_fast (const void a_, const void b_)
…		…
373	HE b = (HE **)b_;	421	HE b = (HE **)b_;
374		422
375	STRLEN la = HeKLEN (a);	423	STRLEN la = HeKLEN (a);
376	STRLEN lb = HeKLEN (b);	424	STRLEN lb = HeKLEN (b);
377		425
378	if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))	426	if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
379	cmp = la - lb;	427	cmp = lb - la;
380		428
381	return cmp;	429	return cmp;
382	}	430	}
383		431
384	// compare hash entries, used when some keys are sv's or utf-x	432	// compare hash entries, used when some keys are sv's or utf-x
385	static int	433	static int
386	he_cmp_slow (const void a, const void b)	434	he_cmp_slow (const void a, const void b)
387	{	435	{
388	return sv_cmp (HeSVKEY_force ((HE )a), HeSVKEY_force ((HE **)b));	436	return sv_cmp (HeSVKEY_force ((HE )b), HeSVKEY_force ((HE **)a));
389	}	437	}
390		438
391	static void	439	static void
392	encode_hv (enc_t enc, HV hv)	440	encode_hv (enc_t enc, HV hv)
393	{	441	{
394	int count, i;	442	HE *he;
395		443
396	if (enc->indent >= enc->maxdepth)	444	if (enc->indent >= enc->maxdepth)
397	croak ("data structure too deep (hit recursion limit)");	445	croak ("data structure too deep (hit recursion limit)");
398		446
399	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;	447	encode_ch (enc, '{');
400		448
401	if ((count = hv_iterinit (hv)))
402	{
403	// for canonical output we have to sort by keys first	449	// for canonical output we have to sort by keys first
404	// actually, this is mostly due to the stupid so-called	450	// actually, this is mostly due to the stupid so-called
405	// security workaround added somewhere in 5.8.x.	451	// security workaround added somewhere in 5.8.x.
406	// that randomises hash orderings	452	// that randomises hash orderings
407	if (enc->json.flags & F_CANONICAL)	453	if (enc->json.flags & F_CANONICAL)
		454	{
		455	int count = hv_iterinit (hv);
		456
		457	if (SvMAGICAL (hv))
408	{	458	{
		459	// need to count by iterating. could improve by dynamically building the vector below
		460	// but I don't care for the speed of this special case.
		461	// note also that we will run into undefined behaviour when the two iterations
		462	// do not result in the same count, something I might care for in some later release.
		463
		464	count = 0;
		465	while (hv_iternext (hv))
		466	++count;
		467
		468	hv_iterinit (hv);
		469	}
		470
		471	if (count)
		472	{
409	int fast = 1;	473	int i, fast = 1;
410	HE *he;
411	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	474	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
412	HE *hes = _alloca (count sizeof (HE));	475	HE *hes = _alloca (count sizeof (HE));
413	#else	476	#else
414	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode	477	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
415	#endif	478	#endif
…		…
442		505
443	FREETMPS;	506	FREETMPS;
444	LEAVE;	507	LEAVE;
445	}	508	}
446		509
447	for (i = 0; i < count; ++i)	510	encode_nl (enc); ++enc->indent;
		511
		512	while (count--)
448	{	513	{
449	encode_indent (enc);	514	encode_indent (enc);
		515	he = hes [count];
450	encode_he (enc, hes [i]);	516	encode_hk (enc, he);
		517	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
451		518
452	if (i < count - 1)	519	if (count)
453	encode_comma (enc);	520	encode_comma (enc);
454	}	521	}
455		522
456	encode_nl (enc);	523	encode_nl (enc); --enc->indent; encode_indent (enc);
457	}	524	}
		525	}
458	else	526	else
		527	{
		528	if (hv_iterinit (hv) \|\| SvMAGICAL (hv))
		529	if ((he = hv_iternext (hv)))
459	{	530	{
460	HE *he = hv_iternext (hv);	531	encode_nl (enc); ++enc->indent;
461		532
462	for (;;)	533	for (;;)
463	{	534	{
464	encode_indent (enc);	535	encode_indent (enc);
465	encode_he (enc, he);	536	encode_hk (enc, he);
		537	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
466		538
467	if (!(he = hv_iternext (hv)))	539	if (!(he = hv_iternext (hv)))
468	break;	540	break;
469		541
470	encode_comma (enc);	542	encode_comma (enc);
471	}	543	}
472		544
473	encode_nl (enc);	545	encode_nl (enc); --enc->indent; encode_indent (enc);
474	}	546	}
475	}	547	}
476		548
477	--enc->indent; encode_indent (enc); encode_ch (enc, '}');	549	encode_ch (enc, '}');
478	}	550	}
479		551
480	// encode objects, arrays and special \0=false and \1=true values.	552	// encode objects, arrays and special \0=false and \1=true values.
481	static void	553	static void
482	encode_rv (enc_t enc, SV sv)	554	encode_rv (enc_t enc, SV sv)
…		…
590	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	662	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
591	enc->cur += strlen (enc->cur);	663	enc->cur += strlen (enc->cur);
592	}	664	}
593	else if (SvIOKp (sv))	665	else if (SvIOKp (sv))
594	{	666	{
595	// we assume we can always read an IV as a UV	667	// we assume we can always read an IV as a UV and vice versa
596	if (SvUV (sv) & ~(UV)0x7fff)	668	// we assume two's complement
597	{	669	// we assume no aliasing issues in the union
598	// large integer, use the (rather slow) snprintf way.	670	if (SvIsUV (sv) ? SvUVX (sv) <= 59000
599	need (enc, sizeof (UV) * 3);	671	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
600	enc->cur +=
601	SvIsUV(sv)
602	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
603	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
604	}
605	else
606	{	672	{
607	// optimise the "small number case"	673	// optimise the "small number case"
608	// code will likely be branchless and use only a single multiplication	674	// code will likely be branchless and use only a single multiplication
		675	// works for numbers up to 59074
609	I32 i = SvIV (sv);	676	I32 i = SvIVX (sv);
610	U32 u;	677	U32 u;
611	char digit, nz = 0;	678	char digit, nz = 0;
612		679
613	need (enc, 6);	680	need (enc, 6);
614		681
…		…
620		687
621	// now output digit by digit, each time masking out the integer part	688	// now output digit by digit, each time masking out the integer part
622	// and multiplying by 5 while moving the decimal point one to the right,	689	// and multiplying by 5 while moving the decimal point one to the right,
623	// resulting in a net multiplication by 10.	690	// resulting in a net multiplication by 10.
624	// we always write the digit to memory but conditionally increment	691	// we always write the digit to memory but conditionally increment
625	// the pointer, to ease the usage of conditional move instructions.	692	// the pointer, to enable the use of conditional move instructions.
626	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffff) 5;	693	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffffUL) 5;
627	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffff) 5;	694	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffffUL) 5;
628	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffff) 5;	695	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffffUL) 5;
629	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffff) 5;	696	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffffUL) 5;
630	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'	697	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
		698	}
		699	else
		700	{
		701	// large integer, use the (rather slow) snprintf way.
		702	need (enc, IVUV_MAXCHARS);
		703	enc->cur +=
		704	SvIsUV(sv)
		705	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
		706	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
631	}	707	}
632	}	708	}
633	else if (SvROK (sv))	709	else if (SvROK (sv))
634	encode_rv (enc, SvRV (sv));	710	encode_rv (enc, SvRV (sv));
635	else if (!SvOK (sv))	711	else if (!SvOK (sv))
…		…
651	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	727	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
652	enc.cur = SvPVX (enc.sv);	728	enc.cur = SvPVX (enc.sv);
653	enc.end = SvEND (enc.sv);	729	enc.end = SvEND (enc.sv);
654	enc.indent = 0;	730	enc.indent = 0;
655	enc.maxdepth = DEC_DEPTH (enc.json.flags);	731	enc.maxdepth = DEC_DEPTH (enc.json.flags);
		732	enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
		733	: enc.json.flags & F_LATIN1 ? 0x000100UL
		734	: 0x110000UL;
656		735
657	SvPOK_only (enc.sv);	736	SvPOK_only (enc.sv);
658	encode_sv (&enc, scalar);	737	encode_sv (&enc, scalar);
659		738
660	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	739	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
…		…
681	JSON json;	760	JSON json;
682	U32 depth; // recursion depth	761	U32 depth; // recursion depth
683	U32 maxdepth; // recursion depth limit	762	U32 maxdepth; // recursion depth limit
684	} dec_t;	763	} dec_t;
685		764
686	inline void	765	INLINE void
		766	decode_comment (dec_t *dec)
		767	{
		768	// only '#'-style comments allowed a.t.m.
		769
		770	while (dec->cur && dec->cur != 0x0a && *dec->cur != 0x0d)
		771	++dec->cur;
		772	}
		773
		774	INLINE void
687	decode_ws (dec_t *dec)	775	decode_ws (dec_t *dec)
688	{	776	{
689	for (;;)	777	for (;;)
690	{	778	{
691	char ch = *dec->cur;	779	char ch = *dec->cur;
692		780
693	if (ch > 0x20	781	if (ch > 0x20)
		782	{
		783	if (expect_false (ch == '#'))
		784	{
		785	if (dec->json.flags & F_RELAXED)
		786	decode_comment (dec);
		787	else
		788	break;
		789	}
		790	else
		791	break;
		792	}
694	\|\| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))	793	else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
695	break;	794	break; // parse error, but let higher level handle it, gives better error messages
696		795
697	++dec->cur;	796	++dec->cur;
698	}	797	}
699	}	798	}
700		799
…		…
806		905
807	if (hi >= 0x80)	906	if (hi >= 0x80)
808	{	907	{
809	utf8 = 1;	908	utf8 = 1;
810		909
811	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);	910	cur = encode_utf8 (cur, hi);
812	}	911	}
813	else	912	else
814	*cur++ = hi;	913	*cur++ = hi;
815	}	914	}
816	break;	915	break;
…		…
818	default:	917	default:
819	--dec_cur;	918	--dec_cur;
820	ERR ("illegal backslash escape sequence in string");	919	ERR ("illegal backslash escape sequence in string");
821	}	920	}
822	}	921	}
823	else if (expect_true (ch >= 0x20 && ch <= 0x7f))	922	else if (expect_true (ch >= 0x20 && ch < 0x80))
824	*cur++ = ch;	923	*cur++ = ch;
825	else if (ch >= 0x80)	924	else if (ch >= 0x80)
826	{	925	{
827	STRLEN clen;	926	STRLEN clen;
828	UV uch;	927	UV uch;
…		…
951		1050
952	if (!is_nv)	1051	if (!is_nv)
953	{	1052	{
954	int len = dec->cur - start;	1053	int len = dec->cur - start;
955		1054
956	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so	1055	// special case the rather common 1..5-digit-int case
957	if (*start == '-')	1056	if (*start == '-')
958	switch (len)	1057	switch (len)
959	{	1058	{
960	case 2: return newSViv (-( start [1] - '0' * 1));	1059	case 2: return newSViv (-( start [1] - '0' * 1));
961	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));	1060	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
962	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));	1061	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
963	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));	1062	case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
		1063	case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
964	}	1064	}
965	else	1065	else
966	switch (len)	1066	switch (len)
967	{	1067	{
968	case 1: return newSViv ( start [0] - '0' * 1);	1068	case 1: return newSViv ( start [0] - '0' * 1);
969	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);	1069	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
970	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);	1070	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
971	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);	1071	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		1072	case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
972	}	1073	}
973		1074
974	{	1075	{
975	UV uv;	1076	UV uv;
976	int numtype = grok_number (start, len, &uv);	1077	int numtype = grok_number (start, len, &uv);
…		…
1037		1138
1038	if (*dec->cur != ',')	1139	if (*dec->cur != ',')
1039	ERR (", or ] expected while parsing array");	1140	ERR (", or ] expected while parsing array");
1040		1141
1041	++dec->cur;	1142	++dec->cur;
		1143
		1144	decode_ws (dec);
		1145
		1146	if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
		1147	{
		1148	++dec->cur;
		1149	break;
		1150	}
1042	}	1151	}
1043		1152
1044	DEC_DEC_DEPTH;	1153	DEC_DEC_DEPTH;
1045	return newRV_noinc ((SV *)av);	1154	return newRV_noinc ((SV *)av);
1046		1155
…		…
1062	if (*dec->cur == '}')	1171	if (*dec->cur == '}')
1063	++dec->cur;	1172	++dec->cur;
1064	else	1173	else
1065	for (;;)	1174	for (;;)
1066	{	1175	{
1067	decode_ws (dec); EXPECT_CH ('"');	1176	EXPECT_CH ('"');
1068		1177
1069	// heuristic: assume that	1178	// heuristic: assume that
1070	// a) decode_str + hv_store_ent are abysmally slow.	1179	// a) decode_str + hv_store_ent are abysmally slow.
1071	// b) most hash keys are short, simple ascii text.	1180	// b) most hash keys are short, simple ascii text.
1072	// => try to "fast-match" such strings to avoid	1181	// => try to "fast-match" such strings to avoid
…		…
1076	char *p = dec->cur;	1185	char *p = dec->cur;
1077	char *e = p + 24; // only try up to 24 bytes	1186	char *e = p + 24; // only try up to 24 bytes
1078		1187
1079	for (;;)	1188	for (;;)
1080	{	1189	{
1081	// the >= 0x80 is true on most architectures	1190	// the >= 0x80 is false on most architectures
1082	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')	1191	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
1083	{	1192	{
1084	// slow path, back up and use decode_str	1193	// slow path, back up and use decode_str
1085	SV *key = decode_str (dec);	1194	SV *key = decode_str (dec);
1086	if (!key)	1195	if (!key)
1087	goto fail;	1196	goto fail;
1088		1197
1089	decode_ws (dec); EXPECT_CH (':');	1198	decode_ws (dec); EXPECT_CH (':');
1090		1199
		1200	decode_ws (dec);
1091	value = decode_sv (dec);	1201	value = decode_sv (dec);
1092	if (!value)	1202	if (!value)
1093	{	1203	{
1094	SvREFCNT_dec (key);	1204	SvREFCNT_dec (key);
1095	goto fail;	1205	goto fail;
…		…
1107	int len = p - key;	1217	int len = p - key;
1108	dec->cur = p + 1;	1218	dec->cur = p + 1;
1109		1219
1110	decode_ws (dec); EXPECT_CH (':');	1220	decode_ws (dec); EXPECT_CH (':');
1111		1221
		1222	decode_ws (dec);
1112	value = decode_sv (dec);	1223	value = decode_sv (dec);
1113	if (!value)	1224	if (!value)
1114	goto fail;	1225	goto fail;
1115		1226
1116	hv_store (hv, key, len, value, 0);	1227	hv_store (hv, key, len, value, 0);
…		…
1132		1243
1133	if (*dec->cur != ',')	1244	if (*dec->cur != ',')
1134	ERR (", or } expected while parsing object/hash");	1245	ERR (", or } expected while parsing object/hash");
1135		1246
1136	++dec->cur;	1247	++dec->cur;
		1248
		1249	decode_ws (dec);
		1250
		1251	if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
		1252	{
		1253	++dec->cur;
		1254	break;
		1255	}
1137	}	1256	}
1138		1257
1139	DEC_DEC_DEPTH;	1258	DEC_DEC_DEPTH;
1140	sv = newRV_noinc ((SV *)hv);	1259	sv = newRV_noinc ((SV *)hv);
1141		1260
…		…
1206	}	1325	}
1207		1326
1208	static SV *	1327	static SV *
1209	decode_sv (dec_t *dec)	1328	decode_sv (dec_t *dec)
1210	{	1329	{
1211	decode_ws (dec);
1212
1213	// the beauty of JSON: you need exactly one character lookahead	1330	// the beauty of JSON: you need exactly one character lookahead
1214	// to parse anything.	1331	// to parse everything.
1215	switch (*dec->cur)	1332	switch (*dec->cur)
1216	{	1333	{
1217	case '"': ++dec->cur; return decode_str (dec);	1334	case '"': ++dec->cur; return decode_str (dec);
1218	case '[': ++dec->cur; return decode_av (dec);	1335	case '[': ++dec->cur; return decode_av (dec);
1219	case '{': ++dec->cur; return decode_hv (dec);	1336	case '{': ++dec->cur; return decode_hv (dec);
1220		1337
1221	case '-':	1338	case '-':
1222	case '0': case '1': case '2': case '3': case '4':	1339	case '0': case '1': case '2': case '3': case '4':
1223	case '5': case '6': case '7': case '8': case '9':	1340	case '5': case '6': case '7': case '8': case '9':
1224	return decode_num (dec);	1341	return decode_num (dec);
…		…
1270	fail:	1387	fail:
1271	return 0;	1388	return 0;
1272	}	1389	}
1273		1390
1274	static SV *	1391	static SV *
1275	decode_json (SV string, JSON json, UV *offset_return)	1392	decode_json (SV string, JSON json, STRLEN *offset_return)
1276	{	1393	{
1277	dec_t dec;	1394	dec_t dec;
1278	UV offset;	1395	STRLEN offset;
1279	SV *sv;	1396	SV *sv;
1280		1397
1281	SvGETMAGIC (string);	1398	SvGETMAGIC (string);
1282	SvUPGRADE (string, SVt_PV);	1399	SvUPGRADE (string, SVt_PV);
1283		1400
…		…
1301		1418
1302	if (dec.json.cb_object \|\| dec.json.cb_sk_object)	1419	if (dec.json.cb_object \|\| dec.json.cb_sk_object)
1303	dec.json.flags \|= F_HOOK;	1420	dec.json.flags \|= F_HOOK;
1304		1421
1305	*dec.end = 0; // this should basically be a nop, too, but make sure it's there	1422	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
		1423
		1424	decode_ws (&dec);
1306	sv = decode_sv (&dec);	1425	sv = decode_sv (&dec);
1307		1426
1308	if (!(offset_return \|\| !sv))	1427	if (!(offset_return \|\| !sv))
1309	{	1428	{
1310	// check for trailing garbage	1429	// check for trailing garbage
…		…
1354		1473
1355	return sv;	1474	return sv;
1356	}	1475	}
1357		1476
1358	/////////////////////////////////////////////////////////////////////////////	1477	/////////////////////////////////////////////////////////////////////////////
		1478	// incremental parser
		1479
		1480	static void
		1481	incr_parse (JSON *self)
		1482	{
		1483	const char *p = SvPVX (self->incr_text) + self->incr_pos;
		1484
		1485	for (;;)
		1486	{
		1487	//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
		1488	switch (self->incr_mode)
		1489	{
		1490	// only used for intiial whitespace skipping
		1491	case INCR_M_WS:
		1492	for (;;)
		1493	{
		1494	if (*p > 0x20)
		1495	{
		1496	self->incr_mode = INCR_M_JSON;
		1497	goto incr_m_json;
		1498	}
		1499	else if (!*p)
		1500	goto interrupt;
		1501
		1502	++p;
		1503	}
		1504
		1505	// skip a single char inside a string (for \\-processing)
		1506	case INCR_M_BS:
		1507	if (!*p)
		1508	goto interrupt;
		1509
		1510	++p;
		1511	self->incr_mode = INCR_M_STR;
		1512	goto incr_m_str;
		1513
		1514	// inside a string
		1515	case INCR_M_STR:
		1516	incr_m_str:
		1517	for (;;)
		1518	{
		1519	if (*p == '"')
		1520	{
		1521	++p;
		1522	self->incr_mode = INCR_M_JSON;
		1523
		1524	if (!self->incr_nest)
		1525	goto interrupt;
		1526
		1527	goto incr_m_json;
		1528	}
		1529	else if (*p == '\\')
		1530	{
		1531	++p; // "virtually" consumes character after \
		1532
		1533	if (!*p) // if at end of string we have to switch modes
		1534	{
		1535	self->incr_mode = INCR_M_BS;
		1536	goto interrupt;
		1537	}
		1538	}
		1539	else if (!*p)
		1540	goto interrupt;
		1541
		1542	++p;
		1543	}
		1544
		1545	// after initial ws, outside string
		1546	case INCR_M_JSON:
		1547	incr_m_json:
		1548	for (;;)
		1549	{
		1550	switch (*p++)
		1551	{
		1552	case 0:
		1553	--p;
		1554	goto interrupt;
		1555
		1556	case 0x09:
		1557	case 0x0a:
		1558	case 0x0d:
		1559	case 0x20:
		1560	if (!self->incr_nest)
		1561	{
		1562	--p; // do not eat the whitespace, let the next round do it
		1563	goto interrupt;
		1564	}
		1565	break;
		1566
		1567	case '"':
		1568	self->incr_mode = INCR_M_STR;
		1569	goto incr_m_str;
		1570
		1571	case '[':
		1572	case '{':
		1573	++self->incr_nest;
		1574	break;
		1575
		1576	case ']':
		1577	case '}':
		1578	if (!--self->incr_nest)
		1579	goto interrupt;
		1580	}
		1581	}
		1582	}
		1583
		1584	modechange:
		1585	;
		1586	}
		1587
		1588	interrupt:
		1589	self->incr_pos = p - SvPVX (self->incr_text);
		1590	//printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
		1591	}
		1592
		1593	/////////////////////////////////////////////////////////////////////////////
1359	// XS interface functions	1594	// XS interface functions
1360		1595
1361	MODULE = JSON::XS PACKAGE = JSON::XS	1596	MODULE = JSON::XS PACKAGE = JSON::XS
1362		1597
1363	BOOT:	1598	BOOT:
…		…
1390	{	1625	{
1391	SV *pv = NEWSV (0, sizeof (JSON));	1626	SV *pv = NEWSV (0, sizeof (JSON));
1392	SvPOK_only (pv);	1627	SvPOK_only (pv);
1393	Zero (SvPVX (pv), 1, JSON);	1628	Zero (SvPVX (pv), 1, JSON);
1394	((JSON *)SvPVX (pv))->flags = F_DEFAULT;	1629	((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1395	XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));	1630	XPUSHs (sv_2mortal (sv_bless (
		1631	newRV_noinc (pv),
		1632	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
		1633	)));
1396	}	1634	}
1397		1635
1398	void ascii (JSON *self, int enable = 1)	1636	void ascii (JSON *self, int enable = 1)
1399	ALIAS:	1637	ALIAS:
1400	ascii = F_ASCII	1638	ascii = F_ASCII
…		…
1407	pretty = F_PRETTY	1645	pretty = F_PRETTY
1408	allow_nonref = F_ALLOW_NONREF	1646	allow_nonref = F_ALLOW_NONREF
1409	shrink = F_SHRINK	1647	shrink = F_SHRINK
1410	allow_blessed = F_ALLOW_BLESSED	1648	allow_blessed = F_ALLOW_BLESSED
1411	convert_blessed = F_CONV_BLESSED	1649	convert_blessed = F_CONV_BLESSED
		1650	relaxed = F_RELAXED
1412	PPCODE:	1651	PPCODE:
1413	{	1652	{
1414	if (enable)	1653	if (enable)
1415	self->flags \|= ix;	1654	self->flags \|= ix;
1416	else	1655	else
1417	self->flags &= ~ix;	1656	self->flags &= ~ix;
1418		1657
1419	XPUSHs (ST (0));	1658	XPUSHs (ST (0));
1420	}	1659	}
1421		1660
		1661	void get_ascii (JSON *self)
		1662	ALIAS:
		1663	get_ascii = F_ASCII
		1664	get_latin1 = F_LATIN1
		1665	get_utf8 = F_UTF8
		1666	get_indent = F_INDENT
		1667	get_canonical = F_CANONICAL
		1668	get_space_before = F_SPACE_BEFORE
		1669	get_space_after = F_SPACE_AFTER
		1670	get_allow_nonref = F_ALLOW_NONREF
		1671	get_shrink = F_SHRINK
		1672	get_allow_blessed = F_ALLOW_BLESSED
		1673	get_convert_blessed = F_CONV_BLESSED
		1674	get_relaxed = F_RELAXED
		1675	PPCODE:
		1676	XPUSHs (boolSV (self->flags & ix));
		1677
1422	void max_depth (JSON *self, UV max_depth = 0x80000000UL)	1678	void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1423	PPCODE:	1679	PPCODE:
1424	{	1680	{
1425	UV log2 = 0;	1681	UV log2 = 0;
1426		1682
…		…
1432	self->flags = self->flags & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);	1688	self->flags = self->flags & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
1433		1689
1434	XPUSHs (ST (0));	1690	XPUSHs (ST (0));
1435	}	1691	}
1436		1692
		1693	U32 get_max_depth (JSON *self)
		1694	CODE:
		1695	RETVAL = DEC_DEPTH (self->flags);
		1696	OUTPUT:
		1697	RETVAL
		1698
1437	void max_size (JSON *self, UV max_size = 0)	1699	void max_size (JSON *self, UV max_size = 0)
1438	PPCODE:	1700	PPCODE:
1439	{	1701	{
1440	UV log2 = 0;	1702	UV log2 = 0;
1441		1703
…		…
1447		1709
1448	self->flags = self->flags & ~F_MAXSIZE \| (log2 << S_MAXSIZE);	1710	self->flags = self->flags & ~F_MAXSIZE \| (log2 << S_MAXSIZE);
1449		1711
1450	XPUSHs (ST (0));	1712	XPUSHs (ST (0));
1451	}	1713	}
		1714
		1715	int get_max_size (JSON *self)
		1716	CODE:
		1717	RETVAL = DEC_SIZE (self->flags);
		1718	OUTPUT:
		1719	RETVAL
1452		1720
1453	void filter_json_object (JSON self, SV cb = &PL_sv_undef)	1721	void filter_json_object (JSON self, SV cb = &PL_sv_undef)
1454	PPCODE:	1722	PPCODE:
1455	{	1723	{
1456	SvREFCNT_dec (self->cb_object);	1724	SvREFCNT_dec (self->cb_object);
…		…
1490	XPUSHs (decode_json (jsonstr, self, 0));	1758	XPUSHs (decode_json (jsonstr, self, 0));
1491		1759
1492	void decode_prefix (JSON self, SV jsonstr)	1760	void decode_prefix (JSON self, SV jsonstr)
1493	PPCODE:	1761	PPCODE:
1494	{	1762	{
1495	UV offset;	1763	STRLEN offset;
1496	EXTEND (SP, 2);	1764	EXTEND (SP, 2);
1497	PUSHs (decode_json (jsonstr, self, &offset));	1765	PUSHs (decode_json (jsonstr, self, &offset));
1498	PUSHs (sv_2mortal (newSVuv (offset)));	1766	PUSHs (sv_2mortal (newSVuv (offset)));
1499	}	1767	}
		1768
		1769	void incr_parse (JSON self, SV jsonstr = 0)
		1770	PPCODE:
		1771	{
		1772	if (!self->incr_text)
		1773	self->incr_text = newSVpvn ("", 0);
		1774
		1775	// append data, if any
		1776	if (jsonstr)
		1777	{
		1778	if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
		1779	{
		1780	/* utf-8-ness differs, need to upgrade */
		1781	sv_utf8_upgrade (self->incr_text);
		1782
		1783	if (self->incr_pos)
		1784	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
		1785	- (U8 *)SvPVX (self->incr_text);
		1786	}
		1787
		1788	{
		1789	STRLEN len;
		1790	const char *str = SvPV (jsonstr, len);
		1791	SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
		1792	Move (str, SvEND (self->incr_text), len, char);
		1793	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
		1794	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
		1795	}
		1796	}
		1797
		1798	if (GIMME_V != G_VOID)
		1799	do
		1800	{
		1801	STRLEN offset;
		1802
		1803	incr_parse (self);
		1804
		1805	if (!INCR_DONE (self))
		1806	break;
		1807
		1808	XPUSHs (decode_json (self->incr_text, self, &offset));
		1809
		1810	sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
		1811	self->incr_pos -= offset;
		1812	self->incr_nest = 0;
		1813	self->incr_mode = 0;
		1814	}
		1815	while (GIMME_V == G_ARRAY);
		1816	}
		1817
		1818	SV incr_text (JSON self)
		1819	ATTRS: lvalue
		1820	CODE:
		1821	{
		1822	if (self->incr_pos)
		1823	croak ("incr_text can only be called after a successful incr_parse call in scalar context %d", self->incr_pos);//D
		1824
		1825	RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
		1826	}
		1827	OUTPUT:
		1828	RETVAL
1500		1829
1501	void DESTROY (JSON *self)	1830	void DESTROY (JSON *self)
1502	CODE:	1831	CODE:
1503	SvREFCNT_dec (self->cb_sk_object);	1832	SvREFCNT_dec (self->cb_sk_object);
1504	SvREFCNT_dec (self->cb_object);	1833	SvREFCNT_dec (self->cb_object);
		1834	SvREFCNT_dec (self->incr_text);
1505		1835
1506	PROTOTYPES: ENABLE	1836	PROTOTYPES: ENABLE
1507		1837
1508	void to_json (SV *scalar)	1838	void encode_json (SV *scalar)
		1839	ALIAS:
		1840	to_json_ = 0
		1841	encode_json = F_UTF8
1509	PPCODE:	1842	PPCODE:
1510	{	1843	{
1511	JSON json = { F_DEFAULT \| F_UTF8 };	1844	JSON json = { F_DEFAULT \| ix };
1512	XPUSHs (encode_json (scalar, &json));	1845	XPUSHs (encode_json (scalar, &json));
1513	}	1846	}
1514		1847
1515	void from_json (SV *jsonstr)	1848	void decode_json (SV *jsonstr)
		1849	ALIAS:
		1850	from_json_ = 0
		1851	decode_json = F_UTF8
1516	PPCODE:	1852	PPCODE:
1517	{	1853	{
1518	JSON json = { F_DEFAULT \| F_UTF8 };	1854	JSON json = { F_DEFAULT \| ix };
1519	XPUSHs (decode_json (jsonstr, &json, 0));	1855	XPUSHs (decode_json (jsonstr, &json, 0));
1520	}	1856	}
1521		1857
		1858

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.60 by root, Mon Aug 13 16:19:13 2007 UTC vs. Revision 1.78 by root, Tue Mar 25 19:31:02 2008 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.60 by root, Mon Aug 13 16:19:13 2007 UTC vs.
Revision 1.78 by root, Tue Mar 25 19:31:02 2008 UTC