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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.20 by root, Sun Mar 25 21:53:19 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"
		9
		10	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		11	# define snprintf _snprintf // C compilers have this in stdio.h
		12	#endif
8		13
9	#define F_ASCII 0x00000001UL	14	#define F_ASCII 0x00000001UL
		15	#define F_LATIN1 0x00000002UL
10	#define F_UTF8 0x00000002UL	16	#define F_UTF8 0x00000004UL
11	#define F_INDENT 0x00000004UL	17	#define F_INDENT 0x00000008UL
12	#define F_CANONICAL 0x00000008UL	18	#define F_CANONICAL 0x00000010UL
13	#define F_SPACE_BEFORE 0x00000010UL	19	#define F_SPACE_BEFORE 0x00000020UL
14	#define F_SPACE_AFTER 0x00000020UL	20	#define F_SPACE_AFTER 0x00000040UL
15	#define F_ALLOW_NONREF 0x00000080UL	21	#define F_ALLOW_NONREF 0x00000100UL
16	#define F_SHRINK 0x00000100UL	22	#define F_SHRINK 0x00000200UL
17	#define F_MAXDEPTH 0xf8000000UL	23	#define F_MAXDEPTH 0xf8000000UL
18	#define S_MAXDEPTH 27	24	#define S_MAXDEPTH 27
19		25
20	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))	26	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21		27
22	// F_SELFCONVERT? <=> to_json/toJson	28	// F_SELFCONVERT? <=> to_json/toJson
23	// F_BLESSED? <=> { $__class__$ => }	29	// F_BLESSED? <=> { $__class__$ => }
24		30
25	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	31	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
26	#define F_DEFAULT (13UL << S_MAXDEPTH)	32	#define F_DEFAULT (9UL << S_MAXDEPTH)
27		33
28	#define INIT_SIZE 32 // initial scalar size to be allocated	34	#define INIT_SIZE 32 // initial scalar size to be allocated
29	#define INDENT_STEP 3 // spaces per indentation level	35	#define INDENT_STEP 3 // spaces per indentation level
30		36
31	#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32	#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
33		38
34	#define SB do {	39	#define SB do {
35	#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)
36		52
37	static HV *json_stash; // JSON::XS::	53	static HV *json_stash; // JSON::XS::
38		54
39	/////////////////////////////////////////////////////////////////////////////	55	/////////////////////////////////////////////////////////////////////////////
40	// utility functions	56	// utility functions
…		…
65	// 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
66	// case of an error.	82	// case of an error.
67	// we special-case "safe" characters from U+80 .. U+7FF,	83	// we special-case "safe" characters from U+80 .. U+7FF,
68	// but use the very good perl function to parse anything else.	84	// but use the very good perl function to parse anything else.
69	// note that we never call this function for a ascii codepoints	85	// note that we never call this function for a ascii codepoints
70	static UV	86	inline UV
71	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)	87	decode_utf8 (unsigned char s, STRLEN len, STRLEN clen)
72	{	88	{
73	if (s[0] > 0xdf \|\| s[0] < 0xc2)	89	if (expect_false (s[0] > 0xdf \|\| s[0] < 0xc2))
74	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);	90	return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)	91	else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76	{	92	{
77	*clen = 2;	93	*clen = 2;
78	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);	94	return ((s[0] & 0x1f) << 6) \| (s[1] & 0x3f);
79	}	95	}
80	else	96	else
		97	{
		98	*clen = (STRLEN)-1;
81	return (UV)-1;	99	return (UV)-1;
		100	}
82	}	101	}
83		102
84	/////////////////////////////////////////////////////////////////////////////	103	/////////////////////////////////////////////////////////////////////////////
85	// encoder	104	// encoder
86		105
…		…
93	U32 flags; // F_*	112	U32 flags; // F_*
94	U32 indent; // indentation level	113	U32 indent; // indentation level
95	U32 maxdepth; // max. indentation/recursion level	114	U32 maxdepth; // max. indentation/recursion level
96	} enc_t;	115	} enc_t;
97		116
98	static void	117	inline void
99	need (enc_t *enc, STRLEN len)	118	need (enc_t *enc, STRLEN len)
100	{	119	{
101	if (enc->cur + len >= enc->end)	120	if (expect_false (enc->cur + len >= enc->end))
102	{	121	{
103	STRLEN cur = enc->cur - SvPVX (enc->sv);	122	STRLEN cur = enc->cur - SvPVX (enc->sv);
104	SvGROW (enc->sv, cur + len + 1);	123	SvGROW (enc->sv, cur + len + 1);
105	enc->cur = SvPVX (enc->sv) + cur;	124	enc->cur = SvPVX (enc->sv) + cur;
106	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);	125	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
107	}	126	}
108	}	127	}
109		128
110	static void	129	inline void
111	encode_ch (enc_t *enc, char ch)	130	encode_ch (enc_t *enc, char ch)
112	{	131	{
113	need (enc, 1);	132	need (enc, 1);
114	*enc->cur++ = ch;	133	*enc->cur++ = ch;
115	}	134	}
…		…
123		142
124	while (str < end)	143	while (str < end)
125	{	144	{
126	unsigned char ch = (unsigned char )str;	145	unsigned char ch = (unsigned char )str;
127		146
128	if (ch >= 0x20 && ch < 0x80) // most common case	147	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
129	{	148	{
130	if (ch == '"') // but with slow exceptions	149	if (expect_false (ch == '"')) // but with slow exceptions
131	{	150	{
132	need (enc, len += 1);	151	need (enc, len += 1);
133	*enc->cur++ = '\\';	152	*enc->cur++ = '\\';
134	*enc->cur++ = '"';	153	*enc->cur++ = '"';
135	}	154	}
136	else if (ch == '\\')	155	else if (expect_false (ch == '\\'))
137	{	156	{
138	need (enc, len += 1);	157	need (enc, len += 1);
139	*enc->cur++ = '\\';	158	*enc->cur++ = '\\';
140	*enc->cur++ = '\\';	159	*enc->cur++ = '\\';
141	}	160	}
…		…
159	STRLEN clen;	178	STRLEN clen;
160	UV uch;	179	UV uch;
161		180
162	if (is_utf8)	181	if (is_utf8)
163	{	182	{
164	//uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165	uch = decode_utf8 (str, end - str, &clen);	183	uch = decode_utf8 (str, end - str, &clen);
166	if (clen == (STRLEN)-1)	184	if (clen == (STRLEN)-1)
167	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);
168	}	186	}
169	else	187	else
…		…
173	}	191	}
174		192
175	if (uch > 0x10FFFFUL)	193	if (uch > 0x10FFFFUL)
176	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);
177		195
178	if (uch < 0x80 \|\| enc->flags & F_ASCII)	196	if (uch < 0x80 \|\| enc->flags & F_ASCII \|\| (enc->flags & F_LATIN1 && uch > 0xFF))
179	{	197	{
180	if (uch > 0xFFFFUL)	198	if (uch > 0xFFFFUL)
181	{	199	{
182	need (enc, len += 11);	200	need (enc, len += 11);
183	sprintf (enc->cur, "\\u%04x\\u%04x",	201	sprintf (enc->cur, "\\u%04x\\u%04x",
…		…
197	*enc->cur++ = hexdigit [(uch >> 0) & 15];	215	*enc->cur++ = hexdigit [(uch >> 0) & 15];
198	}	216	}
199		217
200	str += clen;	218	str += clen;
201	}	219	}
		220	else if (enc->flags & F_LATIN1)
		221	{
		222	*enc->cur++ = uch;
		223	str += clen;
		224	}
202	else if (is_utf8)	225	else if (is_utf8)
203	{	226	{
204	need (enc, len += clen);	227	need (enc, len += clen);
205	do	228	do
206	{	229	{
…		…
208	}	231	}
209	while (--clen);	232	while (--clen);
210	}	233	}
211	else	234	else
212	{	235	{
213	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
214	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	237	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
215	++str;	238	++str;
216	}	239	}
217	}	240	}
218	}	241	}
…		…
220		243
221	--len;	244	--len;
222	}	245	}
223	}	246	}
224		247
225	static void	248	inline void
226	encode_indent (enc_t *enc)	249	encode_indent (enc_t *enc)
227	{	250	{
228	if (enc->flags & F_INDENT)	251	if (enc->flags & F_INDENT)
229	{	252	{
230	int spaces = enc->indent * INDENT_STEP;	253	int spaces = enc->indent * INDENT_STEP;
…		…
233	memset (enc->cur, ' ', spaces);	256	memset (enc->cur, ' ', spaces);
234	enc->cur += spaces;	257	enc->cur += spaces;
235	}	258	}
236	}	259	}
237		260
238	static void	261	inline void
239	encode_space (enc_t *enc)	262	encode_space (enc_t *enc)
240	{	263	{
241	need (enc, 1);	264	need (enc, 1);
242	encode_ch (enc, ' ');	265	encode_ch (enc, ' ');
243	}	266	}
244		267
245	static void	268	inline void
246	encode_nl (enc_t *enc)	269	encode_nl (enc_t *enc)
247	{	270	{
248	if (enc->flags & F_INDENT)	271	if (enc->flags & F_INDENT)
249	{	272	{
250	need (enc, 1);	273	need (enc, 1);
251	encode_ch (enc, '\n');	274	encode_ch (enc, '\n');
252	}	275	}
253	}	276	}
254		277
255	static void	278	inline void
256	encode_comma (enc_t *enc)	279	encode_comma (enc_t *enc)
257	{	280	{
258	encode_ch (enc, ',');	281	encode_ch (enc, ',');
259		282
260	if (enc->flags & F_INDENT)	283	if (enc->flags & F_INDENT)
…		…
267		290
268	static void	291	static void
269	encode_av (enc_t enc, AV av)	292	encode_av (enc_t enc, AV av)
270	{	293	{
271	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)");
272		298
273	encode_ch (enc, '['); encode_nl (enc);	299	encode_ch (enc, '['); encode_nl (enc);
274	++enc->indent;	300	++enc->indent;
275		301
276	for (i = 0; i <= len; ++i)	302	for (i = 0; i <= len; ++i)
…		…
343	static void	369	static void
344	encode_hv (enc_t enc, HV hv)	370	encode_hv (enc_t enc, HV hv)
345	{	371	{
346	int count, i;	372	int count, i;
347		373
		374	if (enc->indent >= enc->maxdepth)
		375	croak ("data structure too deep (hit recursion limit)");
		376
348	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;	377	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
349		378
350	if ((count = hv_iterinit (hv)))	379	if ((count = hv_iterinit (hv)))
351	{	380	{
352	// for canonical output we have to sort by keys first	381	// for canonical output we have to sort by keys first
353	// actually, this is mostly due to the stupid so-called	382	// actually, this is mostly due to the stupid so-called
354	// security workaround added somewhere in 5.8.x.	383	// security workaround added somewhere in 5.8.x.
355	// that randomises hash orderings	384	// that randomises hash orderings
356	if (enc->flags & F_CANONICAL)	385	if (enc->flags & F_CANONICAL)
357	{	386	{
358	HE he, hes [count]; // if your compiler dies here, you need to enable C99 mode
359	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
360		394
361	i = 0;	395	i = 0;
362	while ((he = hv_iternext (hv)))	396	while ((he = hv_iternext (hv)))
363	{	397	{
364	hes [i++] = he;	398	hes [i++] = he;
…		…
399		433
400	encode_nl (enc);	434	encode_nl (enc);
401	}	435	}
402	else	436	else
403	{	437	{
404	SV *sv;
405	HE *he = hv_iternext (hv);	438	HE *he = hv_iternext (hv);
406		439
407	for (;;)	440	for (;;)
408	{	441	{
409	encode_indent (enc);	442	encode_indent (enc);
…		…
418	encode_nl (enc);	451	encode_nl (enc);
419	}	452	}
420	}	453	}
421		454
422	--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))));
423	}	484	}
424		485
425	static void	486	static void
426	encode_sv (enc_t enc, SV sv)	487	encode_sv (enc_t enc, SV sv)
427	{	488	{
…		…
435	encode_str (enc, str, len, SvUTF8 (sv));	496	encode_str (enc, str, len, SvUTF8 (sv));
436	encode_ch (enc, '"');	497	encode_ch (enc, '"');
437	}	498	}
438	else if (SvNOKp (sv))	499	else if (SvNOKp (sv))
439	{	500	{
		501	// trust that perl will do the right thing w.r.t. JSON syntax.
440	need (enc, NV_DIG + 32);	502	need (enc, NV_DIG + 32);
441	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);	503	Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
442	enc->cur += strlen (enc->cur);	504	enc->cur += strlen (enc->cur);
443	}	505	}
444	else if (SvIOKp (sv))	506	else if (SvIOKp (sv))
445	{	507	{
446	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);
447	enc->cur +=	513	enc->cur +=
448	SvIsUV(sv)	514	SvIsUV(sv)
449	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	515	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
450	: 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	}
451	}	545	}
452	else if (SvROK (sv))	546	else if (SvROK (sv))
453	{	547	encode_rv (enc, SvRV (sv));
454	SV *rv = SvRV (sv);
455
456	if (enc->indent >= enc->maxdepth)
457	croak ("data structure too deep (hit recursion limit)");
458
459	switch (SvTYPE (rv))
460	{
461	case SVt_PVAV: encode_av (enc, (AV *)rv); break;
462	case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
463
464	default:
465	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
466	SvPV_nolen (sv));
467	}
468	}
469	else if (!SvOK (sv))	548	else if (!SvOK (sv))
470	encode_str (enc, "null", 4, 0);	549	encode_str (enc, "null", 4, 0);
471	else	550	else
472	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",
473	SvPV_nolen (sv), SvFLAGS (sv));	552	SvPV_nolen (sv), SvFLAGS (sv));
474	}	553	}
475		554
476	static SV *	555	static SV *
477	encode_json (SV *scalar, U32 flags)	556	encode_json (SV *scalar, U32 flags)
478	{	557	{
		558	enc_t enc;
		559
479	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	560	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
480	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)");
481		562
482	enc_t enc;
483	enc.flags = flags;	563	enc.flags = flags;
484	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	564	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
485	enc.cur = SvPVX (enc.sv);	565	enc.cur = SvPVX (enc.sv);
486	enc.end = SvEND (enc.sv);	566	enc.end = SvEND (enc.sv);
487	enc.indent = 0;	567	enc.indent = 0;
488	enc.maxdepth = DEC_DEPTH (flags);	568	enc.maxdepth = DEC_DEPTH (flags);
489		569
490	SvPOK_only (enc.sv);	570	SvPOK_only (enc.sv);
491	encode_sv (&enc, scalar);	571	encode_sv (&enc, scalar);
492		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
493	if (!(flags & (F_ASCII \| F_UTF8)))	576	if (!(flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
494	SvUTF8_on (enc.sv);	577	SvUTF8_on (enc.sv);
495
496	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
497		578
498	if (enc.flags & F_SHRINK)	579	if (enc.flags & F_SHRINK)
499	shrink (enc.sv);	580	shrink (enc.sv);
500		581
501	return enc.sv;	582	return enc.sv;
…		…
513	U32 flags; // F_*	594	U32 flags; // F_*
514	U32 depth; // recursion depth	595	U32 depth; // recursion depth
515	U32 maxdepth; // recursion depth limit	596	U32 maxdepth; // recursion depth limit
516	} dec_t;	597	} dec_t;
517		598
518	static void	599	inline void
519	decode_ws (dec_t *dec)	600	decode_ws (dec_t *dec)
520	{	601	{
521	for (;;)	602	for (;;)
522	{	603	{
523	char ch = *dec->cur;	604	char ch = *dec->cur;
…		…
549	decode_4hex (dec_t *dec)	630	decode_4hex (dec_t *dec)
550	{	631	{
551	signed char d1, d2, d3, d4;	632	signed char d1, d2, d3, d4;
552	unsigned char cur = (unsigned char )dec->cur;	633	unsigned char cur = (unsigned char )dec->cur;
553		634
554	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");
555	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");
556	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");
557	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");
558		639
559	dec->cur += 4;	640	dec->cur += 4;
560		641
561	return ((UV)d1) << 12	642	return ((UV)d1) << 12
562	\| ((UV)d2) << 8	643	\| ((UV)d2) << 8
…		…
570	static SV *	651	static SV *
571	decode_str (dec_t *dec)	652	decode_str (dec_t *dec)
572	{	653	{
573	SV *sv = 0;	654	SV *sv = 0;
574	int utf8 = 0;	655	int utf8 = 0;
		656	char *dec_cur = dec->cur;
575		657
576	do	658	do
577	{	659	{
578	char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];	660	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
579	char *cur = buf;	661	char *cur = buf;
580		662
581	do	663	do
582	{	664	{
583	unsigned char ch = (unsigned char )dec->cur++;	665	unsigned char ch = (unsigned char )dec_cur++;
584		666
585	if (ch == '"')	667	if (expect_false (ch == '"'))
586	{	668	{
587	--dec->cur;	669	--dec_cur;
588	break;	670	break;
589	}	671	}
590	else if (ch == '\\')	672	else if (expect_false (ch == '\\'))
591	{	673	{
592	switch (*dec->cur)	674	switch (*dec_cur)
593	{	675	{
594	case '\\':	676	case '\\':
595	case '/':	677	case '/':
596	case '"': cur++ = dec->cur++; break;	678	case '"': cur++ = dec_cur++; break;
597		679
598	case 'b': ++dec->cur; *cur++ = '\010'; break;	680	case 'b': ++dec_cur; *cur++ = '\010'; break;
599	case 't': ++dec->cur; *cur++ = '\011'; break;	681	case 't': ++dec_cur; *cur++ = '\011'; break;
600	case 'n': ++dec->cur; *cur++ = '\012'; break;	682	case 'n': ++dec_cur; *cur++ = '\012'; break;
601	case 'f': ++dec->cur; *cur++ = '\014'; break;	683	case 'f': ++dec_cur; *cur++ = '\014'; break;
602	case 'r': ++dec->cur; *cur++ = '\015'; break;	684	case 'r': ++dec_cur; *cur++ = '\015'; break;
603		685
604	case 'u':	686	case 'u':
605	{	687	{
606	UV lo, hi;	688	UV lo, hi;
607	++dec->cur;	689	++dec_cur;
608		690
		691	dec->cur = dec_cur;
609	hi = decode_4hex (dec);	692	hi = decode_4hex (dec);
		693	dec_cur = dec->cur;
610	if (hi == (UV)-1)	694	if (hi == (UV)-1)
611	goto fail;	695	goto fail;
612		696
613	// possibly a surrogate pair	697	// possibly a surrogate pair
614	if (hi >= 0xd800)	698	if (hi >= 0xd800)
615	if (hi < 0xdc00)	699	if (hi < 0xdc00)
616	{	700	{
617	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')	701	if (dec_cur [0] != '\\' \|\| dec_cur [1] != 'u')
618	ERR ("missing low surrogate character in surrogate pair");	702	ERR ("missing low surrogate character in surrogate pair");
619		703
620	dec->cur += 2;	704	dec_cur += 2;
621		705
		706	dec->cur = dec_cur;
622	lo = decode_4hex (dec);	707	lo = decode_4hex (dec);
		708	dec_cur = dec->cur;
623	if (lo == (UV)-1)	709	if (lo == (UV)-1)
624	goto fail;	710	goto fail;
625		711
626	if (lo < 0xdc00 \|\| lo >= 0xe000)	712	if (lo < 0xdc00 \|\| lo >= 0xe000)
627	ERR ("surrogate pair expected");	713	ERR ("surrogate pair expected");
…		…
641	*cur++ = hi;	727	*cur++ = hi;
642	}	728	}
643	break;	729	break;
644		730
645	default:	731	default:
646	--dec->cur;	732	--dec_cur;
647	ERR ("illegal backslash escape sequence in string");	733	ERR ("illegal backslash escape sequence in string");
648	}	734	}
649	}	735	}
650	else if (ch >= 0x20 && ch <= 0x7f)	736	else if (expect_true (ch >= 0x20 && ch <= 0x7f))
651	*cur++ = ch;	737	*cur++ = ch;
652	else if (ch >= 0x80)	738	else if (ch >= 0x80)
653	{	739	{
654	--dec->cur;
655
656	STRLEN clen;	740	STRLEN clen;
		741	UV uch;
		742
		743	--dec_cur;
		744
657	UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);	745	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
658	if (clen == (STRLEN)-1)	746	if (clen == (STRLEN)-1)
659	ERR ("malformed UTF-8 character in JSON string");	747	ERR ("malformed UTF-8 character in JSON string");
660		748
661	do	749	do
662	{
663	cur++ = dec->cur++;	750	cur++ = dec_cur++;
664	}
665	while (--clen);	751	while (--clen);
666		752
667	utf8 = 1;	753	utf8 = 1;
668	}	754	}
669	else if (!ch)
670	ERR ("unexpected end of string while parsing json string");
671	else	755	else
		756	{
		757	--dec_cur;
		758
		759	if (!ch)
		760	ERR ("unexpected end of string while parsing JSON string");
		761	else
672	ERR ("invalid character encountered");	762	ERR ("invalid character encountered while parsing JSON string");
673		763	}
674	}	764	}
675	while (cur < buf + SHORT_STRING_LEN);	765	while (cur < buf + SHORT_STRING_LEN);
676		766
		767	{
677	STRLEN len = cur - buf;	768	STRLEN len = cur - buf;
678		769
679	if (sv)	770	if (sv)
680	{	771	{
681	SvGROW (sv, SvCUR (sv) + len + 1);	772	SvGROW (sv, SvCUR (sv) + len + 1);
682	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	773	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
683	SvCUR_set (sv, SvCUR (sv) + len);	774	SvCUR_set (sv, SvCUR (sv) + len);
684	}	775	}
685	else	776	else
686	sv = newSVpvn (buf, len);	777	sv = newSVpvn (buf, len);
687	}	778	}
		779	}
688	while (*dec->cur != '"');	780	while (*dec_cur != '"');
689		781
690	++dec->cur;	782	++dec_cur;
691		783
692	if (sv)	784	if (sv)
693	{	785	{
694	SvPOK_only (sv);	786	SvPOK_only (sv);
695	*SvEND (sv) = 0;	787	*SvEND (sv) = 0;
…		…
698	SvUTF8_on (sv);	790	SvUTF8_on (sv);
699	}	791	}
700	else	792	else
701	sv = newSVpvn ("", 0);	793	sv = newSVpvn ("", 0);
702		794
		795	dec->cur = dec_cur;
703	return sv;	796	return sv;
704		797
705	fail:	798	fail:
		799	dec->cur = dec_cur;
706	return 0;	800	return 0;
707	}	801	}
708		802
709	static SV *	803	static SV *
710	decode_num (dec_t *dec)	804	decode_num (dec_t *dec)
…		…
768	is_nv = 1;	862	is_nv = 1;
769	}	863	}
770		864
771	if (!is_nv)	865	if (!is_nv)
772	{	866	{
773	UV uv;	867	// special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
774	int numtype = grok_number (start, dec->cur - start, &uv);	868	if (*start == '-')
775	if (numtype & IS_NUMBER_IN_UV)	869	switch (dec->cur - start)
776	if (numtype & IS_NUMBER_NEG)
777	{	870	{
778	if (uv < (UV)IV_MIN)	871	case 2: return newSViv (-( start [1] - '0' * 1));
779	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));
780	}	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	}
781	else	894	else
782	return newSVuv (uv);	895	return newSVuv (uv);
		896
		897	// here would likely be the place for bigint support
783	}	898	}
		899	}
784		900
		901	// if we ever support bigint or bigfloat, this is the place for bigfloat
785	return newSVnv (Atof (start));	902	return newSVnv (Atof (start));
786		903
787	fail:	904	fail:
788	return 0;	905	return 0;
789	}	906	}
…		…
890		1007
891	static SV *	1008	static SV *
892	decode_sv (dec_t *dec)	1009	decode_sv (dec_t *dec)
893	{	1010	{
894	decode_ws (dec);	1011	decode_ws (dec);
		1012
		1013	// the beauty of JSON: you need exactly one character lookahead
		1014	// to parse anything.
895	switch (*dec->cur)	1015	switch (*dec->cur)
896	{	1016	{
897	case '"': ++dec->cur; return decode_str (dec);	1017	case '"': ++dec->cur; return decode_str (dec);
898	case '[': ++dec->cur; return decode_av (dec);	1018	case '[': ++dec->cur; return decode_av (dec);
899	case '{': ++dec->cur; return decode_hv (dec);	1019	case '{': ++dec->cur; return decode_hv (dec);
…		…
935	ERR ("'null' expected");	1055	ERR ("'null' expected");
936		1056
937	break;	1057	break;
938		1058
939	default:	1059	default:
940	ERR ("malformed json string, neither array, object, number, string or atom");	1060	ERR ("malformed JSON string, neither array, object, number, string or atom");
941	break;	1061	break;
942	}	1062	}
943		1063
944	fail:	1064	fail:
945	return 0;	1065	return 0;
946	}	1066	}
947		1067
948	static SV *	1068	static SV *
949	decode_json (SV *string, U32 flags)	1069	decode_json (SV string, U32 flags, UV offset_return)
950	{	1070	{
		1071	dec_t dec;
		1072	UV offset;
951	SV *sv;	1073	SV *sv;
		1074
		1075	SvGETMAGIC (string);
		1076	SvUPGRADE (string, SVt_PV);
952		1077
953	if (flags & F_UTF8)	1078	if (flags & F_UTF8)
954	sv_utf8_downgrade (string, 0);	1079	sv_utf8_downgrade (string, 0);
955	else	1080	else
956	sv_utf8_upgrade (string);	1081	sv_utf8_upgrade (string);
957		1082
958	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1083	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
959		1084
960	dec_t dec;
961	dec.flags = flags;	1085	dec.flags = flags;
962	dec.cur = SvPVX (string);	1086	dec.cur = SvPVX (string);
963	dec.end = SvEND (string);	1087	dec.end = SvEND (string);
964	dec.err = 0;	1088	dec.err = 0;
965	dec.depth = 0;	1089	dec.depth = 0;
966	dec.maxdepth = DEC_DEPTH (dec.flags);	1090	dec.maxdepth = DEC_DEPTH (dec.flags);
967		1091
968	*dec.end = 0; // this should basically be a nop, too, but make sure its there	1092	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
969	sv = decode_sv (&dec);	1093	sv = decode_sv (&dec);
970		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
971	if (!sv)	1118	if (!sv)
972	{	1119	{
973	IV offset = dec.flags & F_UTF8
974	? dec.cur - SvPVX (string)
975	: utf8_distance (dec.cur, SvPVX (string));
976	SV *uni = sv_newmortal ();	1120	SV *uni = sv_newmortal ();
977		1121
978	// horrible hack to silence warning inside pv_uni_display	1122	// horrible hack to silence warning inside pv_uni_display
979	COP cop = *PL_curcop;	1123	COP cop = *PL_curcop;
980	cop.cop_warnings = pWARN_NONE;	1124	cop.cop_warnings = pWARN_NONE;
…		…
982	SAVEVPTR (PL_curcop);	1126	SAVEVPTR (PL_curcop);
983	PL_curcop = &cop;	1127	PL_curcop = &cop;
984	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);
985	LEAVE;	1129	LEAVE;
986		1130
987	croak ("%s, at character offset %d (%s)",	1131	croak ("%s, at character offset %d [\"%s\"]",
988	dec.err,	1132	dec.err,
989	(int)offset,	1133	(int)offset,
990	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1134	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
991	}	1135	}
992		1136
…		…
1004	MODULE = JSON::XS PACKAGE = JSON::XS	1148	MODULE = JSON::XS PACKAGE = JSON::XS
1005		1149
1006	BOOT:	1150	BOOT:
1007	{	1151	{
1008	int i;	1152	int i;
1009
1010	memset (decode_hexdigit, 0xff, 256);
1011		1153
1012	for (i = 0; i < 256; ++i)	1154	for (i = 0; i < 256; ++i)
1013	decode_hexdigit [i] =	1155	decode_hexdigit [i] =
1014	i >= '0' && i <= '9' ? i - '0'	1156	i >= '0' && i <= '9' ? i - '0'
1015	: i >= 'a' && i <= 'f' ? i - 'a' + 10	1157	: i >= 'a' && i <= 'f' ? i - 'a' + 10
…		…
1028	RETVAL	1170	RETVAL
1029		1171
1030	SV ascii (SV self, int enable = 1)	1172	SV ascii (SV self, int enable = 1)
1031	ALIAS:	1173	ALIAS:
1032	ascii = F_ASCII	1174	ascii = F_ASCII
		1175	latin1 = F_LATIN1
1033	utf8 = F_UTF8	1176	utf8 = F_UTF8
1034	indent = F_INDENT	1177	indent = F_INDENT
1035	canonical = F_CANONICAL	1178	canonical = F_CANONICAL
1036	space_before = F_SPACE_BEFORE	1179	space_before = F_SPACE_BEFORE
1037	space_after = F_SPACE_AFTER	1180	space_after = F_SPACE_AFTER
…		…
1049	RETVAL = newSVsv (self);	1192	RETVAL = newSVsv (self);
1050	}	1193	}
1051	OUTPUT:	1194	OUTPUT:
1052	RETVAL	1195	RETVAL
1053		1196
1054	SV max_depth (SV self, int max_depth = 0x80000000UL)	1197	SV max_depth (SV self, UV max_depth = 0x80000000UL)
1055	CODE:	1198	CODE:
1056	{	1199	{
1057	UV *uv = SvJSON (self);	1200	UV *uv = SvJSON (self);
1058	UV log2 = 0;	1201	UV log2 = 0;
1059		1202
…		…
1073	PPCODE:	1216	PPCODE:
1074	XPUSHs (encode_json (scalar, *SvJSON (self)));	1217	XPUSHs (encode_json (scalar, *SvJSON (self)));
1075		1218
1076	void decode (SV self, SV jsonstr)	1219	void decode (SV self, SV jsonstr)
1077	PPCODE:	1220	PPCODE:
1078	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	}
1079		1231
1080	PROTOTYPES: ENABLE	1232	PROTOTYPES: ENABLE
1081		1233
1082	void to_json (SV *scalar)	1234	void to_json (SV *scalar)
1083	ALIAS:	1235	ALIAS:
…		…
1087		1239
1088	void from_json (SV *jsonstr)	1240	void from_json (SV *jsonstr)
1089	ALIAS:	1241	ALIAS:
1090	jsonToObj = 0	1242	jsonToObj = 0
1091	PPCODE:	1243	PPCODE:
1092	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8));	1244	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8, 0));
1093		1245

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.20 by root, Sun Mar 25 21:53:19 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 1.20 by root, Sun Mar 25 21:53:19 2007 UTC vs.
Revision 1.40 by root, Tue Jun 12 01:27:02 2007 UTC