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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.75 by root, Wed Mar 19 15:37:54 2008 UTC vs.
Revision 1.103 by root, Wed Jan 6 08:01:39 2010 UTC

…		…
12	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	12	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
13	# define snprintf _snprintf // C compilers have this in stdio.h	13	# define snprintf _snprintf // C compilers have this in stdio.h
14	#endif	14	#endif
15		15
16	// some old perls do not have this, try to make it work, no	16	// some old perls do not have this, try to make it work, no
17	// guarentees, though. if it breaks, you get to keep the pieces.	17	// guarantees, though. if it breaks, you get to keep the pieces.
18	#ifndef UTF8_MAXBYTES	18	#ifndef UTF8_MAXBYTES
19	# define UTF8_MAXBYTES 13	19	# define UTF8_MAXBYTES 13
20	#endif	20	#endif
21		21
		22	// three extra for rounding, sign, and end of string
22	#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)	23	#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
23		24
24	#define F_ASCII 0x00000001UL	25	#define F_ASCII 0x00000001UL
25	#define F_LATIN1 0x00000002UL	26	#define F_LATIN1 0x00000002UL
26	#define F_UTF8 0x00000004UL	27	#define F_UTF8 0x00000004UL
27	#define F_INDENT 0x00000008UL	28	#define F_INDENT 0x00000008UL
…		…
31	#define F_ALLOW_NONREF 0x00000100UL	32	#define F_ALLOW_NONREF 0x00000100UL
32	#define F_SHRINK 0x00000200UL	33	#define F_SHRINK 0x00000200UL
33	#define F_ALLOW_BLESSED 0x00000400UL	34	#define F_ALLOW_BLESSED 0x00000400UL
34	#define F_CONV_BLESSED 0x00000800UL	35	#define F_CONV_BLESSED 0x00000800UL
35	#define F_RELAXED 0x00001000UL	36	#define F_RELAXED 0x00001000UL
36		37	#define F_ALLOW_UNKNOWN 0x00002000UL
37	#define F_MAXDEPTH 0xf8000000UL
38	#define S_MAXDEPTH 27
39	#define F_MAXSIZE 0x01f00000UL
40	#define S_MAXSIZE 20
41	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing	38	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
42		39
43	#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
44	#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
45
46	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	40	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
47	#define F_DEFAULT (9UL << S_MAXDEPTH)
48		41
49	#define INIT_SIZE 32 // initial scalar size to be allocated	42	#define INIT_SIZE 32 // initial scalar size to be allocated
50	#define INDENT_STEP 3 // spaces per indentation level	43	#define INDENT_STEP 3 // spaces per indentation level
51		44
52	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size	45	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
…		…
67		60
68	#define IN_RANGE_INC(type,val,beg,end) \	61	#define IN_RANGE_INC(type,val,beg,end) \
69	((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \	62	((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
70	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))	63	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
71		64
		65	#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
		66
72	#ifdef USE_ITHREADS	67	#ifdef USE_ITHREADS
73	# define JSON_SLOW 1	68	# define JSON_SLOW 1
74	# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))	69	# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
75	#else	70	#else
76	# define JSON_SLOW 0	71	# define JSON_SLOW 0
…		…
78	#endif	73	#endif
79		74
80	static HV json_stash, json_boolean_stash; // JSON::XS::	75	static HV json_stash, json_boolean_stash; // JSON::XS::
81	static SV json_true, json_false;	76	static SV json_true, json_false;
82		77
		78	enum {
		79	INCR_M_WS = 0, // initial whitespace skipping, must be 0
		80	INCR_M_STR, // inside string
		81	INCR_M_BS, // inside backslash
		82	INCR_M_C0, // inside comment in initial whitespace sequence
		83	INCR_M_C1, // inside comment in other places
		84	INCR_M_JSON // outside anything, count nesting
		85	};
		86
		87	#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
		88
83	typedef struct {	89	typedef struct {
84	U32 flags;	90	U32 flags;
		91	U32 max_depth;
		92	STRLEN max_size;
		93
85	SV *cb_object;	94	SV *cb_object;
86	HV *cb_sk_object;	95	HV *cb_sk_object;
		96
		97	// for the incremental parser
		98	SV *incr_text; // the source text so far
		99	STRLEN incr_pos; // the current offset into the text
		100	int incr_nest; // {[]}-nesting level
		101	unsigned char incr_mode;
87	} JSON;	102	} JSON;
		103
		104	INLINE void
		105	json_init (JSON *json)
		106	{
		107	Zero (json, 1, JSON);
		108	json->max_depth = 512;
		109	}
88		110
89	/////////////////////////////////////////////////////////////////////////////	111	/////////////////////////////////////////////////////////////////////////////
90	// utility functions	112	// utility functions
91		113
		114	INLINE SV *
		115	get_bool (const char *name)
		116	{
		117	SV *sv = get_sv (name, 1);
		118
		119	SvREADONLY_on (sv);
		120	SvREADONLY_on (SvRV (sv));
		121
		122	return sv;
		123	}
		124
92	INLINE void	125	INLINE void
93	shrink (SV *sv)	126	shrink (SV *sv)
94	{	127	{
95	sv_utf8_downgrade (sv, 1);	128	sv_utf8_downgrade (sv, 1);
		129
96	if (SvLEN (sv) > SvCUR (sv) + 1)	130	if (SvLEN (sv) > SvCUR (sv) + 1)
97	{	131	{
98	#ifdef SvPV_shrink_to_cur	132	#ifdef SvPV_shrink_to_cur
99	SvPV_shrink_to_cur (sv);	133	SvPV_shrink_to_cur (sv);
100	#elif defined (SvPV_renew)	134	#elif defined (SvPV_renew)
…		…
144	*s++ = 0x80 \| ( ch & 0x3f);	178	*s++ = 0x80 \| ( ch & 0x3f);
145		179
146	return s;	180	return s;
147	}	181	}
148		182
		183	// convert offset pointer to character index, sv must be string
		184	static STRLEN
		185	ptr_to_index (SV sv, char offset)
		186	{
		187	return SvUTF8 (sv)
		188	? utf8_distance (offset, SvPVX (sv))
		189	: offset - SvPVX (sv);
		190	}
		191
149	/////////////////////////////////////////////////////////////////////////////	192	/////////////////////////////////////////////////////////////////////////////
150	// encoder	193	// encoder
151		194
152	// structure used for encoding JSON	195	// structure used for encoding JSON
153	typedef struct	196	typedef struct
…		…
155	char *cur; // SvPVX (sv) + current output position	198	char *cur; // SvPVX (sv) + current output position
156	char *end; // SvEND (sv)	199	char *end; // SvEND (sv)
157	SV *sv; // result scalar	200	SV *sv; // result scalar
158	JSON json;	201	JSON json;
159	U32 indent; // indentation level	202	U32 indent; // indentation level
160	U32 maxdepth; // max. indentation/recursion level
161	UV limit; // escape character values >= this value when encoding	203	UV limit; // escape character values >= this value when encoding
162	} enc_t;	204	} enc_t;
163		205
164	INLINE void	206	INLINE void
165	need (enc_t *enc, STRLEN len)	207	need (enc_t *enc, STRLEN len)
166	{	208	{
167	if (expect_false (enc->cur + len >= enc->end))	209	if (expect_false (enc->cur + len >= enc->end))
168	{	210	{
169	STRLEN cur = enc->cur - SvPVX (enc->sv);	211	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
170	SvGROW (enc->sv, cur + len + 1);	212	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
171	enc->cur = SvPVX (enc->sv) + cur;	213	enc->cur = SvPVX (enc->sv) + cur;
172	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	214	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
173	}	215	}
174	}	216	}
175		217
…		…
250	(int)((uch - 0x10000) % 0x400 + 0xDC00));	292	(int)((uch - 0x10000) % 0x400 + 0xDC00));
251	enc->cur += 12;	293	enc->cur += 12;
252	}	294	}
253	else	295	else
254	{	296	{
255	static char hexdigit [16] = "0123456789abcdef";
256	need (enc, len += 5);	297	need (enc, len += 5);
257	*enc->cur++ = '\\';	298	*enc->cur++ = '\\';
258	*enc->cur++ = 'u';	299	*enc->cur++ = 'u';
259	*enc->cur++ = hexdigit [ uch >> 12 ];	300	*enc->cur++ = PL_hexdigit [ uch >> 12 ];
260	*enc->cur++ = hexdigit [(uch >> 8) & 15];	301	*enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
261	*enc->cur++ = hexdigit [(uch >> 4) & 15];	302	*enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
262	*enc->cur++ = hexdigit [(uch >> 0) & 15];	303	*enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
263	}	304	}
264		305
265	str += clen;	306	str += clen;
266	}	307	}
267	else if (enc->json.flags & F_LATIN1)	308	else if (enc->json.flags & F_LATIN1)
…		…
338	static void	379	static void
339	encode_av (enc_t enc, AV av)	380	encode_av (enc_t enc, AV av)
340	{	381	{
341	int i, len = av_len (av);	382	int i, len = av_len (av);
342		383
343	if (enc->indent >= enc->maxdepth)	384	if (enc->indent >= enc->json.max_depth)
344	croak ("data structure too deep (hit recursion limit)");	385	croak (ERR_NESTING_EXCEEDED);
345		386
346	encode_ch (enc, '[');	387	encode_ch (enc, '[');
347		388
348	if (len >= 0)	389	if (len >= 0)
349	{	390	{
…		…
423		464
424	static void	465	static void
425	encode_hv (enc_t enc, HV hv)	466	encode_hv (enc_t enc, HV hv)
426	{	467	{
427	HE *he;	468	HE *he;
428	int count;
429		469
430	if (enc->indent >= enc->maxdepth)	470	if (enc->indent >= enc->json.max_depth)
431	croak ("data structure too deep (hit recursion limit)");	471	croak (ERR_NESTING_EXCEEDED);
432		472
433	encode_ch (enc, '{');	473	encode_ch (enc, '{');
434		474
435	// for canonical output we have to sort by keys first	475	// for canonical output we have to sort by keys first
436	// actually, this is mostly due to the stupid so-called	476	// actually, this is mostly due to the stupid so-called
437	// security workaround added somewhere in 5.8.x.	477	// security workaround added somewhere in 5.8.x
438	// that randomises hash orderings	478	// that randomises hash orderings
439	if (enc->json.flags & F_CANONICAL)	479	if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
440	{	480	{
441	int count = hv_iterinit (hv);	481	int count = hv_iterinit (hv);
442		482
443	if (SvMAGICAL (hv))	483	if (SvMAGICAL (hv))
444	{	484	{
…		…
617		657
618	if (len == 1 && *pv == '1')	658	if (len == 1 && *pv == '1')
619	encode_str (enc, "true", 4, 0);	659	encode_str (enc, "true", 4, 0);
620	else if (len == 1 && *pv == '0')	660	else if (len == 1 && *pv == '0')
621	encode_str (enc, "false", 5, 0);	661	encode_str (enc, "false", 5, 0);
		662	else if (enc->json.flags & F_ALLOW_UNKNOWN)
		663	encode_str (enc, "null", 4, 0);
622	else	664	else
623	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",	665	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
624	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	666	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
625	}	667	}
		668	else if (enc->json.flags & F_ALLOW_UNKNOWN)
		669	encode_str (enc, "null", 4, 0);
626	else	670	else
627	croak ("encountered %s, but JSON can only represent references to arrays or hashes",	671	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
628	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	672	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
629	}	673	}
630		674
…		…
692	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));	736	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
693	}	737	}
694	}	738	}
695	else if (SvROK (sv))	739	else if (SvROK (sv))
696	encode_rv (enc, SvRV (sv));	740	encode_rv (enc, SvRV (sv));
697	else if (!SvOK (sv))	741	else if (!SvOK (sv) \|\| enc->json.flags & F_ALLOW_UNKNOWN)
698	encode_str (enc, "null", 4, 0);	742	encode_str (enc, "null", 4, 0);
699	else	743	else
700	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	744	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
701	SvPV_nolen (sv), SvFLAGS (sv));	745	SvPV_nolen (sv), SvFLAGS (sv));
702	}	746	}
…		…
712	enc.json = *json;	756	enc.json = *json;
713	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	757	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
714	enc.cur = SvPVX (enc.sv);	758	enc.cur = SvPVX (enc.sv);
715	enc.end = SvEND (enc.sv);	759	enc.end = SvEND (enc.sv);
716	enc.indent = 0;	760	enc.indent = 0;
717	enc.maxdepth = DEC_DEPTH (enc.json.flags);
718	enc.limit = enc.json.flags & F_ASCII ? 0x000080UL	761	enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
719	: enc.json.flags & F_LATIN1 ? 0x000100UL	762	: enc.json.flags & F_LATIN1 ? 0x000100UL
720	: 0x110000UL;	763	: 0x110000UL;
721		764
722	SvPOK_only (enc.sv);	765	SvPOK_only (enc.sv);
723	encode_sv (&enc, scalar);	766	encode_sv (&enc, scalar);
		767	encode_nl (&enc);
724		768
725	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	769	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
726	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	770	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
727		771
728	if (!(enc.json.flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))	772	if (!(enc.json.flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
…		…
789	if (*dec->cur != ch) \	833	if (*dec->cur != ch) \
790	ERR (# ch " expected"); \	834	ERR (# ch " expected"); \
791	++dec->cur; \	835	++dec->cur; \
792	SE	836	SE
793		837
794	#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")	838	#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
795	#define DEC_DEC_DEPTH --dec->depth	839	#define DEC_DEC_DEPTH --dec->depth
796		840
797	static SV decode_sv (dec_t dec);	841	static SV decode_sv (dec_t dec);
798		842
799	static signed char decode_hexdigit[256];	843	static signed char decode_hexdigit[256];
…		…
908	else if (expect_true (ch >= 0x20 && ch < 0x80))	952	else if (expect_true (ch >= 0x20 && ch < 0x80))
909	*cur++ = ch;	953	*cur++ = ch;
910	else if (ch >= 0x80)	954	else if (ch >= 0x80)
911	{	955	{
912	STRLEN clen;	956	STRLEN clen;
913	UV uch;
914		957
915	--dec_cur;	958	--dec_cur;
916		959
917	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);	960	decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
918	if (clen == (STRLEN)-1)	961	if (clen == (STRLEN)-1)
919	ERR ("malformed UTF-8 character in JSON string");	962	ERR ("malformed UTF-8 character in JSON string");
920		963
921	do	964	do
922	cur++ = dec_cur++;	965	cur++ = dec_cur++;
…		…
939	{	982	{
940	STRLEN len = cur - buf;	983	STRLEN len = cur - buf;
941		984
942	if (sv)	985	if (sv)
943	{	986	{
944	SvGROW (sv, SvCUR (sv) + len + 1);	987	STRLEN cur = SvCUR (sv);
		988
		989	if (SvLEN (sv) <= cur + len)
		990	SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
		991
945	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	992	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
946	SvCUR_set (sv, SvCUR (sv) + len);	993	SvCUR_set (sv, SvCUR (sv) + len);
947	}	994	}
948	else	995	else
949	sv = newSVpvn (buf, len);	996	sv = newSVpvn (buf, len);
…		…
1040		1087
1041	// special case the rather common 1..5-digit-int case	1088	// special case the rather common 1..5-digit-int case
1042	if (*start == '-')	1089	if (*start == '-')
1043	switch (len)	1090	switch (len)
1044	{	1091	{
1045	case 2: return newSViv (-( start [1] - '0' * 1));	1092	case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1046	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));	1093	case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1047	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));	1094	case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1048	case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));	1095	case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1049	case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));	1096	case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1050	}	1097	}
1051	else	1098	else
1052	switch (len)	1099	switch (len)
1053	{	1100	{
1054	case 1: return newSViv ( start [0] - '0' * 1);	1101	case 1: return newSViv ( start [0] - '0' * 1);
1055	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);	1102	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1056	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);	1103	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1057	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);	1104	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1058	case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);	1105	case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1059	}	1106	}
1060		1107
1061	{	1108	{
1062	UV uv;	1109	UV uv;
1063	int numtype = grok_number (start, len, &uv);	1110	int numtype = grok_number (start, len, &uv);
…		…
1171	char *p = dec->cur;	1218	char *p = dec->cur;
1172	char *e = p + 24; // only try up to 24 bytes	1219	char *e = p + 24; // only try up to 24 bytes
1173		1220
1174	for (;;)	1221	for (;;)
1175	{	1222	{
1176	// the >= 0x80 is true on most architectures	1223	// the >= 0x80 is false on most architectures
1177	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')	1224	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
1178	{	1225	{
1179	// slow path, back up and use decode_str	1226	// slow path, back up and use decode_str
1180	SV *key = decode_str (dec);	1227	SV *key = decode_str (dec);
1181	if (!key)	1228	if (!key)
…		…
1329	case 't':	1376	case 't':
1330	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1377	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1331	{	1378	{
1332	dec->cur += 4;	1379	dec->cur += 4;
1333	#if JSON_SLOW	1380	#if JSON_SLOW
1334	json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);	1381	json_true = get_bool ("JSON::XS::true");
1335	#endif	1382	#endif
1336	return SvREFCNT_inc (json_true);	1383	return newSVsv (json_true);
1337	}	1384	}
1338	else	1385	else
1339	ERR ("'true' expected");	1386	ERR ("'true' expected");
1340		1387
1341	break;	1388	break;
…		…
1343	case 'f':	1390	case 'f':
1344	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1391	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1345	{	1392	{
1346	dec->cur += 5;	1393	dec->cur += 5;
1347	#if JSON_SLOW	1394	#if JSON_SLOW
1348	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);	1395	json_false = get_bool ("JSON::XS::false");
1349	#endif	1396	#endif
1350	return SvREFCNT_inc (json_false);	1397	return newSVsv (json_false);
1351	}	1398	}
1352	else	1399	else
1353	ERR ("'false' expected");	1400	ERR ("'false' expected");
1354		1401
1355	break;	1402	break;
…		…
1373	fail:	1420	fail:
1374	return 0;	1421	return 0;
1375	}	1422	}
1376		1423
1377	static SV *	1424	static SV *
1378	decode_json (SV string, JSON json, UV *offset_return)	1425	decode_json (SV string, JSON json, char **offset_return)
1379	{	1426	{
1380	dec_t dec;	1427	dec_t dec;
1381	UV offset;
1382	SV *sv;	1428	SV *sv;
1383		1429
		1430	/* work around bugs in 5.10 where manipulating magic values
		1431	* will perl ignore the magic in subsequent accesses
		1432	*/
1384	SvGETMAGIC (string);	1433	/SvGETMAGIC (string);/
		1434	if (SvMAGICAL (string))
		1435	string = sv_2mortal (newSVsv (string));
		1436
1385	SvUPGRADE (string, SVt_PV);	1437	SvUPGRADE (string, SVt_PV);
1386		1438
1387	if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))	1439	/* work around a bug in perl 5.10, which causes SvCUR to fail an
		1440	* assertion with -DDEBUGGING, although SvCUR is documented to
		1441	* return the xpv_cur field which certainly exists after upgrading.
		1442	* according to nicholas clark, calling SvPOK fixes this.
		1443	* But it doesn't fix it, so try another workaround, call SvPV_nolen
		1444	* and hope for the best.
		1445	* Damnit, SvPV_nolen still trips over yet another assertion. This
		1446	* assertion business is seriously broken, try yet another workaround
		1447	* for the broken -DDEBUGGING.
		1448	*/
		1449	{
		1450	#ifdef DEBUGGING
		1451	STRLEN offset = SvOK (string) ? sv_len (string) : 0;
		1452	#else
		1453	STRLEN offset = SvCUR (string);
		1454	#endif
		1455
		1456	if (offset > json->max_size && json->max_size)
1388	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",	1457	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1389	(unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));	1458	(unsigned long)SvCUR (string), (unsigned long)json->max_size);
		1459	}
1390		1460
1391	if (json->flags & F_UTF8)	1461	if (json->flags & F_UTF8)
1392	sv_utf8_downgrade (string, 0);	1462	sv_utf8_downgrade (string, 0);
1393	else	1463	else
1394	sv_utf8_upgrade (string);	1464	sv_utf8_upgrade (string);
1395		1465
1396	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1466	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1397		1467
1398	dec.json = *json;	1468	dec.json = *json;
1399	dec.cur = SvPVX (string);	1469	dec.cur = SvPVX (string);
1400	dec.end = SvEND (string);	1470	dec.end = SvEND (string);
1401	dec.err = 0;	1471	dec.err = 0;
1402	dec.depth = 0;	1472	dec.depth = 0;
1403	dec.maxdepth = DEC_DEPTH (dec.json.flags);
1404		1473
1405	if (dec.json.cb_object \|\| dec.json.cb_sk_object)	1474	if (dec.json.cb_object \|\| dec.json.cb_sk_object)
1406	dec.json.flags \|= F_HOOK;	1475	dec.json.flags \|= F_HOOK;
1407		1476
1408	*dec.end = 0; // this should basically be a nop, too, but make sure it's there	1477	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
1409		1478
1410	decode_ws (&dec);	1479	decode_ws (&dec);
1411	sv = decode_sv (&dec);	1480	sv = decode_sv (&dec);
		1481
		1482	if (offset_return)
		1483	*offset_return = dec.cur;
1412		1484
1413	if (!(offset_return \|\| !sv))	1485	if (!(offset_return \|\| !sv))
1414	{	1486	{
1415	// check for trailing garbage	1487	// check for trailing garbage
1416	decode_ws (&dec);	1488	decode_ws (&dec);
…		…
1419	{	1491	{
1420	dec.err = "garbage after JSON object";	1492	dec.err = "garbage after JSON object";
1421	SvREFCNT_dec (sv);	1493	SvREFCNT_dec (sv);
1422	sv = 0;	1494	sv = 0;
1423	}	1495	}
1424	}
1425
1426	if (offset_return \|\| !sv)
1427	{
1428	offset = dec.json.flags & F_UTF8
1429	? dec.cur - SvPVX (string)
1430	: utf8_distance (dec.cur, SvPVX (string));
1431
1432	if (offset_return)
1433	*offset_return = offset;
1434	}	1496	}
1435		1497
1436	if (!sv)	1498	if (!sv)
1437	{	1499	{
1438	SV *uni = sv_newmortal ();	1500	SV *uni = sv_newmortal ();
…		…
1444	SAVEVPTR (PL_curcop);	1506	SAVEVPTR (PL_curcop);
1445	PL_curcop = &cop;	1507	PL_curcop = &cop;
1446	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1508	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1447	LEAVE;	1509	LEAVE;
1448		1510
1449	croak ("%s, at character offset %d [\"%s\"]",	1511	croak ("%s, at character offset %d (before \"%s\")",
1450	dec.err,	1512	dec.err,
1451	(int)offset,	1513	ptr_to_index (string, dec.cur),
1452	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1514	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1453	}	1515	}
1454		1516
1455	sv = sv_2mortal (sv);	1517	sv = sv_2mortal (sv);
1456		1518
1457	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))	1519	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1458	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");	1520	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1459		1521
1460	return sv;	1522	return sv;
		1523	}
		1524
		1525	/////////////////////////////////////////////////////////////////////////////
		1526	// incremental parser
		1527
		1528	static void
		1529	incr_parse (JSON *self)
		1530	{
		1531	const char *p = SvPVX (self->incr_text) + self->incr_pos;
		1532
		1533	// the state machine here is a bit convoluted and could be simplified a lot
		1534	// but this would make it slower, so...
		1535
		1536	for (;;)
		1537	{
		1538	//printf ("loop pod %d p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), p, p, self->incr_mode, self->incr_nest);//D
		1539	switch (self->incr_mode)
		1540	{
		1541	// only used for initial whitespace skipping
		1542	case INCR_M_WS:
		1543	for (;;)
		1544	{
		1545	if (*p > 0x20)
		1546	{
		1547	if (*p == '#')
		1548	{
		1549	self->incr_mode = INCR_M_C0;
		1550	goto incr_m_c;
		1551	}
		1552	else
		1553	{
		1554	self->incr_mode = INCR_M_JSON;
		1555	goto incr_m_json;
		1556	}
		1557	}
		1558	else if (!*p)
		1559	goto interrupt;
		1560
		1561	++p;
		1562	}
		1563
		1564	// skip a single char inside a string (for \\-processing)
		1565	case INCR_M_BS:
		1566	if (!*p)
		1567	goto interrupt;
		1568
		1569	++p;
		1570	self->incr_mode = INCR_M_STR;
		1571	goto incr_m_str;
		1572
		1573	// inside #-style comments
		1574	case INCR_M_C0:
		1575	case INCR_M_C1:
		1576	incr_m_c:
		1577	for (;;)
		1578	{
		1579	if (*p == '\n')
		1580	{
		1581	self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
		1582	break;
		1583	}
		1584	else if (!*p)
		1585	goto interrupt;
		1586
		1587	++p;
		1588	}
		1589
		1590	break;
		1591
		1592	// inside a string
		1593	case INCR_M_STR:
		1594	incr_m_str:
		1595	for (;;)
		1596	{
		1597	if (*p == '"')
		1598	{
		1599	++p;
		1600	self->incr_mode = INCR_M_JSON;
		1601
		1602	if (!self->incr_nest)
		1603	goto interrupt;
		1604
		1605	goto incr_m_json;
		1606	}
		1607	else if (*p == '\\')
		1608	{
		1609	++p; // "virtually" consumes character after \
		1610
		1611	if (!*p) // if at end of string we have to switch modes
		1612	{
		1613	self->incr_mode = INCR_M_BS;
		1614	goto interrupt;
		1615	}
		1616	}
		1617	else if (!*p)
		1618	goto interrupt;
		1619
		1620	++p;
		1621	}
		1622
		1623	// after initial ws, outside string
		1624	case INCR_M_JSON:
		1625	incr_m_json:
		1626	for (;;)
		1627	{
		1628	switch (*p++)
		1629	{
		1630	case 0:
		1631	--p;
		1632	goto interrupt;
		1633
		1634	case 0x09:
		1635	case 0x0a:
		1636	case 0x0d:
		1637	case 0x20:
		1638	if (!self->incr_nest)
		1639	{
		1640	--p; // do not eat the whitespace, let the next round do it
		1641	goto interrupt;
		1642	}
		1643	break;
		1644
		1645	case '"':
		1646	self->incr_mode = INCR_M_STR;
		1647	goto incr_m_str;
		1648
		1649	case '[':
		1650	case '{':
		1651	if (++self->incr_nest > self->max_depth)
		1652	croak (ERR_NESTING_EXCEEDED);
		1653	break;
		1654
		1655	case ']':
		1656	case '}':
		1657	if (--self->incr_nest <= 0)
		1658	goto interrupt;
		1659	break;
		1660
		1661	case '#':
		1662	self->incr_mode = INCR_M_C1;
		1663	goto incr_m_c;
		1664	}
		1665	}
		1666	}
		1667
		1668	modechange:
		1669	;
		1670	}
		1671
		1672	interrupt:
		1673	self->incr_pos = p - SvPVX (self->incr_text);
		1674	//printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
		1675	//printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1461	}	1676	}
1462		1677
1463	/////////////////////////////////////////////////////////////////////////////	1678	/////////////////////////////////////////////////////////////////////////////
1464	// XS interface functions	1679	// XS interface functions
1465		1680
…		…
1477	: -1;	1692	: -1;
1478		1693
1479	json_stash = gv_stashpv ("JSON::XS" , 1);	1694	json_stash = gv_stashpv ("JSON::XS" , 1);
1480	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);	1695	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1481		1696
1482	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );	1697	json_true = get_bool ("JSON::XS::true");
1483	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);	1698	json_false = get_bool ("JSON::XS::false");
		1699
		1700	CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1484	}	1701	}
1485		1702
1486	PROTOTYPES: DISABLE	1703	PROTOTYPES: DISABLE
1487		1704
1488	void CLONE (...)	1705	void CLONE (...)
…		…
1493	void new (char *klass)	1710	void new (char *klass)
1494	PPCODE:	1711	PPCODE:
1495	{	1712	{
1496	SV *pv = NEWSV (0, sizeof (JSON));	1713	SV *pv = NEWSV (0, sizeof (JSON));
1497	SvPOK_only (pv);	1714	SvPOK_only (pv);
1498	Zero (SvPVX (pv), 1, JSON);	1715	json_init ((JSON *)SvPVX (pv));
1499	((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1500	XPUSHs (sv_2mortal (sv_bless (	1716	XPUSHs (sv_2mortal (sv_bless (
1501	newRV_noinc (pv),	1717	newRV_noinc (pv),
1502	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)	1718	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1503	)));	1719	)));
1504	}	1720	}
…		…
1516	allow_nonref = F_ALLOW_NONREF	1732	allow_nonref = F_ALLOW_NONREF
1517	shrink = F_SHRINK	1733	shrink = F_SHRINK
1518	allow_blessed = F_ALLOW_BLESSED	1734	allow_blessed = F_ALLOW_BLESSED
1519	convert_blessed = F_CONV_BLESSED	1735	convert_blessed = F_CONV_BLESSED
1520	relaxed = F_RELAXED	1736	relaxed = F_RELAXED
		1737	allow_unknown = F_ALLOW_UNKNOWN
1521	PPCODE:	1738	PPCODE:
1522	{	1739	{
1523	if (enable)	1740	if (enable)
1524	self->flags \|= ix;	1741	self->flags \|= ix;
1525	else	1742	else
…		…
1540	get_allow_nonref = F_ALLOW_NONREF	1757	get_allow_nonref = F_ALLOW_NONREF
1541	get_shrink = F_SHRINK	1758	get_shrink = F_SHRINK
1542	get_allow_blessed = F_ALLOW_BLESSED	1759	get_allow_blessed = F_ALLOW_BLESSED
1543	get_convert_blessed = F_CONV_BLESSED	1760	get_convert_blessed = F_CONV_BLESSED
1544	get_relaxed = F_RELAXED	1761	get_relaxed = F_RELAXED
		1762	get_allow_unknown = F_ALLOW_UNKNOWN
1545	PPCODE:	1763	PPCODE:
1546	XPUSHs (boolSV (self->flags & ix));	1764	XPUSHs (boolSV (self->flags & ix));
1547		1765
1548	void max_depth (JSON *self, UV max_depth = 0x80000000UL)	1766	void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1549	PPCODE:	1767	PPCODE:
1550	{	1768	self->max_depth = max_depth;
1551	UV log2 = 0;
1552
1553	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1554
1555	while ((1UL << log2) < max_depth)
1556	++log2;
1557
1558	self->flags = self->flags & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
1559
1560	XPUSHs (ST (0));	1769	XPUSHs (ST (0));
1561	}
1562		1770
1563	U32 get_max_depth (JSON *self)	1771	U32 get_max_depth (JSON *self)
1564	CODE:	1772	CODE:
1565	RETVAL = DEC_DEPTH (self->flags);	1773	RETVAL = self->max_depth;
1566	OUTPUT:	1774	OUTPUT:
1567	RETVAL	1775	RETVAL
1568		1776
1569	void max_size (JSON *self, UV max_size = 0)	1777	void max_size (JSON *self, U32 max_size = 0)
1570	PPCODE:	1778	PPCODE:
1571	{	1779	self->max_size = max_size;
1572	UV log2 = 0;
1573
1574	if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1575	if (max_size == 1) max_size = 2;
1576
1577	while ((1UL << log2) < max_size)
1578	++log2;
1579
1580	self->flags = self->flags & ~F_MAXSIZE \| (log2 << S_MAXSIZE);
1581
1582	XPUSHs (ST (0));	1780	XPUSHs (ST (0));
1583	}
1584		1781
1585	int get_max_size (JSON *self)	1782	int get_max_size (JSON *self)
1586	CODE:	1783	CODE:
1587	RETVAL = DEC_SIZE (self->flags);	1784	RETVAL = self->max_size;
1588	OUTPUT:	1785	OUTPUT:
1589	RETVAL	1786	RETVAL
1590		1787
1591	void filter_json_object (JSON self, SV cb = &PL_sv_undef)	1788	void filter_json_object (JSON self, SV cb = &PL_sv_undef)
1592	PPCODE:	1789	PPCODE:
…		…
1628	XPUSHs (decode_json (jsonstr, self, 0));	1825	XPUSHs (decode_json (jsonstr, self, 0));
1629		1826
1630	void decode_prefix (JSON self, SV jsonstr)	1827	void decode_prefix (JSON self, SV jsonstr)
1631	PPCODE:	1828	PPCODE:
1632	{	1829	{
1633	UV offset;	1830	char *offset;
1634	EXTEND (SP, 2);	1831	EXTEND (SP, 2);
1635	PUSHs (decode_json (jsonstr, self, &offset));	1832	PUSHs (decode_json (jsonstr, self, &offset));
1636	PUSHs (sv_2mortal (newSVuv (offset)));	1833	PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
		1834	}
		1835
		1836	void incr_parse (JSON self, SV jsonstr = 0)
		1837	PPCODE:
		1838	{
		1839	if (!self->incr_text)
		1840	self->incr_text = newSVpvn ("", 0);
		1841
		1842	// append data, if any
		1843	if (jsonstr)
		1844	{
		1845	if (SvUTF8 (jsonstr))
		1846	{
		1847	if (!SvUTF8 (self->incr_text))
		1848	{
		1849	/* utf-8-ness differs, need to upgrade */
		1850	sv_utf8_upgrade (self->incr_text);
		1851
		1852	if (self->incr_pos)
		1853	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
		1854	- (U8 *)SvPVX (self->incr_text);
		1855	}
		1856	}
		1857	else if (SvUTF8 (self->incr_text))
		1858	sv_utf8_upgrade (jsonstr);
		1859
		1860	{
		1861	STRLEN len;
		1862	const char *str = SvPV (jsonstr, len);
		1863	STRLEN cur = SvCUR (self->incr_text);
		1864
		1865	if (SvLEN (self->incr_text) <= cur + len)
		1866	SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
		1867
		1868	Move (str, SvEND (self->incr_text), len, char);
		1869	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
		1870	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
		1871	}
		1872	}
		1873
		1874	if (GIMME_V != G_VOID)
		1875	do
		1876	{
		1877	char *offset;
		1878
		1879	if (!INCR_DONE (self))
		1880	{
		1881	incr_parse (self);
		1882
		1883	if (self->incr_pos > self->max_size && self->max_size)
		1884	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1885	(unsigned long)self->incr_pos, (unsigned long)self->max_size);
		1886
		1887	if (!INCR_DONE (self))
		1888	break;
		1889	}
		1890
		1891	XPUSHs (decode_json (self->incr_text, self, &offset));
		1892
		1893	self->incr_pos -= offset - SvPVX (self->incr_text);
		1894	self->incr_nest = 0;
		1895	self->incr_mode = 0;
		1896
		1897	sv_chop (self->incr_text, offset);
		1898	}
		1899	while (GIMME_V == G_ARRAY);
		1900	}
		1901
		1902	SV incr_text (JSON self)
		1903	ATTRS: lvalue
		1904	CODE:
		1905	{
		1906	if (self->incr_pos)
		1907	croak ("incr_text can not be called when the incremental parser already started parsing");
		1908
		1909	RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
		1910	}
		1911	OUTPUT:
		1912	RETVAL
		1913
		1914	void incr_skip (JSON *self)
		1915	CODE:
		1916	{
		1917	if (self->incr_pos)
		1918	{
		1919	sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
		1920	self->incr_pos = 0;
		1921	self->incr_nest = 0;
		1922	self->incr_mode = 0;
		1923	}
		1924	}
		1925
		1926	void incr_reset (JSON *self)
		1927	CODE:
		1928	{
		1929	SvREFCNT_dec (self->incr_text);
		1930	self->incr_text = 0;
		1931	self->incr_pos = 0;
		1932	self->incr_nest = 0;
		1933	self->incr_mode = 0;
1637	}	1934	}
1638		1935
1639	void DESTROY (JSON *self)	1936	void DESTROY (JSON *self)
1640	CODE:	1937	CODE:
1641	SvREFCNT_dec (self->cb_sk_object);	1938	SvREFCNT_dec (self->cb_sk_object);
1642	SvREFCNT_dec (self->cb_object);	1939	SvREFCNT_dec (self->cb_object);
		1940	SvREFCNT_dec (self->incr_text);
1643		1941
1644	PROTOTYPES: ENABLE	1942	PROTOTYPES: ENABLE
1645		1943
1646	void encode_json (SV *scalar)	1944	void encode_json (SV *scalar)
		1945	ALIAS:
		1946	to_json_ = 0
		1947	encode_json = F_UTF8
1647	PPCODE:	1948	PPCODE:
1648	{	1949	{
1649	JSON json = { F_DEFAULT \| F_UTF8 };	1950	JSON json;
		1951	json_init (&json);
		1952	json.flags \|= ix;
1650	XPUSHs (encode_json (scalar, &json));	1953	XPUSHs (encode_json (scalar, &json));
1651	}	1954	}
1652		1955
1653	void decode_json (SV *jsonstr)	1956	void decode_json (SV *jsonstr)
		1957	ALIAS:
		1958	from_json_ = 0
		1959	decode_json = F_UTF8
1654	PPCODE:	1960	PPCODE:
1655	{	1961	{
1656	JSON json = { F_DEFAULT \| F_UTF8 };	1962	JSON json;
		1963	json_init (&json);
		1964	json.flags \|= ix;
1657	XPUSHs (decode_json (jsonstr, &json, 0));	1965	XPUSHs (decode_json (jsonstr, &json, 0));
1658	}	1966	}
1659		1967
		1968

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.75 by root, Wed Mar 19 15:37:54 2008 UTC vs. Revision 1.103 by root, Wed Jan 6 08:01:39 2010 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.75 by root, Wed Mar 19 15:37:54 2008 UTC vs.
Revision 1.103 by root, Wed Jan 6 08:01:39 2010 UTC