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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.70 by root, Wed Mar 19 00:44:54 2008 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
…		…
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
32	#define F_RELAXED 0x00001000UL	35	#define F_RELAXED 0x00001000UL
		36	#define F_ALLOW_UNKNOWN 0x00002000UL
33		37
34	#define F_MAXDEPTH 0xf8000000UL	38	#define F_MAXDEPTH 0xf8000000UL
35	#define S_MAXDEPTH 27	39	#define S_MAXDEPTH 27
36	#define F_MAXSIZE 0x01f00000UL	40	#define F_MAXSIZE 0x01f00000UL
37	#define S_MAXSIZE 20	41	#define S_MAXSIZE 20
…		…
50		54
51	#define SB do {	55	#define SB do {
52	#define SE } while (0)	56	#define SE } while (0)
53		57
54	#if __GNUC__ >= 3	58	#if __GNUC__ >= 3
55	# define expect(expr,value) __builtin_expect ((expr),(value))	59	# define expect(expr,value) __builtin_expect ((expr), (value))
56	# define inline inline	60	# define INLINE static inline
57	#else	61	#else
58	# define expect(expr,value) (expr)	62	# define expect(expr,value) (expr)
59	# define inline static	63	# define INLINE static
60	#endif	64	#endif
61		65
62	#define expect_false(expr) expect ((expr) != 0, 0)	66	#define expect_false(expr) expect ((expr) != 0, 0)
63	#define expect_true(expr) expect ((expr) != 0, 1)	67	#define expect_true(expr) expect ((expr) != 0, 1)
		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)))
64		72
65	#ifdef USE_ITHREADS	73	#ifdef USE_ITHREADS
66	# define JSON_SLOW 1	74	# define JSON_SLOW 1
67	# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))	75	# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
68	#else	76	#else
…		…
71	#endif	79	#endif
72		80
73	static HV json_stash, json_boolean_stash; // JSON::XS::	81	static HV json_stash, json_boolean_stash; // JSON::XS::
74	static SV json_true, json_false;	82	static SV json_true, json_false;
75		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)
		92
76	typedef struct {	93	typedef struct {
77	U32 flags;	94	U32 flags;
78	SV *cb_object;	95	SV *cb_object;
79	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;
80	} JSON;	103	} JSON;
81		104
82	/////////////////////////////////////////////////////////////////////////////	105	/////////////////////////////////////////////////////////////////////////////
83	// utility functions	106	// utility functions
84		107
85	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
86	shrink (SV *sv)	120	shrink (SV *sv)
87	{	121	{
88	sv_utf8_downgrade (sv, 1);	122	sv_utf8_downgrade (sv, 1);
89	if (SvLEN (sv) > SvCUR (sv) + 1)	123	if (SvLEN (sv) > SvCUR (sv) + 1)
90	{	124	{
…		…
99	// 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
100	// case of an error.	134	// case of an error.
101	// we special-case "safe" characters from U+80 .. U+7FF,	135	// we special-case "safe" characters from U+80 .. U+7FF,
102	// but use the very good perl function to parse anything else.	136	// but use the very good perl function to parse anything else.
103	// note that we never call this function for a ascii codepoints	137	// note that we never call this function for a ascii codepoints
104	inline UV	138	INLINE UV
105	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)	139	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
106	{	140	{
107	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))	141	if (expect_true (len >= 2
108	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);	142	&& IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
109	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)	143	&& IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
110	{	144	{
111	*clen = 2;	145	*clen = 2;
112	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);	146	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
113	}	147	}
114	else	148	else
115	{	149	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
116	*clen = (STRLEN)-1;	150	}
117	return (UV)-1;	151
118	}	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;
119	}	174	}
120		175
121	/////////////////////////////////////////////////////////////////////////////	176	/////////////////////////////////////////////////////////////////////////////
122	// encoder	177	// encoder
123		178
…		…
128	char *end; // SvEND (sv)	183	char *end; // SvEND (sv)
129	SV *sv; // result scalar	184	SV *sv; // result scalar
130	JSON json;	185	JSON json;
131	U32 indent; // indentation level	186	U32 indent; // indentation level
132	U32 maxdepth; // max. indentation/recursion level	187	U32 maxdepth; // max. indentation/recursion level
		188	UV limit; // escape character values >= this value when encoding
133	} enc_t;	189	} enc_t;
134		190
135	inline void	191	INLINE void
136	need (enc_t *enc, STRLEN len)	192	need (enc_t *enc, STRLEN len)
137	{	193	{
138	if (expect_false (enc->cur + len >= enc->end))	194	if (expect_false (enc->cur + len >= enc->end))
139	{	195	{
140	STRLEN cur = enc->cur - SvPVX (enc->sv);	196	STRLEN cur = enc->cur - SvPVX (enc->sv);
…		…
142	enc->cur = SvPVX (enc->sv) + cur;	198	enc->cur = SvPVX (enc->sv) + cur;
143	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	199	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
144	}	200	}
145	}	201	}
146		202
147	inline void	203	INLINE void
148	encode_ch (enc_t *enc, char ch)	204	encode_ch (enc_t *enc, char ch)
149	{	205	{
150	need (enc, 1);	206	need (enc, 1);
151	*enc->cur++ = ch;	207	*enc->cur++ = ch;
152	}	208	}
…		…
206	{	262	{
207	uch = ch;	263	uch = ch;
208	clen = 1;	264	clen = 1;
209	}	265	}
210		266
211	if (uch > 0x10FFFFUL)	267	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	{	268	{
216	if (uch > 0xFFFFUL)	269	if (uch >= 0x10000UL)
217	{	270	{
		271	if (uch >= 0x110000UL)
		272	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
		273
218	need (enc, len += 11);	274	need (enc, len += 11);
219	sprintf (enc->cur, "\\u%04x\\u%04x",	275	sprintf (enc->cur, "\\u%04x\\u%04x",
220	(int)((uch - 0x10000) / 0x400 + 0xD800),	276	(int)((uch - 0x10000) / 0x400 + 0xD800),
221	(int)((uch - 0x10000) % 0x400 + 0xDC00));	277	(int)((uch - 0x10000) % 0x400 + 0xDC00));
222	enc->cur += 12;	278	enc->cur += 12;
…		…
250	while (--clen);	306	while (--clen);
251	}	307	}
252	else	308	else
253	{	309	{
254	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
255	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	311	enc->cur = encode_utf8 (enc->cur, uch);
256	++str;	312	++str;
257	}	313	}
258	}	314	}
259	}	315	}
260	}	316	}
261		317
262	--len;	318	--len;
263	}	319	}
264	}	320	}
265		321
266	inline void	322	INLINE void
267	encode_indent (enc_t *enc)	323	encode_indent (enc_t *enc)
268	{	324	{
269	if (enc->json.flags & F_INDENT)	325	if (enc->json.flags & F_INDENT)
270	{	326	{
271	int spaces = enc->indent * INDENT_STEP;	327	int spaces = enc->indent * INDENT_STEP;
…		…
274	memset (enc->cur, ' ', spaces);	330	memset (enc->cur, ' ', spaces);
275	enc->cur += spaces;	331	enc->cur += spaces;
276	}	332	}
277	}	333	}
278		334
279	inline void	335	INLINE void
280	encode_space (enc_t *enc)	336	encode_space (enc_t *enc)
281	{	337	{
282	need (enc, 1);	338	need (enc, 1);
283	encode_ch (enc, ' ');	339	encode_ch (enc, ' ');
284	}	340	}
285		341
286	inline void	342	INLINE void
287	encode_nl (enc_t *enc)	343	encode_nl (enc_t *enc)
288	{	344	{
289	if (enc->json.flags & F_INDENT)	345	if (enc->json.flags & F_INDENT)
290	{	346	{
291	need (enc, 1);	347	need (enc, 1);
292	encode_ch (enc, '\n');	348	encode_ch (enc, '\n');
293	}	349	}
294	}	350	}
295		351
296	inline void	352	INLINE void
297	encode_comma (enc_t *enc)	353	encode_comma (enc_t *enc)
298	{	354	{
299	encode_ch (enc, ',');	355	encode_ch (enc, ',');
300		356
301	if (enc->json.flags & F_INDENT)	357	if (enc->json.flags & F_INDENT)
…		…
394		450
395	static void	451	static void
396	encode_hv (enc_t enc, HV hv)	452	encode_hv (enc_t enc, HV hv)
397	{	453	{
398	HE *he;	454	HE *he;
399	int count;
400		455
401	if (enc->indent >= enc->maxdepth)	456	if (enc->indent >= enc->maxdepth)
402	croak ("data structure too deep (hit recursion limit)");	457	croak ("data structure too deep (hit recursion limit)");
403		458
404	encode_ch (enc, '{');	459	encode_ch (enc, '{');
…		…
588		643
589	if (len == 1 && *pv == '1')	644	if (len == 1 && *pv == '1')
590	encode_str (enc, "true", 4, 0);	645	encode_str (enc, "true", 4, 0);
591	else if (len == 1 && *pv == '0')	646	else if (len == 1 && *pv == '0')
592	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);
593	else	650	else
594	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",
595	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	652	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
596	}	653	}
		654	else if (enc->json.flags & F_ALLOW_UNKNOWN)
		655	encode_str (enc, "null", 4, 0);
597	else	656	else
598	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",
599	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	658	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
600	}	659	}
601		660
…		…
619	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	678	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
620	enc->cur += strlen (enc->cur);	679	enc->cur += strlen (enc->cur);
621	}	680	}
622	else if (SvIOKp (sv))	681	else if (SvIOKp (sv))
623	{	682	{
624	// 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
625	if (SvUV (sv) & ~(UV)0x7fff)	684	// we assume two's complement
626	{	685	// we assume no aliasing issues in the union
627	// large integer, use the (rather slow) snprintf way.	686	if (SvIsUV (sv) ? SvUVX (sv) <= 59000
628	need (enc, sizeof (UV) * 3);	687	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
629	enc->cur +=
630	SvIsUV(sv)
631	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
632	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
633	}
634	else
635	{	688	{
636	// optimise the "small number case"	689	// optimise the "small number case"
637	// 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
638	I32 i = SvIV (sv);	692	I32 i = SvIVX (sv);
639	U32 u;	693	U32 u;
640	char digit, nz = 0;	694	char digit, nz = 0;
641		695
642	need (enc, 6);	696	need (enc, 6);
643		697
…		…
649		703
650	// 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
651	// 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,
652	// resulting in a net multiplication by 10.	706	// resulting in a net multiplication by 10.
653	// we always write the digit to memory but conditionally increment	707	// we always write the digit to memory but conditionally increment
654	// the pointer, to ease the usage of conditional move instructions.	708	// the pointer, to enable the use of conditional move instructions.
655	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;
656	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;
657	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;
658	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;
659	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'
660	}	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	}
661	}	724	}
662	else if (SvROK (sv))	725	else if (SvROK (sv))
663	encode_rv (enc, SvRV (sv));	726	encode_rv (enc, SvRV (sv));
664	else if (!SvOK (sv))	727	else if (!SvOK (sv) \|\| enc->json.flags & F_ALLOW_UNKNOWN)
665	encode_str (enc, "null", 4, 0);	728	encode_str (enc, "null", 4, 0);
666	else	729	else
667	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",
668	SvPV_nolen (sv), SvFLAGS (sv));	731	SvPV_nolen (sv), SvFLAGS (sv));
669	}	732	}
…		…
680	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	743	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
681	enc.cur = SvPVX (enc.sv);	744	enc.cur = SvPVX (enc.sv);
682	enc.end = SvEND (enc.sv);	745	enc.end = SvEND (enc.sv);
683	enc.indent = 0;	746	enc.indent = 0;
684	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;
685		751
686	SvPOK_only (enc.sv);	752	SvPOK_only (enc.sv);
687	encode_sv (&enc, scalar);	753	encode_sv (&enc, scalar);
688		754
689	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	755	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
…		…
710	JSON json;	776	JSON json;
711	U32 depth; // recursion depth	777	U32 depth; // recursion depth
712	U32 maxdepth; // recursion depth limit	778	U32 maxdepth; // recursion depth limit
713	} dec_t;	779	} dec_t;
714		780
715	inline void	781	INLINE void
716	decode_comment (dec_t *dec)	782	decode_comment (dec_t *dec)
717	{	783	{
718	// only '#'-style comments allowed a.t.m.	784	// only '#'-style comments allowed a.t.m.
719		785
720	while (dec->cur && dec->cur != 0x0a && *dec->cur != 0x0d)	786	while (dec->cur && dec->cur != 0x0a && *dec->cur != 0x0d)
721	++dec->cur;	787	++dec->cur;
722	}	788	}
723		789
724	inline void	790	INLINE void
725	decode_ws (dec_t *dec)	791	decode_ws (dec_t *dec)
726	{	792	{
727	for (;;)	793	for (;;)
728	{	794	{
729	char ch = *dec->cur;	795	char ch = *dec->cur;
…		…
855		921
856	if (hi >= 0x80)	922	if (hi >= 0x80)
857	{	923	{
858	utf8 = 1;	924	utf8 = 1;
859		925
860	cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);	926	cur = encode_utf8 (cur, hi);
861	}	927	}
862	else	928	else
863	*cur++ = hi;	929	*cur++ = hi;
864	}	930	}
865	break;	931	break;
…		…
867	default:	933	default:
868	--dec_cur;	934	--dec_cur;
869	ERR ("illegal backslash escape sequence in string");	935	ERR ("illegal backslash escape sequence in string");
870	}	936	}
871	}	937	}
872	else if (expect_true (ch >= 0x20 && ch <= 0x7f))	938	else if (expect_true (ch >= 0x20 && ch < 0x80))
873	*cur++ = ch;	939	*cur++ = ch;
874	else if (ch >= 0x80)	940	else if (ch >= 0x80)
875	{	941	{
876	STRLEN clen;	942	STRLEN clen;
877	UV uch;	943	UV uch;
…		…
1135	char *p = dec->cur;	1201	char *p = dec->cur;
1136	char *e = p + 24; // only try up to 24 bytes	1202	char *e = p + 24; // only try up to 24 bytes
1137		1203
1138	for (;;)	1204	for (;;)
1139	{	1205	{
1140	// the >= 0x80 is true on most architectures	1206	// the >= 0x80 is false on most architectures
1141	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')	1207	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
1142	{	1208	{
1143	// slow path, back up and use decode_str	1209	// slow path, back up and use decode_str
1144	SV *key = decode_str (dec);	1210	SV *key = decode_str (dec);
1145	if (!key)	1211	if (!key)
…		…
1276		1342
1277	static SV *	1343	static SV *
1278	decode_sv (dec_t *dec)	1344	decode_sv (dec_t *dec)
1279	{	1345	{
1280	// the beauty of JSON: you need exactly one character lookahead	1346	// the beauty of JSON: you need exactly one character lookahead
1281	// to parse anything.	1347	// to parse everything.
1282	switch (*dec->cur)	1348	switch (*dec->cur)
1283	{	1349	{
1284	case '"': ++dec->cur; return decode_str (dec);	1350	case '"': ++dec->cur; return decode_str (dec);
1285	case '[': ++dec->cur; return decode_av (dec);	1351	case '[': ++dec->cur; return decode_av (dec);
1286	case '{': ++dec->cur; return decode_hv (dec);	1352	case '{': ++dec->cur; return decode_hv (dec);
1287		1353
1288	case '-':	1354	case '-':
1289	case '0': case '1': case '2': case '3': case '4':	1355	case '0': case '1': case '2': case '3': case '4':
1290	case '5': case '6': case '7': case '8': case '9':	1356	case '5': case '6': case '7': case '8': case '9':
1291	return decode_num (dec);	1357	return decode_num (dec);
…		…
1293	case 't':	1359	case 't':
1294	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1360	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1295	{	1361	{
1296	dec->cur += 4;	1362	dec->cur += 4;
1297	#if JSON_SLOW	1363	#if JSON_SLOW
1298	json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);	1364	json_true = get_bool ("JSON::XS::true");
1299	#endif	1365	#endif
1300	return SvREFCNT_inc (json_true);	1366	return newSVsv (json_true);
1301	}	1367	}
1302	else	1368	else
1303	ERR ("'true' expected");	1369	ERR ("'true' expected");
1304		1370
1305	break;	1371	break;
…		…
1307	case 'f':	1373	case 'f':
1308	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1374	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1309	{	1375	{
1310	dec->cur += 5;	1376	dec->cur += 5;
1311	#if JSON_SLOW	1377	#if JSON_SLOW
1312	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);	1378	json_false = get_bool ("JSON::XS::false");
1313	#endif	1379	#endif
1314	return SvREFCNT_inc (json_false);	1380	return newSVsv (json_false);
1315	}	1381	}
1316	else	1382	else
1317	ERR ("'false' expected");	1383	ERR ("'false' expected");
1318		1384
1319	break;	1385	break;
…		…
1337	fail:	1403	fail:
1338	return 0;	1404	return 0;
1339	}	1405	}
1340		1406
1341	static SV *	1407	static SV *
1342	decode_json (SV string, JSON json, UV *offset_return)	1408	decode_json (SV string, JSON json, STRLEN *offset_return)
1343	{	1409	{
1344	dec_t dec;	1410	dec_t dec;
1345	UV offset;	1411	STRLEN offset;
1346	SV *sv;	1412	SV *sv;
1347		1413
1348	SvGETMAGIC (string);	1414	SvGETMAGIC (string);
1349	SvUPGRADE (string, SVt_PV);	1415	SvUPGRADE (string, SVt_PV);
1350		1416
…		…
1423		1489
1424	return sv;	1490	return sv;
1425	}	1491	}
1426		1492
1427	/////////////////////////////////////////////////////////////////////////////	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	/////////////////////////////////////////////////////////////////////////////
1428	// XS interface functions	1610	// XS interface functions
1429		1611
1430	MODULE = JSON::XS PACKAGE = JSON::XS	1612	MODULE = JSON::XS PACKAGE = JSON::XS
1431		1613
1432	BOOT:	1614	BOOT:
…		…
1441	: -1;	1623	: -1;
1442		1624
1443	json_stash = gv_stashpv ("JSON::XS" , 1);	1625	json_stash = gv_stashpv ("JSON::XS" , 1);
1444	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);	1626	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1445		1627
1446	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );	1628	json_true = get_bool ("JSON::XS::true");
1447	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);	1629	json_false = get_bool ("JSON::XS::false");
1448	}	1630	}
1449		1631
1450	PROTOTYPES: DISABLE	1632	PROTOTYPES: DISABLE
1451		1633
1452	void CLONE (...)	1634	void CLONE (...)
…		…
1480	allow_nonref = F_ALLOW_NONREF	1662	allow_nonref = F_ALLOW_NONREF
1481	shrink = F_SHRINK	1663	shrink = F_SHRINK
1482	allow_blessed = F_ALLOW_BLESSED	1664	allow_blessed = F_ALLOW_BLESSED
1483	convert_blessed = F_CONV_BLESSED	1665	convert_blessed = F_CONV_BLESSED
1484	relaxed = F_RELAXED	1666	relaxed = F_RELAXED
		1667	allow_unknown = F_ALLOW_UNKNOWN
1485	PPCODE:	1668	PPCODE:
1486	{	1669	{
1487	if (enable)	1670	if (enable)
1488	self->flags \|= ix;	1671	self->flags \|= ix;
1489	else	1672	else
…		…
1504	get_allow_nonref = F_ALLOW_NONREF	1687	get_allow_nonref = F_ALLOW_NONREF
1505	get_shrink = F_SHRINK	1688	get_shrink = F_SHRINK
1506	get_allow_blessed = F_ALLOW_BLESSED	1689	get_allow_blessed = F_ALLOW_BLESSED
1507	get_convert_blessed = F_CONV_BLESSED	1690	get_convert_blessed = F_CONV_BLESSED
1508	get_relaxed = F_RELAXED	1691	get_relaxed = F_RELAXED
		1692	get_allow_unknown = F_ALLOW_UNKNOWN
1509	PPCODE:	1693	PPCODE:
1510	XPUSHs (boolSV (self->flags & ix));	1694	XPUSHs (boolSV (self->flags & ix));
1511		1695
1512	void max_depth (JSON *self, UV max_depth = 0x80000000UL)	1696	void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1513	PPCODE:	1697	PPCODE:
…		…
1592	XPUSHs (decode_json (jsonstr, self, 0));	1776	XPUSHs (decode_json (jsonstr, self, 0));
1593		1777
1594	void decode_prefix (JSON self, SV jsonstr)	1778	void decode_prefix (JSON self, SV jsonstr)
1595	PPCODE:	1779	PPCODE:
1596	{	1780	{
1597	UV offset;	1781	STRLEN offset;
1598	EXTEND (SP, 2);	1782	EXTEND (SP, 2);
1599	PUSHs (decode_json (jsonstr, self, &offset));	1783	PUSHs (decode_json (jsonstr, self, &offset));
1600	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	}
1601	}	1860	}
1602		1861
1603	void DESTROY (JSON *self)	1862	void DESTROY (JSON *self)
1604	CODE:	1863	CODE:
1605	SvREFCNT_dec (self->cb_sk_object);	1864	SvREFCNT_dec (self->cb_sk_object);
1606	SvREFCNT_dec (self->cb_object);	1865	SvREFCNT_dec (self->cb_object);
		1866	SvREFCNT_dec (self->incr_text);
1607		1867
1608	PROTOTYPES: ENABLE	1868	PROTOTYPES: ENABLE
1609		1869
1610	void encode_json (SV *scalar)	1870	void encode_json (SV *scalar)
		1871	ALIAS:
		1872	to_json_ = 0
		1873	encode_json = F_UTF8
1611	PPCODE:	1874	PPCODE:
1612	{	1875	{
1613	JSON json = { F_DEFAULT \| F_UTF8 };	1876	JSON json = { F_DEFAULT \| ix };
1614	XPUSHs (encode_json (scalar, &json));	1877	XPUSHs (encode_json (scalar, &json));
1615	}	1878	}
1616		1879
1617	void decode_json (SV *jsonstr)	1880	void decode_json (SV *jsonstr)
		1881	ALIAS:
		1882	from_json_ = 0
		1883	decode_json = F_UTF8
1618	PPCODE:	1884	PPCODE:
1619	{	1885	{
1620	JSON json = { F_DEFAULT \| F_UTF8 };	1886	JSON json = { F_DEFAULT \| ix };
1621	XPUSHs (decode_json (jsonstr, &json, 0));	1887	XPUSHs (decode_json (jsonstr, &json, 0));
1622	}	1888	}
1623		1889
		1890

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.70 by root, Wed Mar 19 00:44:54 2008 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 1.70 by root, Wed Mar 19 00:44:54 2008 UTC vs.
Revision 1.84 by root, Thu Mar 27 06:37:35 2008 UTC