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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.13 by root, Sat Mar 24 22:55:16 2007 UTC vs.
Revision 1.38 by root, Mon Jun 11 03:18:07 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 256 // 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	}
…		…
166	}	192	}
167		193
168	if (uch > 0x10FFFFUL)	194	if (uch > 0x10FFFFUL)
169	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);	195	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
170		196
171	if (uch < 0x80 \|\| enc->flags & F_ASCII)	197	if (uch < 0x80 \|\| enc->flags & F_ASCII \|\| (enc->flags & F_LATIN1 && uch > 0xFF))
172	{	198	{
173	if (uch > 0xFFFFUL)	199	if (uch > 0xFFFFUL)
174	{	200	{
175	need (enc, len += 11);	201	need (enc, len += 11);
176	sprintf (enc->cur, "\\u%04x\\u%04x",	202	sprintf (enc->cur, "\\u%04x\\u%04x",
…		…
190	*enc->cur++ = hexdigit [(uch >> 0) & 15];	216	*enc->cur++ = hexdigit [(uch >> 0) & 15];
191	}	217	}
192		218
193	str += clen;	219	str += clen;
194	}	220	}
		221	else if (enc->flags & F_LATIN1)
		222	{
		223	*enc->cur++ = uch;
		224	str += clen;
		225	}
195	else if (is_utf8)	226	else if (is_utf8)
196	{	227	{
197	need (enc, len += clen);	228	need (enc, len += clen);
198	do	229	do
199	{	230	{
…		…
201	}	232	}
202	while (--clen);	233	while (--clen);
203	}	234	}
204	else	235	else
205	{	236	{
206	need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed	237	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
207	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);	238	enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
208	++str;	239	++str;
209	}	240	}
210	}	241	}
211	}	242	}
…		…
213		244
214	--len;	245	--len;
215	}	246	}
216	}	247	}
217		248
218	static void	249	inline void
219	encode_indent (enc_t *enc)	250	encode_indent (enc_t *enc)
220	{	251	{
221	if (enc->flags & F_INDENT)	252	if (enc->flags & F_INDENT)
222	{	253	{
223	int spaces = enc->indent * INDENT_STEP;	254	int spaces = enc->indent * INDENT_STEP;
…		…
226	memset (enc->cur, ' ', spaces);	257	memset (enc->cur, ' ', spaces);
227	enc->cur += spaces;	258	enc->cur += spaces;
228	}	259	}
229	}	260	}
230		261
231	static void	262	inline void
232	encode_space (enc_t *enc)	263	encode_space (enc_t *enc)
233	{	264	{
234	need (enc, 1);	265	need (enc, 1);
235	encode_ch (enc, ' ');	266	encode_ch (enc, ' ');
236	}	267	}
237		268
238	static void	269	inline void
239	encode_nl (enc_t *enc)	270	encode_nl (enc_t *enc)
240	{	271	{
241	if (enc->flags & F_INDENT)	272	if (enc->flags & F_INDENT)
242	{	273	{
243	need (enc, 1);	274	need (enc, 1);
244	encode_ch (enc, '\n');	275	encode_ch (enc, '\n');
245	}	276	}
246	}	277	}
247		278
248	static void	279	inline void
249	encode_comma (enc_t *enc)	280	encode_comma (enc_t *enc)
250	{	281	{
251	encode_ch (enc, ',');	282	encode_ch (enc, ',');
252		283
253	if (enc->flags & F_INDENT)	284	if (enc->flags & F_INDENT)
…		…
260		291
261	static void	292	static void
262	encode_av (enc_t enc, AV av)	293	encode_av (enc_t enc, AV av)
263	{	294	{
264	int i, len = av_len (av);	295	int i, len = av_len (av);
		296
		297	if (enc->indent >= enc->maxdepth)
		298	croak ("data structure too deep (hit recursion limit)");
265		299
266	encode_ch (enc, '['); encode_nl (enc);	300	encode_ch (enc, '['); encode_nl (enc);
267	++enc->indent;	301	++enc->indent;
268		302
269	for (i = 0; i <= len; ++i)	303	for (i = 0; i <= len; ++i)
…		…
336	static void	370	static void
337	encode_hv (enc_t enc, HV hv)	371	encode_hv (enc_t enc, HV hv)
338	{	372	{
339	int count, i;	373	int count, i;
340		374
		375	if (enc->indent >= enc->maxdepth)
		376	croak ("data structure too deep (hit recursion limit)");
		377
341	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;	378	encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
342		379
343	if ((count = hv_iterinit (hv)))	380	if ((count = hv_iterinit (hv)))
344	{	381	{
345	// for canonical output we have to sort by keys first	382	// for canonical output we have to sort by keys first
346	// actually, this is mostly due to the stupid so-called	383	// actually, this is mostly due to the stupid so-called
347	// security workaround added somewhere in 5.8.x.	384	// security workaround added somewhere in 5.8.x.
348	// that randomises hash orderings	385	// that randomises hash orderings
349	if (enc->flags & F_CANONICAL)	386	if (enc->flags & F_CANONICAL)
350	{	387	{
351	HE he, hes [count]; // if your compiler dies here, you need to enable C99 mode
352	int fast = 1;	388	int fast = 1;
		389	HE *he;
		390	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
		391	HE *hes = _alloca (count sizeof (HE));
		392	#else
		393	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
		394	#endif
353		395
354	i = 0;	396	i = 0;
355	while ((he = hv_iternext (hv)))	397	while ((he = hv_iternext (hv)))
356	{	398	{
357	hes [i++] = he;	399	hes [i++] = he;
…		…
392		434
393	encode_nl (enc);	435	encode_nl (enc);
394	}	436	}
395	else	437	else
396	{	438	{
397	SV *sv;
398	HE *he = hv_iternext (hv);	439	HE *he = hv_iternext (hv);
399		440
400	for (;;)	441	for (;;)
401	{	442	{
402	encode_indent (enc);	443	encode_indent (enc);
…		…
411	encode_nl (enc);	452	encode_nl (enc);
412	}	453	}
413	}	454	}
414		455
415	--enc->indent; encode_indent (enc); encode_ch (enc, '}');	456	--enc->indent; encode_indent (enc); encode_ch (enc, '}');
		457	}
		458
		459	// encode objects, arrays and special \0=false and \1=true values.
		460	static void
		461	encode_rv (enc_t enc, SV sv)
		462	{
		463	svtype svt;
		464
		465	SvGETMAGIC (sv);
		466	svt = SvTYPE (sv);
		467
		468	if (svt == SVt_PVHV)
		469	encode_hv (enc, (HV *)sv);
		470	else if (svt == SVt_PVAV)
		471	encode_av (enc, (AV *)sv);
		472	else if (svt < SVt_PVAV)
		473	{
		474	if (SvNIOK (sv) && SvIV (sv) == 0)
		475	encode_str (enc, "false", 5, 0);
		476	else if (SvNIOK (sv) && SvIV (sv) == 1)
		477	encode_str (enc, "true", 4, 0);
		478	else
		479	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
		480	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
		481	}
		482	else
		483	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
		484	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416	}	485	}
417		486
418	static void	487	static void
419	encode_sv (enc_t enc, SV sv)	488	encode_sv (enc_t enc, SV sv)
420	{	489	{
…		…
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	need (enc, sizeof (UV) * 3);
440	enc->cur +=	512	enc->cur +=
441	SvIsUV(sv)	513	SvIsUV(sv)
442	? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))	514	? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
443	: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));	515	: snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
		516	}
		517	else
		518	{
		519	// optimise the "small number case"
		520	// code will likely be branchless and use only a single multiplication
		521	I32 i = SvIV (sv);
		522	U32 u;
		523
		524	need (enc, 6);
		525
		526	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
		527	u = i < 0 ? -i : i;
		528
		529	// convert to 4.28 fixed-point representation
		530	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
		531
		532	char digit, nz = 0;
		533
		534	digit = u >> 28; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0xfffffff) 5;
		535	digit = u >> 27; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x7ffffff) 5;
		536	digit = u >> 26; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x3ffffff) 5;
		537	digit = u >> 25; enc->cur = digit + '0'; enc->cur += (nz = nz \|\| digit); u = (u & 0x1ffffff) 5;
		538	digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1;
		539	}
444	}	540	}
445	else if (SvROK (sv))	541	else if (SvROK (sv))
446	{	542	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))	543	else if (!SvOK (sv))
463	encode_str (enc, "null", 4, 0);	544	encode_str (enc, "null", 4, 0);
464	else	545	else
465	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	546	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466	SvPV_nolen (sv), SvFLAGS (sv));	547	SvPV_nolen (sv), SvFLAGS (sv));
467	}	548	}
468		549
469	static SV *	550	static SV *
470	encode_json (SV *scalar, UV flags)	551	encode_json (SV *scalar, U32 flags)
471	{	552	{
		553	enc_t enc;
		554
472	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))	555	if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
473	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");	556	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474		557
475	enc_t enc;
476	enc.flags = flags;	558	enc.flags = flags;
477	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	559	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478	enc.cur = SvPVX (enc.sv);	560	enc.cur = SvPVX (enc.sv);
479	enc.end = SvEND (enc.sv);	561	enc.end = SvEND (enc.sv);
480	enc.indent = 0;	562	enc.indent = 0;
481	enc.max_depth = 0x7fffffffUL;	563	enc.maxdepth = DEC_DEPTH (flags);
482		564
483	SvPOK_only (enc.sv);	565	SvPOK_only (enc.sv);
484	encode_sv (&enc, scalar);	566	encode_sv (&enc, scalar);
485		567
		568	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
		569	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
		570
486	if (!(flags & (F_ASCII \| F_UTF8)))	571	if (!(flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
487	SvUTF8_on (enc.sv);	572	SvUTF8_on (enc.sv);
488
489	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
490		573
491	if (enc.flags & F_SHRINK)	574	if (enc.flags & F_SHRINK)
492	shrink (enc.sv);	575	shrink (enc.sv);
493		576
494	return enc.sv;	577	return enc.sv;
…		…
501	typedef struct	584	typedef struct
502	{	585	{
503	char *cur; // current parser pointer	586	char *cur; // current parser pointer
504	char *end; // end of input string	587	char *end; // end of input string
505	const char *err; // parse error, if != 0	588	const char *err; // parse error, if != 0
506	UV flags; // F_*	589	U32 flags; // F_*
		590	U32 depth; // recursion depth
		591	U32 maxdepth; // recursion depth limit
507	} dec_t;	592	} dec_t;
508		593
509	static void	594	inline void
510	decode_ws (dec_t *dec)	595	decode_ws (dec_t *dec)
511	{	596	{
512	for (;;)	597	for (;;)
513	{	598	{
514	char ch = *dec->cur;	599	char ch = *dec->cur;
…		…
520	++dec->cur;	605	++dec->cur;
521	}	606	}
522	}	607	}
523		608
524	#define ERR(reason) SB dec->err = reason; goto fail; SE	609	#define ERR(reason) SB dec->err = reason; goto fail; SE
		610
525	#define EXPECT_CH(ch) SB \	611	#define EXPECT_CH(ch) SB \
526	if (*dec->cur != ch) \	612	if (*dec->cur != ch) \
527	ERR (# ch " expected"); \	613	ERR (# ch " expected"); \
528	++dec->cur; \	614	++dec->cur; \
529	SE	615	SE
530		616
		617	#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
		618	#define DEC_DEC_DEPTH --dec->depth
		619
531	static SV decode_sv (dec_t dec);	620	static SV decode_sv (dec_t dec);
532		621
533	static signed char decode_hexdigit[256];	622	static signed char decode_hexdigit[256];
534		623
535	static UV	624	static UV
536	decode_4hex (dec_t *dec)	625	decode_4hex (dec_t *dec)
537	{	626	{
538	signed char d1, d2, d3, d4;	627	signed char d1, d2, d3, d4;
539	unsigned char cur = (unsigned char )dec->cur;	628	unsigned char cur = (unsigned char )dec->cur;
540		629
541	d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");	630	d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected");
542	d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");	631	d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected");
543	d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");	632	d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected");
544	d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");	633	d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected");
545		634
546	dec->cur += 4;	635	dec->cur += 4;
547		636
548	return ((UV)d1) << 12	637	return ((UV)d1) << 12
549	\| ((UV)d2) << 8	638	\| ((UV)d2) << 8
…		…
557	static SV *	646	static SV *
558	decode_str (dec_t *dec)	647	decode_str (dec_t *dec)
559	{	648	{
560	SV *sv = 0;	649	SV *sv = 0;
561	int utf8 = 0;	650	int utf8 = 0;
		651	char *dec_cur = dec->cur;
562		652
563	do	653	do
564	{	654	{
565	char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];	655	char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
566	char *cur = buf;	656	char *cur = buf;
567		657
568	do	658	do
569	{	659	{
570	unsigned char ch = (unsigned char )dec->cur++;	660	unsigned char ch = (unsigned char )dec_cur++;
571		661
572	if (ch == '"')	662	if (expect_false (ch == '"'))
573	{	663	{
574	--dec->cur;	664	--dec_cur;
575	break;	665	break;
576	}	666	}
577	else if (ch == '\\')	667	else if (expect_false (ch == '\\'))
578	{	668	{
579	switch (*dec->cur)	669	switch (*dec_cur)
580	{	670	{
581	case '\\':	671	case '\\':
582	case '/':	672	case '/':
583	case '"': cur++ = dec->cur++; break;	673	case '"': cur++ = dec_cur++; break;
584		674
585	case 'b': ++dec->cur; *cur++ = '\010'; break;	675	case 'b': ++dec_cur; *cur++ = '\010'; break;
586	case 't': ++dec->cur; *cur++ = '\011'; break;	676	case 't': ++dec_cur; *cur++ = '\011'; break;
587	case 'n': ++dec->cur; *cur++ = '\012'; break;	677	case 'n': ++dec_cur; *cur++ = '\012'; break;
588	case 'f': ++dec->cur; *cur++ = '\014'; break;	678	case 'f': ++dec_cur; *cur++ = '\014'; break;
589	case 'r': ++dec->cur; *cur++ = '\015'; break;	679	case 'r': ++dec_cur; *cur++ = '\015'; break;
590		680
591	case 'u':	681	case 'u':
592	{	682	{
593	UV lo, hi;	683	UV lo, hi;
594	++dec->cur;	684	++dec_cur;
595		685
		686	dec->cur = dec_cur;
596	hi = decode_4hex (dec);	687	hi = decode_4hex (dec);
		688	dec_cur = dec->cur;
597	if (hi == (UV)-1)	689	if (hi == (UV)-1)
598	goto fail;	690	goto fail;
599		691
600	// possibly a surrogate pair	692	// possibly a surrogate pair
601	if (hi >= 0xd800)	693	if (hi >= 0xd800)
602	if (hi < 0xdc00)	694	if (hi < 0xdc00)
603	{	695	{
604	if (dec->cur [0] != '\\' \|\| dec->cur [1] != 'u')	696	if (dec_cur [0] != '\\' \|\| dec_cur [1] != 'u')
605	ERR ("missing low surrogate character in surrogate pair");	697	ERR ("missing low surrogate character in surrogate pair");
606		698
607	dec->cur += 2;	699	dec_cur += 2;
608		700
		701	dec->cur = dec_cur;
609	lo = decode_4hex (dec);	702	lo = decode_4hex (dec);
		703	dec_cur = dec->cur;
610	if (lo == (UV)-1)	704	if (lo == (UV)-1)
611	goto fail;	705	goto fail;
612		706
613	if (lo < 0xdc00 \|\| lo >= 0xe000)	707	if (lo < 0xdc00 \|\| lo >= 0xe000)
614	ERR ("surrogate pair expected");	708	ERR ("surrogate pair expected");
…		…
628	*cur++ = hi;	722	*cur++ = hi;
629	}	723	}
630	break;	724	break;
631		725
632	default:	726	default:
633	--dec->cur;	727	--dec_cur;
634	ERR ("illegal backslash escape sequence in string");	728	ERR ("illegal backslash escape sequence in string");
635	}	729	}
636	}	730	}
637	else if (ch >= 0x20 && ch <= 0x7f)	731	else if (expect_true (ch >= 0x20 && ch <= 0x7f))
638	*cur++ = ch;	732	*cur++ = ch;
639	else if (ch >= 0x80)	733	else if (ch >= 0x80)
640	{	734	{
641	--dec->cur;
642
643	STRLEN clen;	735	STRLEN clen;
		736	UV uch;
		737
		738	--dec_cur;
		739
644	UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);	740	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
645	if (clen == (STRLEN)-1)	741	if (clen == (STRLEN)-1)
646	ERR ("malformed UTF-8 character in JSON string");	742	ERR ("malformed UTF-8 character in JSON string");
647		743
648	do	744	do
649	{
650	cur++ = dec->cur++;	745	cur++ = dec_cur++;
651	}
652	while (--clen);	746	while (--clen);
653		747
654	utf8 = 1;	748	utf8 = 1;
655	}	749	}
656	else if (!ch)
657	ERR ("unexpected end of string while parsing json string");
658	else	750	else
		751	{
		752	--dec_cur;
		753
		754	if (!ch)
		755	ERR ("unexpected end of string while parsing JSON string");
		756	else
659	ERR ("invalid character encountered");	757	ERR ("invalid character encountered while parsing JSON string");
660		758	}
661	}	759	}
662	while (cur < buf + SHORT_STRING_LEN);	760	while (cur < buf + SHORT_STRING_LEN);
663		761
		762	{
664	STRLEN len = cur - buf;	763	STRLEN len = cur - buf;
665		764
666	if (sv)	765	if (sv)
667	{	766	{
668	SvGROW (sv, SvCUR (sv) + len + 1);	767	SvGROW (sv, SvCUR (sv) + len + 1);
669	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	768	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670	SvCUR_set (sv, SvCUR (sv) + len);	769	SvCUR_set (sv, SvCUR (sv) + len);
671	}	770	}
672	else	771	else
673	sv = newSVpvn (buf, len);	772	sv = newSVpvn (buf, len);
674	}	773	}
		774	}
675	while (*dec->cur != '"');	775	while (*dec_cur != '"');
676		776
677	++dec->cur;	777	++dec_cur;
678		778
679	if (sv)	779	if (sv)
680	{	780	{
681	SvPOK_only (sv);	781	SvPOK_only (sv);
682	*SvEND (sv) = 0;	782	*SvEND (sv) = 0;
…		…
685	SvUTF8_on (sv);	785	SvUTF8_on (sv);
686	}	786	}
687	else	787	else
688	sv = newSVpvn ("", 0);	788	sv = newSVpvn ("", 0);
689		789
		790	dec->cur = dec_cur;
690	return sv;	791	return sv;
691		792
692	fail:	793	fail:
		794	dec->cur = dec_cur;
693	return 0;	795	return 0;
694	}	796	}
695		797
696	static SV *	798	static SV *
697	decode_num (dec_t *dec)	799	decode_num (dec_t *dec)
…		…
755	is_nv = 1;	857	is_nv = 1;
756	}	858	}
757		859
758	if (!is_nv)	860	if (!is_nv)
759	{	861	{
760	UV uv;	862	// 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);	863	if (*start == '-')
762	if (numtype & IS_NUMBER_IN_UV)	864	switch (dec->cur - start)
763	if (numtype & IS_NUMBER_NEG)
764	{	865	{
765	if (uv < (UV)IV_MIN)	866	case 2: return newSViv (-( start [1] - '0' ));
766	return newSViv (-(IV)uv);	867	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
		868	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
		869	case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
767	}	870	}
		871	else
		872	switch (dec->cur - start)
		873	{
		874	case 1: return newSViv ( start [0] - '0' );
		875	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
		876	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
		877	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
		878	}
		879
		880	{
		881	UV uv;
		882	int numtype = grok_number (start, dec->cur - start, &uv);
		883	if (numtype & IS_NUMBER_IN_UV)
		884	if (numtype & IS_NUMBER_NEG)
		885	{
		886	if (uv < (UV)IV_MIN)
		887	return newSViv (-(IV)uv);
		888	}
768	else	889	else
769	return newSVuv (uv);	890	return newSVuv (uv);
		891	}
770	}	892	}
771		893
772	return newSVnv (Atof (start));	894	return newSVnv (Atof (start));
773		895
774	fail:	896	fail:
…		…
778	static SV *	900	static SV *
779	decode_av (dec_t *dec)	901	decode_av (dec_t *dec)
780	{	902	{
781	AV *av = newAV ();	903	AV *av = newAV ();
782		904
		905	DEC_INC_DEPTH;
783	decode_ws (dec);	906	decode_ws (dec);
		907
784	if (*dec->cur == ']')	908	if (*dec->cur == ']')
785	++dec->cur;	909	++dec->cur;
786	else	910	else
787	for (;;)	911	for (;;)
788	{	912	{
…		…
806	ERR (", or ] expected while parsing array");	930	ERR (", or ] expected while parsing array");
807		931
808	++dec->cur;	932	++dec->cur;
809	}	933	}
810		934
		935	DEC_DEC_DEPTH;
811	return newRV_noinc ((SV *)av);	936	return newRV_noinc ((SV *)av);
812		937
813	fail:	938	fail:
814	SvREFCNT_dec (av);	939	SvREFCNT_dec (av);
		940	DEC_DEC_DEPTH;
815	return 0;	941	return 0;
816	}	942	}
817		943
818	static SV *	944	static SV *
819	decode_hv (dec_t *dec)	945	decode_hv (dec_t *dec)
820	{	946	{
821	HV *hv = newHV ();	947	HV *hv = newHV ();
822		948
		949	DEC_INC_DEPTH;
823	decode_ws (dec);	950	decode_ws (dec);
		951
824	if (*dec->cur == '}')	952	if (*dec->cur == '}')
825	++dec->cur;	953	++dec->cur;
826	else	954	else
827	for (;;)	955	for (;;)
828	{	956	{
…		…
841	{	969	{
842	SvREFCNT_dec (key);	970	SvREFCNT_dec (key);
843	goto fail;	971	goto fail;
844	}	972	}
845		973
846	//TODO: optimise
847	hv_store_ent (hv, key, value, 0);	974	hv_store_ent (hv, key, value, 0);
		975	SvREFCNT_dec (key);
848		976
849	decode_ws (dec);	977	decode_ws (dec);
850		978
851	if (*dec->cur == '}')	979	if (*dec->cur == '}')
852	{	980	{
…		…
858	ERR (", or } expected while parsing object/hash");	986	ERR (", or } expected while parsing object/hash");
859		987
860	++dec->cur;	988	++dec->cur;
861	}	989	}
862		990
		991	DEC_DEC_DEPTH;
863	return newRV_noinc ((SV *)hv);	992	return newRV_noinc ((SV *)hv);
864		993
865	fail:	994	fail:
866	SvREFCNT_dec (hv);	995	SvREFCNT_dec (hv);
		996	DEC_DEC_DEPTH;
867	return 0;	997	return 0;
868	}	998	}
869		999
870	static SV *	1000	static SV *
871	decode_sv (dec_t *dec)	1001	decode_sv (dec_t *dec)
…		…
914	ERR ("'null' expected");	1044	ERR ("'null' expected");
915		1045
916	break;	1046	break;
917		1047
918	default:	1048	default:
919	ERR ("malformed json string, neither array, object, number, string or atom");	1049	ERR ("malformed JSON string, neither array, object, number, string or atom");
920	break;	1050	break;
921	}	1051	}
922		1052
923	fail:	1053	fail:
924	return 0;	1054	return 0;
925	}	1055	}
926		1056
927	static SV *	1057	static SV *
928	decode_json (SV *string, UV flags)	1058	decode_json (SV string, U32 flags, UV offset_return)
929	{	1059	{
		1060	dec_t dec;
		1061	UV offset;
930	SV *sv;	1062	SV *sv;
		1063
		1064	SvGETMAGIC (string);
		1065	SvUPGRADE (string, SVt_PV);
931		1066
932	if (flags & F_UTF8)	1067	if (flags & F_UTF8)
933	sv_utf8_downgrade (string, 0);	1068	sv_utf8_downgrade (string, 0);
934	else	1069	else
935	sv_utf8_upgrade (string);	1070	sv_utf8_upgrade (string);
936		1071
937	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1072	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938		1073
939	dec_t dec;
940	dec.flags = flags;	1074	dec.flags = flags;
941	dec.cur = SvPVX (string);	1075	dec.cur = SvPVX (string);
942	dec.end = SvEND (string);	1076	dec.end = SvEND (string);
943	dec.err = 0;	1077	dec.err = 0;
		1078	dec.depth = 0;
		1079	dec.maxdepth = DEC_DEPTH (dec.flags);
944		1080
		1081	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
945	sv = decode_sv (&dec);	1082	sv = decode_sv (&dec);
946		1083
		1084	if (!(offset_return \|\| !sv))
		1085	{
		1086	// check for trailing garbage
		1087	decode_ws (&dec);
		1088
		1089	if (*dec.cur)
		1090	{
		1091	dec.err = "garbage after JSON object";
		1092	SvREFCNT_dec (sv);
		1093	sv = 0;
		1094	}
		1095	}
		1096
		1097	if (offset_return \|\| !sv)
		1098	{
		1099	offset = dec.flags & F_UTF8
		1100	? dec.cur - SvPVX (string)
		1101	: utf8_distance (dec.cur, SvPVX (string));
		1102
		1103	if (offset_return)
		1104	*offset_return = offset;
		1105	}
		1106
947	if (!sv)	1107	if (!sv)
948	{	1108	{
949	IV offset = dec.flags & F_UTF8
950	? dec.cur - SvPVX (string)
951	: utf8_distance (dec.cur, SvPVX (string));
952	SV *uni = sv_newmortal ();	1109	SV *uni = sv_newmortal ();
953		1110
954	// horrible hack to silence warning inside pv_uni_display	1111	// horrible hack to silence warning inside pv_uni_display
955	COP cop = *PL_curcop;	1112	COP cop = *PL_curcop;
956	cop.cop_warnings = pWARN_NONE;	1113	cop.cop_warnings = pWARN_NONE;
…		…
958	SAVEVPTR (PL_curcop);	1115	SAVEVPTR (PL_curcop);
959	PL_curcop = &cop;	1116	PL_curcop = &cop;
960	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1117	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961	LEAVE;	1118	LEAVE;
962		1119
963	croak ("%s, at character offset %d (%s)",	1120	croak ("%s, at character offset %d [\"%s\"]",
964	dec.err,	1121	dec.err,
965	(int)offset,	1122	(int)offset,
966	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1123	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967	}	1124	}
968		1125
…		…
982	BOOT:	1139	BOOT:
983	{	1140	{
984	int i;	1141	int i;
985		1142
986	memset (decode_hexdigit, 0xff, 256);	1143	memset (decode_hexdigit, 0xff, 256);
		1144
987	for (i = 10; i--; )	1145	for (i = 0; i < 256; ++i)
988	decode_hexdigit ['0' + i] = i;	1146	decode_hexdigit [i] =
989		1147	i >= '0' && i <= '9' ? i - '0'
990	for (i = 7; i--; )	1148	: i >= 'a' && i <= 'f' ? i - 'a' + 10
991	{	1149	: i >= 'A' && i <= 'F' ? i - 'A' + 10
992	decode_hexdigit ['a' + i] = 10 + i;	1150	: -1;
993	decode_hexdigit ['A' + i] = 10 + i;
994	}
995		1151
996	json_stash = gv_stashpv ("JSON::XS", 1);	1152	json_stash = gv_stashpv ("JSON::XS", 1);
997	}	1153	}
998		1154
999	PROTOTYPES: DISABLE	1155	PROTOTYPES: DISABLE
…		…
1005	RETVAL	1161	RETVAL
1006		1162
1007	SV ascii (SV self, int enable = 1)	1163	SV ascii (SV self, int enable = 1)
1008	ALIAS:	1164	ALIAS:
1009	ascii = F_ASCII	1165	ascii = F_ASCII
		1166	latin1 = F_LATIN1
1010	utf8 = F_UTF8	1167	utf8 = F_UTF8
1011	indent = F_INDENT	1168	indent = F_INDENT
1012	canonical = F_CANONICAL	1169	canonical = F_CANONICAL
1013	space_before = F_SPACE_BEFORE	1170	space_before = F_SPACE_BEFORE
1014	space_after = F_SPACE_AFTER	1171	space_after = F_SPACE_AFTER
…		…
1026	RETVAL = newSVsv (self);	1183	RETVAL = newSVsv (self);
1027	}	1184	}
1028	OUTPUT:	1185	OUTPUT:
1029	RETVAL	1186	RETVAL
1030		1187
		1188	SV max_depth (SV self, UV max_depth = 0x80000000UL)
		1189	CODE:
		1190	{
		1191	UV *uv = SvJSON (self);
		1192	UV log2 = 0;
		1193
		1194	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
		1195
		1196	while ((1UL << log2) < max_depth)
		1197	++log2;
		1198
		1199	uv = uv & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
		1200
		1201	RETVAL = newSVsv (self);
		1202	}
		1203	OUTPUT:
		1204	RETVAL
		1205
1031	void encode (SV self, SV scalar)	1206	void encode (SV self, SV scalar)
1032	PPCODE:	1207	PPCODE:
1033	XPUSHs (encode_json (scalar, *SvJSON (self)));	1208	XPUSHs (encode_json (scalar, *SvJSON (self)));
1034		1209
1035	void decode (SV self, SV jsonstr)	1210	void decode (SV self, SV jsonstr)
1036	PPCODE:	1211	PPCODE:
1037	XPUSHs (decode_json (jsonstr, *SvJSON (self)));	1212	XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
		1213
		1214	void decode_prefix (SV self, SV jsonstr)
		1215	PPCODE:
		1216	{
		1217	UV offset;
		1218	EXTEND (SP, 2);
		1219	PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
		1220	PUSHs (sv_2mortal (newSVuv (offset)));
		1221	}
1038		1222
1039	PROTOTYPES: ENABLE	1223	PROTOTYPES: ENABLE
1040		1224
1041	void to_json (SV *scalar)	1225	void to_json (SV *scalar)
		1226	ALIAS:
		1227	objToJson = 0
1042	PPCODE:	1228	PPCODE:
1043	XPUSHs (encode_json (scalar, F_UTF8));	1229	XPUSHs (encode_json (scalar, F_DEFAULT \| F_UTF8));
1044		1230
1045	void from_json (SV *jsonstr)	1231	void from_json (SV *jsonstr)
		1232	ALIAS:
		1233	jsonToObj = 0
1046	PPCODE:	1234	PPCODE:
1047	XPUSHs (decode_json (jsonstr, F_UTF8));	1235	XPUSHs (decode_json (jsonstr, F_DEFAULT \| F_UTF8, 0));
1048		1236

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.13 by root, Sat Mar 24 22:55:16 2007 UTC vs. Revision 1.38 by root, Mon Jun 11 03:18:07 2007 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.13 by root, Sat Mar 24 22:55:16 2007 UTC vs.
Revision 1.38 by root, Mon Jun 11 03:18:07 2007 UTC