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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.103 by root, Wed Jan 6 08:01:39 2010 UTC vs.
Revision 1.143 by root, Mon Jun 21 01:04:30 2021 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	// guarantees, 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
		21	#endif
		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
20	#endif	32	#endif
21		33
22	// three extra for rounding, sign, and end of string	34	// three extra for rounding, sign, and end of string
23	#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)	35	#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
24		36
…		…
33	#define F_SHRINK 0x00000200UL	45	#define F_SHRINK 0x00000200UL
34	#define F_ALLOW_BLESSED 0x00000400UL	46	#define F_ALLOW_BLESSED 0x00000400UL
35	#define F_CONV_BLESSED 0x00000800UL	47	#define F_CONV_BLESSED 0x00000800UL
36	#define F_RELAXED 0x00001000UL	48	#define F_RELAXED 0x00001000UL
37	#define F_ALLOW_UNKNOWN 0x00002000UL	49	#define F_ALLOW_UNKNOWN 0x00002000UL
		50	#define F_ALLOW_TAGS 0x00004000UL
38	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing	51	#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
39		52
40	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER	53	#define F_PRETTY F_INDENT \| F_SPACE_BEFORE \| F_SPACE_AFTER
41		54
42	#define INIT_SIZE 32 // initial scalar size to be allocated	55	#define INIT_SIZE 64 // initial scalar size to be allocated
43	#define INDENT_STEP 3 // spaces per indentation level	56	#define INDENT_STEP 3 // spaces per indentation level
44		57
45	#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)
46		61
47	#define SB do {	62	#define SB do {
48	#define SE } while (0)	63	#define SE } while (0)
49		64
50	#if __GNUC__ >= 3	65	#if __GNUC__ >= 3
…		…
63	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))	78	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
64		79
65	#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"	80	#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
66		81
67	#ifdef USE_ITHREADS	82	#ifdef USE_ITHREADS
68	# define JSON_SLOW 1
69	# 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))
70	#else	86	#else
71	# define JSON_SLOW 0
72	# define JSON_STASH json_stash	87	# define JSON_STASH json_stash
		88	# define BOOL_STASH bool_stash
		89	# define GET_BOOL(value) bool_ ## value
73	#endif	90	#endif
74		91
75	static HV json_stash, json_boolean_stash; // JSON::XS::	92	// the amount of HEs to allocate on the stack, when sorting keys
76	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;
77		98
78	enum {	99	enum {
79	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
80	INCR_M_STR, // inside string	103	INCR_M_STR, // inside string
81	INCR_M_BS, // inside backslash	104	INCR_M_BS, // inside backslash
82	INCR_M_C0, // inside comment in initial whitespace sequence	105	INCR_M_C0, // inside comment in initial whitespace sequence
83	INCR_M_C1, // inside comment in other places	106	INCR_M_C1, // inside comment in other places
84	INCR_M_JSON // outside anything, count nesting	107	INCR_M_JSON // outside anything, count nesting
…		…
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	unsigned char incr_mode;	124	unsigned char incr_mode;
		125
		126	SV v_false, v_true;
102	} JSON;	127	} JSON;
103		128
104	INLINE void	129	INLINE void
105	json_init (JSON *json)	130	json_init (JSON *json)
106	{	131	{
107	Zero (json, 1, JSON);	132	static const JSON init = { F_ALLOW_NONREF, 512 };
108	json->max_depth = 512;	133
		134	*json = init;
109	}	135	}
110		136
111	/////////////////////////////////////////////////////////////////////////////	137	/////////////////////////////////////////////////////////////////////////////
112	// utility functions	138	// utility functions
113		139
…		…
135	SvPV_renew (sv, SvCUR (sv) + 1);	161	SvPV_renew (sv, SvCUR (sv) + 1);
136	#endif	162	#endif
137	}	163	}
138	}	164	}
139		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
140	// 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
141	// case of an error.	192	// case of an error.
142	// we special-case "safe" characters from U+80 .. U+7FF,	193	// we special-case "safe" characters from U+80 .. U+7FF,
143	// but use the very good perl function to parse anything else.	194	// but use the very good perl function to parse anything else.
144	// note that we never call this function for a ascii codepoints	195	// note that we never call this function for a ascii codepoints
145	INLINE UV	196	INLINE UV
…		…
188	? utf8_distance (offset, SvPVX (sv))	239	? utf8_distance (offset, SvPVX (sv))
189	: offset - SvPVX (sv);	240	: offset - SvPVX (sv);
190	}	241	}
191		242
192	/////////////////////////////////////////////////////////////////////////////	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
		381	/////////////////////////////////////////////////////////////////////////////
193	// encoder	382	// encoder
194		383
195	// structure used for encoding JSON	384	// structure used for encoding JSON
196	typedef struct	385	typedef struct
197	{	386	{
…		…
204	} enc_t;	393	} enc_t;
205		394
206	INLINE void	395	INLINE void
207	need (enc_t *enc, STRLEN len)	396	need (enc_t *enc, STRLEN len)
208	{	397	{
209	if (expect_false (enc->cur + len >= enc->end))	398	if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
210	{	399	{
211	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	400	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
212	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	401	char *buf = json_sv_grow (enc->sv, cur, len);
213	enc->cur = SvPVX (enc->sv) + cur;	402	enc->cur = buf + cur;
214	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	403	enc->end = buf + SvLEN (enc->sv) - 1;
215	}	404	}
216	}	405	}
217		406
218	INLINE void	407	INLINE void
219	encode_ch (enc_t *enc, char ch)	408	encode_ch (enc_t *enc, char ch)
…		…
235		424
236	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case	425	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
237	{	426	{
238	if (expect_false (ch == '"')) // but with slow exceptions	427	if (expect_false (ch == '"')) // but with slow exceptions
239	{	428	{
240	need (enc, len += 1);	429	need (enc, len + 1);
241	*enc->cur++ = '\\';	430	*enc->cur++ = '\\';
242	*enc->cur++ = '"';	431	*enc->cur++ = '"';
243	}	432	}
244	else if (expect_false (ch == '\\'))	433	else if (expect_false (ch == '\\'))
245	{	434	{
246	need (enc, len += 1);	435	need (enc, len + 1);
247	*enc->cur++ = '\\';	436	*enc->cur++ = '\\';
248	*enc->cur++ = '\\';	437	*enc->cur++ = '\\';
249	}	438	}
250	else	439	else
251	*enc->cur++ = ch;	440	*enc->cur++ = ch;
…		…
254	}	443	}
255	else	444	else
256	{	445	{
257	switch (ch)	446	switch (ch)
258	{	447	{
259	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;
260	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;
261	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;
262	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;
263	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;
264		453
265	default:	454	default:
266	{	455	{
267	STRLEN clen;	456	STRLEN clen;
268	UV uch;	457	UV uch;
…		…
284	if (uch >= 0x10000UL)	473	if (uch >= 0x10000UL)
285	{	474	{
286	if (uch >= 0x110000UL)	475	if (uch >= 0x110000UL)
287	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);
288		477
289	need (enc, len += 11);	478	need (enc, len + 11);
290	sprintf (enc->cur, "\\u%04x\\u%04x",	479	sprintf (enc->cur, "\\u%04x\\u%04x",
291	(int)((uch - 0x10000) / 0x400 + 0xD800),	480	(int)((uch - 0x10000) / 0x400 + 0xD800),
292	(int)((uch - 0x10000) % 0x400 + 0xDC00));	481	(int)((uch - 0x10000) % 0x400 + 0xDC00));
293	enc->cur += 12;	482	enc->cur += 12;
294	}	483	}
295	else	484	else
296	{	485	{
297	need (enc, len += 5);	486	need (enc, len + 5);
298	*enc->cur++ = '\\';	487	*enc->cur++ = '\\';
299	*enc->cur++ = 'u';	488	*enc->cur++ = 'u';
300	*enc->cur++ = PL_hexdigit [ uch >> 12 ];	489	*enc->cur++ = PL_hexdigit [ uch >> 12 ];
301	*enc->cur++ = PL_hexdigit [(uch >> 8) & 15];	490	*enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
302	*enc->cur++ = PL_hexdigit [(uch >> 4) & 15];	491	*enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
…		…
310	*enc->cur++ = uch;	499	*enc->cur++ = uch;
311	str += clen;	500	str += clen;
312	}	501	}
313	else if (is_utf8)	502	else if (is_utf8)
314	{	503	{
315	need (enc, len += clen);	504	need (enc, len + clen);
316	do	505	do
317	{	506	{
318	enc->cur++ = str++;	507	enc->cur++ = str++;
319	}	508	}
320	while (--clen);	509	while (--clen);
321	}	510	}
322	else	511	else
323	{	512	{
324	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
325	enc->cur = encode_utf8 (enc->cur, uch);	514	enc->cur = encode_utf8 (enc->cur, uch);
326	++str;	515	++str;
327	}	516	}
328	}	517	}
329	}	518	}
…		…
383		572
384	if (enc->indent >= enc->json.max_depth)	573	if (enc->indent >= enc->json.max_depth)
385	croak (ERR_NESTING_EXCEEDED);	574	croak (ERR_NESTING_EXCEEDED);
386		575
387	encode_ch (enc, '[');	576	encode_ch (enc, '[');
388		577
389	if (len >= 0)	578	if (len >= 0)
390	{	579	{
391	encode_nl (enc); ++enc->indent;	580	encode_nl (enc); ++enc->indent;
392		581
393	for (i = 0; i <= len; ++i)	582	for (i = 0; i <= len; ++i)
…		…
405	encode_comma (enc);	594	encode_comma (enc);
406	}	595	}
407		596
408	encode_nl (enc); --enc->indent; encode_indent (enc);	597	encode_nl (enc); --enc->indent; encode_indent (enc);
409	}	598	}
410		599
411	encode_ch (enc, ']');	600	encode_ch (enc, ']');
412	}	601	}
413		602
414	static void	603	static void
415	encode_hk (enc_t enc, HE he)	604	encode_hk (enc_t enc, HE he)
…		…
419	if (HeKLEN (he) == HEf_SVKEY)	608	if (HeKLEN (he) == HEf_SVKEY)
420	{	609	{
421	SV *sv = HeSVKEY (he);	610	SV *sv = HeSVKEY (he);
422	STRLEN len;	611	STRLEN len;
423	char *str;	612	char *str;
424		613
425	SvGETMAGIC (sv);	614	SvGETMAGIC (sv);
426	str = SvPV (sv, len);	615	str = SvPV (sv, len);
427		616
428	encode_str (enc, str, len, SvUTF8 (sv));	617	encode_str (enc, str, len, SvUTF8 (sv));
429	}	618	}
…		…
495	}	684	}
496		685
497	if (count)	686	if (count)
498	{	687	{
499	int i, fast = 1;	688	int i, fast = 1;
500	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	689	HE *hes_stack [STACK_HES];
501	HE *hes = _alloca (count sizeof (HE));	690	HE **hes = hes_stack;
502	#else	691
503	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode	692	// allocate larger arrays on the heap
504	#endif	693	if (count > STACK_HES)
		694	{
		695	SV sv = sv_2mortal (NEWSV (0, count sizeof (*hes)));
		696	hes = (HE **)SvPVX (sv);
		697	}
505		698
506	i = 0;	699	i = 0;
507	while ((he = hv_iternext (hv)))	700	while ((he = hv_iternext (hv)))
508	{	701	{
509	hes [i++] = he;	702	hes [i++] = he;
…		…
578	// encode objects, arrays and special \0=false and \1=true values.	771	// encode objects, arrays and special \0=false and \1=true values.
579	static void	772	static void
580	encode_rv (enc_t enc, SV sv)	773	encode_rv (enc_t enc, SV sv)
581	{	774	{
582	svtype svt;	775	svtype svt;
		776	GV *method;
583		777
584	SvGETMAGIC (sv);	778	SvGETMAGIC (sv);
585	svt = SvTYPE (sv);	779	svt = SvTYPE (sv);
586		780
587	if (expect_false (SvOBJECT (sv)))	781	if (expect_false (SvOBJECT (sv)))
588	{	782	{
589	HV *stash = !JSON_SLOW \|\| json_boolean_stash	783	HV *stash = SvSTASH (sv);
590	? json_boolean_stash
591	: gv_stashpv ("JSON::XS::Boolean", 1);
592		784
593	if (SvSTASH (sv) == stash)	785	if (stash == bool_stash)
594	{	786	{
595	if (SvIV (sv))
596	encode_str (enc, "true", 4, 0);	787	if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
597	else
598	encode_str (enc, "false", 5, 0);	788	else encode_str (enc, "false", 5, 0);
599	}	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);
600	else	864	else
601	{
602	#if 0
603	if (0 && sv_derived_from (rv, "JSON::Literal"))
604	{
605	// not yet
606	}
607	#endif
608	if (enc->json.flags & F_CONV_BLESSED)
609	{
610	// we re-bless the reference to get overload and other niceties right
611	GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
612
613	if (to_json)
614	{
615	dSP;
616
617	ENTER; SAVETMPS; PUSHMARK (SP);
618	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
619
620	// calling with G_SCALAR ensures that we always get a 1 return value
621	PUTBACK;
622	call_sv ((SV *)GvCV (to_json), G_SCALAR);
623	SPAGAIN;
624
625	// catch this surprisingly common error
626	if (SvROK (TOPs) && SvRV (TOPs) == sv)
627	croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
628
629	sv = POPs;
630	PUTBACK;
631
632	encode_sv (enc, sv);
633
634	FREETMPS; LEAVE;
635	}
636	else if (enc->json.flags & F_ALLOW_BLESSED)
637	encode_str (enc, "null", 4, 0);
638	else
639	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
640	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
641	}
642	else if (enc->json.flags & F_ALLOW_BLESSED)
643	encode_str (enc, "null", 4, 0);
644	else
645	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)",
646	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	866	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
647	}
648	}	867	}
649	else if (svt == SVt_PVHV)	868	else if (svt == SVt_PVHV)
650	encode_hv (enc, (HV *)sv);	869	encode_hv (enc, (HV *)sv);
651	else if (svt == SVt_PVAV)	870	else if (svt == SVt_PVAV)
652	encode_av (enc, (AV *)sv);	871	encode_av (enc, (AV *)sv);
653	else if (svt < SVt_PVAV)	872	else if (svt < SVt_PVAV)
654	{	873	{
655	STRLEN len = 0;	874	int bool_type = ref_bool_type (sv);
656	char *pv = svt ? SvPV (sv, len) : 0;
657		875
658	if (len == 1 && *pv == '1')	876	if (bool_type == 1)
659	encode_str (enc, "true", 4, 0);	877	encode_str (enc, "true", 4, 0);
660	else if (len == 1 && *pv == '0')	878	else if (bool_type == 0)
661	encode_str (enc, "false", 5, 0);	879	encode_str (enc, "false", 5, 0);
662	else if (enc->json.flags & F_ALLOW_UNKNOWN)	880	else if (enc->json.flags & F_ALLOW_UNKNOWN)
663	encode_str (enc, "null", 4, 0);	881	encode_str (enc, "null", 4, 0);
664	else	882	else
665	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",
…		…
700	if (SvIsUV (sv) ? SvUVX (sv) <= 59000	918	if (SvIsUV (sv) ? SvUVX (sv) <= 59000
701	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)	919	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
702	{	920	{
703	// optimise the "small number case"	921	// optimise the "small number case"
704	// code will likely be branchless and use only a single multiplication	922	// code will likely be branchless and use only a single multiplication
705	// works for numbers up to 59074	923	// 4.28 works for numbers up to 59074
706	I32 i = SvIVX (sv);	924	I32 i = SvIVX (sv);
707	U32 u;	925	U32 u;
708	char digit, nz = 0;	926	char digit, nz = 0;
709		927
710	need (enc, 6);	928	need (enc, 6);
…		…
728	}	946	}
729	else	947	else
730	{	948	{
731	// large integer, use the (rather slow) snprintf way.	949	// large integer, use the (rather slow) snprintf way.
732	need (enc, IVUV_MAXCHARS);	950	need (enc, IVUV_MAXCHARS);
733	enc->cur +=	951	enc->cur +=
734	SvIsUV(sv)	952	SvIsUV(sv)
735	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))	953	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
736	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));	954	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
737	}	955	}
738	}	956	}
739	else if (SvROK (sv))	957	else if (SvROK (sv))
740	encode_rv (enc, SvRV (sv));	958	encode_rv (enc, SvRV (sv));
741	else if (!SvOK (sv) \|\| enc->json.flags & F_ALLOW_UNKNOWN)	959	else if (!SvOK (sv) \|\| enc->json.flags & F_ALLOW_UNKNOWN)
742	encode_str (enc, "null", 4, 0);	960	encode_str (enc, "null", 4, 0);
743	else	961	else
744	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",
745	SvPV_nolen (sv), SvFLAGS (sv));	963	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
746	}	964	}
747		965
748	static SV *	966	static SV *
749	encode_json (SV scalar, JSON json)	967	encode_json (SV scalar, JSON json)
750	{	968	{
751	enc_t enc;	969	enc_t enc;
752		970
753	if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))	971	if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
754	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)");
755		973
756	enc.json = *json;	974	enc.json = *json;
757	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	975	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
758	enc.cur = SvPVX (enc.sv);	976	enc.cur = SvPVX (enc.sv);
…		…
820	else	1038	else
821	break;	1039	break;
822	}	1040	}
823	else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)	1041	else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
824	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
825		1043	else
826	++dec->cur;	1044	++dec->cur;
827	}	1045	}
828	}	1046	}
829		1047
830	#define ERR(reason) SB dec->err = reason; goto fail; SE	1048	#define ERR(reason) SB dec->err = reason; goto fail; SE
831		1049
…		…
965	cur++ = dec_cur++;	1183	cur++ = dec_cur++;
966	while (--clen);	1184	while (--clen);
967		1185
968	utf8 = 1;	1186	utf8 = 1;
969	}	1187	}
		1188	else if (ch == '\t' && dec->json.flags & F_RELAXED)
		1189	*cur++ = ch;
970	else	1190	else
971	{	1191	{
972	--dec_cur;	1192	--dec_cur;
973		1193
974	if (!ch)	1194	if (!ch)
…		…
984		1204
985	if (sv)	1205	if (sv)
986	{	1206	{
987	STRLEN cur = SvCUR (sv);	1207	STRLEN cur = SvCUR (sv);
988		1208
989	if (SvLEN (sv) <= cur + len)	1209	if (SvLEN (sv) - cur <= len)
990	SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	1210	json_sv_grow (sv, cur, len);
991		1211
992	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	1212	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
993	SvCUR_set (sv, SvCUR (sv) + len);	1213	SvCUR_set (sv, SvCUR (sv) + len);
994	}	1214	}
995	else	1215	else
…		…
1119	}	1339	}
1120		1340
1121	len -= *start == '-' ? 1 : 0;	1341	len -= *start == '-' ? 1 : 0;
1122		1342
1123	// does not fit into IV or UV, try NV	1343	// does not fit into IV or UV, try NV
1124	if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)	1344	if (len <= NV_DIG)
1125	#if defined (LDBL_DIG)
1126	\|\| (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1127	#endif
1128	)
1129	// fits into NV without loss of precision	1345	// fits into NV without loss of precision
1130	return newSVnv (Atof (start));	1346	return newSVnv (json_atof (start));
1131		1347
1132	// everything else fails, convert it to a string	1348	// everything else fails, convert it to a string
1133	return newSVpvn (start, dec->cur - start);	1349	return newSVpvn (start, dec->cur - start);
1134	}	1350	}
1135		1351
1136	// loss of precision here	1352	// loss of precision here
1137	return newSVnv (Atof (start));	1353	return newSVnv (json_atof (start));
1138		1354
1139	fail:	1355	fail:
1140	return 0;	1356	return 0;
1141	}	1357	}
1142		1358
…		…
1166	if (*dec->cur == ']')	1382	if (*dec->cur == ']')
1167	{	1383	{
1168	++dec->cur;	1384	++dec->cur;
1169	break;	1385	break;
1170	}	1386	}
1171		1387
1172	if (*dec->cur != ',')	1388	if (*dec->cur != ',')
1173	ERR (", or ] expected while parsing array");	1389	ERR (", or ] expected while parsing array");
1174		1390
1175	++dec->cur;	1391	++dec->cur;
1176		1392
…		…
1290		1506
1291	DEC_DEC_DEPTH;	1507	DEC_DEC_DEPTH;
1292	sv = newRV_noinc ((SV *)hv);	1508	sv = newRV_noinc ((SV *)hv);
1293		1509
1294	// check filter callbacks	1510	// check filter callbacks
1295	if (dec->json.flags & F_HOOK)	1511	if (expect_false (dec->json.flags & F_HOOK))
1296	{	1512	{
1297	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)	1513	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1298	{	1514	{
1299	HE cb, he;	1515	HE cb, he;
1300		1516
1301	hv_iterinit (hv);	1517	hv_iterinit (hv);
1302	he = hv_iternext (hv);	1518	he = hv_iternext (hv);
1303	hv_iterinit (hv);	1519	hv_iterinit (hv);
1304		1520
1305	// the next line creates a mortal sv each time its called.	1521	// the next line creates a mortal sv each time it's called.
1306	// might want to optimise this for common cases.	1522	// might want to optimise this for common cases.
1307	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);
1308		1524
1309	if (cb)	1525	if (cb)
1310	{	1526	{
1311	dSP;	1527	dSP;
1312	int count;	1528	int count;
1313		1529
1314	ENTER; SAVETMPS; PUSHMARK (SP);	1530	ENTER; SAVETMPS;
		1531	PUSHMARK (SP);
1315	XPUSHs (HeVAL (he));	1532	XPUSHs (HeVAL (he));
		1533	sv_2mortal (sv);
1316		1534
1317	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;	1535	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1318		1536
1319	if (count == 1)	1537	if (count == 1)
1320	{	1538	{
1321	sv = newSVsv (POPs);	1539	sv = newSVsv (POPs);
		1540	PUTBACK;
1322	FREETMPS; LEAVE;	1541	FREETMPS; LEAVE;
1323	return sv;	1542	return sv;
1324	}	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);
1325		1550
1326	FREETMPS; LEAVE;	1551	FREETMPS; LEAVE;
1327	}	1552	}
1328	}	1553	}
1329		1554
1330	if (dec->json.cb_object)	1555	if (dec->json.cb_object)
1331	{	1556	{
1332	dSP;	1557	dSP;
1333	int count;	1558	int count;
1334		1559
1335	ENTER; SAVETMPS; PUSHMARK (SP);	1560	ENTER; SAVETMPS;
		1561	PUSHMARK (SP);
1336	XPUSHs (sv_2mortal (sv));	1562	XPUSHs (sv_2mortal (sv));
1337		1563
1338	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;	1564	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1339		1565
1340	if (count == 1)	1566	if (count == 1)
1341	{
1342	sv = newSVsv (POPs);	1567	sv = newSVsv (POPs);
1343	FREETMPS; LEAVE;	1568	else if (count == 0)
1344	return sv;
1345	}
1346
1347	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
1348	FREETMPS; LEAVE;	1575	FREETMPS; LEAVE;
1349	}	1576	}
1350	}	1577	}
1351		1578
1352	return sv;	1579	return sv;
…		…
1356	DEC_DEC_DEPTH;	1583	DEC_DEC_DEPTH;
1357	return 0;	1584	return 0;
1358	}	1585	}
1359		1586
1360	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 *
1361	decode_sv (dec_t *dec)	1671	decode_sv (dec_t *dec)
1362	{	1672	{
1363	// the beauty of JSON: you need exactly one character lookahead	1673	// the beauty of JSON: you need exactly one character lookahead
1364	// to parse everything.	1674	// to parse everything.
1365	switch (*dec->cur)	1675	switch (*dec->cur)
1366	{	1676	{
1367	case '"': ++dec->cur; return decode_str (dec);	1677	case '"': ++dec->cur; return decode_str (dec);
1368	case '[': ++dec->cur; return decode_av (dec);	1678	case '[': ++dec->cur; return decode_av (dec);
1369	case '{': ++dec->cur; return decode_hv (dec);	1679	case '{': ++dec->cur; return decode_hv (dec);
		1680	case '(': return decode_tag (dec);
1370		1681
1371	case '-':	1682	case '-':
1372	case '0': case '1': case '2': case '3': case '4':	1683	case '0': case '1': case '2': case '3': case '4':
1373	case '5': case '6': case '7': case '8': case '9':	1684	case '5': case '6': case '7': case '8': case '9':
1374	return decode_num (dec);	1685	return decode_num (dec);
1375		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
1376	case 't':	1702	case 't':
1377	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1703	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1378	{	1704	{
1379	dec->cur += 4;	1705	dec->cur += 4;
1380	#if JSON_SLOW	1706
1381	json_true = get_bool ("JSON::XS::true");	1707	if (expect_false (!dec->json.v_true))
1382	#endif	1708	dec->json.v_true = GET_BOOL (true);
		1709
1383	return newSVsv (json_true);	1710	return newSVsv (dec->json.v_true);
1384	}	1711	}
1385	else	1712	else
1386	ERR ("'true' expected");	1713	ERR ("'true' expected");
1387
1388	break;
1389
1390	case 'f':
1391	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1392	{
1393	dec->cur += 5;
1394	#if JSON_SLOW
1395	json_false = get_bool ("JSON::XS::false");
1396	#endif
1397	return newSVsv (json_false);
1398	}
1399	else
1400	ERR ("'false' expected");
1401		1714
1402	break;	1715	break;
1403		1716
1404	case 'n':	1717	case 'n':
1405	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))	1718	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
…		…
1411	ERR ("'null' expected");	1724	ERR ("'null' expected");
1412		1725
1413	break;	1726	break;
1414		1727
1415	default:	1728	default:
1416	ERR ("malformed JSON string, neither array, object, number, string or atom");	1729	ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1417	break;	1730	break;
1418	}	1731	}
1419		1732
1420	fail:	1733	fail:
1421	return 0;	1734	return 0;
1422	}	1735	}
1423		1736
1424	static SV *	1737	static SV *
1425	decode_json (SV string, JSON json, char **offset_return)	1738	decode_json (SV string, JSON json, STRLEN *offset_return)
1426	{	1739	{
1427	dec_t dec;	1740	dec_t dec;
1428	SV *sv;	1741	SV *sv;
1429		1742
1430	/* work around bugs in 5.10 where manipulating magic values	1743	/* work around bugs in 5.10 where manipulating magic values
1431	* will perl ignore the magic in subsequent accesses	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.
1432	*/	1750	*/
1433	/SvGETMAGIC (string);/	1751	/SvGETMAGIC (string);/
1434	if (SvMAGICAL (string))	1752	if (SvMAGICAL (string) \|\| !SvPOK (string) \|\| SvIsCOW_shared_hash (string))
1435	string = sv_2mortal (newSVsv (string));	1753	string = sv_2mortal (newSVsv (string));
1436		1754
1437	SvUPGRADE (string, SVt_PV);	1755	SvUPGRADE (string, SVt_PV);
1438		1756
1439	/* work around a bug in perl 5.10, which causes SvCUR to fail an	1757	/* work around a bug in perl 5.10, which causes SvCUR to fail an
…		…
1456	if (offset > json->max_size && json->max_size)	1774	if (offset > json->max_size && json->max_size)
1457	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",
1458	(unsigned long)SvCUR (string), (unsigned long)json->max_size);	1776	(unsigned long)SvCUR (string), (unsigned long)json->max_size);
1459	}	1777	}
1460		1778
1461	if (json->flags & F_UTF8)	1779	if (DECODE_WANTS_OCTETS (json))
1462	sv_utf8_downgrade (string, 0);	1780	sv_utf8_downgrade (string, 0);
1463	else	1781	else
1464	sv_utf8_upgrade (string);	1782	sv_utf8_upgrade (string);
1465		1783
1466	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1784	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
…		…
1478		1796
1479	decode_ws (&dec);	1797	decode_ws (&dec);
1480	sv = decode_sv (&dec);	1798	sv = decode_sv (&dec);
1481		1799
1482	if (offset_return)	1800	if (offset_return)
1483	*offset_return = dec.cur;	1801	*offset_return = dec.cur - SvPVX (string);
1484		1802	else if (sv)
1485	if (!(offset_return \|\| !sv))
1486	{	1803	{
1487	// check for trailing garbage	1804	// check for trailing garbage
1488	decode_ws (&dec);	1805	decode_ws (&dec);
1489		1806
1490	if (*dec.cur)	1807	if (dec.cur != dec.end)
1491	{	1808	{
1492	dec.err = "garbage after JSON object";	1809	dec.err = "garbage after JSON object";
1493	SvREFCNT_dec (sv);	1810	SvREFCNT_dec (sv);
1494	sv = 0;	1811	sv = 0;
1495	}	1812	}
…		…
1508	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);
1509	LEAVE;	1826	LEAVE;
1510		1827
1511	croak ("%s, at character offset %d (before \"%s\")",	1828	croak ("%s, at character offset %d (before \"%s\")",
1512	dec.err,	1829	dec.err,
1513	ptr_to_index (string, dec.cur),	1830	(int)ptr_to_index (string, dec.cur),
1514	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1831	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1515	}	1832	}
1516		1833
1517	sv = sv_2mortal (sv);	1834	sv = sv_2mortal (sv);
1518		1835
1519	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))	1836	if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1520	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)");
1521		1838
1522	return sv;	1839	return sv;
1523	}	1840	}
1524		1841
…		…
1533	// the state machine here is a bit convoluted and could be simplified a lot	1850	// the state machine here is a bit convoluted and could be simplified a lot
1534	// but this would make it slower, so...	1851	// but this would make it slower, so...
1535		1852
1536	for (;;)	1853	for (;;)
1537	{	1854	{
1538	//printf ("loop pod %d p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), p, p, self->incr_mode, self->incr_nest);//D
1539	switch (self->incr_mode)	1855	switch (self->incr_mode)
1540	{	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
1541	// only used for initial whitespace skipping	1866	// only used for initial whitespace skipping
1542	case INCR_M_WS:	1867	case INCR_M_WS:
1543	for (;;)	1868	for (;;)
1544	{	1869	{
1545	if (*p > 0x20)	1870	if (*p > 0x20)
…		…
1587	++p;	1912	++p;
1588	}	1913	}
1589		1914
1590	break;	1915	break;
1591		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	}
		1950
1592	// inside a string	1951	// inside a string
1593	case INCR_M_STR:	1952	case INCR_M_STR:
1594	incr_m_str:	1953	incr_m_str:
1595	for (;;)	1954	for (;;)
1596	{	1955	{
1597	if (*p == '"')	1956	if (*p == '"')
1598	{	1957	{
1599	++p;	1958	++p;
1600	self->incr_mode = INCR_M_JSON;
1601
1602	if (!self->incr_nest)
1603	goto interrupt;
1604
1605	goto incr_m_json;	1959	goto end_of_scalar;
1606	}	1960	}
1607	else if (*p == '\\')	1961	else if (*p == '\\')
1608	{	1962	{
1609	++p; // "virtually" consumes character after \	1963	++p; // "virtually" consumes character after \
1610		1964
…		…
1640	--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
1641	goto interrupt;	1995	goto interrupt;
1642	}	1996	}
1643	break;	1997	break;
1644		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
1645	case '"':	2014	case '"':
1646	self->incr_mode = INCR_M_STR;	2015	self->incr_mode = INCR_M_STR;
1647	goto incr_m_str;	2016	goto incr_m_str;
1648		2017
1649	case '[':	2018	case '[':
1650	case '{':	2019	case '{':
		2020	case '(':
1651	if (++self->incr_nest > self->max_depth)	2021	if (++self->incr_nest > self->max_depth)
1652	croak (ERR_NESTING_EXCEEDED);	2022	croak (ERR_NESTING_EXCEEDED);
1653	break;	2023	break;
1654		2024
1655	case ']':	2025	case ']':
1656	case '}':	2026	case '}':
1657	if (--self->incr_nest <= 0)	2027	if (--self->incr_nest <= 0)
1658	goto interrupt;	2028	goto interrupt;
		2029	break;
		2030
		2031	case ')':
		2032	--self->incr_nest;
1659	break;	2033	break;
1660		2034
1661	case '#':	2035	case '#':
1662	self->incr_mode = INCR_M_C1;	2036	self->incr_mode = INCR_M_C1;
1663	goto incr_m_c;	2037	goto incr_m_c;
…		…
1689	i >= '0' && i <= '9' ? i - '0'	2063	i >= '0' && i <= '9' ? i - '0'
1690	: i >= 'a' && i <= 'f' ? i - 'a' + 10	2064	: i >= 'a' && i <= 'f' ? i - 'a' + 10
1691	: i >= 'A' && i <= 'F' ? i - 'A' + 10	2065	: i >= 'A' && i <= 'F' ? i - 'A' + 10
1692	: -1;	2066	: -1;
1693		2067
1694	json_stash = gv_stashpv ("JSON::XS" , 1);	2068	json_stash = gv_stashpv ("JSON::XS" , 1);
1695	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);	2069	bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1696
1697	json_true = get_bool ("JSON::XS::true");
1698	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);
1699		2075
1700	CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */	2076	CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1701	}	2077	}
1702		2078
1703	PROTOTYPES: DISABLE	2079	PROTOTYPES: DISABLE
1704		2080
1705	void CLONE (...)	2081	void CLONE (...)
1706	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.
1707	json_stash = 0;	2086	json_stash = 0;
1708	json_boolean_stash = 0;	2087	bool_stash = 0;
		2088	bool_false = 0;
		2089	bool_true = 0;
1709		2090
1710	void new (char *klass)	2091	void new (char *klass)
1711	PPCODE:	2092	PPCODE:
1712	{	2093	{
1713	SV *pv = NEWSV (0, sizeof (JSON));	2094	SV *pv = NEWSV (0, sizeof (JSON));
1714	SvPOK_only (pv);	2095	SvPOK_only (pv);
1715	json_init ((JSON *)SvPVX (pv));	2096	json_init ((JSON *)SvPVX (pv));
1716	XPUSHs (sv_2mortal (sv_bless (	2097	XPUSHs (sv_2mortal (sv_bless (
1717	newRV_noinc (pv),	2098	newRV_noinc (pv),
1718	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)	2099	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1719	)));	2100	)));
1720	}	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	}
1721		2117
1722	void ascii (JSON *self, int enable = 1)	2118	void ascii (JSON *self, int enable = 1)
1723	ALIAS:	2119	ALIAS:
1724	ascii = F_ASCII	2120	ascii = F_ASCII
1725	latin1 = F_LATIN1	2121	latin1 = F_LATIN1
…		…
1733	shrink = F_SHRINK	2129	shrink = F_SHRINK
1734	allow_blessed = F_ALLOW_BLESSED	2130	allow_blessed = F_ALLOW_BLESSED
1735	convert_blessed = F_CONV_BLESSED	2131	convert_blessed = F_CONV_BLESSED
1736	relaxed = F_RELAXED	2132	relaxed = F_RELAXED
1737	allow_unknown = F_ALLOW_UNKNOWN	2133	allow_unknown = F_ALLOW_UNKNOWN
		2134	allow_tags = F_ALLOW_TAGS
1738	PPCODE:	2135	PPCODE:
1739	{	2136	{
1740	if (enable)	2137	if (enable)
1741	self->flags \|= ix;	2138	self->flags \|= ix;
1742	else	2139	else
…		…
1758	get_shrink = F_SHRINK	2155	get_shrink = F_SHRINK
1759	get_allow_blessed = F_ALLOW_BLESSED	2156	get_allow_blessed = F_ALLOW_BLESSED
1760	get_convert_blessed = F_CONV_BLESSED	2157	get_convert_blessed = F_CONV_BLESSED
1761	get_relaxed = F_RELAXED	2158	get_relaxed = F_RELAXED
1762	get_allow_unknown = F_ALLOW_UNKNOWN	2159	get_allow_unknown = F_ALLOW_UNKNOWN
		2160	get_allow_tags = F_ALLOW_TAGS
1763	PPCODE:	2161	PPCODE:
1764	XPUSHs (boolSV (self->flags & ix));	2162	XPUSHs (boolSV (self->flags & ix));
1765		2163
1766	void max_depth (JSON *self, U32 max_depth = 0x80000000UL)	2164	void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1767	PPCODE:	2165	PPCODE:
…		…
1795	}	2193	}
1796		2194
1797	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)
1798	PPCODE:	2196	PPCODE:
1799	{	2197	{
1800	if (!self->cb_sk_object)	2198	if (!self->cb_sk_object)
1801	self->cb_sk_object = newHV ();	2199	self->cb_sk_object = newHV ();
1802		2200
1803	if (SvOK (cb))	2201	if (SvOK (cb))
1804	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);	2202	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1805	else	2203	else
…		…
1816	XPUSHs (ST (0));	2214	XPUSHs (ST (0));
1817	}	2215	}
1818		2216
1819	void encode (JSON self, SV scalar)	2217	void encode (JSON self, SV scalar)
1820	PPCODE:	2218	PPCODE:
1821	XPUSHs (encode_json (scalar, self));	2219	PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
		2220	XPUSHs (scalar);
1822		2221
1823	void decode (JSON self, SV jsonstr)	2222	void decode (JSON self, SV jsonstr)
1824	PPCODE:	2223	PPCODE:
1825	XPUSHs (decode_json (jsonstr, self, 0));	2224	PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
		2225	XPUSHs (jsonstr);
1826		2226
1827	void decode_prefix (JSON self, SV jsonstr)	2227	void decode_prefix (JSON self, SV jsonstr)
1828	PPCODE:	2228	PPCODE:
1829	{	2229	{
1830	char *offset;	2230	SV *sv;
		2231	STRLEN offset;
		2232	PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1831	EXTEND (SP, 2);	2233	EXTEND (SP, 2);
1832	PUSHs (decode_json (jsonstr, self, &offset));	2234	PUSHs (sv);
1833	PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));	2235	PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
1834	}	2236	}
1835		2237
1836	void incr_parse (JSON self, SV jsonstr = 0)	2238	void incr_parse (JSON self, SV jsonstr = 0)
1837	PPCODE:	2239	PPCODE:
1838	{	2240	{
1839	if (!self->incr_text)	2241	if (!self->incr_text)
1840	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	}
1841		2262
1842	// append data, if any	2263	// append data, if any
1843	if (jsonstr)	2264	if (jsonstr)
1844	{	2265	{
		2266	/* make sure both strings have same encoding */
		2267	if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1845	if (SvUTF8 (jsonstr))	2268	if (SvUTF8 (jsonstr))
		2269	sv_utf8_downgrade (jsonstr, 0);
1846	{	2270	else
1847	if (!SvUTF8 (self->incr_text))
1848	{
1849	/* utf-8-ness differs, need to upgrade */
1850	sv_utf8_upgrade (self->incr_text);
1851
1852	if (self->incr_pos)
1853	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1854	- (U8 *)SvPVX (self->incr_text);
1855	}
1856	}
1857	else if (SvUTF8 (self->incr_text))
1858	sv_utf8_upgrade (jsonstr);	2271	sv_utf8_upgrade (jsonstr);
1859		2272
		2273	/* and then just blindly append */
1860	{	2274	{
1861	STRLEN len;	2275	STRLEN len;
1862	const char *str = SvPV (jsonstr, len);	2276	const char *str = SvPV (jsonstr, len);
1863	STRLEN cur = SvCUR (self->incr_text);	2277	STRLEN cur = SvCUR (self->incr_text);
1864		2278
1865	if (SvLEN (self->incr_text) <= cur + len)	2279	if (SvLEN (self->incr_text) - cur <= len)
1866	SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	2280	json_sv_grow (self->incr_text, cur, len);
1867		2281
1868	Move (str, SvEND (self->incr_text), len, char);	2282	Move (str, SvEND (self->incr_text), len, char);
1869	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);	2283	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1870	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there	2284	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1871	}	2285	}
1872	}	2286	}
1873		2287
1874	if (GIMME_V != G_VOID)	2288	if (GIMME_V != G_VOID)
1875	do	2289	do
1876	{	2290	{
		2291	SV *sv;
1877	char *offset;	2292	STRLEN offset;
1878		2293
1879	if (!INCR_DONE (self))	2294	if (!INCR_DONE (self))
1880	{	2295	{
1881	incr_parse (self);	2296	incr_parse (self);
1882		2297
1883	if (self->incr_pos > self->max_size && self->max_size)	2298	if (self->incr_pos > self->max_size && self->max_size)
1884	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",	2299	croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1885	(unsigned long)self->incr_pos, (unsigned long)self->max_size);	2300	(unsigned long)self->incr_pos, (unsigned long)self->max_size);
1886		2301
1887	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
1888	break;	2311	break;
		2312	}
1889	}	2313	}
1890		2314
1891	XPUSHs (decode_json (self->incr_text, self, &offset));	2315	PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
		2316	XPUSHs (sv);
1892		2317
1893	self->incr_pos -= offset - SvPVX (self->incr_text);	2318	self->incr_pos -= offset;
1894	self->incr_nest = 0;	2319	self->incr_nest = 0;
1895	self->incr_mode = 0;	2320	self->incr_mode = 0;
1896		2321
1897	sv_chop (self->incr_text, offset);	2322	sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
1898	}	2323	}
1899	while (GIMME_V == G_ARRAY);	2324	while (GIMME_V == G_ARRAY);
1900	}	2325	}
1901		2326
1902	SV incr_text (JSON self)	2327	SV incr_text (JSON self)
…		…
1933	self->incr_mode = 0;	2358	self->incr_mode = 0;
1934	}	2359	}
1935		2360
1936	void DESTROY (JSON *self)	2361	void DESTROY (JSON *self)
1937	CODE:	2362	CODE:
		2363	SvREFCNT_dec (self->v_false);
		2364	SvREFCNT_dec (self->v_true);
1938	SvREFCNT_dec (self->cb_sk_object);	2365	SvREFCNT_dec (self->cb_sk_object);
1939	SvREFCNT_dec (self->cb_object);	2366	SvREFCNT_dec (self->cb_object);
1940	SvREFCNT_dec (self->incr_text);	2367	SvREFCNT_dec (self->incr_text);
1941		2368
1942	PROTOTYPES: ENABLE	2369	PROTOTYPES: ENABLE
1943		2370
1944	void encode_json (SV *scalar)	2371	void encode_json (SV *scalar)
1945	ALIAS:
1946	to_json_ = 0
1947	encode_json = F_UTF8
1948	PPCODE:	2372	PPCODE:
1949	{	2373	{
1950	JSON json;	2374	JSON json;
1951	json_init (&json);	2375	json_init (&json);
1952	json.flags \|= ix;	2376	json.flags \|= F_UTF8;
1953	XPUSHs (encode_json (scalar, &json));	2377	PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
		2378	XPUSHs (scalar);
1954	}	2379	}
1955		2380
1956	void decode_json (SV *jsonstr)	2381	void decode_json (SV *jsonstr)
1957	ALIAS:
1958	from_json_ = 0
1959	decode_json = F_UTF8
1960	PPCODE:	2382	PPCODE:
1961	{	2383	{
1962	JSON json;	2384	JSON json;
1963	json_init (&json);	2385	json_init (&json);
1964	json.flags \|= ix;	2386	json.flags \|= F_UTF8;
1965	XPUSHs (decode_json (jsonstr, &json, 0));	2387	PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
		2388	XPUSHs (jsonstr);
1966	}	2389	}
1967		2390
1968

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.103 by root, Wed Jan 6 08:01:39 2010 UTC vs. Revision 1.143 by root, Mon Jun 21 01:04:30 2021 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.103 by root, Wed Jan 6 08:01:39 2010 UTC vs.
Revision 1.143 by root, Mon Jun 21 01:04:30 2021 UTC