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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.76 by root, Thu Mar 20 00:56:37 2008 UTC vs.
Revision 1.99 by root, Sat Aug 8 10:06:02 2009 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	#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)	22	#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
…		…
31	#define F_ALLOW_NONREF 0x00000100UL	31	#define F_ALLOW_NONREF 0x00000100UL
32	#define F_SHRINK 0x00000200UL	32	#define F_SHRINK 0x00000200UL
33	#define F_ALLOW_BLESSED 0x00000400UL	33	#define F_ALLOW_BLESSED 0x00000400UL
34	#define F_CONV_BLESSED 0x00000800UL	34	#define F_CONV_BLESSED 0x00000800UL
35	#define F_RELAXED 0x00001000UL	35	#define F_RELAXED 0x00001000UL
36		36	#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	37	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
42		38
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	39	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
47	#define F_DEFAULT (9UL << S_MAXDEPTH)
48		40
49	#define INIT_SIZE 32 // initial scalar size to be allocated	41	#define INIT_SIZE 32 // initial scalar size to be allocated
50	#define INDENT_STEP 3 // spaces per indentation level	42	#define INDENT_STEP 3 // spaces per indentation level
51		43
52	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size	44	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
…		…
67		59
68	#define IN_RANGE_INC(type,val,beg,end) \	60	#define IN_RANGE_INC(type,val,beg,end) \
69	((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \	61	((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
70	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))	62	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
71		63
		64	#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
		65
72	#ifdef USE_ITHREADS	66	#ifdef USE_ITHREADS
73	# define JSON_SLOW 1	67	# define JSON_SLOW 1
74	# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))	68	# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
75	#else	69	#else
76	# define JSON_SLOW 0	70	# define JSON_SLOW 0
…		…
78	#endif	72	#endif
79		73
80	static HV json_stash, json_boolean_stash; // JSON::XS::	74	static HV json_stash, json_boolean_stash; // JSON::XS::
81	static SV json_true, json_false;	75	static SV json_true, json_false;
82		76
		77	enum {
		78	INCR_M_WS = 0, // initial whitespace skipping, must be 0
		79	INCR_M_STR, // inside string
		80	INCR_M_BS, // inside backslash
		81	INCR_M_JSON // outside anything, count nesting
		82	};
		83
		84	#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
		85
83	typedef struct {	86	typedef struct {
84	U32 flags;	87	U32 flags;
		88	U32 max_depth;
		89	STRLEN max_size;
		90
85	SV *cb_object;	91	SV *cb_object;
86	HV *cb_sk_object;	92	HV *cb_sk_object;
		93
		94	// for the incremental parser
		95	SV *incr_text; // the source text so far
		96	STRLEN incr_pos; // the current offset into the text
		97	int incr_nest; // {[]}-nesting level
		98	unsigned char incr_mode;
87	} JSON;	99	} JSON;
		100
		101	INLINE void
		102	json_init (JSON *json)
		103	{
		104	Zero (json, 1, JSON);
		105	json->max_depth = 512;
		106	}
88		107
89	/////////////////////////////////////////////////////////////////////////////	108	/////////////////////////////////////////////////////////////////////////////
90	// utility functions	109	// utility functions
91		110
		111	INLINE SV *
		112	get_bool (const char *name)
		113	{
		114	SV *sv = get_sv (name, 1);
		115
		116	SvREADONLY_on (sv);
		117	SvREADONLY_on (SvRV (sv));
		118
		119	return sv;
		120	}
		121
92	INLINE void	122	INLINE void
93	shrink (SV *sv)	123	shrink (SV *sv)
94	{	124	{
95	sv_utf8_downgrade (sv, 1);	125	sv_utf8_downgrade (sv, 1);
		126
96	if (SvLEN (sv) > SvCUR (sv) + 1)	127	if (SvLEN (sv) > SvCUR (sv) + 1)
97	{	128	{
98	#ifdef SvPV_shrink_to_cur	129	#ifdef SvPV_shrink_to_cur
99	SvPV_shrink_to_cur (sv);	130	SvPV_shrink_to_cur (sv);
100	#elif defined (SvPV_renew)	131	#elif defined (SvPV_renew)
…		…
144	*s++ = 0x80 \| ( ch & 0x3f);	175	*s++ = 0x80 \| ( ch & 0x3f);
145		176
146	return s;	177	return s;
147	}	178	}
148		179
		180	// convert offset pointer to character index, sv must be string
		181	static STRLEN
		182	ptr_to_index (SV sv, char offset)
		183	{
		184	return SvUTF8 (sv)
		185	? utf8_distance (offset, SvPVX (sv))
		186	: offset - SvPVX (sv);
		187	}
		188
149	/////////////////////////////////////////////////////////////////////////////	189	/////////////////////////////////////////////////////////////////////////////
150	// encoder	190	// encoder
151		191
152	// structure used for encoding JSON	192	// structure used for encoding JSON
153	typedef struct	193	typedef struct
…		…
155	char *cur; // SvPVX (sv) + current output position	195	char *cur; // SvPVX (sv) + current output position
156	char *end; // SvEND (sv)	196	char *end; // SvEND (sv)
157	SV *sv; // result scalar	197	SV *sv; // result scalar
158	JSON json;	198	JSON json;
159	U32 indent; // indentation level	199	U32 indent; // indentation level
160	U32 maxdepth; // max. indentation/recursion level
161	UV limit; // escape character values >= this value when encoding	200	UV limit; // escape character values >= this value when encoding
162	} enc_t;	201	} enc_t;
163		202
164	INLINE void	203	INLINE void
165	need (enc_t *enc, STRLEN len)	204	need (enc_t *enc, STRLEN len)
166	{	205	{
167	if (expect_false (enc->cur + len >= enc->end))	206	if (expect_false (enc->cur + len >= enc->end))
168	{	207	{
169	STRLEN cur = enc->cur - SvPVX (enc->sv);	208	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
170	SvGROW (enc->sv, cur + len + 1);	209	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
171	enc->cur = SvPVX (enc->sv) + cur;	210	enc->cur = SvPVX (enc->sv) + cur;
172	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	211	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
173	}	212	}
174	}	213	}
175		214
…		…
250	(int)((uch - 0x10000) % 0x400 + 0xDC00));	289	(int)((uch - 0x10000) % 0x400 + 0xDC00));
251	enc->cur += 12;	290	enc->cur += 12;
252	}	291	}
253	else	292	else
254	{	293	{
255	static char hexdigit [16] = "0123456789abcdef";
256	need (enc, len += 5);	294	need (enc, len += 5);
257	*enc->cur++ = '\\';	295	*enc->cur++ = '\\';
258	*enc->cur++ = 'u';	296	*enc->cur++ = 'u';
259	*enc->cur++ = hexdigit [ uch >> 12 ];	297	*enc->cur++ = PL_hexdigit [ uch >> 12 ];
260	*enc->cur++ = hexdigit [(uch >> 8) & 15];	298	*enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
261	*enc->cur++ = hexdigit [(uch >> 4) & 15];	299	*enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
262	*enc->cur++ = hexdigit [(uch >> 0) & 15];	300	*enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
263	}	301	}
264		302
265	str += clen;	303	str += clen;
266	}	304	}
267	else if (enc->json.flags & F_LATIN1)	305	else if (enc->json.flags & F_LATIN1)
…		…
338	static void	376	static void
339	encode_av (enc_t enc, AV av)	377	encode_av (enc_t enc, AV av)
340	{	378	{
341	int i, len = av_len (av);	379	int i, len = av_len (av);
342		380
343	if (enc->indent >= enc->maxdepth)	381	if (enc->indent >= enc->json.max_depth)
344	croak ("data structure too deep (hit recursion limit)");	382	croak (ERR_NESTING_EXCEEDED);
345		383
346	encode_ch (enc, '[');	384	encode_ch (enc, '[');
347		385
348	if (len >= 0)	386	if (len >= 0)
349	{	387	{
…		…
423		461
424	static void	462	static void
425	encode_hv (enc_t enc, HV hv)	463	encode_hv (enc_t enc, HV hv)
426	{	464	{
427	HE *he;	465	HE *he;
428	int count;
429		466
430	if (enc->indent >= enc->maxdepth)	467	if (enc->indent >= enc->json.max_depth)
431	croak ("data structure too deep (hit recursion limit)");	468	croak (ERR_NESTING_EXCEEDED);
432		469
433	encode_ch (enc, '{');	470	encode_ch (enc, '{');
434		471
435	// for canonical output we have to sort by keys first	472	// for canonical output we have to sort by keys first
436	// actually, this is mostly due to the stupid so-called	473	// actually, this is mostly due to the stupid so-called
437	// security workaround added somewhere in 5.8.x.	474	// security workaround added somewhere in 5.8.x
438	// that randomises hash orderings	475	// that randomises hash orderings
439	if (enc->json.flags & F_CANONICAL)	476	if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
440	{	477	{
441	int count = hv_iterinit (hv);	478	int count = hv_iterinit (hv);
442		479
443	if (SvMAGICAL (hv))	480	if (SvMAGICAL (hv))
444	{	481	{
…		…
617		654
618	if (len == 1 && *pv == '1')	655	if (len == 1 && *pv == '1')
619	encode_str (enc, "true", 4, 0);	656	encode_str (enc, "true", 4, 0);
620	else if (len == 1 && *pv == '0')	657	else if (len == 1 && *pv == '0')
621	encode_str (enc, "false", 5, 0);	658	encode_str (enc, "false", 5, 0);
		659	else if (enc->json.flags & F_ALLOW_UNKNOWN)
		660	encode_str (enc, "null", 4, 0);
622	else	661	else
623	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",	662	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
624	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	663	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
625	}	664	}
		665	else if (enc->json.flags & F_ALLOW_UNKNOWN)
		666	encode_str (enc, "null", 4, 0);
626	else	667	else
627	croak ("encountered %s, but JSON can only represent references to arrays or hashes",	668	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
628	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	669	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
629	}	670	}
630		671
…		…
692	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));	733	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
693	}	734	}
694	}	735	}
695	else if (SvROK (sv))	736	else if (SvROK (sv))
696	encode_rv (enc, SvRV (sv));	737	encode_rv (enc, SvRV (sv));
697	else if (!SvOK (sv))	738	else if (!SvOK (sv) \|\| enc->json.flags & F_ALLOW_UNKNOWN)
698	encode_str (enc, "null", 4, 0);	739	encode_str (enc, "null", 4, 0);
699	else	740	else
700	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	741	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
701	SvPV_nolen (sv), SvFLAGS (sv));	742	SvPV_nolen (sv), SvFLAGS (sv));
702	}	743	}
…		…
712	enc.json = *json;	753	enc.json = *json;
713	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	754	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
714	enc.cur = SvPVX (enc.sv);	755	enc.cur = SvPVX (enc.sv);
715	enc.end = SvEND (enc.sv);	756	enc.end = SvEND (enc.sv);
716	enc.indent = 0;	757	enc.indent = 0;
717	enc.maxdepth = DEC_DEPTH (enc.json.flags);
718	enc.limit = enc.json.flags & F_ASCII ? 0x000080UL	758	enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
719	: enc.json.flags & F_LATIN1 ? 0x000100UL	759	: enc.json.flags & F_LATIN1 ? 0x000100UL
720	: 0x110000UL;	760	: 0x110000UL;
721		761
722	SvPOK_only (enc.sv);	762	SvPOK_only (enc.sv);
…		…
789	if (*dec->cur != ch) \	829	if (*dec->cur != ch) \
790	ERR (# ch " expected"); \	830	ERR (# ch " expected"); \
791	++dec->cur; \	831	++dec->cur; \
792	SE	832	SE
793		833
794	#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")	834	#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
795	#define DEC_DEC_DEPTH --dec->depth	835	#define DEC_DEC_DEPTH --dec->depth
796		836
797	static SV decode_sv (dec_t dec);	837	static SV decode_sv (dec_t dec);
798		838
799	static signed char decode_hexdigit[256];	839	static signed char decode_hexdigit[256];
…		…
939	{	979	{
940	STRLEN len = cur - buf;	980	STRLEN len = cur - buf;
941		981
942	if (sv)	982	if (sv)
943	{	983	{
944	SvGROW (sv, SvCUR (sv) + len + 1);	984	STRLEN cur = SvCUR (sv);
		985
		986	if (SvLEN (sv) <= cur + len)
		987	SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
		988
945	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	989	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
946	SvCUR_set (sv, SvCUR (sv) + len);	990	SvCUR_set (sv, SvCUR (sv) + len);
947	}	991	}
948	else	992	else
949	sv = newSVpvn (buf, len);	993	sv = newSVpvn (buf, len);
…		…
1040		1084
1041	// special case the rather common 1..5-digit-int case	1085	// special case the rather common 1..5-digit-int case
1042	if (*start == '-')	1086	if (*start == '-')
1043	switch (len)	1087	switch (len)
1044	{	1088	{
1045	case 2: return newSViv (-( start [1] - '0' * 1));	1089	case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1046	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));	1090	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));	1091	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));	1092	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));	1093	case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1050	}	1094	}
1051	else	1095	else
1052	switch (len)	1096	switch (len)
1053	{	1097	{
1054	case 1: return newSViv ( start [0] - '0' * 1);	1098	case 1: return newSViv ( start [0] - '0' * 1);
1055	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);	1099	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);	1100	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);	1101	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);	1102	case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1059	}	1103	}
1060		1104
1061	{	1105	{
1062	UV uv;	1106	UV uv;
1063	int numtype = grok_number (start, len, &uv);	1107	int numtype = grok_number (start, len, &uv);
…		…
1171	char *p = dec->cur;	1215	char *p = dec->cur;
1172	char *e = p + 24; // only try up to 24 bytes	1216	char *e = p + 24; // only try up to 24 bytes
1173		1217
1174	for (;;)	1218	for (;;)
1175	{	1219	{
1176	// the >= 0x80 is true on most architectures	1220	// the >= 0x80 is false on most architectures
1177	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')	1221	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
1178	{	1222	{
1179	// slow path, back up and use decode_str	1223	// slow path, back up and use decode_str
1180	SV *key = decode_str (dec);	1224	SV *key = decode_str (dec);
1181	if (!key)	1225	if (!key)
…		…
1329	case 't':	1373	case 't':
1330	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1374	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1331	{	1375	{
1332	dec->cur += 4;	1376	dec->cur += 4;
1333	#if JSON_SLOW	1377	#if JSON_SLOW
1334	json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);	1378	json_true = get_bool ("JSON::XS::true");
1335	#endif	1379	#endif
1336	return SvREFCNT_inc (json_true);	1380	return newSVsv (json_true);
1337	}	1381	}
1338	else	1382	else
1339	ERR ("'true' expected");	1383	ERR ("'true' expected");
1340		1384
1341	break;	1385	break;
…		…
1343	case 'f':	1387	case 'f':
1344	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))	1388	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1345	{	1389	{
1346	dec->cur += 5;	1390	dec->cur += 5;
1347	#if JSON_SLOW	1391	#if JSON_SLOW
1348	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);	1392	json_false = get_bool ("JSON::XS::false");
1349	#endif	1393	#endif
1350	return SvREFCNT_inc (json_false);	1394	return newSVsv (json_false);
1351	}	1395	}
1352	else	1396	else
1353	ERR ("'false' expected");	1397	ERR ("'false' expected");
1354		1398
1355	break;	1399	break;
…		…
1373	fail:	1417	fail:
1374	return 0;	1418	return 0;
1375	}	1419	}
1376		1420
1377	static SV *	1421	static SV *
1378	decode_json (SV string, JSON json, UV *offset_return)	1422	decode_json (SV string, JSON json, char **offset_return)
1379	{	1423	{
1380	dec_t dec;	1424	dec_t dec;
1381	UV offset;
1382	SV *sv;	1425	SV *sv;
1383		1426
		1427	/* work around bugs in 5.10 where manipulating magic values
		1428	* will perl ignore the magic in subsequent accesses
		1429	*/
1384	SvGETMAGIC (string);	1430	/SvGETMAGIC (string);/
		1431	if (SvMAGICAL (string))
		1432	string = sv_2mortal (newSVsv (string));
		1433
1385	SvUPGRADE (string, SVt_PV);	1434	SvUPGRADE (string, SVt_PV);
1386		1435
1387	if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))	1436	/* work around a bug in perl 5.10, which causes SvCUR to fail an
		1437	* assertion with -DDEBUGGING, although SvCUR is documented to
		1438	* return the xpv_cur field which certainly exists after upgrading.
		1439	* according to nicholas clark, calling SvPOK fixes this.
		1440	* But it doesn't fix it, so try another workaround, call SvPV_nolen
		1441	* and hope for the best.
		1442	* Damnit, SvPV_nolen still trips over yet another assertion. This
		1443	* assertion business is seriously broken, try yet another workaround
		1444	* for the broken -DDEBUGGING.
		1445	*/
		1446	{
		1447	#ifdef DEBUGGING
		1448	STRLEN offset = SvOK (string) ? sv_len (string) : 0;
		1449	#else
		1450	STRLEN offset = SvCUR (string);
		1451	#endif
		1452
		1453	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",	1454	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));	1455	(unsigned long)SvCUR (string), (unsigned long)json->max_size);
		1456	}
1390		1457
1391	if (json->flags & F_UTF8)	1458	if (json->flags & F_UTF8)
1392	sv_utf8_downgrade (string, 0);	1459	sv_utf8_downgrade (string, 0);
1393	else	1460	else
1394	sv_utf8_upgrade (string);	1461	sv_utf8_upgrade (string);
1395		1462
1396	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1463	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1397		1464
1398	dec.json = *json;	1465	dec.json = *json;
1399	dec.cur = SvPVX (string);	1466	dec.cur = SvPVX (string);
1400	dec.end = SvEND (string);	1467	dec.end = SvEND (string);
1401	dec.err = 0;	1468	dec.err = 0;
1402	dec.depth = 0;	1469	dec.depth = 0;
1403	dec.maxdepth = DEC_DEPTH (dec.json.flags);
1404		1470
1405	if (dec.json.cb_object \|\| dec.json.cb_sk_object)	1471	if (dec.json.cb_object \|\| dec.json.cb_sk_object)
1406	dec.json.flags \|= F_HOOK;	1472	dec.json.flags \|= F_HOOK;
1407		1473
1408	*dec.end = 0; // this should basically be a nop, too, but make sure it's there	1474	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
1409		1475
1410	decode_ws (&dec);	1476	decode_ws (&dec);
1411	sv = decode_sv (&dec);	1477	sv = decode_sv (&dec);
		1478
		1479	if (offset_return)
		1480	*offset_return = dec.cur;
1412		1481
1413	if (!(offset_return \|\| !sv))	1482	if (!(offset_return \|\| !sv))
1414	{	1483	{
1415	// check for trailing garbage	1484	// check for trailing garbage
1416	decode_ws (&dec);	1485	decode_ws (&dec);
…		…
1419	{	1488	{
1420	dec.err = "garbage after JSON object";	1489	dec.err = "garbage after JSON object";
1421	SvREFCNT_dec (sv);	1490	SvREFCNT_dec (sv);
1422	sv = 0;	1491	sv = 0;
1423	}	1492	}
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	}	1493	}
1435		1494
1436	if (!sv)	1495	if (!sv)
1437	{	1496	{
1438	SV *uni = sv_newmortal ();	1497	SV *uni = sv_newmortal ();
…		…
1444	SAVEVPTR (PL_curcop);	1503	SAVEVPTR (PL_curcop);
1445	PL_curcop = &cop;	1504	PL_curcop = &cop;
1446	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1505	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1447	LEAVE;	1506	LEAVE;
1448		1507
1449	croak ("%s, at character offset %d [\"%s\"]",	1508	croak ("%s, at character offset %d (before \"%s\")",
1450	dec.err,	1509	dec.err,
1451	(int)offset,	1510	ptr_to_index (string, dec.cur),
1452	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1511	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1453	}	1512	}
1454		1513
1455	sv = sv_2mortal (sv);	1514	sv = sv_2mortal (sv);
1456		1515
1457	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))	1516	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)");	1517	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1459		1518
1460	return sv;	1519	return sv;
		1520	}
		1521
		1522	/////////////////////////////////////////////////////////////////////////////
		1523	// incremental parser
		1524
		1525	static void
		1526	incr_parse (JSON *self)
		1527	{
		1528	const char *p = SvPVX (self->incr_text) + self->incr_pos;
		1529
		1530	for (;;)
		1531	{
		1532	//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
		1533	switch (self->incr_mode)
		1534	{
		1535	// only used for intiial whitespace skipping
		1536	case INCR_M_WS:
		1537	for (;;)
		1538	{
		1539	if (*p > 0x20)
		1540	{
		1541	self->incr_mode = INCR_M_JSON;
		1542	goto incr_m_json;
		1543	}
		1544	else if (!*p)
		1545	goto interrupt;
		1546
		1547	++p;
		1548	}
		1549
		1550	// skip a single char inside a string (for \\-processing)
		1551	case INCR_M_BS:
		1552	if (!*p)
		1553	goto interrupt;
		1554
		1555	++p;
		1556	self->incr_mode = INCR_M_STR;
		1557	goto incr_m_str;
		1558
		1559	// inside a string
		1560	case INCR_M_STR:
		1561	incr_m_str:
		1562	for (;;)
		1563	{
		1564	if (*p == '"')
		1565	{
		1566	++p;
		1567	self->incr_mode = INCR_M_JSON;
		1568
		1569	if (!self->incr_nest)
		1570	goto interrupt;
		1571
		1572	goto incr_m_json;
		1573	}
		1574	else if (*p == '\\')
		1575	{
		1576	++p; // "virtually" consumes character after \
		1577
		1578	if (!*p) // if at end of string we have to switch modes
		1579	{
		1580	self->incr_mode = INCR_M_BS;
		1581	goto interrupt;
		1582	}
		1583	}
		1584	else if (!*p)
		1585	goto interrupt;
		1586
		1587	++p;
		1588	}
		1589
		1590	// after initial ws, outside string
		1591	case INCR_M_JSON:
		1592	incr_m_json:
		1593	for (;;)
		1594	{
		1595	switch (*p++)
		1596	{
		1597	case 0:
		1598	--p;
		1599	goto interrupt;
		1600
		1601	case 0x09:
		1602	case 0x0a:
		1603	case 0x0d:
		1604	case 0x20:
		1605	if (!self->incr_nest)
		1606	{
		1607	--p; // do not eat the whitespace, let the next round do it
		1608	goto interrupt;
		1609	}
		1610	break;
		1611
		1612	case '"':
		1613	self->incr_mode = INCR_M_STR;
		1614	goto incr_m_str;
		1615
		1616	case '[':
		1617	case '{':
		1618	if (++self->incr_nest > self->max_depth)
		1619	croak (ERR_NESTING_EXCEEDED);
		1620	break;
		1621
		1622	case ']':
		1623	case '}':
		1624	if (--self->incr_nest <= 0)
		1625	goto interrupt;
		1626	}
		1627	}
		1628	}
		1629
		1630	modechange:
		1631	;
		1632	}
		1633
		1634	interrupt:
		1635	self->incr_pos = p - SvPVX (self->incr_text);
		1636	//printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1461	}	1637	}
1462		1638
1463	/////////////////////////////////////////////////////////////////////////////	1639	/////////////////////////////////////////////////////////////////////////////
1464	// XS interface functions	1640	// XS interface functions
1465		1641
…		…
1477	: -1;	1653	: -1;
1478		1654
1479	json_stash = gv_stashpv ("JSON::XS" , 1);	1655	json_stash = gv_stashpv ("JSON::XS" , 1);
1480	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);	1656	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1481		1657
1482	json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );	1658	json_true = get_bool ("JSON::XS::true");
1483	json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);	1659	json_false = get_bool ("JSON::XS::false");
		1660
		1661	CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1484	}	1662	}
1485		1663
1486	PROTOTYPES: DISABLE	1664	PROTOTYPES: DISABLE
1487		1665
1488	void CLONE (...)	1666	void CLONE (...)
…		…
1493	void new (char *klass)	1671	void new (char *klass)
1494	PPCODE:	1672	PPCODE:
1495	{	1673	{
1496	SV *pv = NEWSV (0, sizeof (JSON));	1674	SV *pv = NEWSV (0, sizeof (JSON));
1497	SvPOK_only (pv);	1675	SvPOK_only (pv);
1498	Zero (SvPVX (pv), 1, JSON);	1676	json_init ((JSON *)SvPVX (pv));
1499	((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1500	XPUSHs (sv_2mortal (sv_bless (	1677	XPUSHs (sv_2mortal (sv_bless (
1501	newRV_noinc (pv),	1678	newRV_noinc (pv),
1502	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)	1679	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1503	)));	1680	)));
1504	}	1681	}
…		…
1516	allow_nonref = F_ALLOW_NONREF	1693	allow_nonref = F_ALLOW_NONREF
1517	shrink = F_SHRINK	1694	shrink = F_SHRINK
1518	allow_blessed = F_ALLOW_BLESSED	1695	allow_blessed = F_ALLOW_BLESSED
1519	convert_blessed = F_CONV_BLESSED	1696	convert_blessed = F_CONV_BLESSED
1520	relaxed = F_RELAXED	1697	relaxed = F_RELAXED
		1698	allow_unknown = F_ALLOW_UNKNOWN
1521	PPCODE:	1699	PPCODE:
1522	{	1700	{
1523	if (enable)	1701	if (enable)
1524	self->flags \|= ix;	1702	self->flags \|= ix;
1525	else	1703	else
…		…
1540	get_allow_nonref = F_ALLOW_NONREF	1718	get_allow_nonref = F_ALLOW_NONREF
1541	get_shrink = F_SHRINK	1719	get_shrink = F_SHRINK
1542	get_allow_blessed = F_ALLOW_BLESSED	1720	get_allow_blessed = F_ALLOW_BLESSED
1543	get_convert_blessed = F_CONV_BLESSED	1721	get_convert_blessed = F_CONV_BLESSED
1544	get_relaxed = F_RELAXED	1722	get_relaxed = F_RELAXED
		1723	get_allow_unknown = F_ALLOW_UNKNOWN
1545	PPCODE:	1724	PPCODE:
1546	XPUSHs (boolSV (self->flags & ix));	1725	XPUSHs (boolSV (self->flags & ix));
1547		1726
1548	void max_depth (JSON *self, UV max_depth = 0x80000000UL)	1727	void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1549	PPCODE:	1728	PPCODE:
1550	{	1729	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));	1730	XPUSHs (ST (0));
1561	}
1562		1731
1563	U32 get_max_depth (JSON *self)	1732	U32 get_max_depth (JSON *self)
1564	CODE:	1733	CODE:
1565	RETVAL = DEC_DEPTH (self->flags);	1734	RETVAL = self->max_depth;
1566	OUTPUT:	1735	OUTPUT:
1567	RETVAL	1736	RETVAL
1568		1737
1569	void max_size (JSON *self, UV max_size = 0)	1738	void max_size (JSON *self, U32 max_size = 0)
1570	PPCODE:	1739	PPCODE:
1571	{	1740	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));	1741	XPUSHs (ST (0));
1583	}
1584		1742
1585	int get_max_size (JSON *self)	1743	int get_max_size (JSON *self)
1586	CODE:	1744	CODE:
1587	RETVAL = DEC_SIZE (self->flags);	1745	RETVAL = self->max_size;
1588	OUTPUT:	1746	OUTPUT:
1589	RETVAL	1747	RETVAL
1590		1748
1591	void filter_json_object (JSON self, SV cb = &PL_sv_undef)	1749	void filter_json_object (JSON self, SV cb = &PL_sv_undef)
1592	PPCODE:	1750	PPCODE:
…		…
1628	XPUSHs (decode_json (jsonstr, self, 0));	1786	XPUSHs (decode_json (jsonstr, self, 0));
1629		1787
1630	void decode_prefix (JSON self, SV jsonstr)	1788	void decode_prefix (JSON self, SV jsonstr)
1631	PPCODE:	1789	PPCODE:
1632	{	1790	{
1633	UV offset;	1791	char *offset;
1634	EXTEND (SP, 2);	1792	EXTEND (SP, 2);
1635	PUSHs (decode_json (jsonstr, self, &offset));	1793	PUSHs (decode_json (jsonstr, self, &offset));
1636	PUSHs (sv_2mortal (newSVuv (offset)));	1794	PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
		1795	}
		1796
		1797	void incr_parse (JSON self, SV jsonstr = 0)
		1798	PPCODE:
		1799	{
		1800	if (!self->incr_text)
		1801	self->incr_text = newSVpvn ("", 0);
		1802
		1803	// append data, if any
		1804	if (jsonstr)
		1805	{
		1806	if (SvUTF8 (jsonstr))
		1807	{
		1808	if (!SvUTF8 (self->incr_text))
		1809	{
		1810	/* utf-8-ness differs, need to upgrade */
		1811	sv_utf8_upgrade (self->incr_text);
		1812
		1813	if (self->incr_pos)
		1814	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
		1815	- (U8 *)SvPVX (self->incr_text);
		1816	}
		1817	}
		1818	else if (SvUTF8 (self->incr_text))
		1819	sv_utf8_upgrade (jsonstr);
		1820
		1821	{
		1822	STRLEN len;
		1823	const char *str = SvPV (jsonstr, len);
		1824	STRLEN cur = SvCUR (self->incr_text);
		1825
		1826	if (SvLEN (self->incr_text) <= cur + len)
		1827	SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
		1828
		1829	Move (str, SvEND (self->incr_text), len, char);
		1830	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
		1831	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
		1832	}
		1833	}
		1834
		1835	if (GIMME_V != G_VOID)
		1836	do
		1837	{
		1838	char *offset;
		1839
		1840	if (!INCR_DONE (self))
		1841	{
		1842	incr_parse (self);
		1843
		1844	if (self->incr_pos > self->max_size && self->max_size)
		1845	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		1846	(unsigned long)self->incr_pos, (unsigned long)self->max_size);
		1847
		1848	if (!INCR_DONE (self))
		1849	break;
		1850	}
		1851
		1852	XPUSHs (decode_json (self->incr_text, self, &offset));
		1853
		1854	self->incr_pos -= offset - SvPVX (self->incr_text);
		1855	self->incr_nest = 0;
		1856	self->incr_mode = 0;
		1857
		1858	sv_chop (self->incr_text, offset);
		1859	}
		1860	while (GIMME_V == G_ARRAY);
		1861	}
		1862
		1863	SV incr_text (JSON self)
		1864	ATTRS: lvalue
		1865	CODE:
		1866	{
		1867	if (self->incr_pos)
		1868	croak ("incr_text can not be called when the incremental parser already started parsing");
		1869
		1870	RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
		1871	}
		1872	OUTPUT:
		1873	RETVAL
		1874
		1875	void incr_skip (JSON *self)
		1876	CODE:
		1877	{
		1878	if (self->incr_pos)
		1879	{
		1880	sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
		1881	self->incr_pos = 0;
		1882	self->incr_nest = 0;
		1883	self->incr_mode = 0;
		1884	}
		1885	}
		1886
		1887	void incr_reset (JSON *self)
		1888	CODE:
		1889	{
		1890	SvREFCNT_dec (self->incr_text);
		1891	self->incr_text = 0;
		1892	self->incr_pos = 0;
		1893	self->incr_nest = 0;
		1894	self->incr_mode = 0;
1637	}	1895	}
1638		1896
1639	void DESTROY (JSON *self)	1897	void DESTROY (JSON *self)
1640	CODE:	1898	CODE:
1641	SvREFCNT_dec (self->cb_sk_object);	1899	SvREFCNT_dec (self->cb_sk_object);
1642	SvREFCNT_dec (self->cb_object);	1900	SvREFCNT_dec (self->cb_object);
		1901	SvREFCNT_dec (self->incr_text);
1643		1902
1644	PROTOTYPES: ENABLE	1903	PROTOTYPES: ENABLE
1645		1904
1646	void encode_json (SV *scalar)	1905	void encode_json (SV *scalar)
1647	ALIAS:	1906	ALIAS:
1648	to_json_ = 0	1907	to_json_ = 0
1649	encode_json = F_UTF8	1908	encode_json = F_UTF8
1650	PPCODE:	1909	PPCODE:
1651	{	1910	{
1652	JSON json = { F_DEFAULT \| ix };	1911	JSON json;
		1912	json_init (&json);
		1913	json.flags \|= ix;
1653	XPUSHs (encode_json (scalar, &json));	1914	XPUSHs (encode_json (scalar, &json));
1654	}	1915	}
1655		1916
1656	void decode_json (SV *jsonstr)	1917	void decode_json (SV *jsonstr)
1657	ALIAS:	1918	ALIAS:
1658	from_json_ = 0	1919	from_json_ = 0
1659	decode_json = F_UTF8	1920	decode_json = F_UTF8
1660	PPCODE:	1921	PPCODE:
1661	{	1922	{
1662	JSON json = { F_DEFAULT \| ix };	1923	JSON json;
		1924	json_init (&json);
		1925	json.flags \|= ix;
1663	XPUSHs (decode_json (jsonstr, &json, 0));	1926	XPUSHs (decode_json (jsonstr, &json, 0));
1664	}	1927	}
1665		1928
		1929

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.76 by root, Thu Mar 20 00:56:37 2008 UTC vs. Revision 1.99 by root, Sat Aug 8 10:06:02 2009 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.76 by root, Thu Mar 20 00:56:37 2008 UTC vs.
Revision 1.99 by root, Sat Aug 8 10:06:02 2009 UTC