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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.83 by root, Wed Mar 26 22:54:38 2008 UTC vs.
Revision 1.144 by root, Wed Nov 9 11:47:12 2022 UTC

…		…
6	#include <string.h>	6	#include <string.h>
7	#include <stdlib.h>	7	#include <stdlib.h>
8	#include <stdio.h>	8	#include <stdio.h>
9	#include <limits.h>	9	#include <limits.h>
10	#include <float.h>	10	#include <float.h>
		11	#include <inttypes.h>
11		12
12	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	13	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)
13	# define snprintf _snprintf // C compilers have this in stdio.h	14	# define snprintf _snprintf // C compilers have this in stdio.h
14	#endif	15	#endif
15		16
16	// some old perls do not have this, try to make it work, no	17	// 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.	18	// guarantees, though. if it breaks, you get to keep the pieces.
18	#ifndef UTF8_MAXBYTES	19	#ifndef UTF8_MAXBYTES
19	# define UTF8_MAXBYTES 13	20	# define UTF8_MAXBYTES 13
20	#endif	21	#endif
21		22
		23	// compatibility with perl <5.18
		24	#ifndef HvNAMELEN_get
		25	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
		26	#endif
		27	#ifndef HvNAMELEN
		28	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
		29	#endif
		30	#ifndef HvNAMEUTF8
		31	# define HvNAMEUTF8(hv) 0
		32	#endif
		33
		34	// three extra for rounding, sign, and end of string
22	#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)	35	#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
23		36
24	#define F_ASCII 0x00000001UL	37	#define F_ASCII 0x00000001UL
25	#define F_LATIN1 0x00000002UL	38	#define F_LATIN1 0x00000002UL
26	#define F_UTF8 0x00000004UL	39	#define F_UTF8 0x00000004UL
27	#define F_INDENT 0x00000008UL	40	#define F_INDENT 0x00000008UL
…		…
31	#define F_ALLOW_NONREF 0x00000100UL	44	#define F_ALLOW_NONREF 0x00000100UL
32	#define F_SHRINK 0x00000200UL	45	#define F_SHRINK 0x00000200UL
33	#define F_ALLOW_BLESSED 0x00000400UL	46	#define F_ALLOW_BLESSED 0x00000400UL
34	#define F_CONV_BLESSED 0x00000800UL	47	#define F_CONV_BLESSED 0x00000800UL
35	#define F_RELAXED 0x00001000UL	48	#define F_RELAXED 0x00001000UL
36		49	#define F_ALLOW_UNKNOWN 0x00002000UL
37	#define F_MAXDEPTH 0xf8000000UL	50	#define F_ALLOW_TAGS 0x00004000UL
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	51	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
42		52
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	53	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
47	#define F_DEFAULT (9UL << S_MAXDEPTH)
48		54
49	#define INIT_SIZE 32 // initial scalar size to be allocated	55	#define INIT_SIZE 64 // initial scalar size to be allocated
50	#define INDENT_STEP 3 // spaces per indentation level	56	#define INDENT_STEP 3 // spaces per indentation level
51		57
52	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size	58	#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
		59
		60	#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
53		61
54	#define SB do {	62	#define SB do {
55	#define SE } while (0)	63	#define SE } while (0)
56		64
57	#if __GNUC__ >= 3	65	#if __GNUC__ >= 3
…		…
67		75
68	#define IN_RANGE_INC(type,val,beg,end) \	76	#define IN_RANGE_INC(type,val,beg,end) \
69	((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \	77	((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
70	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))	78	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
71		79
		80	#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
		81
72	#ifdef USE_ITHREADS	82	#ifdef USE_ITHREADS
73	# define JSON_SLOW 1
74	# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))	83	# define JSON_STASH (expect_true (json_stash) ? json_stash : gv_stashpv ("JSON::XS", 1))
		84	# define BOOL_STASH (expect_true (bool_stash) ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1))
		85	# define GET_BOOL(value) (expect_true (bool_ ## value) ? bool_ ## value : get_bool ("Types::Serialiser::" # value))
75	#else	86	#else
76	# define JSON_SLOW 0
77	# define JSON_STASH json_stash	87	# define JSON_STASH json_stash
		88	# define BOOL_STASH bool_stash
		89	# define GET_BOOL(value) bool_ ## value
78	#endif	90	#endif
79		91
80	static HV json_stash, json_boolean_stash; // JSON::XS::	92	// the amount of HEs to allocate on the stack, when sorting keys
81	static SV json_true, json_false;	93	#define STACK_HES 64
		94
		95	static HV json_stash, bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
		96	static SV bool_false, bool_true;
		97	static SV *sv_json;
82		98
83	enum {	99	enum {
84	INCR_M_WS = 0, // initial whitespace skipping, must be 0	100	INCR_M_WS = 0, // initial whitespace skipping, must be 0
		101	INCR_M_TFN, // inside true/false/null
		102	INCR_M_NUM, // inside number
85	INCR_M_STR, // inside string	103	INCR_M_STR, // inside string
86	INCR_M_BS, // inside backslash	104	INCR_M_BS, // inside backslash
		105	INCR_M_C0, // inside comment in initial whitespace sequence
		106	INCR_M_C1, // inside comment in other places
87	INCR_M_JSON // outside anything, count nesting	107	INCR_M_JSON // outside anything, count nesting
88	};	108	};
89		109
90	#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON)	110	#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
91		111
92	typedef struct {	112	typedef struct {
93	U32 flags;	113	U32 flags;
		114	U32 max_depth;
		115	STRLEN max_size;
		116
94	SV *cb_object;	117	SV *cb_object;
95	HV *cb_sk_object;	118	HV *cb_sk_object;
96		119
97	// for the incremental parser	120	// for the incremental parser
98	SV *incr_text; // the source text so far	121	SV *incr_text; // the source text so far
99	STRLEN incr_pos; // the current offset into the text	122	STRLEN incr_pos; // the current offset into the text
100	int incr_nest; // {[]}-nesting level	123	int incr_nest; // {[]}-nesting level
101	int incr_mode;	124	unsigned char incr_mode;
		125
		126	SV v_false, v_true;
102	} JSON;	127	} JSON;
		128
		129	INLINE void
		130	json_init (JSON *json)
		131	{
		132	static const JSON init = { F_ALLOW_NONREF, 512 };
		133
		134	*json = init;
		135	}
103		136
104	/////////////////////////////////////////////////////////////////////////////	137	/////////////////////////////////////////////////////////////////////////////
105	// utility functions	138	// utility functions
106		139
107	INLINE SV *	140	INLINE SV *
…		…
117		150
118	INLINE void	151	INLINE void
119	shrink (SV *sv)	152	shrink (SV *sv)
120	{	153	{
121	sv_utf8_downgrade (sv, 1);	154	sv_utf8_downgrade (sv, 1);
		155
122	if (SvLEN (sv) > SvCUR (sv) + 1)	156	if (SvLEN (sv) > SvCUR (sv) + 1)
123	{	157	{
124	#ifdef SvPV_shrink_to_cur	158	#ifdef SvPV_shrink_to_cur
125	SvPV_shrink_to_cur (sv);	159	SvPV_shrink_to_cur (sv);
126	#elif defined (SvPV_renew)	160	#elif defined (SvPV_renew)
127	SvPV_renew (sv, SvCUR (sv) + 1);	161	SvPV_renew (sv, SvCUR (sv) + 1);
128	#endif	162	#endif
129	}	163	}
130	}	164	}
131		165
		166	/* adds two STRLENs together, slow, and with paranoia */
		167	static STRLEN
		168	strlen_sum (STRLEN l1, STRLEN l2)
		169	{
		170	size_t sum = l1 + l2;
		171
		172	if (sum < (size_t)l2 \|\| sum != (size_t)(STRLEN)sum)
		173	croak ("JSON::XS: string size overflow");
		174
		175	return sum;
		176	}
		177
		178	/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
		179	static char *
		180	json_sv_grow (SV *sv, size_t len1, size_t len2)
		181	{
		182	len1 = strlen_sum (len1, len2);
		183	len1 = strlen_sum (len1, len1 >> 1);
		184
		185	if (len1 > 4096 - 24)
		186	len1 = (len1 \| 4095) - 24;
		187
		188	return SvGROW (sv, len1);
		189	}
		190
132	// decode an utf-8 character and return it, or (UV)-1 in	191	// decode a utf-8 character and return it, or (UV)-1 in
133	// case of an error.	192	// case of an error.
134	// we special-case "safe" characters from U+80 .. U+7FF,	193	// we special-case "safe" characters from U+80 .. U+7FF,
135	// but use the very good perl function to parse anything else.	194	// but use the very good perl function to parse anything else.
136	// note that we never call this function for a ascii codepoints	195	// note that we never call this function for a ascii codepoints
137	INLINE UV	196	INLINE UV
…		…
170	*s++ = 0x80 \| ( ch & 0x3f);	229	*s++ = 0x80 \| ( ch & 0x3f);
171		230
172	return s;	231	return s;
173	}	232	}
174		233
		234	// convert offset pointer to character index, sv must be string
		235	static STRLEN
		236	ptr_to_index (SV sv, char offset)
		237	{
		238	return SvUTF8 (sv)
		239	? utf8_distance (offset, SvPVX (sv))
		240	: offset - SvPVX (sv);
		241	}
		242
		243	/////////////////////////////////////////////////////////////////////////////
		244	// fp hell
		245
		246	// scan a group of digits, and a trailing exponent
		247	static void
		248	json_atof_scan1 (const char s, NV accum, int *expo, int postdp, int maxdepth)
		249	{
		250	UV uaccum = 0;
		251	int eaccum = 0;
		252
		253	// if we recurse too deep, skip all remaining digits
		254	// to avoid a stack overflow attack
		255	if (expect_false (--maxdepth <= 0))
		256	while (((U8)*s - '0') < 10)
		257	++s;
		258
		259	for (;;)
		260	{
		261	U8 dig = (U8)*s - '0';
		262
		263	if (expect_false (dig >= 10))
		264	{
		265	if (dig == (U8)((U8)'.' - (U8)'0'))
		266	{
		267	++s;
		268	json_atof_scan1 (s, accum, expo, 1, maxdepth);
		269	}
		270	else if ((dig \| ' ') == 'e' - '0')
		271	{
		272	int exp2 = 0;
		273	int neg = 0;
		274
		275	++s;
		276
		277	if (*s == '-')
		278	{
		279	++s;
		280	neg = 1;
		281	}
		282	else if (*s == '+')
		283	++s;
		284
		285	while ((dig = (U8)*s - '0') < 10)
		286	exp2 = exp2 * 10 + *s++ - '0';
		287
		288	*expo += neg ? -exp2 : exp2;
		289	}
		290
		291	break;
		292	}
		293
		294	++s;
		295
		296	uaccum = uaccum * 10 + dig;
		297	++eaccum;
		298
		299	// if we have too many digits, then recurse for more
		300	// we actually do this for rather few digits
		301	if (uaccum >= (UV_MAX - 9) / 10)
		302	{
		303	if (postdp) *expo -= eaccum;
		304	json_atof_scan1 (s, accum, expo, postdp, maxdepth);
		305	if (postdp) *expo += eaccum;
		306
		307	break;
		308	}
		309	}
		310
		311	// this relies greatly on the quality of the pow ()
		312	// implementation of the platform, but a good
		313	// implementation is hard to beat.
		314	// (IEEE 754 conformant ones are required to be exact)
		315	if (postdp) *expo -= eaccum;
		316	accum += uaccum Perl_pow (10., *expo);
		317	*expo += eaccum;
		318	}
		319
		320	static NV
		321	json_atof (const char *s)
		322	{
		323	NV accum = 0.;
		324	int expo = 0;
		325	int neg = 0;
		326
		327	if (*s == '-')
		328	{
		329	++s;
		330	neg = 1;
		331	}
		332
		333	// a recursion depth of ten gives us >>500 bits
		334	json_atof_scan1 (s, &accum, &expo, 0, 10);
		335
		336	return neg ? -accum : accum;
		337	}
		338
		339	// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
		340	static int
		341	ref_bool_type (SV *sv)
		342	{
		343	svtype svt = SvTYPE (sv);
		344
		345	if (svt < SVt_PVAV)
		346	{
		347	STRLEN len = 0;
		348	char *pv = svt ? SvPV (sv, len) : 0;
		349
		350	if (len == 1)
		351	if (*pv == '1')
		352	return 1;
		353	else if (*pv == '0')
		354	return 0;
		355	}
		356
		357	return -1;
		358	}
		359
		360	// returns whether scalar is not a reference in the sense of allow_nonref
		361	static int
		362	json_nonref (SV *scalar)
		363	{
		364	if (!SvROK (scalar))
		365	return 1;
		366
		367	scalar = SvRV (scalar);
		368
		369	if (SvTYPE (scalar) >= SVt_PVMG)
		370	{
		371	if (SvSTASH (scalar) == bool_stash)
		372	return 1;
		373
		374	if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
		375	return 1;
		376	}
		377
		378	return 0;
		379	}
		380
175	/////////////////////////////////////////////////////////////////////////////	381	/////////////////////////////////////////////////////////////////////////////
176	// encoder	382	// encoder
177		383
178	// structure used for encoding JSON	384	// structure used for encoding JSON
179	typedef struct	385	typedef struct
…		…
181	char *cur; // SvPVX (sv) + current output position	387	char *cur; // SvPVX (sv) + current output position
182	char *end; // SvEND (sv)	388	char *end; // SvEND (sv)
183	SV *sv; // result scalar	389	SV *sv; // result scalar
184	JSON json;	390	JSON json;
185	U32 indent; // indentation level	391	U32 indent; // indentation level
186	U32 maxdepth; // max. indentation/recursion level
187	UV limit; // escape character values >= this value when encoding	392	UV limit; // escape character values >= this value when encoding
188	} enc_t;	393	} enc_t;
189		394
190	INLINE void	395	INLINE void
191	need (enc_t *enc, STRLEN len)	396	need (enc_t *enc, STRLEN len)
192	{	397	{
193	if (expect_false (enc->cur + len >= enc->end))	398	if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
194	{	399	{
195	STRLEN cur = enc->cur - SvPVX (enc->sv);	400	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
196	SvGROW (enc->sv, cur + len + 1);	401	char *buf = json_sv_grow (enc->sv, cur, len);
197	enc->cur = SvPVX (enc->sv) + cur;	402	enc->cur = buf + cur;
198	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	403	enc->end = buf + SvLEN (enc->sv) - 1;
199	}	404	}
200	}	405	}
201		406
202	INLINE void	407	INLINE void
203	encode_ch (enc_t *enc, char ch)	408	encode_ch (enc_t *enc, char ch)
…		…
219		424
220	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case	425	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
221	{	426	{
222	if (expect_false (ch == '"')) // but with slow exceptions	427	if (expect_false (ch == '"')) // but with slow exceptions
223	{	428	{
224	need (enc, len += 1);	429	need (enc, len + 1);
225	*enc->cur++ = '\\';	430	*enc->cur++ = '\\';
226	*enc->cur++ = '"';	431	*enc->cur++ = '"';
227	}	432	}
228	else if (expect_false (ch == '\\'))	433	else if (expect_false (ch == '\\'))
229	{	434	{
230	need (enc, len += 1);	435	need (enc, len + 1);
231	*enc->cur++ = '\\';	436	*enc->cur++ = '\\';
232	*enc->cur++ = '\\';	437	*enc->cur++ = '\\';
233	}	438	}
234	else	439	else
235	*enc->cur++ = ch;	440	*enc->cur++ = ch;
…		…
238	}	443	}
239	else	444	else
240	{	445	{
241	switch (ch)	446	switch (ch)
242	{	447	{
243	case '\010': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 'b'; ++str; break;	448	case '\010': need (enc, len + 1); enc->cur++ = '\\'; enc->cur++ = 'b'; ++str; break;
244	case '\011': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 't'; ++str; break;	449	case '\011': need (enc, len + 1); enc->cur++ = '\\'; enc->cur++ = 't'; ++str; break;
245	case '\012': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 'n'; ++str; break;	450	case '\012': need (enc, len + 1); enc->cur++ = '\\'; enc->cur++ = 'n'; ++str; break;
246	case '\014': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 'f'; ++str; break;	451	case '\014': need (enc, len + 1); enc->cur++ = '\\'; enc->cur++ = 'f'; ++str; break;
247	case '\015': need (enc, len += 1); enc->cur++ = '\\'; enc->cur++ = 'r'; ++str; break;	452	case '\015': need (enc, len + 1); enc->cur++ = '\\'; enc->cur++ = 'r'; ++str; break;
248		453
249	default:	454	default:
250	{	455	{
251	STRLEN clen;	456	STRLEN clen;
252	UV uch;	457	UV uch;
…		…
268	if (uch >= 0x10000UL)	473	if (uch >= 0x10000UL)
269	{	474	{
270	if (uch >= 0x110000UL)	475	if (uch >= 0x110000UL)
271	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);	476	croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
272		477
273	need (enc, len += 11);	478	need (enc, len + 11);
274	sprintf (enc->cur, "\\u%04x\\u%04x",	479	sprintf (enc->cur, "\\u%04x\\u%04x",
275	(int)((uch - 0x10000) / 0x400 + 0xD800),	480	(int)((uch - 0x10000) / 0x400 + 0xD800),
276	(int)((uch - 0x10000) % 0x400 + 0xDC00));	481	(int)((uch - 0x10000) % 0x400 + 0xDC00));
277	enc->cur += 12;	482	enc->cur += 12;
278	}	483	}
279	else	484	else
280	{	485	{
281	static char hexdigit [16] = "0123456789abcdef";
282	need (enc, len += 5);	486	need (enc, len + 5);
283	*enc->cur++ = '\\';	487	*enc->cur++ = '\\';
284	*enc->cur++ = 'u';	488	*enc->cur++ = 'u';
285	*enc->cur++ = hexdigit [ uch >> 12 ];	489	*enc->cur++ = PL_hexdigit [ uch >> 12 ];
286	*enc->cur++ = hexdigit [(uch >> 8) & 15];	490	*enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
287	*enc->cur++ = hexdigit [(uch >> 4) & 15];	491	*enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
288	*enc->cur++ = hexdigit [(uch >> 0) & 15];	492	*enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
289	}	493	}
290		494
291	str += clen;	495	str += clen;
292	}	496	}
293	else if (enc->json.flags & F_LATIN1)	497	else if (enc->json.flags & F_LATIN1)
…		…
295	*enc->cur++ = uch;	499	*enc->cur++ = uch;
296	str += clen;	500	str += clen;
297	}	501	}
298	else if (is_utf8)	502	else if (is_utf8)
299	{	503	{
300	need (enc, len += clen);	504	need (enc, len + clen);
301	do	505	do
302	{	506	{
303	enc->cur++ = str++;	507	enc->cur++ = str++;
304	}	508	}
305	while (--clen);	509	while (--clen);
306	}	510	}
307	else	511	else
308	{	512	{
309	need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed	513	need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
310	enc->cur = encode_utf8 (enc->cur, uch);	514	enc->cur = encode_utf8 (enc->cur, uch);
311	++str;	515	++str;
312	}	516	}
313	}	517	}
314	}	518	}
…		…
364	static void	568	static void
365	encode_av (enc_t enc, AV av)	569	encode_av (enc_t enc, AV av)
366	{	570	{
367	int i, len = av_len (av);	571	int i, len = av_len (av);
368		572
369	if (enc->indent >= enc->maxdepth)	573	if (enc->indent >= enc->json.max_depth)
370	croak ("data structure too deep (hit recursion limit)");	574	croak (ERR_NESTING_EXCEEDED);
371		575
372	encode_ch (enc, '[');	576	encode_ch (enc, '[');
373		577
374	if (len >= 0)	578	if (len >= 0)
375	{	579	{
376	encode_nl (enc); ++enc->indent;	580	encode_nl (enc); ++enc->indent;
377		581
378	for (i = 0; i <= len; ++i)	582	for (i = 0; i <= len; ++i)
…		…
390	encode_comma (enc);	594	encode_comma (enc);
391	}	595	}
392		596
393	encode_nl (enc); --enc->indent; encode_indent (enc);	597	encode_nl (enc); --enc->indent; encode_indent (enc);
394	}	598	}
395		599
396	encode_ch (enc, ']');	600	encode_ch (enc, ']');
397	}	601	}
398		602
399	static void	603	static void
400	encode_hk (enc_t enc, HE he)	604	encode_hk (enc_t enc, HE he)
…		…
404	if (HeKLEN (he) == HEf_SVKEY)	608	if (HeKLEN (he) == HEf_SVKEY)
405	{	609	{
406	SV *sv = HeSVKEY (he);	610	SV *sv = HeSVKEY (he);
407	STRLEN len;	611	STRLEN len;
408	char *str;	612	char *str;
409		613
410	SvGETMAGIC (sv);	614	SvGETMAGIC (sv);
411	str = SvPV (sv, len);	615	str = SvPV (sv, len);
412		616
413	encode_str (enc, str, len, SvUTF8 (sv));	617	encode_str (enc, str, len, SvUTF8 (sv));
414	}	618	}
…		…
450	static void	654	static void
451	encode_hv (enc_t enc, HV hv)	655	encode_hv (enc_t enc, HV hv)
452	{	656	{
453	HE *he;	657	HE *he;
454		658
455	if (enc->indent >= enc->maxdepth)	659	if (enc->indent >= enc->json.max_depth)
456	croak ("data structure too deep (hit recursion limit)");	660	croak (ERR_NESTING_EXCEEDED);
457		661
458	encode_ch (enc, '{');	662	encode_ch (enc, '{');
459		663
460	// for canonical output we have to sort by keys first	664	// for canonical output we have to sort by keys first
461	// actually, this is mostly due to the stupid so-called	665	// actually, this is mostly due to the stupid so-called
462	// security workaround added somewhere in 5.8.x.	666	// security workaround added somewhere in 5.8.x
463	// that randomises hash orderings	667	// that randomises hash orderings
464	if (enc->json.flags & F_CANONICAL)	668	if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
465	{	669	{
466	int count = hv_iterinit (hv);	670	int count = hv_iterinit (hv);
467		671
468	if (SvMAGICAL (hv))	672	if (SvMAGICAL (hv))
469	{	673	{
…		…
480	}	684	}
481		685
482	if (count)	686	if (count)
483	{	687	{
484	int i, fast = 1;	688	int i, fast = 1;
485	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	689	HE *hes_stack [STACK_HES];
486	HE *hes = _alloca (count sizeof (HE));	690	HE **hes = hes_stack;
487	#else	691
488	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode	692	// allocate larger arrays on the heap
489	#endif	693	if (count > STACK_HES)
		694	{
		695	SV sv = sv_2mortal (NEWSV (0, count sizeof (*hes)));
		696	hes = (HE **)SvPVX (sv);
		697	}
490		698
491	i = 0;	699	i = 0;
492	while ((he = hv_iternext (hv)))	700	while ((he = hv_iternext (hv)))
493	{	701	{
494	hes [i++] = he;	702	hes [i++] = he;
…		…
563	// encode objects, arrays and special \0=false and \1=true values.	771	// encode objects, arrays and special \0=false and \1=true values.
564	static void	772	static void
565	encode_rv (enc_t enc, SV sv)	773	encode_rv (enc_t enc, SV sv)
566	{	774	{
567	svtype svt;	775	svtype svt;
		776	GV *method;
568		777
569	SvGETMAGIC (sv);	778	SvGETMAGIC (sv);
570	svt = SvTYPE (sv);	779	svt = SvTYPE (sv);
571		780
572	if (expect_false (SvOBJECT (sv)))	781	if (expect_false (SvOBJECT (sv)))
573	{	782	{
574	HV *stash = !JSON_SLOW \|\| json_boolean_stash	783	HV *stash = SvSTASH (sv);
575	? json_boolean_stash
576	: gv_stashpv ("JSON::XS::Boolean", 1);
577		784
578	if (SvSTASH (sv) == stash)	785	if (stash == bool_stash)
579	{	786	{
580	if (SvIV (sv))
581	encode_str (enc, "true", 4, 0);	787	if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
582	else
583	encode_str (enc, "false", 5, 0);	788	else encode_str (enc, "false", 5, 0);
584	}	789	}
		790	else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
		791	{
		792	int count;
		793	dSP;
		794
		795	ENTER; SAVETMPS;
		796	PUSHMARK (SP);
		797	EXTEND (SP, 2);
		798	// we re-bless the reference to get overload and other niceties right
		799	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
		800	PUSHs (sv_json);
		801
		802	PUTBACK;
		803	count = call_sv ((SV *)GvCV (method), G_ARRAY);
		804	SPAGAIN;
		805
		806	// catch this surprisingly common error
		807	if (SvROK (TOPs) && SvRV (TOPs) == sv)
		808	croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
		809
		810	encode_ch (enc, '(');
		811	encode_ch (enc, '"');
		812	encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
		813	encode_ch (enc, '"');
		814	encode_ch (enc, ')');
		815	encode_ch (enc, '[');
		816
		817	if (count)
		818	{
		819	int i;
		820
		821	for (i = 0; i < count - 1; ++i)
		822	{
		823	encode_sv (enc, SP[i + 1 - count]);
		824	encode_ch (enc, ',');
		825	}
		826
		827	encode_sv (enc, TOPs);
		828	SP -= count;
		829	}
		830
		831	PUTBACK;
		832
		833	encode_ch (enc, ']');
		834
		835	FREETMPS; LEAVE;
		836	}
		837	else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
		838	{
		839	dSP;
		840
		841	ENTER; SAVETMPS;
		842	PUSHMARK (SP);
		843	// we re-bless the reference to get overload and other niceties right
		844	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
		845
		846	// calling with G_SCALAR ensures that we always get a 1 return value
		847	PUTBACK;
		848	call_sv ((SV *)GvCV (method), G_SCALAR);
		849	SPAGAIN;
		850
		851	// catch this surprisingly common error
		852	if (SvROK (TOPs) && SvRV (TOPs) == sv)
		853	croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
		854
		855	sv = POPs;
		856	PUTBACK;
		857
		858	encode_sv (enc, sv);
		859
		860	FREETMPS; LEAVE;
		861	}
		862	else if (enc->json.flags & F_ALLOW_BLESSED)
		863	encode_str (enc, "null", 4, 0);
585	else	864	else
586	{
587	#if 0
588	if (0 && sv_derived_from (rv, "JSON::Literal"))
589	{
590	// not yet
591	}
592	#endif
593	if (enc->json.flags & F_CONV_BLESSED)
594	{
595	// we re-bless the reference to get overload and other niceties right
596	GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
597
598	if (to_json)
599	{
600	dSP;
601
602	ENTER; SAVETMPS; PUSHMARK (SP);
603	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
604
605	// calling with G_SCALAR ensures that we always get a 1 return value
606	PUTBACK;
607	call_sv ((SV *)GvCV (to_json), G_SCALAR);
608	SPAGAIN;
609
610	// catch this surprisingly common error
611	if (SvROK (TOPs) && SvRV (TOPs) == sv)
612	croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
613
614	sv = POPs;
615	PUTBACK;
616
617	encode_sv (enc, sv);
618
619	FREETMPS; LEAVE;
620	}
621	else if (enc->json.flags & F_ALLOW_BLESSED)
622	encode_str (enc, "null", 4, 0);
623	else
624	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
625	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
626	}
627	else if (enc->json.flags & F_ALLOW_BLESSED)
628	encode_str (enc, "null", 4, 0);
629	else
630	croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",	865	croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
631	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	866	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
632	}
633	}	867	}
634	else if (svt == SVt_PVHV)	868	else if (svt == SVt_PVHV)
635	encode_hv (enc, (HV *)sv);	869	encode_hv (enc, (HV *)sv);
636	else if (svt == SVt_PVAV)	870	else if (svt == SVt_PVAV)
637	encode_av (enc, (AV *)sv);	871	encode_av (enc, (AV *)sv);
638	else if (svt < SVt_PVAV)	872	else if (svt < SVt_PVAV)
639	{	873	{
640	STRLEN len = 0;	874	int bool_type = ref_bool_type (sv);
641	char *pv = svt ? SvPV (sv, len) : 0;
642		875
643	if (len == 1 && *pv == '1')	876	if (bool_type == 1)
644	encode_str (enc, "true", 4, 0);	877	encode_str (enc, "true", 4, 0);
645	else if (len == 1 && *pv == '0')	878	else if (bool_type == 0)
646	encode_str (enc, "false", 5, 0);	879	encode_str (enc, "false", 5, 0);
		880	else if (enc->json.flags & F_ALLOW_UNKNOWN)
		881	encode_str (enc, "null", 4, 0);
647	else	882	else
648	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",	883	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
649	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	884	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
650	}	885	}
		886	else if (enc->json.flags & F_ALLOW_UNKNOWN)
		887	encode_str (enc, "null", 4, 0);
651	else	888	else
652	croak ("encountered %s, but JSON can only represent references to arrays or hashes",	889	croak ("encountered %s, but JSON can only represent references to arrays or hashes",
653	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	890	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
654	}	891	}
655		892
…		…
681	if (SvIsUV (sv) ? SvUVX (sv) <= 59000	918	if (SvIsUV (sv) ? SvUVX (sv) <= 59000
682	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)	919	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
683	{	920	{
684	// optimise the "small number case"	921	// optimise the "small number case"
685	// code will likely be branchless and use only a single multiplication	922	// code will likely be branchless and use only a single multiplication
686	// works for numbers up to 59074	923	// 4.28 works for numbers up to 59074
687	I32 i = SvIVX (sv);	924	I32 i = SvIVX (sv);
688	U32 u;	925	U32 u;
689	char digit, nz = 0;	926	char digit, nz = 0;
690		927
691	need (enc, 6);	928	need (enc, 6);
692		929
693	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;	930	*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
694	u = i < 0 ? -i : i;	931	u = i < 0 ? -i : i; // not undefined due to range check above
695		932
696	// convert to 4.28 fixed-point representation	933	// convert to 4.28 fixed-point representation
697	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits	934	u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
698		935
699	// now output digit by digit, each time masking out the integer part	936	// now output digit by digit, each time masking out the integer part
…		…
709	}	946	}
710	else	947	else
711	{	948	{
712	// large integer, use the (rather slow) snprintf way.	949	// large integer, use the (rather slow) snprintf way.
713	need (enc, IVUV_MAXCHARS);	950	need (enc, IVUV_MAXCHARS);
714	enc->cur +=	951	enc->cur +=
715	SvIsUV(sv)	952	SvIsUV(sv)
716	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))	953	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
717	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));	954	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
718	}	955	}
719	}	956	}
720	else if (SvROK (sv))	957	else if (SvROK (sv))
721	encode_rv (enc, SvRV (sv));	958	encode_rv (enc, SvRV (sv));
722	else if (!SvOK (sv))	959	else if (!SvOK (sv) \|\| enc->json.flags & F_ALLOW_UNKNOWN)
723	encode_str (enc, "null", 4, 0);	960	encode_str (enc, "null", 4, 0);
724	else	961	else
725	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",	962	croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
726	SvPV_nolen (sv), SvFLAGS (sv));	963	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
727	}	964	}
728		965
729	static SV *	966	static SV *
730	encode_json (SV scalar, JSON json)	967	encode_json (SV scalar, JSON json)
731	{	968	{
732	enc_t enc;	969	enc_t enc;
733		970
734	if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))	971	if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
735	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");	972	croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
736		973
737	enc.json = *json;	974	enc.json = *json;
738	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	975	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
739	enc.cur = SvPVX (enc.sv);	976	enc.cur = SvPVX (enc.sv);
740	enc.end = SvEND (enc.sv);	977	enc.end = SvEND (enc.sv);
741	enc.indent = 0;	978	enc.indent = 0;
742	enc.maxdepth = DEC_DEPTH (enc.json.flags);
743	enc.limit = enc.json.flags & F_ASCII ? 0x000080UL	979	enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
744	: enc.json.flags & F_LATIN1 ? 0x000100UL	980	: enc.json.flags & F_LATIN1 ? 0x000100UL
745	: 0x110000UL;	981	: 0x110000UL;
746		982
747	SvPOK_only (enc.sv);	983	SvPOK_only (enc.sv);
748	encode_sv (&enc, scalar);	984	encode_sv (&enc, scalar);
		985	encode_nl (&enc);
749		986
750	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	987	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
751	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	988	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
752		989
753	if (!(enc.json.flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))	990	if (!(enc.json.flags & (F_ASCII \| F_LATIN1 \| F_UTF8)))
…		…
801	else	1038	else
802	break;	1039	break;
803	}	1040	}
804	else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)	1041	else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
805	break; // parse error, but let higher level handle it, gives better error messages	1042	break; // parse error, but let higher level handle it, gives better error messages
806		1043	else
807	++dec->cur;	1044	++dec->cur;
808	}	1045	}
809	}	1046	}
810		1047
811	#define ERR(reason) SB dec->err = reason; goto fail; SE	1048	#define ERR(reason) SB dec->err = reason; goto fail; SE
812		1049
…		…
814	if (*dec->cur != ch) \	1051	if (*dec->cur != ch) \
815	ERR (# ch " expected"); \	1052	ERR (# ch " expected"); \
816	++dec->cur; \	1053	++dec->cur; \
817	SE	1054	SE
818		1055
819	#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")	1056	#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
820	#define DEC_DEC_DEPTH --dec->depth	1057	#define DEC_DEC_DEPTH --dec->depth
821		1058
822	static SV decode_sv (dec_t dec);	1059	static SV decode_sv (dec_t dec);
823		1060
824	static signed char decode_hexdigit[256];	1061	static signed char decode_hexdigit[256];
…		…
933	else if (expect_true (ch >= 0x20 && ch < 0x80))	1170	else if (expect_true (ch >= 0x20 && ch < 0x80))
934	*cur++ = ch;	1171	*cur++ = ch;
935	else if (ch >= 0x80)	1172	else if (ch >= 0x80)
936	{	1173	{
937	STRLEN clen;	1174	STRLEN clen;
938	UV uch;
939		1175
940	--dec_cur;	1176	--dec_cur;
941		1177
942	uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);	1178	decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
943	if (clen == (STRLEN)-1)	1179	if (clen == (STRLEN)-1)
944	ERR ("malformed UTF-8 character in JSON string");	1180	ERR ("malformed UTF-8 character in JSON string");
945		1181
946	do	1182	do
947	cur++ = dec_cur++;	1183	cur++ = dec_cur++;
948	while (--clen);	1184	while (--clen);
949		1185
950	utf8 = 1;	1186	utf8 = 1;
951	}	1187	}
		1188	else if (ch == '\t' && dec->json.flags & F_RELAXED)
		1189	*cur++ = ch;
952	else	1190	else
953	{	1191	{
954	--dec_cur;	1192	--dec_cur;
955		1193
956	if (!ch)	1194	if (!ch)
…		…
964	{	1202	{
965	STRLEN len = cur - buf;	1203	STRLEN len = cur - buf;
966		1204
967	if (sv)	1205	if (sv)
968	{	1206	{
969	SvGROW (sv, SvCUR (sv) + len + 1);	1207	STRLEN cur = SvCUR (sv);
		1208
		1209	if (SvLEN (sv) - cur <= len)
		1210	json_sv_grow (sv, cur, len);
		1211
970	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	1212	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
971	SvCUR_set (sv, SvCUR (sv) + len);	1213	SvCUR_set (sv, SvCUR (sv) + len);
972	}	1214	}
973	else	1215	else
974	sv = newSVpvn (buf, len);	1216	sv = newSVpvn (buf, len);
…		…
1065		1307
1066	// special case the rather common 1..5-digit-int case	1308	// special case the rather common 1..5-digit-int case
1067	if (*start == '-')	1309	if (*start == '-')
1068	switch (len)	1310	switch (len)
1069	{	1311	{
1070	case 2: return newSViv (-( start [1] - '0' * 1));	1312	case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1071	case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));	1313	case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1072	case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));	1314	case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1073	case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));	1315	case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1074	case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));	1316	case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1075	}	1317	}
1076	else	1318	else
1077	switch (len)	1319	switch (len)
1078	{	1320	{
1079	case 1: return newSViv ( start [0] - '0' * 1);	1321	case 1: return newSViv ( start [0] - '0' * 1);
1080	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);	1322	case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1081	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);	1323	case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1082	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);	1324	case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1083	case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);	1325	case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1084	}	1326	}
1085		1327
1086	{	1328	{
1087	UV uv;	1329	UV uv;
1088	int numtype = grok_number (start, len, &uv);	1330	int numtype = grok_number (start, len, &uv);
…		…
1097	}	1339	}
1098		1340
1099	len -= *start == '-' ? 1 : 0;	1341	len -= *start == '-' ? 1 : 0;
1100		1342
1101	// does not fit into IV or UV, try NV	1343	// does not fit into IV or UV, try NV
1102	if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)	1344	if (len <= NV_DIG)
1103	#if defined (LDBL_DIG)
1104	\|\| (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1105	#endif
1106	)
1107	// fits into NV without loss of precision	1345	// fits into NV without loss of precision
1108	return newSVnv (Atof (start));	1346	return newSVnv (json_atof (start));
1109		1347
1110	// everything else fails, convert it to a string	1348	// everything else fails, convert it to a string
1111	return newSVpvn (start, dec->cur - start);	1349	return newSVpvn (start, dec->cur - start);
1112	}	1350	}
1113		1351
1114	// loss of precision here	1352	// loss of precision here
1115	return newSVnv (Atof (start));	1353	return newSVnv (json_atof (start));
1116		1354
1117	fail:	1355	fail:
1118	return 0;	1356	return 0;
1119	}	1357	}
1120		1358
…		…
1144	if (*dec->cur == ']')	1382	if (*dec->cur == ']')
1145	{	1383	{
1146	++dec->cur;	1384	++dec->cur;
1147	break;	1385	break;
1148	}	1386	}
1149		1387
1150	if (*dec->cur != ',')	1388	if (*dec->cur != ',')
1151	ERR (", or ] expected while parsing array");	1389	ERR (", or ] expected while parsing array");
1152		1390
1153	++dec->cur;	1391	++dec->cur;
1154		1392
…		…
1268		1506
1269	DEC_DEC_DEPTH;	1507	DEC_DEC_DEPTH;
1270	sv = newRV_noinc ((SV *)hv);	1508	sv = newRV_noinc ((SV *)hv);
1271		1509
1272	// check filter callbacks	1510	// check filter callbacks
1273	if (dec->json.flags & F_HOOK)	1511	if (expect_false (dec->json.flags & F_HOOK))
1274	{	1512	{
1275	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)	1513	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1276	{	1514	{
1277	HE cb, he;	1515	HE cb, he;
1278		1516
1279	hv_iterinit (hv);	1517	hv_iterinit (hv);
1280	he = hv_iternext (hv);	1518	he = hv_iternext (hv);
1281	hv_iterinit (hv);	1519	hv_iterinit (hv);
1282		1520
1283	// the next line creates a mortal sv each time its called.	1521	// the next line creates a mortal sv each time it's called.
1284	// might want to optimise this for common cases.	1522	// might want to optimise this for common cases.
1285	cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);	1523	cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1286		1524
1287	if (cb)	1525	if (cb)
1288	{	1526	{
1289	dSP;	1527	dSP;
1290	int count;	1528	int count;
1291		1529
1292	ENTER; SAVETMPS; PUSHMARK (SP);	1530	ENTER; SAVETMPS;
		1531	PUSHMARK (SP);
1293	XPUSHs (HeVAL (he));	1532	XPUSHs (HeVAL (he));
		1533	sv_2mortal (sv);
1294		1534
1295	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;	1535	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1296		1536
1297	if (count == 1)	1537	if (count == 1)
1298	{	1538	{
1299	sv = newSVsv (POPs);	1539	sv = newSVsv (POPs);
		1540	PUTBACK;
1300	FREETMPS; LEAVE;	1541	FREETMPS; LEAVE;
1301	return sv;	1542	return sv;
1302	}	1543	}
		1544	else if (count)
		1545	croak ("filter_json_single_key_object callbacks must not return more than one scalar");
		1546
		1547	PUTBACK;
		1548
		1549	SvREFCNT_inc (sv);
1303		1550
1304	FREETMPS; LEAVE;	1551	FREETMPS; LEAVE;
1305	}	1552	}
1306	}	1553	}
1307		1554
1308	if (dec->json.cb_object)	1555	if (dec->json.cb_object)
1309	{	1556	{
1310	dSP;	1557	dSP;
1311	int count;	1558	int count;
1312		1559
1313	ENTER; SAVETMPS; PUSHMARK (SP);	1560	ENTER; SAVETMPS;
		1561	PUSHMARK (SP);
1314	XPUSHs (sv_2mortal (sv));	1562	XPUSHs (sv_2mortal (sv));
1315		1563
1316	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;	1564	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1317		1565
1318	if (count == 1)	1566	if (count == 1)
1319	{
1320	sv = newSVsv (POPs);	1567	sv = newSVsv (POPs);
1321	FREETMPS; LEAVE;	1568	else if (count == 0)
1322	return sv;
1323	}
1324
1325	SvREFCNT_inc (sv);	1569	SvREFCNT_inc (sv);
		1570	else
		1571	croak ("filter_json_object callbacks must not return more than one scalar");
		1572
		1573	PUTBACK;
		1574
1326	FREETMPS; LEAVE;	1575	FREETMPS; LEAVE;
1327	}	1576	}
1328	}	1577	}
1329		1578
1330	return sv;	1579	return sv;
…		…
1334	DEC_DEC_DEPTH;	1583	DEC_DEC_DEPTH;
1335	return 0;	1584	return 0;
1336	}	1585	}
1337		1586
1338	static SV *	1587	static SV *
		1588	decode_tag (dec_t *dec)
		1589	{
		1590	SV *tag = 0;
		1591	SV *val = 0;
		1592
		1593	if (!(dec->json.flags & F_ALLOW_TAGS))
		1594	ERR ("malformed JSON string, neither array, object, number, string or atom");
		1595
		1596	++dec->cur;
		1597
		1598	decode_ws (dec);
		1599
		1600	tag = decode_sv (dec);
		1601	if (!tag)
		1602	goto fail;
		1603
		1604	if (!SvPOK (tag))
		1605	ERR ("malformed JSON string, (tag) must be a string");
		1606
		1607	decode_ws (dec);
		1608
		1609	if (*dec->cur != ')')
		1610	ERR (") expected after tag");
		1611
		1612	++dec->cur;
		1613
		1614	decode_ws (dec);
		1615
		1616	val = decode_sv (dec);
		1617	if (!val)
		1618	goto fail;
		1619
		1620	if (!SvROK (val) \|\| SvTYPE (SvRV (val)) != SVt_PVAV)
		1621	ERR ("malformed JSON string, tag value must be an array");
		1622
		1623	{
		1624	AV av = (AV )SvRV (val);
		1625	int i, len = av_len (av) + 1;
		1626	HV *stash = gv_stashsv (tag, 0);
		1627	SV *sv;
		1628
		1629	if (!stash)
		1630	ERR ("cannot decode perl-object (package does not exist)");
		1631
		1632	GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
		1633
		1634	if (!method)
		1635	ERR ("cannot decode perl-object (package does not have a THAW method)");
		1636
		1637	dSP;
		1638
		1639	ENTER; SAVETMPS;
		1640	PUSHMARK (SP);
		1641	EXTEND (SP, len + 2);
		1642	// we re-bless the reference to get overload and other niceties right
		1643	PUSHs (tag);
		1644	PUSHs (sv_json);
		1645
		1646	for (i = 0; i < len; ++i)
		1647	PUSHs (*av_fetch (av, i, 1));
		1648
		1649	PUTBACK;
		1650	call_sv ((SV *)GvCV (method), G_SCALAR);
		1651	SPAGAIN;
		1652
		1653	SvREFCNT_dec (tag);
		1654	SvREFCNT_dec (val);
		1655	sv = SvREFCNT_inc (POPs);
		1656
		1657	PUTBACK;
		1658
		1659	FREETMPS; LEAVE;
		1660
		1661	return sv;
		1662	}
		1663
		1664	fail:
		1665	SvREFCNT_dec (tag);
		1666	SvREFCNT_dec (val);
		1667	return 0;
		1668	}
		1669
		1670	static SV *
1339	decode_sv (dec_t *dec)	1671	decode_sv (dec_t *dec)
1340	{	1672	{
1341	// the beauty of JSON: you need exactly one character lookahead	1673	// the beauty of JSON: you need exactly one character lookahead
1342	// to parse everything.	1674	// to parse everything.
1343	switch (*dec->cur)	1675	switch (*dec->cur)
1344	{	1676	{
1345	case '"': ++dec->cur; return decode_str (dec);	1677	case '"': ++dec->cur; return decode_str (dec);
1346	case '[': ++dec->cur; return decode_av (dec);	1678	case '[': ++dec->cur; return decode_av (dec);
1347	case '{': ++dec->cur; return decode_hv (dec);	1679	case '{': ++dec->cur; return decode_hv (dec);
		1680	case '(': return decode_tag (dec);
1348		1681
1349	case '-':	1682	case '-':
1350	case '0': case '1': case '2': case '3': case '4':	1683	case '0': case '1': case '2': case '3': case '4':
1351	case '5': case '6': case '7': case '8': case '9':	1684	case '5': case '6': case '7': case '8': case '9':
1352	return decode_num (dec);	1685	return decode_num (dec);
1353		1686
		1687	case 'f':
		1688	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
		1689	{
		1690	dec->cur += 5;
		1691
		1692	if (expect_false (!dec->json.v_false))
		1693	dec->json.v_false = GET_BOOL (false);
		1694
		1695	return newSVsv (dec->json.v_false);
		1696	}
		1697	else
		1698	ERR ("'false' expected");
		1699
		1700	break;
		1701
1354	case 't':	1702	case 't':
1355	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1703	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1356	{	1704	{
1357	dec->cur += 4;	1705	dec->cur += 4;
1358	#if JSON_SLOW	1706
1359	json_true = get_bool ("JSON::XS::true");	1707	if (expect_false (!dec->json.v_true))
1360	#endif	1708	dec->json.v_true = GET_BOOL (true);
		1709
1361	return newSVsv (json_true);	1710	return newSVsv (dec->json.v_true);
1362	}	1711	}
1363	else	1712	else
1364	ERR ("'true' expected");	1713	ERR ("'true' expected");
1365
1366	break;
1367
1368	case 'f':
1369	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1370	{
1371	dec->cur += 5;
1372	#if JSON_SLOW
1373	json_false = get_bool ("JSON::XS::false");
1374	#endif
1375	return newSVsv (json_false);
1376	}
1377	else
1378	ERR ("'false' expected");
1379		1714
1380	break;	1715	break;
1381		1716
1382	case 'n':	1717	case 'n':
1383	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))	1718	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
…		…
1389	ERR ("'null' expected");	1724	ERR ("'null' expected");
1390		1725
1391	break;	1726	break;
1392		1727
1393	default:	1728	default:
1394	ERR ("malformed JSON string, neither array, object, number, string or atom");	1729	ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1395	break;	1730	break;
1396	}	1731	}
1397		1732
1398	fail:	1733	fail:
1399	return 0;	1734	return 0;
…		…
1401		1736
1402	static SV *	1737	static SV *
1403	decode_json (SV string, JSON json, STRLEN *offset_return)	1738	decode_json (SV string, JSON json, STRLEN *offset_return)
1404	{	1739	{
1405	dec_t dec;	1740	dec_t dec;
1406	STRLEN offset;
1407	SV *sv;	1741	SV *sv;
1408		1742
		1743	/* work around bugs in 5.10 where manipulating magic values
		1744	* makes perl ignore the magic in subsequent accesses.
		1745	* also make a copy of non-PV values, to get them into a clean
		1746	* state (SvPV should do that, but it's buggy, see below).
		1747	*
		1748	* SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
		1749	* as reported by Reini Urban.
		1750	*/
1409	SvGETMAGIC (string);	1751	/SvGETMAGIC (string);/
		1752	if (SvMAGICAL (string) \|\| !SvPOK (string) \|\| SvIsCOW_shared_hash (string))
		1753	string = sv_2mortal (newSVsv (string));
		1754
1410	SvUPGRADE (string, SVt_PV);	1755	SvUPGRADE (string, SVt_PV);
1411		1756
1412	if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))	1757	/* work around a bug in perl 5.10, which causes SvCUR to fail an
		1758	* assertion with -DDEBUGGING, although SvCUR is documented to
		1759	* return the xpv_cur field which certainly exists after upgrading.
		1760	* according to nicholas clark, calling SvPOK fixes this.
		1761	* But it doesn't fix it, so try another workaround, call SvPV_nolen
		1762	* and hope for the best.
		1763	* Damnit, SvPV_nolen still trips over yet another assertion. This
		1764	* assertion business is seriously broken, try yet another workaround
		1765	* for the broken -DDEBUGGING.
		1766	*/
		1767	{
		1768	#ifdef DEBUGGING
		1769	STRLEN offset = SvOK (string) ? sv_len (string) : 0;
		1770	#else
		1771	STRLEN offset = SvCUR (string);
		1772	#endif
		1773
		1774	if (offset > json->max_size && json->max_size)
1413	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",	1775	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1414	(unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));	1776	(unsigned long)SvCUR (string), (unsigned long)json->max_size);
		1777	}
1415		1778
1416	if (json->flags & F_UTF8)	1779	if (DECODE_WANTS_OCTETS (json))
1417	sv_utf8_downgrade (string, 0);	1780	sv_utf8_downgrade (string, 0);
1418	else	1781	else
1419	sv_utf8_upgrade (string);	1782	sv_utf8_upgrade (string);
1420		1783
1421	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1784	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1422		1785
1423	dec.json = *json;	1786	dec.json = *json;
1424	dec.cur = SvPVX (string);	1787	dec.cur = SvPVX (string);
1425	dec.end = SvEND (string);	1788	dec.end = SvEND (string);
1426	dec.err = 0;	1789	dec.err = 0;
1427	dec.depth = 0;	1790	dec.depth = 0;
1428	dec.maxdepth = DEC_DEPTH (dec.json.flags);
1429		1791
1430	if (dec.json.cb_object \|\| dec.json.cb_sk_object)	1792	if (dec.json.cb_object \|\| dec.json.cb_sk_object)
1431	dec.json.flags \|= F_HOOK;	1793	dec.json.flags \|= F_HOOK;
1432		1794
1433	*dec.end = 0; // this should basically be a nop, too, but make sure it's there	1795	*dec.end = 0; // this should basically be a nop, too, but make sure it's there
1434		1796
1435	decode_ws (&dec);	1797	decode_ws (&dec);
1436	sv = decode_sv (&dec);	1798	sv = decode_sv (&dec);
1437		1799
1438	if (!(offset_return \|\| !sv))	1800	if (offset_return)
		1801	*offset_return = dec.cur - SvPVX (string);
		1802	else if (sv)
1439	{	1803	{
1440	// check for trailing garbage	1804	// check for trailing garbage
1441	decode_ws (&dec);	1805	decode_ws (&dec);
1442		1806
1443	if (*dec.cur)	1807	if (dec.cur != dec.end)
1444	{	1808	{
1445	dec.err = "garbage after JSON object";	1809	dec.err = "garbage after JSON object";
1446	SvREFCNT_dec (sv);	1810	SvREFCNT_dec (sv);
1447	sv = 0;	1811	sv = 0;
1448	}	1812	}
1449	}
1450
1451	if (offset_return \|\| !sv)
1452	{
1453	offset = dec.json.flags & F_UTF8
1454	? dec.cur - SvPVX (string)
1455	: utf8_distance (dec.cur, SvPVX (string));
1456
1457	if (offset_return)
1458	*offset_return = offset;
1459	}	1813	}
1460		1814
1461	if (!sv)	1815	if (!sv)
1462	{	1816	{
1463	SV *uni = sv_newmortal ();	1817	SV *uni = sv_newmortal ();
…		…
1469	SAVEVPTR (PL_curcop);	1823	SAVEVPTR (PL_curcop);
1470	PL_curcop = &cop;	1824	PL_curcop = &cop;
1471	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);	1825	pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1472	LEAVE;	1826	LEAVE;
1473		1827
1474	croak ("%s, at character offset %d [\"%s\"]",	1828	croak ("%s, at character offset %d (before \"%s\")",
1475	dec.err,	1829	dec.err,
1476	(int)offset,	1830	(int)ptr_to_index (string, dec.cur),
1477	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1831	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1478	}	1832	}
1479		1833
1480	sv = sv_2mortal (sv);	1834	sv = sv_2mortal (sv);
1481		1835
1482	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))	1836	if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1483	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");	1837	croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1484		1838
1485	return sv;	1839	return sv;
1486	}	1840	}
1487		1841
…		…
1491	static void	1845	static void
1492	incr_parse (JSON *self)	1846	incr_parse (JSON *self)
1493	{	1847	{
1494	const char *p = SvPVX (self->incr_text) + self->incr_pos;	1848	const char *p = SvPVX (self->incr_text) + self->incr_pos;
1495		1849
		1850	// the state machine here is a bit convoluted and could be simplified a lot
		1851	// but this would make it slower, so...
		1852
1496	for (;;)	1853	for (;;)
1497	{	1854	{
1498	//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
1499	switch (self->incr_mode)	1855	switch (self->incr_mode)
1500	{	1856	{
		1857	// reached end of a scalar, see if we are inside a nested structure or not
		1858	end_of_scalar:
		1859	self->incr_mode = INCR_M_JSON;
		1860
		1861	if (self->incr_nest) // end of a scalar inside array, object or tag
		1862	goto incr_m_json;
		1863	else // end of scalar outside structure, json text ends here
		1864	goto interrupt;
		1865
1501	// only used for intiial whitespace skipping	1866	// only used for initial whitespace skipping
1502	case INCR_M_WS:	1867	case INCR_M_WS:
1503	for (;;)	1868	for (;;)
1504	{	1869	{
1505	if (*p > 0x20)	1870	if (*p > 0x20)
1506	{	1871	{
		1872	if (*p == '#')
		1873	{
		1874	self->incr_mode = INCR_M_C0;
		1875	goto incr_m_c;
		1876	}
		1877	else
		1878	{
1507	self->incr_mode = INCR_M_JSON;	1879	self->incr_mode = INCR_M_JSON;
1508	goto incr_m_json;	1880	goto incr_m_json;
		1881	}
1509	}	1882	}
1510	else if (!*p)	1883	else if (!*p)
1511	goto interrupt;	1884	goto interrupt;
1512		1885
1513	++p;	1886	++p;
…		…
1519	goto interrupt;	1892	goto interrupt;
1520		1893
1521	++p;	1894	++p;
1522	self->incr_mode = INCR_M_STR;	1895	self->incr_mode = INCR_M_STR;
1523	goto incr_m_str;	1896	goto incr_m_str;
		1897
		1898	// inside #-style comments
		1899	case INCR_M_C0:
		1900	case INCR_M_C1:
		1901	incr_m_c:
		1902	for (;;)
		1903	{
		1904	if (*p == '\n')
		1905	{
		1906	self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
		1907	break;
		1908	}
		1909	else if (!*p)
		1910	goto interrupt;
		1911
		1912	++p;
		1913	}
		1914
		1915	break;
		1916
		1917	// inside true/false/null
		1918	case INCR_M_TFN:
		1919	incr_m_tfn:
		1920	for (;;)
		1921	switch (*p++)
		1922	{
		1923	case 'r': case 'u': case 'e': // tRUE, falsE, nUll
		1924	case 'a': case 'l': case 's': // fALSe, nuLL
		1925	// allowed
		1926	break;
		1927
		1928	default:
		1929	--p;
		1930	goto end_of_scalar;
		1931	}
		1932
		1933	// inside a number
		1934	case INCR_M_NUM:
		1935	incr_m_num:
		1936	for (;;)
		1937	switch (*p++)
		1938	{
		1939	case 'e': case 'E': case '.': case '+':
		1940	case '-':
		1941	case '0': case '1': case '2': case '3': case '4':
		1942	case '5': case '6': case '7': case '8': case '9':
		1943	// allowed
		1944	break;
		1945
		1946	default:
		1947	--p;
		1948	goto end_of_scalar;
		1949	}
1524		1950
1525	// inside a string	1951	// inside a string
1526	case INCR_M_STR:	1952	case INCR_M_STR:
1527	incr_m_str:	1953	incr_m_str:
1528	for (;;)	1954	for (;;)
1529	{	1955	{
1530	if (*p == '"')	1956	if (*p == '"')
1531	{	1957	{
1532	++p;	1958	++p;
1533	self->incr_mode = INCR_M_JSON;
1534
1535	if (!self->incr_nest)
1536	goto interrupt;
1537
1538	goto incr_m_json;	1959	goto end_of_scalar;
1539	}	1960	}
1540	else if (*p == '\\')	1961	else if (*p == '\\')
1541	{	1962	{
1542	++p; // "virtually" consumes character after \	1963	++p; // "virtually" consumes character after \
1543		1964
…		…
1573	--p; // do not eat the whitespace, let the next round do it	1994	--p; // do not eat the whitespace, let the next round do it
1574	goto interrupt;	1995	goto interrupt;
1575	}	1996	}
1576	break;	1997	break;
1577		1998
		1999	// the following three blocks handle scalars. this makes the parser
		2000	// more strict than required inside arrays or objects, and could
		2001	// be moved to a special case on the toplevel (except strings)
		2002	case 't':
		2003	case 'f':
		2004	case 'n':
		2005	self->incr_mode = INCR_M_TFN;
		2006	goto incr_m_tfn;
		2007
		2008	case '-':
		2009	case '0': case '1': case '2': case '3': case '4':
		2010	case '5': case '6': case '7': case '8': case '9':
		2011	self->incr_mode = INCR_M_NUM;
		2012	goto incr_m_num;
		2013
1578	case '"':	2014	case '"':
1579	self->incr_mode = INCR_M_STR;	2015	self->incr_mode = INCR_M_STR;
1580	goto incr_m_str;	2016	goto incr_m_str;
1581		2017
1582	case '[':	2018	case '[':
1583	case '{':	2019	case '{':
		2020	case '(':
1584	++self->incr_nest;	2021	if (++self->incr_nest > self->max_depth)
		2022	croak (ERR_NESTING_EXCEEDED);
1585	break;	2023	break;
1586		2024
1587	case ']':	2025	case ']':
1588	case '}':	2026	case '}':
1589	if (!--self->incr_nest)	2027	if (--self->incr_nest <= 0)
1590	goto interrupt;	2028	goto interrupt;
		2029	break;
		2030
		2031	case ')':
		2032	--self->incr_nest;
		2033	break;
		2034
		2035	case '#':
		2036	self->incr_mode = INCR_M_C1;
		2037	goto incr_m_c;
1591	}	2038	}
1592	}	2039	}
1593	}	2040	}
1594		2041
1595	modechange:	2042	modechange:
1596	;	2043	;
1597	}	2044	}
1598		2045
1599	interrupt:	2046	interrupt:
1600	self->incr_pos = p - SvPVX (self->incr_text);	2047	self->incr_pos = p - SvPVX (self->incr_text);
		2048	//printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1601	//printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D	2049	//printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1602	}	2050	}
1603		2051
1604	/////////////////////////////////////////////////////////////////////////////	2052	/////////////////////////////////////////////////////////////////////////////
1605	// XS interface functions	2053	// XS interface functions
…		…
1615	i >= '0' && i <= '9' ? i - '0'	2063	i >= '0' && i <= '9' ? i - '0'
1616	: i >= 'a' && i <= 'f' ? i - 'a' + 10	2064	: i >= 'a' && i <= 'f' ? i - 'a' + 10
1617	: i >= 'A' && i <= 'F' ? i - 'A' + 10	2065	: i >= 'A' && i <= 'F' ? i - 'A' + 10
1618	: -1;	2066	: -1;
1619		2067
1620	json_stash = gv_stashpv ("JSON::XS" , 1);	2068	json_stash = gv_stashpv ("JSON::XS" , 1);
1621	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);	2069	bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1622
1623	json_true = get_bool ("JSON::XS::true");
1624	json_false = get_bool ("JSON::XS::false");	2070	bool_false = get_bool ("Types::Serialiser::false");
		2071	bool_true = get_bool ("Types::Serialiser::true");
		2072
		2073	sv_json = newSVpv ("JSON", 0);
		2074	SvREADONLY_on (sv_json);
		2075
		2076	CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1625	}	2077	}
1626		2078
1627	PROTOTYPES: DISABLE	2079	PROTOTYPES: DISABLE
1628		2080
1629	void CLONE (...)	2081	void CLONE (...)
1630	CODE:	2082	CODE:
		2083	// as long as these writes are atomic, the race should not matter
		2084	// as existing threads either already use 0, or use the old value,
		2085	// which is sitll correct for the initial thread.
1631	json_stash = 0;	2086	json_stash = 0;
1632	json_boolean_stash = 0;	2087	bool_stash = 0;
		2088	bool_false = 0;
		2089	bool_true = 0;
1633		2090
1634	void new (char *klass)	2091	void new (char *klass)
1635	PPCODE:	2092	PPCODE:
1636	{	2093	{
1637	SV *pv = NEWSV (0, sizeof (JSON));	2094	SV *pv = NEWSV (0, sizeof (JSON));
1638	SvPOK_only (pv);	2095	SvPOK_only (pv);
1639	Zero (SvPVX (pv), 1, JSON);	2096	json_init ((JSON *)SvPVX (pv));
1640	((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1641	XPUSHs (sv_2mortal (sv_bless (	2097	XPUSHs (sv_2mortal (sv_bless (
1642	newRV_noinc (pv),	2098	newRV_noinc (pv),
1643	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)	2099	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1644	)));	2100	)));
1645	}	2101	}
		2102
		2103	void boolean_values (JSON self, SV v_false = 0, SV *v_true = 0)
		2104	PPCODE:
		2105	self->v_false = newSVsv (v_false);
		2106	self->v_true = newSVsv (v_true);
		2107	XPUSHs (ST (0));
		2108
		2109	void get_boolean_values (JSON *self)
		2110	PPCODE:
		2111	if (self->v_false && self->v_true)
		2112	{
		2113	EXTEND (SP, 2);
		2114	PUSHs (self->v_false);
		2115	PUSHs (self->v_true);
		2116	}
1646		2117
1647	void ascii (JSON *self, int enable = 1)	2118	void ascii (JSON *self, int enable = 1)
1648	ALIAS:	2119	ALIAS:
1649	ascii = F_ASCII	2120	ascii = F_ASCII
1650	latin1 = F_LATIN1	2121	latin1 = F_LATIN1
…		…
1657	allow_nonref = F_ALLOW_NONREF	2128	allow_nonref = F_ALLOW_NONREF
1658	shrink = F_SHRINK	2129	shrink = F_SHRINK
1659	allow_blessed = F_ALLOW_BLESSED	2130	allow_blessed = F_ALLOW_BLESSED
1660	convert_blessed = F_CONV_BLESSED	2131	convert_blessed = F_CONV_BLESSED
1661	relaxed = F_RELAXED	2132	relaxed = F_RELAXED
		2133	allow_unknown = F_ALLOW_UNKNOWN
		2134	allow_tags = F_ALLOW_TAGS
1662	PPCODE:	2135	PPCODE:
1663	{	2136	{
1664	if (enable)	2137	if (enable)
1665	self->flags \|= ix;	2138	self->flags \|= ix;
1666	else	2139	else
…		…
1681	get_allow_nonref = F_ALLOW_NONREF	2154	get_allow_nonref = F_ALLOW_NONREF
1682	get_shrink = F_SHRINK	2155	get_shrink = F_SHRINK
1683	get_allow_blessed = F_ALLOW_BLESSED	2156	get_allow_blessed = F_ALLOW_BLESSED
1684	get_convert_blessed = F_CONV_BLESSED	2157	get_convert_blessed = F_CONV_BLESSED
1685	get_relaxed = F_RELAXED	2158	get_relaxed = F_RELAXED
		2159	get_allow_unknown = F_ALLOW_UNKNOWN
		2160	get_allow_tags = F_ALLOW_TAGS
1686	PPCODE:	2161	PPCODE:
1687	XPUSHs (boolSV (self->flags & ix));	2162	XPUSHs (boolSV (self->flags & ix));
1688		2163
1689	void max_depth (JSON *self, UV max_depth = 0x80000000UL)	2164	void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1690	PPCODE:	2165	PPCODE:
1691	{	2166	self->max_depth = max_depth;
1692	UV log2 = 0;
1693
1694	if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1695
1696	while ((1UL << log2) < max_depth)
1697	++log2;
1698
1699	self->flags = self->flags & ~F_MAXDEPTH \| (log2 << S_MAXDEPTH);
1700
1701	XPUSHs (ST (0));	2167	XPUSHs (ST (0));
1702	}
1703		2168
1704	U32 get_max_depth (JSON *self)	2169	U32 get_max_depth (JSON *self)
1705	CODE:	2170	CODE:
1706	RETVAL = DEC_DEPTH (self->flags);	2171	RETVAL = self->max_depth;
1707	OUTPUT:	2172	OUTPUT:
1708	RETVAL	2173	RETVAL
1709		2174
1710	void max_size (JSON *self, UV max_size = 0)	2175	void max_size (JSON *self, U32 max_size = 0)
1711	PPCODE:	2176	PPCODE:
1712	{	2177	self->max_size = max_size;
1713	UV log2 = 0;
1714
1715	if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1716	if (max_size == 1) max_size = 2;
1717
1718	while ((1UL << log2) < max_size)
1719	++log2;
1720
1721	self->flags = self->flags & ~F_MAXSIZE \| (log2 << S_MAXSIZE);
1722
1723	XPUSHs (ST (0));	2178	XPUSHs (ST (0));
1724	}
1725		2179
1726	int get_max_size (JSON *self)	2180	int get_max_size (JSON *self)
1727	CODE:	2181	CODE:
1728	RETVAL = DEC_SIZE (self->flags);	2182	RETVAL = self->max_size;
1729	OUTPUT:	2183	OUTPUT:
1730	RETVAL	2184	RETVAL
1731		2185
1732	void filter_json_object (JSON self, SV cb = &PL_sv_undef)	2186	void filter_json_object (JSON self, SV cb = &PL_sv_undef)
1733	PPCODE:	2187	PPCODE:
…		…
1739	}	2193	}
1740		2194
1741	void filter_json_single_key_object (JSON self, SV key, SV *cb = &PL_sv_undef)	2195	void filter_json_single_key_object (JSON self, SV key, SV *cb = &PL_sv_undef)
1742	PPCODE:	2196	PPCODE:
1743	{	2197	{
1744	if (!self->cb_sk_object)	2198	if (!self->cb_sk_object)
1745	self->cb_sk_object = newHV ();	2199	self->cb_sk_object = newHV ();
1746		2200
1747	if (SvOK (cb))	2201	if (SvOK (cb))
1748	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);	2202	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1749	else	2203	else
…		…
1760	XPUSHs (ST (0));	2214	XPUSHs (ST (0));
1761	}	2215	}
1762		2216
1763	void encode (JSON self, SV scalar)	2217	void encode (JSON self, SV scalar)
1764	PPCODE:	2218	PPCODE:
1765	XPUSHs (encode_json (scalar, self));	2219	PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
		2220	XPUSHs (scalar);
1766		2221
1767	void decode (JSON self, SV jsonstr)	2222	void decode (JSON self, SV jsonstr)
1768	PPCODE:	2223	PPCODE:
1769	XPUSHs (decode_json (jsonstr, self, 0));	2224	PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
		2225	XPUSHs (jsonstr);
1770		2226
1771	void decode_prefix (JSON self, SV jsonstr)	2227	void decode_prefix (JSON self, SV jsonstr)
1772	PPCODE:	2228	PPCODE:
1773	{	2229	{
		2230	SV *sv;
1774	STRLEN offset;	2231	STRLEN offset;
		2232	PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1775	EXTEND (SP, 2);	2233	EXTEND (SP, 2);
1776	PUSHs (decode_json (jsonstr, self, &offset));	2234	PUSHs (sv);
1777	PUSHs (sv_2mortal (newSVuv (offset)));	2235	PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
1778	}	2236	}
1779		2237
1780	void incr_parse (JSON self, SV jsonstr = 0)	2238	void incr_parse (JSON self, SV jsonstr = 0)
1781	PPCODE:	2239	PPCODE:
1782	{	2240	{
1783	if (!self->incr_text)	2241	if (!self->incr_text)
1784	self->incr_text = newSVpvn ("", 0);	2242	self->incr_text = newSVpvn ("", 0);
		2243
		2244	/* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
		2245	if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
		2246	if (DECODE_WANTS_OCTETS (self))
		2247	{
		2248	if (self->incr_pos)
		2249	self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
		2250	(U8 *)SvPVX (self->incr_text) + self->incr_pos);
		2251
		2252	sv_utf8_downgrade (self->incr_text, 0);
		2253	}
		2254	else
		2255	{
		2256	sv_utf8_upgrade (self->incr_text);
		2257
		2258	if (self->incr_pos)
		2259	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
		2260	- (U8 *)SvPVX (self->incr_text);
		2261	}
1785		2262
1786	// append data, if any	2263	// append data, if any
1787	if (jsonstr)	2264	if (jsonstr)
1788	{	2265	{
		2266	/* make sure both strings have same encoding */
1789	if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))	2267	if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
		2268	if (SvUTF8 (jsonstr))
		2269	sv_utf8_downgrade (jsonstr, 0);
1790	{	2270	else
1791	/* utf-8-ness differs, need to upgrade */
1792	sv_utf8_upgrade (self->incr_text);	2271	sv_utf8_upgrade (jsonstr);
1793		2272
1794	if (self->incr_pos)	2273	/* and then just blindly append */
1795	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1796	- (U8 *)SvPVX (self->incr_text);
1797	}
1798
1799	{	2274	{
1800	STRLEN len;	2275	STRLEN len;
1801	const char *str = SvPV (jsonstr, len);	2276	const char *str = SvPV (jsonstr, len);
1802	SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);	2277	STRLEN cur = SvCUR (self->incr_text);
		2278
		2279	if (SvLEN (self->incr_text) - cur <= len)
		2280	json_sv_grow (self->incr_text, cur, len);
		2281
1803	Move (str, SvEND (self->incr_text), len, char);	2282	Move (str, SvEND (self->incr_text), len, char);
1804	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);	2283	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1805	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there	2284	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1806	}	2285	}
1807	}	2286	}
1808		2287
1809	if (GIMME_V != G_VOID)	2288	if (GIMME_V != G_VOID)
1810	do	2289	do
1811	{	2290	{
		2291	SV *sv;
1812	STRLEN offset;	2292	STRLEN offset;
1813		2293
1814	if (!INCR_DONE (self))	2294	if (!INCR_DONE (self))
1815	{	2295	{
1816	incr_parse (self);	2296	incr_parse (self);
		2297
		2298	if (self->incr_pos > self->max_size && self->max_size)
		2299	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
		2300	(unsigned long)self->incr_pos, (unsigned long)self->max_size);
		2301
1817	if (!INCR_DONE (self))	2302	if (!INCR_DONE (self))
		2303	{
		2304	// as an optimisation, do not accumulate white space in the incr buffer
		2305	if (self->incr_mode == INCR_M_WS && self->incr_pos)
		2306	{
		2307	self->incr_pos = 0;
		2308	SvCUR_set (self->incr_text, 0);
		2309	}
		2310
1818	break;	2311	break;
		2312	}
1819	}	2313	}
1820		2314
1821	XPUSHs (decode_json (self->incr_text, self, &offset));	2315	PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
		2316	XPUSHs (sv);
1822		2317
1823	sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1824	self->incr_pos -= offset;	2318	self->incr_pos -= offset;
1825	self->incr_nest = 0;	2319	self->incr_nest = 0;
1826	self->incr_mode = 0;	2320	self->incr_mode = 0;
		2321
		2322	sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
1827	}	2323	}
1828	while (GIMME_V == G_ARRAY);	2324	while (GIMME_V == G_ARRAY);
1829	}	2325	}
1830		2326
1831	SV incr_text (JSON self)	2327	SV incr_text (JSON self)
…		…
1850	self->incr_nest = 0;	2346	self->incr_nest = 0;
1851	self->incr_mode = 0;	2347	self->incr_mode = 0;
1852	}	2348	}
1853	}	2349	}
1854		2350
		2351	void incr_reset (JSON *self)
		2352	CODE:
		2353	{
		2354	SvREFCNT_dec (self->incr_text);
		2355	self->incr_text = 0;
		2356	self->incr_pos = 0;
		2357	self->incr_nest = 0;
		2358	self->incr_mode = 0;
		2359	}
		2360
1855	void DESTROY (JSON *self)	2361	void DESTROY (JSON *self)
1856	CODE:	2362	CODE:
		2363	SvREFCNT_dec (self->v_false);
		2364	SvREFCNT_dec (self->v_true);
1857	SvREFCNT_dec (self->cb_sk_object);	2365	SvREFCNT_dec (self->cb_sk_object);
1858	SvREFCNT_dec (self->cb_object);	2366	SvREFCNT_dec (self->cb_object);
1859	SvREFCNT_dec (self->incr_text);	2367	SvREFCNT_dec (self->incr_text);
1860		2368
1861	PROTOTYPES: ENABLE	2369	PROTOTYPES: ENABLE
1862		2370
1863	void encode_json (SV *scalar)	2371	void encode_json (SV *scalar)
1864	ALIAS:
1865	to_json_ = 0
1866	encode_json = F_UTF8
1867	PPCODE:	2372	PPCODE:
1868	{	2373	{
1869	JSON json = { F_DEFAULT \| ix };	2374	JSON json;
1870	XPUSHs (encode_json (scalar, &json));	2375	json_init (&json);
		2376	json.flags \|= F_UTF8;
		2377	PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
		2378	XPUSHs (scalar);
1871	}	2379	}
1872		2380
1873	void decode_json (SV *jsonstr)	2381	void decode_json (SV *jsonstr)
1874	ALIAS:
1875	from_json_ = 0
1876	decode_json = F_UTF8
1877	PPCODE:	2382	PPCODE:
1878	{	2383	{
1879	JSON json = { F_DEFAULT \| ix };	2384	JSON json;
1880	XPUSHs (decode_json (jsonstr, &json, 0));	2385	json_init (&json);
		2386	json.flags \|= F_UTF8;
		2387	PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
		2388	XPUSHs (jsonstr);
1881	}	2389	}
1882		2390
1883

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.83 by root, Wed Mar 26 22:54:38 2008 UTC vs. Revision 1.144 by root, Wed Nov 9 11:47:12 2022 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.83 by root, Wed Mar 26 22:54:38 2008 UTC vs.
Revision 1.144 by root, Wed Nov 9 11:47:12 2022 UTC