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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.15 by root, Sun Mar 25 02:32:40 2007 UTC vs.
Revision 1.40 by root, Tue Jun 12 01:27:02 2007 UTC

…		…
3	#include "XSUB.h"	3	#include "XSUB.h"
4		4
5	#include "assert.h"	5	#include "assert.h"
6	#include "string.h"	6	#include "string.h"
7	#include "stdlib.h"	7	#include "stdlib.h"
		8	#include "stdio.h"
8		9
		10	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		11	# define snprintf _snprintf // C compilers have this in stdio.h
		12	#endif
		13
9	#define F_ASCII 0x00000001	14	#define F_ASCII 0x00000001UL
		15	#define F_LATIN1 0x00000002UL
10	#define F_UTF8 0x00000002	16	#define F_UTF8 0x00000004UL
11	#define F_INDENT 0x00000004	17	#define F_INDENT 0x00000008UL
12	#define F_CANONICAL 0x00000008	18	#define F_CANONICAL 0x00000010UL
13	#define F_SPACE_BEFORE 0x00000010	19	#define F_SPACE_BEFORE 0x00000020UL
14	#define F_SPACE_AFTER 0x00000020	20	#define F_SPACE_AFTER 0x00000040UL
15	#define F_ALLOW_NONREF 0x00000080	21	#define F_ALLOW_NONREF 0x00000100UL
16	#define F_SHRINK 0x00000100	22	#define F_SHRINK 0x00000200UL
		23	#define F_MAXDEPTH 0xf8000000UL
		24	#define S_MAXDEPTH 27
		25
		26	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
		27
		28	// F_SELFCONVERT? <=> to_json/toJson
		29	// F_BLESSED? <=> { $__class__$ => }
17		30
18	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	31	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
19	#define F_DEFAULT 0	32	#define F_DEFAULT (9UL << S_MAXDEPTH)
20		33
21	#define INIT_SIZE 32 // initial scalar size to be allocated	34	#define INIT_SIZE 32 // initial scalar size to be allocated
22	#define INDENT_STEP 3 // spaces per indentation level	35	#define INDENT_STEP 3 // spaces per indentation level
23		36
24	#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
25	#define SHORT_STRING_LEN 512 // special-case strings of up to this size	37	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
26		38
27	#define SB do {	39	#define SB do {
28	#define SE } while (0)	40	#define SE } while (0)
		41
		42	#if __GNUC__ >= 3
		43	# define expect(expr,value) __builtin_expect ((expr),(value))
		44	# define inline inline
		45	#else
		46	# define expect(expr,value) (expr)
		47	# define inline static
		48	#endif
		49
		50	#define expect_false(expr) expect ((expr) != 0, 0)
		51	#define expect_true(expr) expect ((expr) != 0, 1)
29		52
30	static HV *json_stash; // JSON::XS::	53	static HV *json_stash; // JSON::XS::
31		54
32	/////////////////////////////////////////////////////////////////////////////	55	/////////////////////////////////////////////////////////////////////////////
33	// utility functions	56	// utility functions
…		…
58	// decode an utf-8 character and return it, or (UV)-1 in	81	// decode an utf-8 character and return it, or (UV)-1 in
59	// case of an error.	82	// case of an error.
60	// we special-case "safe" characters from U+80 .. U+7FF,	83	// we special-case "safe" characters from U+80 .. U+7FF,
61	// but use the very good perl function to parse anything else.	84	// but use the very good perl function to parse anything else.
62	// note that we never call this function for a ascii codepoints	85	// note that we never call this function for a ascii codepoints
63	static UV	86	inline UV
64	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)	87	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
65	{	88	{
66	if (s[0] > 0xdf \|\| s[0] < 0xc2)	89	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))
67	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);	90	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
68	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)	91	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
69	{	92	{
70	*clen = 2;	93	*clen = 2;
71	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);	94	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
72	}	95	}
73	else	96	else
		97	{
		98	*clen = (STRLEN)-1;
74	return (UV)-1;	99	return (UV)-1;
		100	}
75	}	101	}
76		102
77	/////////////////////////////////////////////////////////////////////////////	103	/////////////////////////////////////////////////////////////////////////////
78	// encoder	104	// encoder
79		105
…		…
81	typedef struct	107	typedef struct
82	{	108	{
83	char *cur; // SvPVX (sv) + current output position	109	char *cur; // SvPVX (sv) + current output position
84	char *end; // SvEND (sv)	110	char *end; // SvEND (sv)
85	SV *sv; // result scalar	111	SV *sv; // result scalar
86	UV flags; // F_*	112	U32 flags; // F_*
87	int indent; // indentation level	113	U32 indent; // indentation level
88	int max_depth; // max. recursion level	114	U32 maxdepth; // max. indentation/recursion level
89	} enc_t;	115	} enc_t;
90		116
91	static void	117	inline void
92	need (enc_t *enc, STRLEN len)	118	need (enc_t *enc, STRLEN len)
93	{	119	{
94	if (enc->cur + len >= enc->end)	120	if (expect_false (enc->cur + len >= enc->end))
95	{	121	{
96	STRLEN cur = enc->cur - SvPVX (enc->sv);	122	STRLEN cur = enc->cur - SvPVX (enc->sv);
97	SvGROW (enc->sv, cur + len + 1);	123	SvGROW (enc->sv, cur + len + 1);
98	enc->cur = SvPVX (enc->sv) + cur;	124	enc->cur = SvPVX (enc->sv) + cur;
99	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);	125	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
100	}	126	}
101	}	127	}
102		128
103	static void	129	inline void
104	encode_ch (enc_t *enc, char ch)	130	encode_ch (enc_t *enc, char ch)
105	{	131	{
106	need (enc, 1);	132	need (enc, 1);
107	*enc->cur++ = ch;	133	*enc->cur++ = ch;
108	}	134	}
…		…
116		142
117	while (str < end)	143	while (str < end)
118	{	144	{
119	unsigned char ch = (unsigned char )str;	145	unsigned char ch = (unsigned char )str;
120		146
121	if (ch >= 0x20 && ch < 0x80) // most common case	147	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
122	{	148	{
123	if (ch == '"') // but with slow exceptions	149	if (expect_false (ch == '"')) // but with slow exceptions
124	{	150	{
125	need (enc, len += 1);	151	need (enc, len += 1);
126	*enc->cur++ = '\\';	152	*enc->cur++ = '\\';
127	*enc->cur++ = '"';	153	*enc->cur++ = '"';
128	}	154	}
129	else if (ch == '\\')	155	else if (expect_false (ch == '\\'))
130	{	156	{
131	need (enc, len += 1);	157	need (enc, len += 1);
132	*enc->cur++ = '\\';	158	*enc->cur++ = '\\';
133	*enc->cur++ = '\\';	159	*enc->cur++ = '\\';
134	}	160	}
…		…
152	STRLEN clen;	178	STRLEN clen;
153	UV uch;	179	UV uch;
154		180
155	if (is_utf8)	181	if (is_utf8)
156	{	182	{
157	//uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
158	uch = decode_utf8 (str, end - str, &clen);	183	uch = decode_utf8 (str, end - str, &clen);
159	if (clen == (STRLEN)-1)	184	if (clen == (STRLEN)-1)
160	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);	185	croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
161	}	186	}
162	else	187	else
…		…
166	}	191	}
167		192
168	if (uch > 0x10FFFFUL)	193	if (uch > 0x10FFFFUL)
169	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);	194	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
170		195
171	if (uch < 0x80 \|\| enc->flags & F_ASCII)	196	if (uch < 0x80 \|\| enc->flags & F_ASCII \|\| (enc->flags & F_LATIN1 && uch > 0xFF))
172	{	197	{
173	if (uch > 0xFFFFUL)	198	if (uch > 0xFFFFUL)
174	{	199	{
175	need (enc, len += 11);	200	need (enc, len += 11);
176	sprintf (enc->cur, "\\u%04x\\u%04x",	201	sprintf (enc->cur, "\\u%04x\\u%04x",
…		…
190	*enc->cur++ = hexdigit [(uch >> 0) & 15];	215	*enc->cur++ = hexdigit [(uch >> 0) & 15];
191	}	216	}
192		217
193	str += clen;	218	str += clen;
194	}	219	}
		220	else if (enc->flags & F_LATIN1)
		221	{
		222	*enc->cur++ = uch;
		223	str += clen;
		224	}
195	else if (is_utf8)	225	else if (is_utf8)
196	{	226	{
197	need (enc, len += clen);	227	need (enc, len += clen);
198	do	228	do
199	{	229	{
…		…
201	}	231	}
202	while (--clen);	232	while (--clen);
203	}	233	}
204	else	234	else
205	{	235	{
206	need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed	236	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
207	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	237	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
208	++str;	238	++str;
209	}	239	}
210	}	240	}
211	}	241	}
…		…
213		243
214	--len;	244	--len;
215	}	245	}
216	}	246	}
217		247
218	static void	248	inline void
219	encode_indent (enc_t *enc)	249	encode_indent (enc_t *enc)
220	{	250	{
221	if (enc->flags & F_INDENT)	251	if (enc->flags & F_INDENT)
222	{	252	{
223	int spaces = enc->indent * INDENT_STEP;	253	int spaces = enc->indent * INDENT_STEP;
…		…
226	memset (enc->cur, ' ', spaces);	256	memset (enc->cur, ' ', spaces);
227	enc->cur += spaces;	257	enc->cur += spaces;
228	}	258	}
229	}	259	}
230		260
231	static void	261	inline void
232	encode_space (enc_t *enc)	262	encode_space (enc_t *enc)
233	{	263	{
234	need (enc, 1);	264	need (enc, 1);
235	encode_ch (enc, ' ');	265	encode_ch (enc, ' ');
236	}	266	}
237		267
238	static void	268	inline void
239	encode_nl (enc_t *enc)	269	encode_nl (enc_t *enc)
240	{	270	{
241	if (enc->flags & F_INDENT)	271	if (enc->flags & F_INDENT)
242	{	272	{
243	need (enc, 1);	273	need (enc, 1);
244	encode_ch (enc, '\n');	274	encode_ch (enc, '\n');
245	}	275	}
246	}	276	}
247		277
248	static void	278	inline void
249	encode_comma (enc_t *enc)	279	encode_comma (enc_t *enc)
250	{	280	{
251	encode_ch (enc, ',');	281	encode_ch (enc, ',');
252		282
253	if (enc->flags & F_INDENT)	283	if (enc->flags & F_INDENT)
…		…
260		290
261	static void	291	static void
262	encode_av (enc_t enc, AV av)	292	encode_av (enc_t enc, AV av)
263	{	293	{
264	int i, len = av_len (av);	294	int i, len = av_len (av);
		295
		296	if (enc->indent >= enc->maxdepth)
		297	croak ("data structure too deep (hit recursion limit)");
265		298
266	encode_ch (enc, '['); encode_nl (enc);	299	encode_ch (enc, '['); encode_nl (enc);
267	++enc->indent;	300	++enc->indent;
268		301
269	for (i = 0; i <= len; ++i)	302	for (i = 0; i <= len; ++i)
…		…
336	static void	369	static void
337	encode_hv (enc_t enc, HV hv)	370	encode_hv (enc_t enc, HV hv)
338	{	371	{
339	int count, i;	372	int count, i;
340		373
		374	if (enc->indent >= enc->maxdepth)
		375	croak ("data structure too deep (hit recursion limit)");
		376
341	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;	377	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
342		378
343	if ((count = hv_iterinit (hv)))	379	if ((count = hv_iterinit (hv)))
344	{	380	{
345	// for canonical output we have to sort by keys first	381	// for canonical output we have to sort by keys first
346	// actually, this is mostly due to the stupid so-called	382	// actually, this is mostly due to the stupid so-called
347	// security workaround added somewhere in 5.8.x.	383	// security workaround added somewhere in 5.8.x.
348	// that randomises hash orderings	384	// that randomises hash orderings
349	if (enc->flags & F_CANONICAL)	385	if (enc->flags & F_CANONICAL)
350	{	386	{
351	HE he, hes [count]; // if your compiler dies here, you need to enable C99 mode
352	int fast = 1;	387	int fast = 1;
		388	HE *he;
		389	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		390	HE *hes = _alloca (count sizeof (HE));
		391	#else
		392	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
		393	#endif
353		394
354	i = 0;	395	i = 0;
355	while ((he = hv_iternext (hv)))	396	while ((he = hv_iternext (hv)))
356	{	397	{
357	hes [i++] = he;	398	hes [i++] = he;
…		…
392		433
393	encode_nl (enc);	434	encode_nl (enc);
394	}	435	}
395	else	436	else
396	{	437	{
397	SV *sv;
398	HE *he = hv_iternext (hv);	438	HE *he = hv_iternext (hv);
399		439
400	for (;;)	440	for (;;)
401	{	441	{
402	encode_indent (enc);	442	encode_indent (enc);
…		…
411	encode_nl (enc);	451	encode_nl (enc);
412	}	452	}
413	}	453	}
414		454
415	--enc->indent; encode_indent (enc); encode_ch (enc, '}');	455	--enc->indent; encode_indent (enc); encode_ch (enc, '}');
		456	}
		457
		458	// encode objects, arrays and special \0=false and \1=true values.
		459	static void
		460	encode_rv (enc_t enc, SV sv)
		461	{
		462	svtype svt;
		463
		464	SvGETMAGIC (sv);
		465	svt = SvTYPE (sv);
		466
		467	if (svt == SVt_PVHV)
		468	encode_hv (enc, (HV *)sv);
		469	else if (svt == SVt_PVAV)
		470	encode_av (enc, (AV *)sv);
		471	else if (svt < SVt_PVAV)
		472	{
		473	if (SvNIOK (sv) && SvIV (sv) == 0)
		474	encode_str (enc, "false", 5, 0);
		475	else if (SvNIOK (sv) && SvIV (sv) == 1)
		476	encode_str (enc, "true", 4, 0);
		477	else
		478	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
		479	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		480	}
		481	else
		482	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
		483	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416	}	484	}
417		485
418	static void	486	static void
419	encode_sv (enc_t enc, SV sv)	487	encode_sv (enc_t enc, SV sv)
420	{	488	{
…		…
428	encode_str (enc, str, len, SvUTF8 (sv));	496	encode_str (enc, str, len, SvUTF8 (sv));
429	encode_ch (enc, '"');	497	encode_ch (enc, '"');
430	}	498	}
431	else if (SvNOKp (sv))	499	else if (SvNOKp (sv))
432	{	500	{
		501	// trust that perl will do the right thing w.r.t. JSON syntax.
433	need (enc, NV_DIG + 32);	502	need (enc, NV_DIG + 32);
434	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	503	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
435	enc->cur += strlen (enc->cur);	504	enc->cur += strlen (enc->cur);
436	}	505	}
437	else if (SvIOKp (sv))	506	else if (SvIOKp (sv))
438	{	507	{
439	need (enc, 64);	508	// we assume we can always read an IV as a UV
		509	if (SvUV (sv) & ~(UV)0x7fff)
		510	{
		511	// large integer, use the (rather slow) snprintf way.
		512	need (enc, sizeof (UV) * 3);
440	enc->cur +=	513	enc->cur +=
441	SvIsUV(sv)	514	SvIsUV(sv)
442	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	515	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
443	: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));	516	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
		517	}
		518	else
		519	{
		520	// optimise the "small number case"
		521	// code will likely be branchless and use only a single multiplication
		522	I32 i = SvIV (sv);
		523	U32 u;
		524	char digit, nz = 0;
		525
		526	need (enc, 6);
		527
		528	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
		529	u = i < 0 ? -i : i;
		530
		531	// convert to 4.28 fixed-point representation
		532	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
		533
		534	// now output digit by digit, each time masking out the integer part
		535	// and multiplying by 5 while moving the decimal point one to the right,
		536	// resulting in a net multiplication by 10.
		537	// we always write the digit to memory but conditionally increment
		538	// the pointer, to ease the usage of conditional move instructions.
		539	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffff) 5;
		540	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffff) 5;
		541	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffff) 5;
		542	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffff) 5;
		543	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
		544	}
444	}	545	}
445	else if (SvROK (sv))	546	else if (SvROK (sv))
446	{	547	encode_rv (enc, SvRV (sv));
447	SV *rv = SvRV (sv);
448
449	if (enc->indent >= enc->max_depth)
450	croak ("data structure too deep (hit recursion limit)");
451
452	switch (SvTYPE (rv))
453	{
454	case SVt_PVAV: encode_av (enc, (AV *)rv); break;
455	case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
456
457	default:
458	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
459	SvPV_nolen (sv));
460	}
461	}
462	else if (!SvOK (sv))	548	else if (!SvOK (sv))
463	encode_str (enc, "null", 4, 0);	549	encode_str (enc, "null", 4, 0);
464	else	550	else
465	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	551	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466	SvPV_nolen (sv), SvFLAGS (sv));	552	SvPV_nolen (sv), SvFLAGS (sv));
467	}	553	}
468		554
469	static SV *	555	static SV *
470	encode_json (SV *scalar, UV flags)	556	encode_json (SV *scalar, U32 flags)
471	{	557	{
		558	enc_t enc;
		559
472	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	560	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
473	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");	561	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474		562
475	enc_t enc;
476	enc.flags = flags;	563	enc.flags = flags;
477	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	564	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478	enc.cur = SvPVX (enc.sv);	565	enc.cur = SvPVX (enc.sv);
479	enc.end = SvEND (enc.sv);	566	enc.end = SvEND (enc.sv);
480	enc.indent = 0;	567	enc.indent = 0;
481	enc.max_depth = 0x7fffffffUL;	568	enc.maxdepth = DEC_DEPTH (flags);
482		569
483	SvPOK_only (enc.sv);	570	SvPOK_only (enc.sv);
484	encode_sv (&enc, scalar);	571	encode_sv (&enc, scalar);
485		572
		573	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
		574	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
		575
486	if (!(flags & (F_ASCII \| F_UTF8)))	576	if (!(flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
487	SvUTF8_on (enc.sv);	577	SvUTF8_on (enc.sv);
488
489	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
490		578
491	if (enc.flags & F_SHRINK)	579	if (enc.flags & F_SHRINK)
492	shrink (enc.sv);	580	shrink (enc.sv);
493		581
494	return enc.sv;	582	return enc.sv;
…		…
501	typedef struct	589	typedef struct
502	{	590	{
503	char *cur; // current parser pointer	591	char *cur; // current parser pointer
504	char *end; // end of input string	592	char *end; // end of input string
505	const char *err; // parse error, if != 0	593	const char *err; // parse error, if != 0
506	UV flags; // F_*	594	U32 flags; // F_*
		595	U32 depth; // recursion depth
		596	U32 maxdepth; // recursion depth limit
507	} dec_t;	597	} dec_t;
508		598
509	static void	599	inline void
510	decode_ws (dec_t *dec)	600	decode_ws (dec_t *dec)
511	{	601	{
512	for (;;)	602	for (;;)
513	{	603	{
514	char ch = *dec->cur;	604	char ch = *dec->cur;
…		…
520	++dec->cur;	610	++dec->cur;
521	}	611	}
522	}	612	}
523		613
524	#define ERR(reason) SB dec->err = reason; goto fail; SE	614	#define ERR(reason) SB dec->err = reason; goto fail; SE
		615
525	#define EXPECT_CH(ch) SB \	616	#define EXPECT_CH(ch) SB \
526	if (*dec->cur != ch) \	617	if (*dec->cur != ch) \
527	ERR (# ch " expected"); \	618	ERR (# ch " expected"); \
528	++dec->cur; \	619	++dec->cur; \
529	SE	620	SE
530		621
		622	#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
		623	#define DEC_DEC_DEPTH --dec->depth
		624
531	static SV decode_sv (dec_t dec);	625	static SV decode_sv (dec_t dec);
532		626
533	static signed char decode_hexdigit[256];	627	static signed char decode_hexdigit[256];
534		628
535	static UV	629	static UV
536	decode_4hex (dec_t *dec)	630	decode_4hex (dec_t *dec)
537	{	631	{
538	signed char d1, d2, d3, d4;	632	signed char d1, d2, d3, d4;
539	unsigned char cur = (unsigned char )dec->cur;	633	unsigned char cur = (unsigned char )dec->cur;
540		634
541	d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");	635	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
542	d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");	636	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
543	d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");	637	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
544	d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");	638	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
545		639
546	dec->cur += 4;	640	dec->cur += 4;
547		641
548	return ((UV)d1) << 12	642	return ((UV)d1) << 12
549	\| ((UV)d2) << 8	643	\| ((UV)d2) << 8
…		…
557	static SV *	651	static SV *
558	decode_str (dec_t *dec)	652	decode_str (dec_t *dec)
559	{	653	{
560	SV *sv = 0;	654	SV *sv = 0;
561	int utf8 = 0;	655	int utf8 = 0;
		656	char *dec_cur = dec->cur;
562		657
563	do	658	do
564	{	659	{
565	char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];	660	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
566	char *cur = buf;	661	char *cur = buf;
567		662
568	do	663	do
569	{	664	{
570	unsigned char ch = (unsigned char )dec->cur++;	665	unsigned char ch = (unsigned char )dec_cur++;
571		666
572	if (ch == '"')	667	if (expect_false (ch == '"'))
573	{	668	{
574	--dec->cur;	669	--dec_cur;
575	break;	670	break;
576	}	671	}
577	else if (ch == '\\')	672	else if (expect_false (ch == '\\'))
578	{	673	{
579	switch (*dec->cur)	674	switch (*dec_cur)
580	{	675	{
581	case '\\':	676	case '\\':
582	case '/':	677	case '/':
583	case '"': cur++ = dec->cur++; break;	678	case '"': cur++ = dec_cur++; break;
584		679
585	case 'b': ++dec->cur; *cur++ = '\010'; break;	680	case 'b': ++dec_cur; *cur++ = '\010'; break;
586	case 't': ++dec->cur; *cur++ = '\011'; break;	681	case 't': ++dec_cur; *cur++ = '\011'; break;
587	case 'n': ++dec->cur; *cur++ = '\012'; break;	682	case 'n': ++dec_cur; *cur++ = '\012'; break;
588	case 'f': ++dec->cur; *cur++ = '\014'; break;	683	case 'f': ++dec_cur; *cur++ = '\014'; break;
589	case 'r': ++dec->cur; *cur++ = '\015'; break;	684	case 'r': ++dec_cur; *cur++ = '\015'; break;
590		685
591	case 'u':	686	case 'u':
592	{	687	{
593	UV lo, hi;	688	UV lo, hi;
594	++dec->cur;	689	++dec_cur;
595		690
		691	dec->cur = dec_cur;
596	hi = decode_4hex (dec);	692	hi = decode_4hex (dec);
		693	dec_cur = dec->cur;
597	if (hi == (UV)-1)	694	if (hi == (UV)-1)
598	goto fail;	695	goto fail;
599		696
600	// possibly a surrogate pair	697	// possibly a surrogate pair
601	if (hi >= 0xd800)	698	if (hi >= 0xd800)
602	if (hi < 0xdc00)	699	if (hi < 0xdc00)
603	{	700	{
604	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')	701	if (dec_cur [0] != '\\' \|\| dec_cur [1] != 'u')
605	ERR ("missing low surrogate character in surrogate pair");	702	ERR ("missing low surrogate character in surrogate pair");
606		703
607	dec->cur += 2;	704	dec_cur += 2;
608		705
		706	dec->cur = dec_cur;
609	lo = decode_4hex (dec);	707	lo = decode_4hex (dec);
		708	dec_cur = dec->cur;
610	if (lo == (UV)-1)	709	if (lo == (UV)-1)
611	goto fail;	710	goto fail;
612		711
613	if (lo < 0xdc00 \|\| lo >= 0xe000)	712	if (lo < 0xdc00 \|\| lo >= 0xe000)
614	ERR ("surrogate pair expected");	713	ERR ("surrogate pair expected");
…		…
628	*cur++ = hi;	727	*cur++ = hi;
629	}	728	}
630	break;	729	break;
631		730
632	default:	731	default:
633	--dec->cur;	732	--dec_cur;
634	ERR ("illegal backslash escape sequence in string");	733	ERR ("illegal backslash escape sequence in string");
635	}	734	}
636	}	735	}
637	else if (ch >= 0x20 && ch <= 0x7f)	736	else if (expect_true (ch >= 0x20 && ch <= 0x7f))
638	*cur++ = ch;	737	*cur++ = ch;
639	else if (ch >= 0x80)	738	else if (ch >= 0x80)
640	{	739	{
641	--dec->cur;
642
643	STRLEN clen;	740	STRLEN clen;
		741	UV uch;
		742
		743	--dec_cur;
		744
644	UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);	745	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
645	if (clen == (STRLEN)-1)	746	if (clen == (STRLEN)-1)
646	ERR ("malformed UTF-8 character in JSON string");	747	ERR ("malformed UTF-8 character in JSON string");
647		748
648	do	749	do
649	{
650	cur++ = dec->cur++;	750	cur++ = dec_cur++;
651	}
652	while (--clen);	751	while (--clen);
653		752
654	utf8 = 1;	753	utf8 = 1;
655	}	754	}
656	else if (!ch)
657	ERR ("unexpected end of string while parsing json string");
658	else	755	else
		756	{
		757	--dec_cur;
		758
		759	if (!ch)
		760	ERR ("unexpected end of string while parsing JSON string");
		761	else
659	ERR ("invalid character encountered");	762	ERR ("invalid character encountered while parsing JSON string");
660		763	}
661	}	764	}
662	while (cur < buf + SHORT_STRING_LEN);	765	while (cur < buf + SHORT_STRING_LEN);
663		766
		767	{
664	STRLEN len = cur - buf;	768	STRLEN len = cur - buf;
665		769
666	if (sv)	770	if (sv)
667	{	771	{
668	SvGROW (sv, SvCUR (sv) + len + 1);	772	SvGROW (sv, SvCUR (sv) + len + 1);
669	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	773	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670	SvCUR_set (sv, SvCUR (sv) + len);	774	SvCUR_set (sv, SvCUR (sv) + len);
671	}	775	}
672	else	776	else
673	sv = newSVpvn (buf, len);	777	sv = newSVpvn (buf, len);
674	}	778	}
		779	}
675	while (*dec->cur != '"');	780	while (*dec_cur != '"');
676		781
677	++dec->cur;	782	++dec_cur;
678		783
679	if (sv)	784	if (sv)
680	{	785	{
681	SvPOK_only (sv);	786	SvPOK_only (sv);
682	*SvEND (sv) = 0;	787	*SvEND (sv) = 0;
…		…
685	SvUTF8_on (sv);	790	SvUTF8_on (sv);
686	}	791	}
687	else	792	else
688	sv = newSVpvn ("", 0);	793	sv = newSVpvn ("", 0);
689		794
		795	dec->cur = dec_cur;
690	return sv;	796	return sv;
691		797
692	fail:	798	fail:
		799	dec->cur = dec_cur;
693	return 0;	800	return 0;
694	}	801	}
695		802
696	static SV *	803	static SV *
697	decode_num (dec_t *dec)	804	decode_num (dec_t *dec)
…		…
755	is_nv = 1;	862	is_nv = 1;
756	}	863	}
757		864
758	if (!is_nv)	865	if (!is_nv)
759	{	866	{
760	UV uv;	867	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
761	int numtype = grok_number (start, dec->cur - start, &uv);	868	if (*start == '-')
762	if (numtype & IS_NUMBER_IN_UV)	869	switch (dec->cur - start)
763	if (numtype & IS_NUMBER_NEG)
764	{	870	{
765	if (uv < (UV)IV_MIN)	871	case 2: return newSViv (-( start [1] - '0' * 1));
766	return newSViv (-(IV)uv);	872	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
		873	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
		874	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
767	}	875	}
		876	else
		877	switch (dec->cur - start)
		878	{
		879	case 1: return newSViv ( start [0] - '0' * 1);
		880	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
		881	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
		882	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		883	}
		884
		885	{
		886	UV uv;
		887	int numtype = grok_number (start, dec->cur - start, &uv);
		888	if (numtype & IS_NUMBER_IN_UV)
		889	if (numtype & IS_NUMBER_NEG)
		890	{
		891	if (uv < (UV)IV_MIN)
		892	return newSViv (-(IV)uv);
		893	}
768	else	894	else
769	return newSVuv (uv);	895	return newSVuv (uv);
		896
		897	// here would likely be the place for bigint support
770	}	898	}
		899	}
771		900
		901	// if we ever support bigint or bigfloat, this is the place for bigfloat
772	return newSVnv (Atof (start));	902	return newSVnv (Atof (start));
773		903
774	fail:	904	fail:
775	return 0;	905	return 0;
776	}	906	}
…		…
778	static SV *	908	static SV *
779	decode_av (dec_t *dec)	909	decode_av (dec_t *dec)
780	{	910	{
781	AV *av = newAV ();	911	AV *av = newAV ();
782		912
		913	DEC_INC_DEPTH;
783	decode_ws (dec);	914	decode_ws (dec);
		915
784	if (*dec->cur == ']')	916	if (*dec->cur == ']')
785	++dec->cur;	917	++dec->cur;
786	else	918	else
787	for (;;)	919	for (;;)
788	{	920	{
…		…
806	ERR (", or ] expected while parsing array");	938	ERR (", or ] expected while parsing array");
807		939
808	++dec->cur;	940	++dec->cur;
809	}	941	}
810		942
		943	DEC_DEC_DEPTH;
811	return newRV_noinc ((SV *)av);	944	return newRV_noinc ((SV *)av);
812		945
813	fail:	946	fail:
814	SvREFCNT_dec (av);	947	SvREFCNT_dec (av);
		948	DEC_DEC_DEPTH;
815	return 0;	949	return 0;
816	}	950	}
817		951
818	static SV *	952	static SV *
819	decode_hv (dec_t *dec)	953	decode_hv (dec_t *dec)
820	{	954	{
821	HV *hv = newHV ();	955	HV *hv = newHV ();
822		956
		957	DEC_INC_DEPTH;
823	decode_ws (dec);	958	decode_ws (dec);
		959
824	if (*dec->cur == '}')	960	if (*dec->cur == '}')
825	++dec->cur;	961	++dec->cur;
826	else	962	else
827	for (;;)	963	for (;;)
828	{	964	{
…		…
841	{	977	{
842	SvREFCNT_dec (key);	978	SvREFCNT_dec (key);
843	goto fail;	979	goto fail;
844	}	980	}
845		981
846	//TODO: optimise
847	hv_store_ent (hv, key, value, 0);	982	hv_store_ent (hv, key, value, 0);
		983	SvREFCNT_dec (key);
848		984
849	decode_ws (dec);	985	decode_ws (dec);
850		986
851	if (*dec->cur == '}')	987	if (*dec->cur == '}')
852	{	988	{
…		…
858	ERR (", or } expected while parsing object/hash");	994	ERR (", or } expected while parsing object/hash");
859		995
860	++dec->cur;	996	++dec->cur;
861	}	997	}
862		998
		999	DEC_DEC_DEPTH;
863	return newRV_noinc ((SV *)hv);	1000	return newRV_noinc ((SV *)hv);
864		1001
865	fail:	1002	fail:
866	SvREFCNT_dec (hv);	1003	SvREFCNT_dec (hv);
		1004	DEC_DEC_DEPTH;
867	return 0;	1005	return 0;
868	}	1006	}
869		1007
870	static SV *	1008	static SV *
871	decode_sv (dec_t *dec)	1009	decode_sv (dec_t *dec)
872	{	1010	{
873	decode_ws (dec);	1011	decode_ws (dec);
		1012
		1013	// the beauty of JSON: you need exactly one character lookahead
		1014	// to parse anything.
874	switch (*dec->cur)	1015	switch (*dec->cur)
875	{	1016	{
876	case '"': ++dec->cur; return decode_str (dec);	1017	case '"': ++dec->cur; return decode_str (dec);
877	case '[': ++dec->cur; return decode_av (dec);	1018	case '[': ++dec->cur; return decode_av (dec);
878	case '{': ++dec->cur; return decode_hv (dec);	1019	case '{': ++dec->cur; return decode_hv (dec);
…		…
914	ERR ("'null' expected");	1055	ERR ("'null' expected");
915		1056
916	break;	1057	break;
917		1058
918	default:	1059	default:
919	ERR ("malformed json string, neither array, object, number, string or atom");	1060	ERR ("malformed JSON string, neither array, object, number, string or atom");
920	break;	1061	break;
921	}	1062	}
922		1063
923	fail:	1064	fail:
924	return 0;	1065	return 0;
925	}	1066	}
926		1067
927	static SV *	1068	static SV *
928	decode_json (SV *string, UV flags)	1069	decode_json (SV string, U32 flags, UV offset_return)
929	{	1070	{
		1071	dec_t dec;
		1072	UV offset;
930	SV *sv;	1073	SV *sv;
		1074
		1075	SvGETMAGIC (string);
		1076	SvUPGRADE (string, SVt_PV);
931		1077
932	if (flags & F_UTF8)	1078	if (flags & F_UTF8)
933	sv_utf8_downgrade (string, 0);	1079	sv_utf8_downgrade (string, 0);
934	else	1080	else
935	sv_utf8_upgrade (string);	1081	sv_utf8_upgrade (string);
936		1082
937	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1083	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938		1084
939	dec_t dec;
940	dec.flags = flags;	1085	dec.flags = flags;
941	dec.cur = SvPVX (string);	1086	dec.cur = SvPVX (string);
942	dec.end = SvEND (string);	1087	dec.end = SvEND (string);
943	dec.err = 0;	1088	dec.err = 0;
		1089	dec.depth = 0;
		1090	dec.maxdepth = DEC_DEPTH (dec.flags);
944		1091
		1092	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
945	sv = decode_sv (&dec);	1093	sv = decode_sv (&dec);
946		1094
		1095	if (!(offset_return \|\| !sv))
		1096	{
		1097	// check for trailing garbage
		1098	decode_ws (&dec);
		1099
		1100	if (*dec.cur)
		1101	{
		1102	dec.err = "garbage after JSON object";
		1103	SvREFCNT_dec (sv);
		1104	sv = 0;
		1105	}
		1106	}
		1107
		1108	if (offset_return \|\| !sv)
		1109	{
		1110	offset = dec.flags & F_UTF8
		1111	? dec.cur - SvPVX (string)
		1112	: utf8_distance (dec.cur, SvPVX (string));
		1113
		1114	if (offset_return)
		1115	*offset_return = offset;
		1116	}
		1117
947	if (!sv)	1118	if (!sv)
948	{	1119	{
949	IV offset = dec.flags & F_UTF8
950	? dec.cur - SvPVX (string)
951	: utf8_distance (dec.cur, SvPVX (string));
952	SV *uni = sv_newmortal ();	1120	SV *uni = sv_newmortal ();
953		1121
954	// horrible hack to silence warning inside pv_uni_display	1122	// horrible hack to silence warning inside pv_uni_display
955	COP cop = *PL_curcop;	1123	COP cop = *PL_curcop;
956	cop.cop_warnings = pWARN_NONE;	1124	cop.cop_warnings = pWARN_NONE;
…		…
958	SAVEVPTR (PL_curcop);	1126	SAVEVPTR (PL_curcop);
959	PL_curcop = &cop;	1127	PL_curcop = &cop;
960	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1128	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961	LEAVE;	1129	LEAVE;
962		1130
963	croak ("%s, at character offset %d (%s)",	1131	croak ("%s, at character offset %d [\"%s\"]",
964	dec.err,	1132	dec.err,
965	(int)offset,	1133	(int)offset,
966	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1134	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967	}	1135	}
968		1136
…		…
981		1149
982	BOOT:	1150	BOOT:
983	{	1151	{
984	int i;	1152	int i;
985		1153
986	memset (decode_hexdigit, 0xff, 256);
987	for (i = 10; i--; )	1154	for (i = 0; i < 256; ++i)
988	decode_hexdigit ['0' + i] = i;	1155	decode_hexdigit [i] =
989		1156	i >= '0' && i <= '9' ? i - '0'
990	for (i = 7; i--; )	1157	: i >= 'a' && i <= 'f' ? i - 'a' + 10
991	{	1158	: i >= 'A' && i <= 'F' ? i - 'A' + 10
992	decode_hexdigit ['a' + i] = 10 + i;	1159	: -1;
993	decode_hexdigit ['A' + i] = 10 + i;
994	}
995		1160
996	json_stash = gv_stashpv ("JSON::XS", 1);	1161	json_stash = gv_stashpv ("JSON::XS", 1);
997	}	1162	}
998		1163
999	PROTOTYPES: DISABLE	1164	PROTOTYPES: DISABLE
…		…
1005	RETVAL	1170	RETVAL
1006		1171
1007	SV ascii (SV self, int enable = 1)	1172	SV ascii (SV self, int enable = 1)
1008	ALIAS:	1173	ALIAS:
1009	ascii = F_ASCII	1174	ascii = F_ASCII
		1175	latin1 = F_LATIN1
1010	utf8 = F_UTF8	1176	utf8 = F_UTF8
1011	indent = F_INDENT	1177	indent = F_INDENT
1012	canonical = F_CANONICAL	1178	canonical = F_CANONICAL
1013	space_before = F_SPACE_BEFORE	1179	space_before = F_SPACE_BEFORE
1014	space_after = F_SPACE_AFTER	1180	space_after = F_SPACE_AFTER
…		…
1026	RETVAL = newSVsv (self);	1192	RETVAL = newSVsv (self);
1027	}	1193	}
1028	OUTPUT:	1194	OUTPUT:
1029	RETVAL	1195	RETVAL
1030		1196
		1197	SV max_depth (SV self, UV max_depth = 0x80000000UL)
		1198	CODE:
		1199	{
		1200	UV *uv = SvJSON (self);
		1201	UV log2 = 0;
		1202
		1203	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
		1204
		1205	while ((1UL << log2) < max_depth)
		1206	++log2;
		1207
		1208	uv = uv & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
		1209
		1210	RETVAL = newSVsv (self);
		1211	}
		1212	OUTPUT:
		1213	RETVAL
		1214
1031	void encode (SV self, SV scalar)	1215	void encode (SV self, SV scalar)
1032	PPCODE:	1216	PPCODE:
1033	XPUSHs (encode_json (scalar, *SvJSON (self)));	1217	XPUSHs (encode_json (scalar, *SvJSON (self)));
1034		1218
1035	void decode (SV self, SV jsonstr)	1219	void decode (SV self, SV jsonstr)
1036	PPCODE:	1220	PPCODE:
1037	XPUSHs (decode_json (jsonstr, *SvJSON (self)));	1221	XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
		1222
		1223	void decode_prefix (SV self, SV jsonstr)
		1224	PPCODE:
		1225	{
		1226	UV offset;
		1227	EXTEND (SP, 2);
		1228	PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
		1229	PUSHs (sv_2mortal (newSVuv (offset)));
		1230	}
1038		1231
1039	PROTOTYPES: ENABLE	1232	PROTOTYPES: ENABLE
1040		1233
1041	void to_json (SV *scalar)	1234	void to_json (SV *scalar)
1042	ALIAS:	1235	ALIAS:
1043	objToJson = 0	1236	objToJson = 0
1044	PPCODE:	1237	PPCODE:
1045	XPUSHs (encode_json (scalar, F_UTF8));	1238	XPUSHs (encode_json (scalar, F_DEFAULT \| F_UTF8));
1046		1239
1047	void from_json (SV *jsonstr)	1240	void from_json (SV *jsonstr)
1048	ALIAS:	1241	ALIAS:
1049	jsonToObj = 0	1242	jsonToObj = 0
1050	PPCODE:	1243	PPCODE:
1051	XPUSHs (decode_json (jsonstr, F_UTF8));	1244	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8, 0));
1052		1245

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.15 by root, Sun Mar 25 02:32:40 2007 UTC vs. Revision 1.40 by root, Tue Jun 12 01:27:02 2007 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.15 by root, Sun Mar 25 02:32:40 2007 UTC vs.
Revision 1.40 by root, Tue Jun 12 01:27:02 2007 UTC