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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.86 by root, Tue May 27 05:31:39 2008 UTC vs.
Revision 1.115 by root, Fri Dec 21 05:16:19 2012 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
…		…
41	#define INIT_SIZE 32 // initial scalar size to be allocated	42	#define INIT_SIZE 32 // initial scalar size to be allocated
42	#define INDENT_STEP 3 // spaces per indentation level	43	#define INDENT_STEP 3 // spaces per indentation level
43		44
44	#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
45		46
		47	#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
		48
46	#define SB do {	49	#define SB do {
47	#define SE } while (0)	50	#define SE } while (0)
48		51
49	#if __GNUC__ >= 3	52	#if __GNUC__ >= 3
50	# define expect(expr,value) __builtin_expect ((expr), (value))	53	# define expect(expr,value) __builtin_expect ((expr), (value))
…		…
69	#else	72	#else
70	# define JSON_SLOW 0	73	# define JSON_SLOW 0
71	# define JSON_STASH json_stash	74	# define JSON_STASH json_stash
72	#endif	75	#endif
73		76
		77	// the amount of HEs to allocate on the stack, when sorting keys
		78	#define STACK_HES 64
		79
74	static HV json_stash, json_boolean_stash; // JSON::XS::	80	static HV json_stash, json_boolean_stash; // JSON::XS::
75	static SV json_true, json_false;	81	static SV json_true, json_false;
76		82
77	enum {	83	enum {
78	INCR_M_WS = 0, // initial whitespace skipping, must be 0	84	INCR_M_WS = 0, // initial whitespace skipping, must be 0
79	INCR_M_STR, // inside string	85	INCR_M_STR, // inside string
80	INCR_M_BS, // inside backslash	86	INCR_M_BS, // inside backslash
		87	INCR_M_C0, // inside comment in initial whitespace sequence
		88	INCR_M_C1, // inside comment in other places
81	INCR_M_JSON // outside anything, count nesting	89	INCR_M_JSON // outside anything, count nesting
82	};	90	};
83		91
84	#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON)	92	#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85		93
86	typedef struct {	94	typedef struct {
87	U32 flags;	95	U32 flags;
88	U32 max_depth;	96	U32 max_depth;
89	STRLEN max_size;	97	STRLEN max_size;
…		…
92	HV *cb_sk_object;	100	HV *cb_sk_object;
93		101
94	// for the incremental parser	102	// for the incremental parser
95	SV *incr_text; // the source text so far	103	SV *incr_text; // the source text so far
96	STRLEN incr_pos; // the current offset into the text	104	STRLEN incr_pos; // the current offset into the text
97	unsigned char incr_nest; // {[]}-nesting level	105	int incr_nest; // {[]}-nesting level
98	unsigned char incr_mode;	106	unsigned char incr_mode;
99	} JSON;	107	} JSON;
100		108
101	INLINE void	109	INLINE void
102	json_init (JSON *json)	110	json_init (JSON *json)
…		…
121		129
122	INLINE void	130	INLINE void
123	shrink (SV *sv)	131	shrink (SV *sv)
124	{	132	{
125	sv_utf8_downgrade (sv, 1);	133	sv_utf8_downgrade (sv, 1);
		134
126	if (SvLEN (sv) > SvCUR (sv) + 1)	135	if (SvLEN (sv) > SvCUR (sv) + 1)
127	{	136	{
128	#ifdef SvPV_shrink_to_cur	137	#ifdef SvPV_shrink_to_cur
129	SvPV_shrink_to_cur (sv);	138	SvPV_shrink_to_cur (sv);
130	#elif defined (SvPV_renew)	139	#elif defined (SvPV_renew)
…		…
174	*s++ = 0x80 \| ( ch & 0x3f);	183	*s++ = 0x80 \| ( ch & 0x3f);
175		184
176	return s;	185	return s;
177	}	186	}
178		187
		188	// convert offset pointer to character index, sv must be string
		189	static STRLEN
		190	ptr_to_index (SV sv, char offset)
		191	{
		192	return SvUTF8 (sv)
		193	? utf8_distance (offset, SvPVX (sv))
		194	: offset - SvPVX (sv);
		195	}
		196
		197	/////////////////////////////////////////////////////////////////////////////
		198	// fp hell
		199
		200	// scan a group of digits, and a trailing exponent
		201	static void
		202	json_atof_scan1 (const char s, NV accum, int *expo, int postdp, int maxdepth)
		203	{
		204	UV uaccum = 0;
		205	int eaccum = 0;
		206
		207	// if we recurse too deep, skip all remaining digits
		208	// to avoid a stack overflow attack
		209	if (expect_false (--maxdepth <= 0))
		210	while (((U8)*s - '0') < 10)
		211	++s;
		212
		213	for (;;)
		214	{
		215	U8 dig = (U8)*s - '0';
		216
		217	if (expect_false (dig >= 10))
		218	{
		219	if (dig == (U8)((U8)'.' - (U8)'0'))
		220	{
		221	++s;
		222	json_atof_scan1 (s, accum, expo, 1, maxdepth);
		223	}
		224	else if ((dig \| ' ') == 'e' - '0')
		225	{
		226	int exp2 = 0;
		227	int neg = 0;
		228
		229	++s;
		230
		231	if (*s == '-')
		232	{
		233	++s;
		234	neg = 1;
		235	}
		236	else if (*s == '+')
		237	++s;
		238
		239	while ((dig = (U8)*s - '0') < 10)
		240	exp2 = exp2 * 10 + *s++ - '0';
		241
		242	*expo += neg ? -exp2 : exp2;
		243	}
		244
		245	break;
		246	}
		247
		248	++s;
		249
		250	uaccum = uaccum * 10 + dig;
		251	++eaccum;
		252
		253	// if we have too many digits, then recurse for more
		254	// we actually do this for rather few digits
		255	if (uaccum >= (UV_MAX - 9) / 10)
		256	{
		257	if (postdp) *expo -= eaccum;
		258	json_atof_scan1 (s, accum, expo, postdp, maxdepth);
		259	if (postdp) *expo += eaccum;
		260
		261	break;
		262	}
		263	}
		264
		265	// this relies greatly on the quality of the pow ()
		266	// implementation of the platform, but a good
		267	// implementation is hard to beat.
		268	// (IEEE 754 conformant ones are required to be exact)
		269	if (postdp) *expo -= eaccum;
		270	accum += uaccum Perl_pow (10., *expo);
		271	*expo += eaccum;
		272	}
		273
		274	static NV
		275	json_atof (const char *s)
		276	{
		277	NV accum = 0.;
		278	int expo = 0;
		279	int neg = 0;
		280
		281	if (*s == '-')
		282	{
		283	++s;
		284	neg = 1;
		285	}
		286
		287	// a recursion depth of ten gives us >>500 bits
		288	json_atof_scan1 (s, &accum, &expo, 0, 10);
		289
		290	return neg ? -accum : accum;
		291	}
179	/////////////////////////////////////////////////////////////////////////////	292	/////////////////////////////////////////////////////////////////////////////
180	// encoder	293	// encoder
181		294
182	// structure used for encoding JSON	295	// structure used for encoding JSON
183	typedef struct	296	typedef struct
…		…
193	INLINE void	306	INLINE void
194	need (enc_t *enc, STRLEN len)	307	need (enc_t *enc, STRLEN len)
195	{	308	{
196	if (expect_false (enc->cur + len >= enc->end))	309	if (expect_false (enc->cur + len >= enc->end))
197	{	310	{
198	STRLEN cur = enc->cur - SvPVX (enc->sv);	311	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
199	SvGROW (enc->sv, cur + len + 1);	312	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
200	enc->cur = SvPVX (enc->sv) + cur;	313	enc->cur = SvPVX (enc->sv) + cur;
201	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	314	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
202	}	315	}
203	}	316	}
204		317
…		…
279	(int)((uch - 0x10000) % 0x400 + 0xDC00));	392	(int)((uch - 0x10000) % 0x400 + 0xDC00));
280	enc->cur += 12;	393	enc->cur += 12;
281	}	394	}
282	else	395	else
283	{	396	{
284	static char hexdigit [16] = "0123456789abcdef";
285	need (enc, len += 5);	397	need (enc, len += 5);
286	*enc->cur++ = '\\';	398	*enc->cur++ = '\\';
287	*enc->cur++ = 'u';	399	*enc->cur++ = 'u';
288	*enc->cur++ = hexdigit [ uch >> 12 ];	400	*enc->cur++ = PL_hexdigit [ uch >> 12 ];
289	*enc->cur++ = hexdigit [(uch >> 8) & 15];	401	*enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
290	*enc->cur++ = hexdigit [(uch >> 4) & 15];	402	*enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
291	*enc->cur++ = hexdigit [(uch >> 0) & 15];	403	*enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
292	}	404	}
293		405
294	str += clen;	406	str += clen;
295	}	407	}
296	else if (enc->json.flags & F_LATIN1)	408	else if (enc->json.flags & F_LATIN1)
…		…
460		572
461	encode_ch (enc, '{');	573	encode_ch (enc, '{');
462		574
463	// for canonical output we have to sort by keys first	575	// for canonical output we have to sort by keys first
464	// actually, this is mostly due to the stupid so-called	576	// actually, this is mostly due to the stupid so-called
465	// security workaround added somewhere in 5.8.x.	577	// security workaround added somewhere in 5.8.x
466	// that randomises hash orderings	578	// that randomises hash orderings
467	if (enc->json.flags & F_CANONICAL)	579	if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
468	{	580	{
469	int count = hv_iterinit (hv);	581	int count = hv_iterinit (hv);
470		582
471	if (SvMAGICAL (hv))	583	if (SvMAGICAL (hv))
472	{	584	{
…		…
483	}	595	}
484		596
485	if (count)	597	if (count)
486	{	598	{
487	int i, fast = 1;	599	int i, fast = 1;
488	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	600	HE *hes_stack [STACK_HES];
489	HE *hes = _alloca (count sizeof (HE));	601	HE **hes = hes_stack;
490	#else	602
491	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode	603	// allocate larger arrays on the heap
492	#endif	604	if (count > STACK_HES)
		605	{
		606	SV sv = sv_2mortal (NEWSV (0, count sizeof (*hes)));
		607	hes = (HE **)SvPVX (sv);
		608	}
493		609
494	i = 0;	610	i = 0;
495	while ((he = hv_iternext (hv)))	611	while ((he = hv_iternext (hv)))
496	{	612	{
497	hes [i++] = he;	613	hes [i++] = he;
…		…
716	}	832	}
717	else	833	else
718	{	834	{
719	// large integer, use the (rather slow) snprintf way.	835	// large integer, use the (rather slow) snprintf way.
720	need (enc, IVUV_MAXCHARS);	836	need (enc, IVUV_MAXCHARS);
721	enc->cur +=	837	enc->cur +=
722	SvIsUV(sv)	838	SvIsUV(sv)
723	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))	839	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
724	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));	840	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
725	}	841	}
726	}	842	}
…		…
728	encode_rv (enc, SvRV (sv));	844	encode_rv (enc, SvRV (sv));
729	else if (!SvOK (sv) \|\| enc->json.flags & F_ALLOW_UNKNOWN)	845	else if (!SvOK (sv) \|\| enc->json.flags & F_ALLOW_UNKNOWN)
730	encode_str (enc, "null", 4, 0);	846	encode_str (enc, "null", 4, 0);
731	else	847	else
732	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	848	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
733	SvPV_nolen (sv), SvFLAGS (sv));	849	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
734	}	850	}
735		851
736	static SV *	852	static SV *
737	encode_json (SV scalar, JSON json)	853	encode_json (SV scalar, JSON json)
738	{	854	{
…		…
750	: enc.json.flags & F_LATIN1 ? 0x000100UL	866	: enc.json.flags & F_LATIN1 ? 0x000100UL
751	: 0x110000UL;	867	: 0x110000UL;
752		868
753	SvPOK_only (enc.sv);	869	SvPOK_only (enc.sv);
754	encode_sv (&enc, scalar);	870	encode_sv (&enc, scalar);
		871	encode_nl (&enc);
755		872
756	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	873	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
757	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	874	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
758		875
759	if (!(enc.json.flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))	876	if (!(enc.json.flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
…		…
939	else if (expect_true (ch >= 0x20 && ch < 0x80))	1056	else if (expect_true (ch >= 0x20 && ch < 0x80))
940	*cur++ = ch;	1057	*cur++ = ch;
941	else if (ch >= 0x80)	1058	else if (ch >= 0x80)
942	{	1059	{
943	STRLEN clen;	1060	STRLEN clen;
944	UV uch;
945		1061
946	--dec_cur;	1062	--dec_cur;
947		1063
948	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);	1064	decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
949	if (clen == (STRLEN)-1)	1065	if (clen == (STRLEN)-1)
950	ERR ("malformed UTF-8 character in JSON string");	1066	ERR ("malformed UTF-8 character in JSON string");
951		1067
952	do	1068	do
953	cur++ = dec_cur++;	1069	cur++ = dec_cur++;
…		…
970	{	1086	{
971	STRLEN len = cur - buf;	1087	STRLEN len = cur - buf;
972		1088
973	if (sv)	1089	if (sv)
974	{	1090	{
975	SvGROW (sv, SvCUR (sv) + len + 1);	1091	STRLEN cur = SvCUR (sv);
		1092
		1093	if (SvLEN (sv) <= cur + len)
		1094	SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
		1095
976	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	1096	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
977	SvCUR_set (sv, SvCUR (sv) + len);	1097	SvCUR_set (sv, SvCUR (sv) + len);
978	}	1098	}
979	else	1099	else
980	sv = newSVpvn (buf, len);	1100	sv = newSVpvn (buf, len);
…		…
1071		1191
1072	// special case the rather common 1..5-digit-int case	1192	// special case the rather common 1..5-digit-int case
1073	if (*start == '-')	1193	if (*start == '-')
1074	switch (len)	1194	switch (len)
1075	{	1195	{
1076	case 2: return newSViv (-( start [1] - '0' * 1));	1196	case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1077	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));	1197	case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1078	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));	1198	case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1079	case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));	1199	case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1080	case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));	1200	case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1081	}	1201	}
1082	else	1202	else
1083	switch (len)	1203	switch (len)
1084	{	1204	{
1085	case 1: return newSViv ( start [0] - '0' * 1);	1205	case 1: return newSViv ( start [0] - '0' * 1);
1086	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);	1206	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1087	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);	1207	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1088	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);	1208	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1089	case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);	1209	case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1090	}	1210	}
1091		1211
1092	{	1212	{
1093	UV uv;	1213	UV uv;
1094	int numtype = grok_number (start, len, &uv);	1214	int numtype = grok_number (start, len, &uv);
…		…
1103	}	1223	}
1104		1224
1105	len -= *start == '-' ? 1 : 0;	1225	len -= *start == '-' ? 1 : 0;
1106		1226
1107	// does not fit into IV or UV, try NV	1227	// does not fit into IV or UV, try NV
1108	if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)	1228	if (len <= NV_DIG)
1109	#if defined (LDBL_DIG)
1110	\|\| (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1111	#endif
1112	)
1113	// fits into NV without loss of precision	1229	// fits into NV without loss of precision
1114	return newSVnv (Atof (start));	1230	return newSVnv (json_atof (start));
1115		1231
1116	// everything else fails, convert it to a string	1232	// everything else fails, convert it to a string
1117	return newSVpvn (start, dec->cur - start);	1233	return newSVpvn (start, dec->cur - start);
1118	}	1234	}
1119		1235
1120	// loss of precision here	1236	// loss of precision here
1121	return newSVnv (Atof (start));	1237	return newSVnv (json_atof (start));
1122		1238
1123	fail:	1239	fail:
1124	return 0;	1240	return 0;
1125	}	1241	}
1126		1242
…		…
1295	dSP;	1411	dSP;
1296	int count;	1412	int count;
1297		1413
1298	ENTER; SAVETMPS; PUSHMARK (SP);	1414	ENTER; SAVETMPS; PUSHMARK (SP);
1299	XPUSHs (HeVAL (he));	1415	XPUSHs (HeVAL (he));
		1416	sv_2mortal (sv);
1300		1417
1301	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;	1418	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1302		1419
1303	if (count == 1)	1420	if (count == 1)
1304	{	1421	{
1305	sv = newSVsv (POPs);	1422	sv = newSVsv (POPs);
1306	FREETMPS; LEAVE;	1423	FREETMPS; LEAVE;
1307	return sv;	1424	return sv;
1308	}	1425	}
1309		1426
		1427	SvREFCNT_inc (sv);
1310	FREETMPS; LEAVE;	1428	FREETMPS; LEAVE;
1311	}	1429	}
1312	}	1430	}
1313		1431
1314	if (dec->json.cb_object)	1432	if (dec->json.cb_object)
…		…
1404	fail:	1522	fail:
1405	return 0;	1523	return 0;
1406	}	1524	}
1407		1525
1408	static SV *	1526	static SV *
1409	decode_json (SV string, JSON json, STRLEN *offset_return)	1527	decode_json (SV string, JSON json, char **offset_return)
1410	{	1528	{
1411	dec_t dec;	1529	dec_t dec;
1412	STRLEN offset;
1413	SV *sv;	1530	SV *sv;
1414		1531
		1532	/* work around bugs in 5.10 where manipulating magic values
		1533	* will perl ignore the magic in subsequent accesses.
		1534	* also make a copy of non-PV values, to get them into a clean
		1535	* state (SvPV should do that, but it's buggy, see below).
		1536	*/
1415	SvGETMAGIC (string);	1537	/SvGETMAGIC (string);/
		1538	if (SvMAGICAL (string) \|\| !SvPOK (string))
		1539	string = sv_2mortal (newSVsv (string));
		1540
1416	SvUPGRADE (string, SVt_PV);	1541	SvUPGRADE (string, SVt_PV);
1417		1542
		1543	/* work around a bug in perl 5.10, which causes SvCUR to fail an
		1544	* assertion with -DDEBUGGING, although SvCUR is documented to
		1545	* return the xpv_cur field which certainly exists after upgrading.
		1546	* according to nicholas clark, calling SvPOK fixes this.
		1547	* But it doesn't fix it, so try another workaround, call SvPV_nolen
		1548	* and hope for the best.
		1549	* Damnit, SvPV_nolen still trips over yet another assertion. This
		1550	* assertion business is seriously broken, try yet another workaround
		1551	* for the broken -DDEBUGGING.
		1552	*/
		1553	{
		1554	#ifdef DEBUGGING
		1555	STRLEN offset = SvOK (string) ? sv_len (string) : 0;
		1556	#else
		1557	STRLEN offset = SvCUR (string);
		1558	#endif
		1559
1418	if (SvCUR (string) > json->max_size && json->max_size)	1560	if (offset > json->max_size && json->max_size)
1419	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",	1561	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1420	(unsigned long)SvCUR (string), (unsigned long)json->max_size);	1562	(unsigned long)SvCUR (string), (unsigned long)json->max_size);
		1563	}
1421		1564
1422	if (json->flags & F_UTF8)	1565	if (DECODE_WANTS_OCTETS (json))
1423	sv_utf8_downgrade (string, 0);	1566	sv_utf8_downgrade (string, 0);
1424	else	1567	else
1425	sv_utf8_upgrade (string);	1568	sv_utf8_upgrade (string);
1426		1569
1427	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1570	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
…		…
1437		1580
1438	*dec.end = 0; // this should basically be a nop, too, but make sure it's there	1581	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
1439		1582
1440	decode_ws (&dec);	1583	decode_ws (&dec);
1441	sv = decode_sv (&dec);	1584	sv = decode_sv (&dec);
		1585
		1586	if (offset_return)
		1587	*offset_return = dec.cur;
1442		1588
1443	if (!(offset_return \|\| !sv))	1589	if (!(offset_return \|\| !sv))
1444	{	1590	{
1445	// check for trailing garbage	1591	// check for trailing garbage
1446	decode_ws (&dec);	1592	decode_ws (&dec);
…		…
1449	{	1595	{
1450	dec.err = "garbage after JSON object";	1596	dec.err = "garbage after JSON object";
1451	SvREFCNT_dec (sv);	1597	SvREFCNT_dec (sv);
1452	sv = 0;	1598	sv = 0;
1453	}	1599	}
1454	}
1455
1456	if (offset_return \|\| !sv)
1457	{
1458	offset = dec.json.flags & F_UTF8
1459	? dec.cur - SvPVX (string)
1460	: utf8_distance (dec.cur, SvPVX (string));
1461
1462	if (offset_return)
1463	*offset_return = offset;
1464	}	1600	}
1465		1601
1466	if (!sv)	1602	if (!sv)
1467	{	1603	{
1468	SV *uni = sv_newmortal ();	1604	SV *uni = sv_newmortal ();
…		…
1474	SAVEVPTR (PL_curcop);	1610	SAVEVPTR (PL_curcop);
1475	PL_curcop = &cop;	1611	PL_curcop = &cop;
1476	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1612	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1477	LEAVE;	1613	LEAVE;
1478		1614
1479	croak ("%s, at character offset %d [\"%s\"]",	1615	croak ("%s, at character offset %d (before \"%s\")",
1480	dec.err,	1616	dec.err,
1481	(int)offset,	1617	(int)ptr_to_index (string, dec.cur),
1482	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1618	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1483	}	1619	}
1484		1620
1485	sv = sv_2mortal (sv);	1621	sv = sv_2mortal (sv);
1486		1622
…		…
1495		1631
1496	static void	1632	static void
1497	incr_parse (JSON *self)	1633	incr_parse (JSON *self)
1498	{	1634	{
1499	const char *p = SvPVX (self->incr_text) + self->incr_pos;	1635	const char *p = SvPVX (self->incr_text) + self->incr_pos;
		1636
		1637	// the state machine here is a bit convoluted and could be simplified a lot
		1638	// but this would make it slower, so...
1500		1639
1501	for (;;)	1640	for (;;)
1502	{	1641	{
1503	//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	1642	//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
1504	switch (self->incr_mode)	1643	switch (self->incr_mode)
1505	{	1644	{
1506	// only used for intiial whitespace skipping	1645	// only used for initial whitespace skipping
1507	case INCR_M_WS:	1646	case INCR_M_WS:
1508	for (;;)	1647	for (;;)
1509	{	1648	{
1510	if (*p > 0x20)	1649	if (*p > 0x20)
1511	{	1650	{
		1651	if (*p == '#')
		1652	{
		1653	self->incr_mode = INCR_M_C0;
		1654	goto incr_m_c;
		1655	}
		1656	else
		1657	{
1512	self->incr_mode = INCR_M_JSON;	1658	self->incr_mode = INCR_M_JSON;
1513	goto incr_m_json;	1659	goto incr_m_json;
		1660	}
1514	}	1661	}
1515	else if (!*p)	1662	else if (!*p)
1516	goto interrupt;	1663	goto interrupt;
1517		1664
1518	++p;	1665	++p;
…		…
1524	goto interrupt;	1671	goto interrupt;
1525		1672
1526	++p;	1673	++p;
1527	self->incr_mode = INCR_M_STR;	1674	self->incr_mode = INCR_M_STR;
1528	goto incr_m_str;	1675	goto incr_m_str;
		1676
		1677	// inside #-style comments
		1678	case INCR_M_C0:
		1679	case INCR_M_C1:
		1680	incr_m_c:
		1681	for (;;)
		1682	{
		1683	if (*p == '\n')
		1684	{
		1685	self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
		1686	break;
		1687	}
		1688	else if (!*p)
		1689	goto interrupt;
		1690
		1691	++p;
		1692	}
		1693
		1694	break;
1529		1695
1530	// inside a string	1696	// inside a string
1531	case INCR_M_STR:	1697	case INCR_M_STR:
1532	incr_m_str:	1698	incr_m_str:
1533	for (;;)	1699	for (;;)
…		…
1590	croak (ERR_NESTING_EXCEEDED);	1756	croak (ERR_NESTING_EXCEEDED);
1591	break;	1757	break;
1592		1758
1593	case ']':	1759	case ']':
1594	case '}':	1760	case '}':
1595	if (!--self->incr_nest)	1761	if (--self->incr_nest <= 0)
1596	goto interrupt;	1762	goto interrupt;
		1763	break;
		1764
		1765	case '#':
		1766	self->incr_mode = INCR_M_C1;
		1767	goto incr_m_c;
1597	}	1768	}
1598	}	1769	}
1599	}	1770	}
1600		1771
1601	modechange:	1772	modechange:
1602	;	1773	;
1603	}	1774	}
1604		1775
1605	interrupt:	1776	interrupt:
1606	self->incr_pos = p - SvPVX (self->incr_text);	1777	self->incr_pos = p - SvPVX (self->incr_text);
		1778	//printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1607	//printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D	1779	//printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1608	}	1780	}
1609		1781
1610	/////////////////////////////////////////////////////////////////////////////	1782	/////////////////////////////////////////////////////////////////////////////
1611	// XS interface functions	1783	// XS interface functions
…		…
1626	json_stash = gv_stashpv ("JSON::XS" , 1);	1798	json_stash = gv_stashpv ("JSON::XS" , 1);
1627	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);	1799	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1628		1800
1629	json_true = get_bool ("JSON::XS::true");	1801	json_true = get_bool ("JSON::XS::true");
1630	json_false = get_bool ("JSON::XS::false");	1802	json_false = get_bool ("JSON::XS::false");
		1803
		1804	CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1631	}	1805	}
1632		1806
1633	PROTOTYPES: DISABLE	1807	PROTOTYPES: DISABLE
1634		1808
1635	void CLONE (...)	1809	void CLONE (...)
…		…
1638	json_boolean_stash = 0;	1812	json_boolean_stash = 0;
1639		1813
1640	void new (char *klass)	1814	void new (char *klass)
1641	PPCODE:	1815	PPCODE:
1642	{	1816	{
1643	SV *pv = NEWSV (0, sizeof (JSON));	1817	SV *pv = NEWSV (0, sizeof (JSON));
1644	SvPOK_only (pv);	1818	SvPOK_only (pv);
1645	json_init ((JSON *)SvPVX (pv));	1819	json_init ((JSON *)SvPVX (pv));
1646	XPUSHs (sv_2mortal (sv_bless (	1820	XPUSHs (sv_2mortal (sv_bless (
1647	newRV_noinc (pv),	1821	newRV_noinc (pv),
1648	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)	1822	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
…		…
1725	}	1899	}
1726		1900
1727	void filter_json_single_key_object (JSON self, SV key, SV *cb = &PL_sv_undef)	1901	void filter_json_single_key_object (JSON self, SV key, SV *cb = &PL_sv_undef)
1728	PPCODE:	1902	PPCODE:
1729	{	1903	{
1730	if (!self->cb_sk_object)	1904	if (!self->cb_sk_object)
1731	self->cb_sk_object = newHV ();	1905	self->cb_sk_object = newHV ();
1732		1906
1733	if (SvOK (cb))	1907	if (SvOK (cb))
1734	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);	1908	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1735	else	1909	else
…		…
1746	XPUSHs (ST (0));	1920	XPUSHs (ST (0));
1747	}	1921	}
1748		1922
1749	void encode (JSON self, SV scalar)	1923	void encode (JSON self, SV scalar)
1750	PPCODE:	1924	PPCODE:
1751	XPUSHs (encode_json (scalar, self));	1925	PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
		1926	XPUSHs (scalar);
1752		1927
1753	void decode (JSON self, SV jsonstr)	1928	void decode (JSON self, SV jsonstr)
1754	PPCODE:	1929	PPCODE:
1755	XPUSHs (decode_json (jsonstr, self, 0));	1930	PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
		1931	XPUSHs (jsonstr);
1756		1932
1757	void decode_prefix (JSON self, SV jsonstr)	1933	void decode_prefix (JSON self, SV jsonstr)
1758	PPCODE:	1934	PPCODE:
1759	{	1935	{
1760	STRLEN offset;	1936	SV *sv;
		1937	char *offset;
		1938	PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1761	EXTEND (SP, 2);	1939	EXTEND (SP, 2);
1762	PUSHs (decode_json (jsonstr, self, &offset));	1940	PUSHs (sv);
1763	PUSHs (sv_2mortal (newSVuv (offset)));	1941	PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1764	}	1942	}
1765		1943
1766	void incr_parse (JSON self, SV jsonstr = 0)	1944	void incr_parse (JSON self, SV jsonstr = 0)
1767	PPCODE:	1945	PPCODE:
1768	{	1946	{
1769	if (!self->incr_text)	1947	if (!self->incr_text)
1770	self->incr_text = newSVpvn ("", 0);	1948	self->incr_text = newSVpvn ("", 0);
		1949
		1950	/* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
		1951	if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
		1952	if (DECODE_WANTS_OCTETS (self))
		1953	{
		1954	if (self->incr_pos)
		1955	self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
		1956	(U8 *)SvPVX (self->incr_text) + self->incr_pos);
		1957
		1958	sv_utf8_downgrade (self->incr_text, 0);
		1959	}
		1960	else
		1961	{
		1962	sv_utf8_upgrade (self->incr_text);
		1963
		1964	if (self->incr_pos)
		1965	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
		1966	- (U8 *)SvPVX (self->incr_text);
		1967	}
1771		1968
1772	// append data, if any	1969	// append data, if any
1773	if (jsonstr)	1970	if (jsonstr)
1774	{	1971	{
		1972	/* make sure both strings have same encoding */
1775	if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))	1973	if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
		1974	if (SvUTF8 (jsonstr))
		1975	sv_utf8_downgrade (jsonstr, 0);
1776	{	1976	else
1777	/* utf-8-ness differs, need to upgrade */
1778	sv_utf8_upgrade (self->incr_text);	1977	sv_utf8_upgrade (jsonstr);
1779		1978
1780	if (self->incr_pos)	1979	/* and then just blindly append */
1781	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1782	- (U8 *)SvPVX (self->incr_text);
1783	}
1784
1785	{	1980	{
1786	STRLEN len;	1981	STRLEN len;
1787	const char *str = SvPV (jsonstr, len);	1982	const char *str = SvPV (jsonstr, len);
1788	SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);	1983	STRLEN cur = SvCUR (self->incr_text);
		1984
		1985	if (SvLEN (self->incr_text) <= cur + len)
		1986	SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
		1987
1789	Move (str, SvEND (self->incr_text), len, char);	1988	Move (str, SvEND (self->incr_text), len, char);
1790	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);	1989	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1791	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there	1990	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1792	}	1991	}
1793	}	1992	}
1794		1993
1795	if (GIMME_V != G_VOID)	1994	if (GIMME_V != G_VOID)
1796	do	1995	do
1797	{	1996	{
		1997	SV *sv;
1798	STRLEN offset;	1998	char *offset;
1799		1999
1800	if (!INCR_DONE (self))	2000	if (!INCR_DONE (self))
1801	{	2001	{
1802	incr_parse (self);	2002	incr_parse (self);
1803		2003
1804	if (self->incr_pos > self->max_size && self->max_size)	2004	if (self->incr_pos > self->max_size && self->max_size)
1805	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",	2005	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1806	(unsigned long)self->incr_pos, (unsigned long)self->max_size);	2006	(unsigned long)self->incr_pos, (unsigned long)self->max_size);
1807		2007
1808	if (!INCR_DONE (self))	2008	if (!INCR_DONE (self))
		2009	{
		2010	// as an optimisation, do not accumulate white space in the incr buffer
		2011	if (self->incr_mode == INCR_M_WS && self->incr_pos)
		2012	{
		2013	self->incr_pos = 0;
		2014	SvCUR_set (self->incr_text, 0);
		2015	}
		2016
1809	break;	2017	break;
		2018	}
1810	}	2019	}
1811		2020
1812	XPUSHs (decode_json (self->incr_text, self, &offset));	2021	PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
		2022	XPUSHs (sv);
1813		2023
1814	sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1815	self->incr_pos -= offset;	2024	self->incr_pos -= offset - SvPVX (self->incr_text);
1816	self->incr_nest = 0;	2025	self->incr_nest = 0;
1817	self->incr_mode = 0;	2026	self->incr_mode = 0;
		2027
		2028	sv_chop (self->incr_text, offset);
1818	}	2029	}
1819	while (GIMME_V == G_ARRAY);	2030	while (GIMME_V == G_ARRAY);
1820	}	2031	}
1821		2032
1822	SV incr_text (JSON self)	2033	SV incr_text (JSON self)
…		…
1868	PPCODE:	2079	PPCODE:
1869	{	2080	{
1870	JSON json;	2081	JSON json;
1871	json_init (&json);	2082	json_init (&json);
1872	json.flags \|= ix;	2083	json.flags \|= ix;
1873	XPUSHs (encode_json (scalar, &json));	2084	PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
		2085	XPUSHs (scalar);
1874	}	2086	}
1875		2087
1876	void decode_json (SV *jsonstr)	2088	void decode_json (SV *jsonstr)
1877	ALIAS:	2089	ALIAS:
1878	from_json_ = 0	2090	from_json_ = 0
…		…
1880	PPCODE:	2092	PPCODE:
1881	{	2093	{
1882	JSON json;	2094	JSON json;
1883	json_init (&json);	2095	json_init (&json);
1884	json.flags \|= ix;	2096	json.flags \|= ix;
1885	XPUSHs (decode_json (jsonstr, &json, 0));	2097	PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
		2098	XPUSHs (jsonstr);
1886	}	2099	}
1887		2100
1888

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.86 by root, Tue May 27 05:31:39 2008 UTC vs. Revision 1.115 by root, Fri Dec 21 05:16:19 2012 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.86 by root, Tue May 27 05:31:39 2008 UTC vs.
Revision 1.115 by root, Fri Dec 21 05:16:19 2012 UTC