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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.64 by root, Mon Aug 27 02:03:23 2007 UTC vs.
Revision 1.77 by root, Tue Mar 25 06:37:38 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
…		…
50		53
51	#define SB do {	54	#define SB do {
52	#define SE } while (0)	55	#define SE } while (0)
53		56
54	#if __GNUC__ >= 3	57	#if __GNUC__ >= 3
55	# define expect(expr,value) __builtin_expect ((expr),(value))	58	# define expect(expr,value) __builtin_expect ((expr), (value))
56	# define inline inline	59	# define INLINE static inline
57	#else	60	#else
58	# define expect(expr,value) (expr)	61	# define expect(expr,value) (expr)
59	# define inline static	62	# define INLINE static
60	#endif	63	#endif
61		64
62	#define expect_false(expr) expect ((expr) != 0, 0)	65	#define expect_false(expr) expect ((expr) != 0, 0)
63	#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)))
64		71
65	#ifdef USE_ITHREADS	72	#ifdef USE_ITHREADS
66	# define JSON_SLOW 1	73	# define JSON_SLOW 1
67	# 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))
68	#else	75	#else
…		…
71	#endif	78	#endif
72		79
73	static HV json_stash, json_boolean_stash; // JSON::XS::	80	static HV json_stash, json_boolean_stash; // JSON::XS::
74	static SV json_true, json_false;	81	static SV json_true, json_false;
75		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
76	typedef struct {	92	typedef struct {
77	U32 flags;	93	U32 flags;
78	SV *cb_object;	94	SV *cb_object;
79	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;
80	} JSON;	102	} JSON;
81		103
82	/////////////////////////////////////////////////////////////////////////////	104	/////////////////////////////////////////////////////////////////////////////
83	// utility functions	105	// utility functions
84		106
85	inline void	107	INLINE void
86	shrink (SV *sv)	108	shrink (SV *sv)
87	{	109	{
88	sv_utf8_downgrade (sv, 1);	110	sv_utf8_downgrade (sv, 1);
89	if (SvLEN (sv) > SvCUR (sv) + 1)	111	if (SvLEN (sv) > SvCUR (sv) + 1)
90	{	112	{
…		…
99	// 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
100	// case of an error.	122	// case of an error.
101	// we special-case "safe" characters from U+80 .. U+7FF,	123	// we special-case "safe" characters from U+80 .. U+7FF,
102	// but use the very good perl function to parse anything else.	124	// but use the very good perl function to parse anything else.
103	// note that we never call this function for a ascii codepoints	125	// note that we never call this function for a ascii codepoints
104	inline UV	126	INLINE UV
105	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)	127	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
106	{	128	{
107	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))	129	if (expect_true (len >= 2
108	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);	130	&& IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
109	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)	131	&& IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
110	{	132	{
111	*clen = 2;	133	*clen = 2;
112	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);	134	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
113	}	135	}
114	else	136	else
115	{	137	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
116	*clen = (STRLEN)-1;	138	}
117	return (UV)-1;	139
118	}	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;
119	}	162	}
120		163
121	/////////////////////////////////////////////////////////////////////////////	164	/////////////////////////////////////////////////////////////////////////////
122	// encoder	165	// encoder
123		166
…		…
128	char *end; // SvEND (sv)	171	char *end; // SvEND (sv)
129	SV *sv; // result scalar	172	SV *sv; // result scalar
130	JSON json;	173	JSON json;
131	U32 indent; // indentation level	174	U32 indent; // indentation level
132	U32 maxdepth; // max. indentation/recursion level	175	U32 maxdepth; // max. indentation/recursion level
		176	UV limit; // escape character values >= this value when encoding
133	} enc_t;	177	} enc_t;
134		178
135	inline void	179	INLINE void
136	need (enc_t *enc, STRLEN len)	180	need (enc_t *enc, STRLEN len)
137	{	181	{
138	if (expect_false (enc->cur + len >= enc->end))	182	if (expect_false (enc->cur + len >= enc->end))
139	{	183	{
140	STRLEN cur = enc->cur - SvPVX (enc->sv);	184	STRLEN cur = enc->cur - SvPVX (enc->sv);
…		…
142	enc->cur = SvPVX (enc->sv) + cur;	186	enc->cur = SvPVX (enc->sv) + cur;
143	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	187	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
144	}	188	}
145	}	189	}
146		190
147	inline void	191	INLINE void
148	encode_ch (enc_t *enc, char ch)	192	encode_ch (enc_t *enc, char ch)
149	{	193	{
150	need (enc, 1);	194	need (enc, 1);
151	*enc->cur++ = ch;	195	*enc->cur++ = ch;
152	}	196	}
…		…
206	{	250	{
207	uch = ch;	251	uch = ch;
208	clen = 1;	252	clen = 1;
209	}	253	}
210		254
211	if (uch > 0x10FFFFUL)	255	if (uch < 0x80/0x20/ \|\| uch >= enc->limit)
212	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
213
214	if (uch < 0x80 \|\| enc->json.flags & F_ASCII \|\| (enc->json.flags & F_LATIN1 && uch > 0xFF))
215	{	256	{
216	if (uch > 0xFFFFUL)	257	if (uch >= 0x10000UL)
217	{	258	{
		259	if (uch >= 0x110000UL)
		260	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
		261
218	need (enc, len += 11);	262	need (enc, len += 11);
219	sprintf (enc->cur, "\\u%04x\\u%04x",	263	sprintf (enc->cur, "\\u%04x\\u%04x",
220	(int)((uch - 0x10000) / 0x400 + 0xD800),	264	(int)((uch - 0x10000) / 0x400 + 0xD800),
221	(int)((uch - 0x10000) % 0x400 + 0xDC00));	265	(int)((uch - 0x10000) % 0x400 + 0xDC00));
222	enc->cur += 12;	266	enc->cur += 12;
…		…
250	while (--clen);	294	while (--clen);
251	}	295	}
252	else	296	else
253	{	297	{
254	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
255	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	299	enc->cur = encode_utf8 (enc->cur, uch);
256	++str;	300	++str;
257	}	301	}
258	}	302	}
259	}	303	}
260	}	304	}
261		305
262	--len;	306	--len;
263	}	307	}
264	}	308	}
265		309
266	inline void	310	INLINE void
267	encode_indent (enc_t *enc)	311	encode_indent (enc_t *enc)
268	{	312	{
269	if (enc->json.flags & F_INDENT)	313	if (enc->json.flags & F_INDENT)
270	{	314	{
271	int spaces = enc->indent * INDENT_STEP;	315	int spaces = enc->indent * INDENT_STEP;
…		…
274	memset (enc->cur, ' ', spaces);	318	memset (enc->cur, ' ', spaces);
275	enc->cur += spaces;	319	enc->cur += spaces;
276	}	320	}
277	}	321	}
278		322
279	inline void	323	INLINE void
280	encode_space (enc_t *enc)	324	encode_space (enc_t *enc)
281	{	325	{
282	need (enc, 1);	326	need (enc, 1);
283	encode_ch (enc, ' ');	327	encode_ch (enc, ' ');
284	}	328	}
285		329
286	inline void	330	INLINE void
287	encode_nl (enc_t *enc)	331	encode_nl (enc_t *enc)
288	{	332	{
289	if (enc->json.flags & F_INDENT)	333	if (enc->json.flags & F_INDENT)
290	{	334	{
291	need (enc, 1);	335	need (enc, 1);
292	encode_ch (enc, '\n');	336	encode_ch (enc, '\n');
293	}	337	}
294	}	338	}
295		339
296	inline void	340	INLINE void
297	encode_comma (enc_t *enc)	341	encode_comma (enc_t *enc)
298	{	342	{
299	encode_ch (enc, ',');	343	encode_ch (enc, ',');
300		344
301	if (enc->json.flags & F_INDENT)	345	if (enc->json.flags & F_INDENT)
…		…
312	int i, len = av_len (av);	356	int i, len = av_len (av);
313		357
314	if (enc->indent >= enc->maxdepth)	358	if (enc->indent >= enc->maxdepth)
315	croak ("data structure too deep (hit recursion limit)");	359	croak ("data structure too deep (hit recursion limit)");
316		360
317	encode_ch (enc, '['); encode_nl (enc);	361	encode_ch (enc, '[');
318	++enc->indent;	362
		363	if (len >= 0)
		364	{
		365	encode_nl (enc); ++enc->indent;
319		366
320	for (i = 0; i <= len; ++i)	367	for (i = 0; i <= len; ++i)
321	{	368	{
322	SV **svp = av_fetch (av, i, 0);	369	SV **svp = av_fetch (av, i, 0);
323		370
324	encode_indent (enc);	371	encode_indent (enc);
325		372
326	if (svp)	373	if (svp)
327	encode_sv (enc, *svp);	374	encode_sv (enc, *svp);
328	else	375	else
329	encode_str (enc, "null", 4, 0);	376	encode_str (enc, "null", 4, 0);
330		377
331	if (i < len)	378	if (i < len)
332	encode_comma (enc);	379	encode_comma (enc);
333	}	380	}
334		381
		382	encode_nl (enc); --enc->indent; encode_indent (enc);
		383	}
		384
335	encode_nl (enc);	385	encode_ch (enc, ']');
336
337	--enc->indent;
338	encode_indent (enc); encode_ch (enc, ']');
339	}	386	}
340		387
341	static void	388	static void
342	encode_hk (enc_t enc, HE he)	389	encode_hk (enc_t enc, HE he)
343	{	390	{
…		…
396	int count;	443	int count;
397		444
398	if (enc->indent >= enc->maxdepth)	445	if (enc->indent >= enc->maxdepth)
399	croak ("data structure too deep (hit recursion limit)");	446	croak ("data structure too deep (hit recursion limit)");
400		447
401	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;	448	encode_ch (enc, '{');
402		449
403	// for canonical output we have to sort by keys first	450	// for canonical output we have to sort by keys first
404	// actually, this is mostly due to the stupid so-called	451	// actually, this is mostly due to the stupid so-called
405	// security workaround added somewhere in 5.8.x.	452	// security workaround added somewhere in 5.8.x.
406	// that randomises hash orderings	453	// that randomises hash orderings
…		…
459		506
460	FREETMPS;	507	FREETMPS;
461	LEAVE;	508	LEAVE;
462	}	509	}
463		510
		511	encode_nl (enc); ++enc->indent;
		512
464	while (count--)	513	while (count--)
465	{	514	{
466	encode_indent (enc);	515	encode_indent (enc);
467	he = hes [count];	516	he = hes [count];
468	encode_hk (enc, he);	517	encode_hk (enc, he);
469	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));	518	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
470		519
471	if (count)	520	if (count)
472	encode_comma (enc);	521	encode_comma (enc);
473	}	522	}
		523
		524	encode_nl (enc); --enc->indent; encode_indent (enc);
474	}	525	}
475	}	526	}
476	else	527	else
477	{	528	{
478	if (hv_iterinit (hv) \|\| SvMAGICAL (hv))	529	if (hv_iterinit (hv) \|\| SvMAGICAL (hv))
479	if ((he = hv_iternext (hv)))	530	if ((he = hv_iternext (hv)))
		531	{
		532	encode_nl (enc); ++enc->indent;
		533
480	for (;;)	534	for (;;)
481	{	535	{
482	encode_indent (enc);	536	encode_indent (enc);
483	encode_hk (enc, he);	537	encode_hk (enc, he);
484	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));	538	encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
485		539
486	if (!(he = hv_iternext (hv)))	540	if (!(he = hv_iternext (hv)))
487	break;	541	break;
488		542
489	encode_comma (enc);	543	encode_comma (enc);
490	}	544	}
491	}
492		545
		546	encode_nl (enc); --enc->indent; encode_indent (enc);
		547	}
		548	}
		549
493	encode_nl (enc);	550	encode_ch (enc, '}');
494
495	--enc->indent; encode_indent (enc); encode_ch (enc, '}');
496	}	551	}
497		552
498	// encode objects, arrays and special \0=false and \1=true values.	553	// encode objects, arrays and special \0=false and \1=true values.
499	static void	554	static void
500	encode_rv (enc_t enc, SV sv)	555	encode_rv (enc_t enc, SV sv)
…		…
608	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	663	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
609	enc->cur += strlen (enc->cur);	664	enc->cur += strlen (enc->cur);
610	}	665	}
611	else if (SvIOKp (sv))	666	else if (SvIOKp (sv))
612	{	667	{
613	// we assume we can always read an IV as a UV	668	// we assume we can always read an IV as a UV and vice versa
614	if (SvUV (sv) & ~(UV)0x7fff)	669	// we assume two's complement
615	{	670	// we assume no aliasing issues in the union
616	// large integer, use the (rather slow) snprintf way.	671	if (SvIsUV (sv) ? SvUVX (sv) <= 59000
617	need (enc, sizeof (UV) * 3);	672	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
618	enc->cur +=
619	SvIsUV(sv)
620	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
621	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
622	}
623	else
624	{	673	{
625	// optimise the "small number case"	674	// optimise the "small number case"
626	// code will likely be branchless and use only a single multiplication	675	// code will likely be branchless and use only a single multiplication
		676	// works for numbers up to 59074
627	I32 i = SvIV (sv);	677	I32 i = SvIVX (sv);
628	U32 u;	678	U32 u;
629	char digit, nz = 0;	679	char digit, nz = 0;
630		680
631	need (enc, 6);	681	need (enc, 6);
632		682
…		…
638		688
639	// now output digit by digit, each time masking out the integer part	689	// 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,	690	// and multiplying by 5 while moving the decimal point one to the right,
641	// resulting in a net multiplication by 10.	691	// resulting in a net multiplication by 10.
642	// we always write the digit to memory but conditionally increment	692	// we always write the digit to memory but conditionally increment
643	// the pointer, to ease the usage of conditional move instructions.	693	// the pointer, to enable the use of conditional move instructions.
644	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffff) 5;	694	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffffUL) 5;
645	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffff) 5;	695	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffffUL) 5;
646	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffff) 5;	696	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffffUL) 5;
647	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffff) 5;	697	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffffUL) 5;
648	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'	698	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
		699	}
		700	else
		701	{
		702	// large integer, use the (rather slow) snprintf way.
		703	need (enc, IVUV_MAXCHARS);
		704	enc->cur +=
		705	SvIsUV(sv)
		706	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
		707	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
649	}	708	}
650	}	709	}
651	else if (SvROK (sv))	710	else if (SvROK (sv))
652	encode_rv (enc, SvRV (sv));	711	encode_rv (enc, SvRV (sv));
653	else if (!SvOK (sv))	712	else if (!SvOK (sv))
…		…
669	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	728	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
670	enc.cur = SvPVX (enc.sv);	729	enc.cur = SvPVX (enc.sv);
671	enc.end = SvEND (enc.sv);	730	enc.end = SvEND (enc.sv);
672	enc.indent = 0;	731	enc.indent = 0;
673	enc.maxdepth = DEC_DEPTH (enc.json.flags);	732	enc.maxdepth = DEC_DEPTH (enc.json.flags);
		733	enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
		734	: enc.json.flags & F_LATIN1 ? 0x000100UL
		735	: 0x110000UL;
674		736
675	SvPOK_only (enc.sv);	737	SvPOK_only (enc.sv);
676	encode_sv (&enc, scalar);	738	encode_sv (&enc, scalar);
677		739
678	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	740	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
…		…
699	JSON json;	761	JSON json;
700	U32 depth; // recursion depth	762	U32 depth; // recursion depth
701	U32 maxdepth; // recursion depth limit	763	U32 maxdepth; // recursion depth limit
702	} dec_t;	764	} dec_t;
703		765
704	inline void	766	INLINE void
705	decode_comment (dec_t *dec)	767	decode_comment (dec_t *dec)
706	{	768	{
707	// only '#'-style comments allowed a.t.m.	769	// only '#'-style comments allowed a.t.m.
708		770
709	while (dec->cur && dec->cur != 0x0a && *dec->cur != 0x0d)	771	while (dec->cur && dec->cur != 0x0a && *dec->cur != 0x0d)
710	++dec->cur;	772	++dec->cur;
711	}	773	}
712		774
713	inline void	775	INLINE void
714	decode_ws (dec_t *dec)	776	decode_ws (dec_t *dec)
715	{	777	{
716	for (;;)	778	for (;;)
717	{	779	{
718	char ch = *dec->cur;	780	char ch = *dec->cur;
…		…
844		906
845	if (hi >= 0x80)	907	if (hi >= 0x80)
846	{	908	{
847	utf8 = 1;	909	utf8 = 1;
848		910
849	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);	911	cur = encode_utf8 (cur, hi);
850	}	912	}
851	else	913	else
852	*cur++ = hi;	914	*cur++ = hi;
853	}	915	}
854	break;	916	break;
…		…
856	default:	918	default:
857	--dec_cur;	919	--dec_cur;
858	ERR ("illegal backslash escape sequence in string");	920	ERR ("illegal backslash escape sequence in string");
859	}	921	}
860	}	922	}
861	else if (expect_true (ch >= 0x20 && ch <= 0x7f))	923	else if (expect_true (ch >= 0x20 && ch < 0x80))
862	*cur++ = ch;	924	*cur++ = ch;
863	else if (ch >= 0x80)	925	else if (ch >= 0x80)
864	{	926	{
865	STRLEN clen;	927	STRLEN clen;
866	UV uch;	928	UV uch;
…		…
989		1051
990	if (!is_nv)	1052	if (!is_nv)
991	{	1053	{
992	int len = dec->cur - start;	1054	int len = dec->cur - start;
993		1055
994	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so	1056	// special case the rather common 1..5-digit-int case
995	if (*start == '-')	1057	if (*start == '-')
996	switch (len)	1058	switch (len)
997	{	1059	{
998	case 2: return newSViv (-( start [1] - '0' * 1));	1060	case 2: return newSViv (-( start [1] - '0' * 1));
999	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));	1061	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
1000	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));	1062	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1001	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));	1063	case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
		1064	case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1002	}	1065	}
1003	else	1066	else
1004	switch (len)	1067	switch (len)
1005	{	1068	{
1006	case 1: return newSViv ( start [0] - '0' * 1);	1069	case 1: return newSViv ( start [0] - '0' * 1);
1007	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);	1070	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1008	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);	1071	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1009	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);	1072	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		1073	case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1010	}	1074	}
1011		1075
1012	{	1076	{
1013	UV uv;	1077	UV uv;
1014	int numtype = grok_number (start, len, &uv);	1078	int numtype = grok_number (start, len, &uv);
…		…
1122	char *p = dec->cur;	1186	char *p = dec->cur;
1123	char *e = p + 24; // only try up to 24 bytes	1187	char *e = p + 24; // only try up to 24 bytes
1124		1188
1125	for (;;)	1189	for (;;)
1126	{	1190	{
1127	// the >= 0x80 is true on most architectures	1191	// the >= 0x80 is false on most architectures
1128	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')	1192	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
1129	{	1193	{
1130	// slow path, back up and use decode_str	1194	// slow path, back up and use decode_str
1131	SV *key = decode_str (dec);	1195	SV *key = decode_str (dec);
1132	if (!key)	1196	if (!key)
…		…
1263		1327
1264	static SV *	1328	static SV *
1265	decode_sv (dec_t *dec)	1329	decode_sv (dec_t *dec)
1266	{	1330	{
1267	// the beauty of JSON: you need exactly one character lookahead	1331	// the beauty of JSON: you need exactly one character lookahead
1268	// to parse anything.	1332	// to parse everything.
1269	switch (*dec->cur)	1333	switch (*dec->cur)
1270	{	1334	{
1271	case '"': ++dec->cur; return decode_str (dec);	1335	case '"': ++dec->cur; return decode_str (dec);
1272	case '[': ++dec->cur; return decode_av (dec);	1336	case '[': ++dec->cur; return decode_av (dec);
1273	case '{': ++dec->cur; return decode_hv (dec);	1337	case '{': ++dec->cur; return decode_hv (dec);
1274		1338
1275	case '-':	1339	case '-':
1276	case '0': case '1': case '2': case '3': case '4':	1340	case '0': case '1': case '2': case '3': case '4':
1277	case '5': case '6': case '7': case '8': case '9':	1341	case '5': case '6': case '7': case '8': case '9':
1278	return decode_num (dec);	1342	return decode_num (dec);
…		…
1324	fail:	1388	fail:
1325	return 0;	1389	return 0;
1326	}	1390	}
1327		1391
1328	static SV *	1392	static SV *
1329	decode_json (SV string, JSON json, UV *offset_return)	1393	decode_json (SV string, JSON json, STRLEN *offset_return)
1330	{	1394	{
1331	dec_t dec;	1395	dec_t dec;
1332	UV offset;	1396	STRLEN offset;
1333	SV *sv;	1397	SV *sv;
1334		1398
1335	SvGETMAGIC (string);	1399	SvGETMAGIC (string);
1336	SvUPGRADE (string, SVt_PV);	1400	SvUPGRADE (string, SVt_PV);
1337		1401
…		…
1410		1474
1411	return sv;	1475	return sv;
1412	}	1476	}
1413		1477
1414	/////////////////////////////////////////////////////////////////////////////	1478	/////////////////////////////////////////////////////////////////////////////
		1479	// incremental parser
		1480
		1481	static void
		1482	incr_parse (JSON *self)
		1483	{
		1484	const char *p = SvPVX (self->incr_text) + self->incr_pos;
		1485
		1486	for (;;)
		1487	{
		1488	//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
		1489	switch (self->incr_mode)
		1490	{
		1491	// only used for intiial whitespace skipping
		1492	case INCR_M_WS:
		1493	for (;;)
		1494	{
		1495	if (*p > 0x20)
		1496	{
		1497	self->incr_mode = INCR_M_JSON;
		1498	goto incr_m_json;
		1499	}
		1500	else if (!*p)
		1501	goto interrupt;
		1502
		1503	++p;
		1504	}
		1505
		1506	// skip a single char inside a string (for \\-processing)
		1507	case INCR_M_BS:
		1508	if (!*p)
		1509	goto interrupt;
		1510
		1511	++p;
		1512	self->incr_mode = INCR_M_STR;
		1513	goto incr_m_str;
		1514
		1515	// inside a string
		1516	case INCR_M_STR:
		1517	incr_m_str:
		1518	for (;;)
		1519	{
		1520	if (*p == '"')
		1521	{
		1522	++p;
		1523	self->incr_mode = INCR_M_JSON;
		1524
		1525	if (!self->incr_nest)
		1526	goto interrupt;
		1527
		1528	goto incr_m_json;
		1529	}
		1530	else if (*p == '\\')
		1531	{
		1532	++p; // "virtually" consumes character after \
		1533
		1534	if (!*p) // if at end of string we have to switch modes
		1535	{
		1536	self->incr_mode = INCR_M_BS;
		1537	goto interrupt;
		1538	}
		1539	}
		1540	else if (!*p)
		1541	goto interrupt;
		1542
		1543	++p;
		1544	}
		1545
		1546	// after initial ws, outside string
		1547	case INCR_M_JSON:
		1548	incr_m_json:
		1549	for (;;)
		1550	{
		1551	switch (*p++)
		1552	{
		1553	case 0:
		1554	--p;
		1555	goto interrupt;
		1556
		1557	case 0x09:
		1558	case 0x0a:
		1559	case 0x0d:
		1560	case 0x20:
		1561	if (!self->incr_nest)
		1562	{
		1563	--p; // do not eat the whitespace, let the next round do it
		1564	goto interrupt;
		1565	}
		1566	break;
		1567
		1568	case '"':
		1569	self->incr_mode = INCR_M_STR;
		1570	goto incr_m_str;
		1571
		1572	case '[':
		1573	case '{':
		1574	++self->incr_nest;
		1575	break;
		1576
		1577	case ']':
		1578	case '}':
		1579	if (!--self->incr_nest)
		1580	goto interrupt;
		1581	}
		1582	}
		1583	}
		1584
		1585	modechange:
		1586	;
		1587	}
		1588
		1589	interrupt:
		1590	self->incr_pos = p - SvPVX (self->incr_text);
		1591	//printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
		1592	}
		1593
		1594	/////////////////////////////////////////////////////////////////////////////
1415	// XS interface functions	1595	// XS interface functions
1416		1596
1417	MODULE = JSON::XS PACKAGE = JSON::XS	1597	MODULE = JSON::XS PACKAGE = JSON::XS
1418		1598
1419	BOOT:	1599	BOOT:
…		…
1446	{	1626	{
1447	SV *pv = NEWSV (0, sizeof (JSON));	1627	SV *pv = NEWSV (0, sizeof (JSON));
1448	SvPOK_only (pv);	1628	SvPOK_only (pv);
1449	Zero (SvPVX (pv), 1, JSON);	1629	Zero (SvPVX (pv), 1, JSON);
1450	((JSON *)SvPVX (pv))->flags = F_DEFAULT;	1630	((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1451	XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));	1631	XPUSHs (sv_2mortal (sv_bless (
		1632	newRV_noinc (pv),
		1633	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
		1634	)));
1452	}	1635	}
1453		1636
1454	void ascii (JSON *self, int enable = 1)	1637	void ascii (JSON *self, int enable = 1)
1455	ALIAS:	1638	ALIAS:
1456	ascii = F_ASCII	1639	ascii = F_ASCII
…		…
1474	self->flags &= ~ix;	1657	self->flags &= ~ix;
1475		1658
1476	XPUSHs (ST (0));	1659	XPUSHs (ST (0));
1477	}	1660	}
1478		1661
		1662	void get_ascii (JSON *self)
		1663	ALIAS:
		1664	get_ascii = F_ASCII
		1665	get_latin1 = F_LATIN1
		1666	get_utf8 = F_UTF8
		1667	get_indent = F_INDENT
		1668	get_canonical = F_CANONICAL
		1669	get_space_before = F_SPACE_BEFORE
		1670	get_space_after = F_SPACE_AFTER
		1671	get_allow_nonref = F_ALLOW_NONREF
		1672	get_shrink = F_SHRINK
		1673	get_allow_blessed = F_ALLOW_BLESSED
		1674	get_convert_blessed = F_CONV_BLESSED
		1675	get_relaxed = F_RELAXED
		1676	PPCODE:
		1677	XPUSHs (boolSV (self->flags & ix));
		1678
1479	void max_depth (JSON *self, UV max_depth = 0x80000000UL)	1679	void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1480	PPCODE:	1680	PPCODE:
1481	{	1681	{
1482	UV log2 = 0;	1682	UV log2 = 0;
1483		1683
…		…
1489	self->flags = self->flags & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);	1689	self->flags = self->flags & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
1490		1690
1491	XPUSHs (ST (0));	1691	XPUSHs (ST (0));
1492	}	1692	}
1493		1693
		1694	U32 get_max_depth (JSON *self)
		1695	CODE:
		1696	RETVAL = DEC_DEPTH (self->flags);
		1697	OUTPUT:
		1698	RETVAL
		1699
1494	void max_size (JSON *self, UV max_size = 0)	1700	void max_size (JSON *self, UV max_size = 0)
1495	PPCODE:	1701	PPCODE:
1496	{	1702	{
1497	UV log2 = 0;	1703	UV log2 = 0;
1498		1704
…		…
1504		1710
1505	self->flags = self->flags & ~F_MAXSIZE \| (log2 << S_MAXSIZE);	1711	self->flags = self->flags & ~F_MAXSIZE \| (log2 << S_MAXSIZE);
1506		1712
1507	XPUSHs (ST (0));	1713	XPUSHs (ST (0));
1508	}	1714	}
		1715
		1716	int get_max_size (JSON *self)
		1717	CODE:
		1718	RETVAL = DEC_SIZE (self->flags);
		1719	OUTPUT:
		1720	RETVAL
1509		1721
1510	void filter_json_object (JSON self, SV cb = &PL_sv_undef)	1722	void filter_json_object (JSON self, SV cb = &PL_sv_undef)
1511	PPCODE:	1723	PPCODE:
1512	{	1724	{
1513	SvREFCNT_dec (self->cb_object);	1725	SvREFCNT_dec (self->cb_object);
…		…
1547	XPUSHs (decode_json (jsonstr, self, 0));	1759	XPUSHs (decode_json (jsonstr, self, 0));
1548		1760
1549	void decode_prefix (JSON self, SV jsonstr)	1761	void decode_prefix (JSON self, SV jsonstr)
1550	PPCODE:	1762	PPCODE:
1551	{	1763	{
1552	UV offset;	1764	STRLEN offset;
1553	EXTEND (SP, 2);	1765	EXTEND (SP, 2);
1554	PUSHs (decode_json (jsonstr, self, &offset));	1766	PUSHs (decode_json (jsonstr, self, &offset));
1555	PUSHs (sv_2mortal (newSVuv (offset)));	1767	PUSHs (sv_2mortal (newSVuv (offset)));
1556	}	1768	}
		1769
		1770	void incr_parse (JSON self, SV jsonstr = 0)
		1771	PPCODE:
		1772	{
		1773	if (!self->incr_text)
		1774	self->incr_text = newSVpvn ("", 0);
		1775
		1776	// append data, if any
		1777	if (jsonstr)
		1778	{
		1779	if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
		1780	{
		1781	/* utf-8-ness differs, need to upgrade */
		1782	sv_utf8_upgrade (self->incr_text);
		1783
		1784	if (self->incr_pos)
		1785	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
		1786	- (U8 *)SvPVX (self->incr_text);
		1787	}
		1788
		1789	{
		1790	STRLEN len;
		1791	const char *str = SvPV (jsonstr, len);
		1792	SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
		1793	Move (str, SvEND (self->incr_text), len, char);
		1794	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
		1795	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
		1796	}
		1797	}
		1798
		1799	if (GIMME_V != G_VOID)
		1800	do
		1801	{
		1802	STRLEN offset;
		1803
		1804	incr_parse (self);
		1805
		1806	if (!INCR_DONE (self))
		1807	break;
		1808
		1809	XPUSHs (decode_json (self->incr_text, self, &offset));
		1810
		1811	sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
		1812	self->incr_pos -= offset;
		1813	self->incr_nest = 0;
		1814	self->incr_mode = 0;
		1815	}
		1816	while (GIMME_V == G_ARRAY);
		1817	}
		1818
		1819	SV incr_text (JSON self)
		1820	ATTRS: lvalue
		1821	CODE:
		1822	{
		1823	if (self->incr_pos)
		1824	croak ("incr_text can only be called after a successful incr_parse call in scalar context %d", self->incr_pos);//D
		1825
		1826	RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
		1827	}
		1828	OUTPUT:
		1829	RETVAL
1557		1830
1558	void DESTROY (JSON *self)	1831	void DESTROY (JSON *self)
1559	CODE:	1832	CODE:
1560	SvREFCNT_dec (self->cb_sk_object);	1833	SvREFCNT_dec (self->cb_sk_object);
1561	SvREFCNT_dec (self->cb_object);	1834	SvREFCNT_dec (self->cb_object);
		1835	SvREFCNT_dec (self->incr_text);
1562		1836
1563	PROTOTYPES: ENABLE	1837	PROTOTYPES: ENABLE
1564		1838
1565	void to_json (SV *scalar)	1839	void encode_json (SV *scalar)
		1840	ALIAS:
		1841	to_json_ = 0
		1842	encode_json = F_UTF8
1566	PPCODE:	1843	PPCODE:
1567	{	1844	{
1568	JSON json = { F_DEFAULT \| F_UTF8 };	1845	JSON json = { F_DEFAULT \| ix };
1569	XPUSHs (encode_json (scalar, &json));	1846	XPUSHs (encode_json (scalar, &json));
1570	}	1847	}
1571		1848
1572	void from_json (SV *jsonstr)	1849	void decode_json (SV *jsonstr)
		1850	ALIAS:
		1851	from_json_ = 0
		1852	decode_json = F_UTF8
1573	PPCODE:	1853	PPCODE:
1574	{	1854	{
1575	JSON json = { F_DEFAULT \| F_UTF8 };	1855	JSON json = { F_DEFAULT \| ix };
1576	XPUSHs (decode_json (jsonstr, &json, 0));	1856	XPUSHs (decode_json (jsonstr, &json, 0));
1577	}	1857	}
1578		1858
		1859

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.64 by root, Mon Aug 27 02:03:23 2007 UTC vs. Revision 1.77 by root, Tue Mar 25 06:37:38 2008 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.64 by root, Mon Aug 27 02:03:23 2007 UTC vs.
Revision 1.77 by root, Tue Mar 25 06:37:38 2008 UTC