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

Comparing JSON-XS/XS.xs (file contents):
Revision 1.106 by root, Tue Jan 19 01:36:34 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
…		…
258	}	309	}
259		310
260	// this relies greatly on the quality of the pow ()	311	// this relies greatly on the quality of the pow ()
261	// implementation of the platform, but a good	312	// implementation of the platform, but a good
262	// implementation is hard to beat.	313	// implementation is hard to beat.
		314	// (IEEE 754 conformant ones are required to be exact)
263	if (postdp) *expo -= eaccum;	315	if (postdp) *expo -= eaccum;
264	accum += uaccum Perl_pow (10., *expo);	316	accum += uaccum Perl_pow (10., *expo);
265	*expo += eaccum;	317	*expo += eaccum;
266	}	318	}
267		319
…		…
281	// a recursion depth of ten gives us >>500 bits	333	// a recursion depth of ten gives us >>500 bits
282	json_atof_scan1 (s, &accum, &expo, 0, 10);	334	json_atof_scan1 (s, &accum, &expo, 0, 10);
283		335
284	return neg ? -accum : accum;	336	return neg ? -accum : accum;
285	}	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
286	/////////////////////////////////////////////////////////////////////////////	381	/////////////////////////////////////////////////////////////////////////////
287	// encoder	382	// encoder
288		383
289	// structure used for encoding JSON	384	// structure used for encoding JSON
290	typedef struct	385	typedef struct
…		…
298	} enc_t;	393	} enc_t;
299		394
300	INLINE void	395	INLINE void
301	need (enc_t *enc, STRLEN len)	396	need (enc_t *enc, STRLEN len)
302	{	397	{
303	if (expect_false (enc->cur + len >= enc->end))	398	if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
304	{	399	{
305	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	400	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
306	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	401	char *buf = json_sv_grow (enc->sv, cur, len);
307	enc->cur = SvPVX (enc->sv) + cur;	402	enc->cur = buf + cur;
308	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	403	enc->end = buf + SvLEN (enc->sv) - 1;
309	}	404	}
310	}	405	}
311		406
312	INLINE void	407	INLINE void
313	encode_ch (enc_t *enc, char ch)	408	encode_ch (enc_t *enc, char ch)
…		…
329		424
330	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case	425	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
331	{	426	{
332	if (expect_false (ch == '"')) // but with slow exceptions	427	if (expect_false (ch == '"')) // but with slow exceptions
333	{	428	{
334	need (enc, len += 1);	429	need (enc, len + 1);
335	*enc->cur++ = '\\';	430	*enc->cur++ = '\\';
336	*enc->cur++ = '"';	431	*enc->cur++ = '"';
337	}	432	}
338	else if (expect_false (ch == '\\'))	433	else if (expect_false (ch == '\\'))
339	{	434	{
340	need (enc, len += 1);	435	need (enc, len + 1);
341	*enc->cur++ = '\\';	436	*enc->cur++ = '\\';
342	*enc->cur++ = '\\';	437	*enc->cur++ = '\\';
343	}	438	}
344	else	439	else
345	*enc->cur++ = ch;	440	*enc->cur++ = ch;
…		…
348	}	443	}
349	else	444	else
350	{	445	{
351	switch (ch)	446	switch (ch)
352	{	447	{
353	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;
354	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;
355	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;
356	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;
357	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;
358		453
359	default:	454	default:
360	{	455	{
361	STRLEN clen;	456	STRLEN clen;
362	UV uch;	457	UV uch;
…		…
378	if (uch >= 0x10000UL)	473	if (uch >= 0x10000UL)
379	{	474	{
380	if (uch >= 0x110000UL)	475	if (uch >= 0x110000UL)
381	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);
382		477
383	need (enc, len += 11);	478	need (enc, len + 11);
384	sprintf (enc->cur, "\\u%04x\\u%04x",	479	sprintf (enc->cur, "\\u%04x\\u%04x",
385	(int)((uch - 0x10000) / 0x400 + 0xD800),	480	(int)((uch - 0x10000) / 0x400 + 0xD800),
386	(int)((uch - 0x10000) % 0x400 + 0xDC00));	481	(int)((uch - 0x10000) % 0x400 + 0xDC00));
387	enc->cur += 12;	482	enc->cur += 12;
388	}	483	}
389	else	484	else
390	{	485	{
391	need (enc, len += 5);	486	need (enc, len + 5);
392	*enc->cur++ = '\\';	487	*enc->cur++ = '\\';
393	*enc->cur++ = 'u';	488	*enc->cur++ = 'u';
394	*enc->cur++ = PL_hexdigit [ uch >> 12 ];	489	*enc->cur++ = PL_hexdigit [ uch >> 12 ];
395	*enc->cur++ = PL_hexdigit [(uch >> 8) & 15];	490	*enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
396	*enc->cur++ = PL_hexdigit [(uch >> 4) & 15];	491	*enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
…		…
404	*enc->cur++ = uch;	499	*enc->cur++ = uch;
405	str += clen;	500	str += clen;
406	}	501	}
407	else if (is_utf8)	502	else if (is_utf8)
408	{	503	{
409	need (enc, len += clen);	504	need (enc, len + clen);
410	do	505	do
411	{	506	{
412	enc->cur++ = str++;	507	enc->cur++ = str++;
413	}	508	}
414	while (--clen);	509	while (--clen);
415	}	510	}
416	else	511	else
417	{	512	{
418	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
419	enc->cur = encode_utf8 (enc->cur, uch);	514	enc->cur = encode_utf8 (enc->cur, uch);
420	++str;	515	++str;
421	}	516	}
422	}	517	}
423	}	518	}
…		…
477		572
478	if (enc->indent >= enc->json.max_depth)	573	if (enc->indent >= enc->json.max_depth)
479	croak (ERR_NESTING_EXCEEDED);	574	croak (ERR_NESTING_EXCEEDED);
480		575
481	encode_ch (enc, '[');	576	encode_ch (enc, '[');
482		577
483	if (len >= 0)	578	if (len >= 0)
484	{	579	{
485	encode_nl (enc); ++enc->indent;	580	encode_nl (enc); ++enc->indent;
486		581
487	for (i = 0; i <= len; ++i)	582	for (i = 0; i <= len; ++i)
…		…
499	encode_comma (enc);	594	encode_comma (enc);
500	}	595	}
501		596
502	encode_nl (enc); --enc->indent; encode_indent (enc);	597	encode_nl (enc); --enc->indent; encode_indent (enc);
503	}	598	}
504		599
505	encode_ch (enc, ']');	600	encode_ch (enc, ']');
506	}	601	}
507		602
508	static void	603	static void
509	encode_hk (enc_t enc, HE he)	604	encode_hk (enc_t enc, HE he)
…		…
513	if (HeKLEN (he) == HEf_SVKEY)	608	if (HeKLEN (he) == HEf_SVKEY)
514	{	609	{
515	SV *sv = HeSVKEY (he);	610	SV *sv = HeSVKEY (he);
516	STRLEN len;	611	STRLEN len;
517	char *str;	612	char *str;
518		613
519	SvGETMAGIC (sv);	614	SvGETMAGIC (sv);
520	str = SvPV (sv, len);	615	str = SvPV (sv, len);
521		616
522	encode_str (enc, str, len, SvUTF8 (sv));	617	encode_str (enc, str, len, SvUTF8 (sv));
523	}	618	}
…		…
589	}	684	}
590		685
591	if (count)	686	if (count)
592	{	687	{
593	int i, fast = 1;	688	int i, fast = 1;
594	#if defined(__BORLANDC__) \|\| defined(_MSC_VER)	689	HE *hes_stack [STACK_HES];
595	HE *hes = _alloca (count sizeof (HE));	690	HE **hes = hes_stack;
596	#else	691
597	HE *hes [count]; // if your compiler dies here, you need to enable C99 mode	692	// allocate larger arrays on the heap
598	#endif	693	if (count > STACK_HES)
		694	{
		695	SV sv = sv_2mortal (NEWSV (0, count sizeof (*hes)));
		696	hes = (HE **)SvPVX (sv);
		697	}
599		698
600	i = 0;	699	i = 0;
601	while ((he = hv_iternext (hv)))	700	while ((he = hv_iternext (hv)))
602	{	701	{
603	hes [i++] = he;	702	hes [i++] = he;
…		…
672	// encode objects, arrays and special \0=false and \1=true values.	771	// encode objects, arrays and special \0=false and \1=true values.
673	static void	772	static void
674	encode_rv (enc_t enc, SV sv)	773	encode_rv (enc_t enc, SV sv)
675	{	774	{
676	svtype svt;	775	svtype svt;
		776	GV *method;
677		777
678	SvGETMAGIC (sv);	778	SvGETMAGIC (sv);
679	svt = SvTYPE (sv);	779	svt = SvTYPE (sv);
680		780
681	if (expect_false (SvOBJECT (sv)))	781	if (expect_false (SvOBJECT (sv)))
682	{	782	{
683	HV *stash = !JSON_SLOW \|\| json_boolean_stash	783	HV *stash = SvSTASH (sv);
684	? json_boolean_stash
685	: gv_stashpv ("JSON::XS::Boolean", 1);
686		784
687	if (SvSTASH (sv) == stash)	785	if (stash == bool_stash)
688	{	786	{
689	if (SvIV (sv))
690	encode_str (enc, "true", 4, 0);	787	if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
691	else
692	encode_str (enc, "false", 5, 0);	788	else encode_str (enc, "false", 5, 0);
693	}	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);
694	else	864	else
695	{
696	#if 0
697	if (0 && sv_derived_from (rv, "JSON::Literal"))
698	{
699	// not yet
700	}
701	#endif
702	if (enc->json.flags & F_CONV_BLESSED)
703	{
704	// we re-bless the reference to get overload and other niceties right
705	GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
706
707	if (to_json)
708	{
709	dSP;
710
711	ENTER; SAVETMPS; PUSHMARK (SP);
712	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
713
714	// calling with G_SCALAR ensures that we always get a 1 return value
715	PUTBACK;
716	call_sv ((SV *)GvCV (to_json), G_SCALAR);
717	SPAGAIN;
718
719	// catch this surprisingly common error
720	if (SvROK (TOPs) && SvRV (TOPs) == sv)
721	croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
722
723	sv = POPs;
724	PUTBACK;
725
726	encode_sv (enc, sv);
727
728	FREETMPS; LEAVE;
729	}
730	else if (enc->json.flags & F_ALLOW_BLESSED)
731	encode_str (enc, "null", 4, 0);
732	else
733	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
734	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
735	}
736	else if (enc->json.flags & F_ALLOW_BLESSED)
737	encode_str (enc, "null", 4, 0);
738	else
739	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)",
740	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	866	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
741	}
742	}	867	}
743	else if (svt == SVt_PVHV)	868	else if (svt == SVt_PVHV)
744	encode_hv (enc, (HV *)sv);	869	encode_hv (enc, (HV *)sv);
745	else if (svt == SVt_PVAV)	870	else if (svt == SVt_PVAV)
746	encode_av (enc, (AV *)sv);	871	encode_av (enc, (AV *)sv);
747	else if (svt < SVt_PVAV)	872	else if (svt < SVt_PVAV)
748	{	873	{
749	STRLEN len = 0;	874	int bool_type = ref_bool_type (sv);
750	char *pv = svt ? SvPV (sv, len) : 0;
751		875
752	if (len == 1 && *pv == '1')	876	if (bool_type == 1)
753	encode_str (enc, "true", 4, 0);	877	encode_str (enc, "true", 4, 0);
754	else if (len == 1 && *pv == '0')	878	else if (bool_type == 0)
755	encode_str (enc, "false", 5, 0);	879	encode_str (enc, "false", 5, 0);
756	else if (enc->json.flags & F_ALLOW_UNKNOWN)	880	else if (enc->json.flags & F_ALLOW_UNKNOWN)
757	encode_str (enc, "null", 4, 0);	881	encode_str (enc, "null", 4, 0);
758	else	882	else
759	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",
…		…
794	if (SvIsUV (sv) ? SvUVX (sv) <= 59000	918	if (SvIsUV (sv) ? SvUVX (sv) <= 59000
795	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)	919	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
796	{	920	{
797	// optimise the "small number case"	921	// optimise the "small number case"
798	// code will likely be branchless and use only a single multiplication	922	// code will likely be branchless and use only a single multiplication
799	// works for numbers up to 59074	923	// 4.28 works for numbers up to 59074
800	I32 i = SvIVX (sv);	924	I32 i = SvIVX (sv);
801	U32 u;	925	U32 u;
802	char digit, nz = 0;	926	char digit, nz = 0;
803		927
804	need (enc, 6);	928	need (enc, 6);
…		…
822	}	946	}
823	else	947	else
824	{	948	{
825	// large integer, use the (rather slow) snprintf way.	949	// large integer, use the (rather slow) snprintf way.
826	need (enc, IVUV_MAXCHARS);	950	need (enc, IVUV_MAXCHARS);
827	enc->cur +=	951	enc->cur +=
828	SvIsUV(sv)	952	SvIsUV(sv)
829	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))	953	? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
830	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));	954	: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
831	}	955	}
832	}	956	}
833	else if (SvROK (sv))	957	else if (SvROK (sv))
834	encode_rv (enc, SvRV (sv));	958	encode_rv (enc, SvRV (sv));
835	else if (!SvOK (sv) \|\| enc->json.flags & F_ALLOW_UNKNOWN)	959	else if (!SvOK (sv) \|\| enc->json.flags & F_ALLOW_UNKNOWN)
836	encode_str (enc, "null", 4, 0);	960	encode_str (enc, "null", 4, 0);
837	else	961	else
838	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",
839	SvPV_nolen (sv), SvFLAGS (sv));	963	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
840	}	964	}
841		965
842	static SV *	966	static SV *
843	encode_json (SV scalar, JSON json)	967	encode_json (SV scalar, JSON json)
844	{	968	{
845	enc_t enc;	969	enc_t enc;
846		970
847	if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))	971	if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
848	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)");
849		973
850	enc.json = *json;	974	enc.json = *json;
851	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	975	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
852	enc.cur = SvPVX (enc.sv);	976	enc.cur = SvPVX (enc.sv);
…		…
914	else	1038	else
915	break;	1039	break;
916	}	1040	}
917	else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)	1041	else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
918	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
919		1043	else
920	++dec->cur;	1044	++dec->cur;
921	}	1045	}
922	}	1046	}
923		1047
924	#define ERR(reason) SB dec->err = reason; goto fail; SE	1048	#define ERR(reason) SB dec->err = reason; goto fail; SE
925		1049
…		…
1059	cur++ = dec_cur++;	1183	cur++ = dec_cur++;
1060	while (--clen);	1184	while (--clen);
1061		1185
1062	utf8 = 1;	1186	utf8 = 1;
1063	}	1187	}
		1188	else if (ch == '\t' && dec->json.flags & F_RELAXED)
		1189	*cur++ = ch;
1064	else	1190	else
1065	{	1191	{
1066	--dec_cur;	1192	--dec_cur;
1067		1193
1068	if (!ch)	1194	if (!ch)
…		…
1078		1204
1079	if (sv)	1205	if (sv)
1080	{	1206	{
1081	STRLEN cur = SvCUR (sv);	1207	STRLEN cur = SvCUR (sv);
1082		1208
1083	if (SvLEN (sv) <= cur + len)	1209	if (SvLEN (sv) - cur <= len)
1084	SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	1210	json_sv_grow (sv, cur, len);
1085		1211
1086	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	1212	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1087	SvCUR_set (sv, SvCUR (sv) + len);	1213	SvCUR_set (sv, SvCUR (sv) + len);
1088	}	1214	}
1089	else	1215	else
…		…
1256	if (*dec->cur == ']')	1382	if (*dec->cur == ']')
1257	{	1383	{
1258	++dec->cur;	1384	++dec->cur;
1259	break;	1385	break;
1260	}	1386	}
1261		1387
1262	if (*dec->cur != ',')	1388	if (*dec->cur != ',')
1263	ERR (", or ] expected while parsing array");	1389	ERR (", or ] expected while parsing array");
1264		1390
1265	++dec->cur;	1391	++dec->cur;
1266		1392
…		…
1380		1506
1381	DEC_DEC_DEPTH;	1507	DEC_DEC_DEPTH;
1382	sv = newRV_noinc ((SV *)hv);	1508	sv = newRV_noinc ((SV *)hv);
1383		1509
1384	// check filter callbacks	1510	// check filter callbacks
1385	if (dec->json.flags & F_HOOK)	1511	if (expect_false (dec->json.flags & F_HOOK))
1386	{	1512	{
1387	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)	1513	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1388	{	1514	{
1389	HE cb, he;	1515	HE cb, he;
1390		1516
1391	hv_iterinit (hv);	1517	hv_iterinit (hv);
1392	he = hv_iternext (hv);	1518	he = hv_iternext (hv);
1393	hv_iterinit (hv);	1519	hv_iterinit (hv);
1394		1520
1395	// the next line creates a mortal sv each time its called.	1521	// the next line creates a mortal sv each time it's called.
1396	// might want to optimise this for common cases.	1522	// might want to optimise this for common cases.
1397	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);
1398		1524
1399	if (cb)	1525	if (cb)
1400	{	1526	{
1401	dSP;	1527	dSP;
1402	int count;	1528	int count;
1403		1529
1404	ENTER; SAVETMPS; PUSHMARK (SP);	1530	ENTER; SAVETMPS;
		1531	PUSHMARK (SP);
1405	XPUSHs (HeVAL (he));	1532	XPUSHs (HeVAL (he));
		1533	sv_2mortal (sv);
1406		1534
1407	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;	1535	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1408		1536
1409	if (count == 1)	1537	if (count == 1)
1410	{	1538	{
1411	sv = newSVsv (POPs);	1539	sv = newSVsv (POPs);
		1540	PUTBACK;
1412	FREETMPS; LEAVE;	1541	FREETMPS; LEAVE;
1413	return sv;	1542	return sv;
1414	}	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);
1415		1550
1416	FREETMPS; LEAVE;	1551	FREETMPS; LEAVE;
1417	}	1552	}
1418	}	1553	}
1419		1554
1420	if (dec->json.cb_object)	1555	if (dec->json.cb_object)
1421	{	1556	{
1422	dSP;	1557	dSP;
1423	int count;	1558	int count;
1424		1559
1425	ENTER; SAVETMPS; PUSHMARK (SP);	1560	ENTER; SAVETMPS;
		1561	PUSHMARK (SP);
1426	XPUSHs (sv_2mortal (sv));	1562	XPUSHs (sv_2mortal (sv));
1427		1563
1428	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;	1564	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1429		1565
1430	if (count == 1)	1566	if (count == 1)
1431	{
1432	sv = newSVsv (POPs);	1567	sv = newSVsv (POPs);
1433	FREETMPS; LEAVE;	1568	else if (count == 0)
1434	return sv;
1435	}
1436
1437	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
1438	FREETMPS; LEAVE;	1575	FREETMPS; LEAVE;
1439	}	1576	}
1440	}	1577	}
1441		1578
1442	return sv;	1579	return sv;
…		…
1446	DEC_DEC_DEPTH;	1583	DEC_DEC_DEPTH;
1447	return 0;	1584	return 0;
1448	}	1585	}
1449		1586
1450	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 *
1451	decode_sv (dec_t *dec)	1671	decode_sv (dec_t *dec)
1452	{	1672	{
1453	// the beauty of JSON: you need exactly one character lookahead	1673	// the beauty of JSON: you need exactly one character lookahead
1454	// to parse everything.	1674	// to parse everything.
1455	switch (*dec->cur)	1675	switch (*dec->cur)
1456	{	1676	{
1457	case '"': ++dec->cur; return decode_str (dec);	1677	case '"': ++dec->cur; return decode_str (dec);
1458	case '[': ++dec->cur; return decode_av (dec);	1678	case '[': ++dec->cur; return decode_av (dec);
1459	case '{': ++dec->cur; return decode_hv (dec);	1679	case '{': ++dec->cur; return decode_hv (dec);
		1680	case '(': return decode_tag (dec);
1460		1681
1461	case '-':	1682	case '-':
1462	case '0': case '1': case '2': case '3': case '4':	1683	case '0': case '1': case '2': case '3': case '4':
1463	case '5': case '6': case '7': case '8': case '9':	1684	case '5': case '6': case '7': case '8': case '9':
1464	return decode_num (dec);	1685	return decode_num (dec);
1465		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
1466	case 't':	1702	case 't':
1467	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1703	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1468	{	1704	{
1469	dec->cur += 4;	1705	dec->cur += 4;
1470	#if JSON_SLOW	1706
1471	json_true = get_bool ("JSON::XS::true");	1707	if (expect_false (!dec->json.v_true))
1472	#endif	1708	dec->json.v_true = GET_BOOL (true);
		1709
1473	return newSVsv (json_true);	1710	return newSVsv (dec->json.v_true);
1474	}	1711	}
1475	else	1712	else
1476	ERR ("'true' expected");	1713	ERR ("'true' expected");
1477
1478	break;
1479
1480	case 'f':
1481	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1482	{
1483	dec->cur += 5;
1484	#if JSON_SLOW
1485	json_false = get_bool ("JSON::XS::false");
1486	#endif
1487	return newSVsv (json_false);
1488	}
1489	else
1490	ERR ("'false' expected");
1491		1714
1492	break;	1715	break;
1493		1716
1494	case 'n':	1717	case 'n':
1495	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))	1718	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
…		…
1501	ERR ("'null' expected");	1724	ERR ("'null' expected");
1502		1725
1503	break;	1726	break;
1504		1727
1505	default:	1728	default:
1506	ERR ("malformed JSON string, neither array, object, number, string or atom");	1729	ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1507	break;	1730	break;
1508	}	1731	}
1509		1732
1510	fail:	1733	fail:
1511	return 0;	1734	return 0;
1512	}	1735	}
1513		1736
1514	static SV *	1737	static SV *
1515	decode_json (SV string, JSON json, char **offset_return)	1738	decode_json (SV string, JSON json, STRLEN *offset_return)
1516	{	1739	{
1517	dec_t dec;	1740	dec_t dec;
1518	SV *sv;	1741	SV *sv;
1519		1742
1520	/* work around bugs in 5.10 where manipulating magic values	1743	/* work around bugs in 5.10 where manipulating magic values
1521	* 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.
1522	*/	1750	*/
1523	/SvGETMAGIC (string);/	1751	/SvGETMAGIC (string);/
1524	if (SvMAGICAL (string))	1752	if (SvMAGICAL (string) \|\| !SvPOK (string) \|\| SvIsCOW_shared_hash (string))
1525	string = sv_2mortal (newSVsv (string));	1753	string = sv_2mortal (newSVsv (string));
1526		1754
1527	SvUPGRADE (string, SVt_PV);	1755	SvUPGRADE (string, SVt_PV);
1528		1756
1529	/* 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
…		…
1546	if (offset > json->max_size && json->max_size)	1774	if (offset > json->max_size && json->max_size)
1547	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",
1548	(unsigned long)SvCUR (string), (unsigned long)json->max_size);	1776	(unsigned long)SvCUR (string), (unsigned long)json->max_size);
1549	}	1777	}
1550		1778
1551	if (json->flags & F_UTF8)	1779	if (DECODE_WANTS_OCTETS (json))
1552	sv_utf8_downgrade (string, 0);	1780	sv_utf8_downgrade (string, 0);
1553	else	1781	else
1554	sv_utf8_upgrade (string);	1782	sv_utf8_upgrade (string);
1555		1783
1556	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP	1784	SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
…		…
1568		1796
1569	decode_ws (&dec);	1797	decode_ws (&dec);
1570	sv = decode_sv (&dec);	1798	sv = decode_sv (&dec);
1571		1799
1572	if (offset_return)	1800	if (offset_return)
1573	*offset_return = dec.cur;	1801	*offset_return = dec.cur - SvPVX (string);
1574		1802	else if (sv)
1575	if (!(offset_return \|\| !sv))
1576	{	1803	{
1577	// check for trailing garbage	1804	// check for trailing garbage
1578	decode_ws (&dec);	1805	decode_ws (&dec);
1579		1806
1580	if (*dec.cur)	1807	if (dec.cur != dec.end)
1581	{	1808	{
1582	dec.err = "garbage after JSON object";	1809	dec.err = "garbage after JSON object";
1583	SvREFCNT_dec (sv);	1810	SvREFCNT_dec (sv);
1584	sv = 0;	1811	sv = 0;
1585	}	1812	}
…		…
1598	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);
1599	LEAVE;	1826	LEAVE;
1600		1827
1601	croak ("%s, at character offset %d (before \"%s\")",	1828	croak ("%s, at character offset %d (before \"%s\")",
1602	dec.err,	1829	dec.err,
1603	ptr_to_index (string, dec.cur),	1830	(int)ptr_to_index (string, dec.cur),
1604	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");	1831	dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1605	}	1832	}
1606		1833
1607	sv = sv_2mortal (sv);	1834	sv = sv_2mortal (sv);
1608		1835
1609	if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))	1836	if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1610	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)");
1611		1838
1612	return sv;	1839	return sv;
1613	}	1840	}
1614		1841
…		…
1623	// 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
1624	// but this would make it slower, so...	1851	// but this would make it slower, so...
1625		1852
1626	for (;;)	1853	for (;;)
1627	{	1854	{
1628	//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
1629	switch (self->incr_mode)	1855	switch (self->incr_mode)
1630	{	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
1631	// only used for initial whitespace skipping	1866	// only used for initial whitespace skipping
1632	case INCR_M_WS:	1867	case INCR_M_WS:
1633	for (;;)	1868	for (;;)
1634	{	1869	{
1635	if (*p > 0x20)	1870	if (*p > 0x20)
…		…
1677	++p;	1912	++p;
1678	}	1913	}
1679		1914
1680	break;	1915	break;
1681		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
1682	// inside a string	1951	// inside a string
1683	case INCR_M_STR:	1952	case INCR_M_STR:
1684	incr_m_str:	1953	incr_m_str:
1685	for (;;)	1954	for (;;)
1686	{	1955	{
1687	if (*p == '"')	1956	if (*p == '"')
1688	{	1957	{
1689	++p;	1958	++p;
1690	self->incr_mode = INCR_M_JSON;
1691
1692	if (!self->incr_nest)
1693	goto interrupt;
1694
1695	goto incr_m_json;	1959	goto end_of_scalar;
1696	}	1960	}
1697	else if (*p == '\\')	1961	else if (*p == '\\')
1698	{	1962	{
1699	++p; // "virtually" consumes character after \	1963	++p; // "virtually" consumes character after \
1700		1964
…		…
1730	--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
1731	goto interrupt;	1995	goto interrupt;
1732	}	1996	}
1733	break;	1997	break;
1734		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
1735	case '"':	2014	case '"':
1736	self->incr_mode = INCR_M_STR;	2015	self->incr_mode = INCR_M_STR;
1737	goto incr_m_str;	2016	goto incr_m_str;
1738		2017
1739	case '[':	2018	case '[':
1740	case '{':	2019	case '{':
		2020	case '(':
1741	if (++self->incr_nest > self->max_depth)	2021	if (++self->incr_nest > self->max_depth)
1742	croak (ERR_NESTING_EXCEEDED);	2022	croak (ERR_NESTING_EXCEEDED);
1743	break;	2023	break;
1744		2024
1745	case ']':	2025	case ']':
1746	case '}':	2026	case '}':
1747	if (--self->incr_nest <= 0)	2027	if (--self->incr_nest <= 0)
1748	goto interrupt;	2028	goto interrupt;
		2029	break;
		2030
		2031	case ')':
		2032	--self->incr_nest;
1749	break;	2033	break;
1750		2034
1751	case '#':	2035	case '#':
1752	self->incr_mode = INCR_M_C1;	2036	self->incr_mode = INCR_M_C1;
1753	goto incr_m_c;	2037	goto incr_m_c;
…		…
1779	i >= '0' && i <= '9' ? i - '0'	2063	i >= '0' && i <= '9' ? i - '0'
1780	: i >= 'a' && i <= 'f' ? i - 'a' + 10	2064	: i >= 'a' && i <= 'f' ? i - 'a' + 10
1781	: i >= 'A' && i <= 'F' ? i - 'A' + 10	2065	: i >= 'A' && i <= 'F' ? i - 'A' + 10
1782	: -1;	2066	: -1;
1783		2067
1784	json_stash = gv_stashpv ("JSON::XS" , 1);	2068	json_stash = gv_stashpv ("JSON::XS" , 1);
1785	json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);	2069	bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1786
1787	json_true = get_bool ("JSON::XS::true");
1788	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);
1789		2075
1790	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 */
1791	}	2077	}
1792		2078
1793	PROTOTYPES: DISABLE	2079	PROTOTYPES: DISABLE
1794		2080
1795	void CLONE (...)	2081	void CLONE (...)
1796	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.
1797	json_stash = 0;	2086	json_stash = 0;
1798	json_boolean_stash = 0;	2087	bool_stash = 0;
		2088	bool_false = 0;
		2089	bool_true = 0;
1799		2090
1800	void new (char *klass)	2091	void new (char *klass)
1801	PPCODE:	2092	PPCODE:
1802	{	2093	{
1803	SV *pv = NEWSV (0, sizeof (JSON));	2094	SV *pv = NEWSV (0, sizeof (JSON));
1804	SvPOK_only (pv);	2095	SvPOK_only (pv);
1805	json_init ((JSON *)SvPVX (pv));	2096	json_init ((JSON *)SvPVX (pv));
1806	XPUSHs (sv_2mortal (sv_bless (	2097	XPUSHs (sv_2mortal (sv_bless (
1807	newRV_noinc (pv),	2098	newRV_noinc (pv),
1808	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)	2099	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1809	)));	2100	)));
1810	}	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	}
1811		2117
1812	void ascii (JSON *self, int enable = 1)	2118	void ascii (JSON *self, int enable = 1)
1813	ALIAS:	2119	ALIAS:
1814	ascii = F_ASCII	2120	ascii = F_ASCII
1815	latin1 = F_LATIN1	2121	latin1 = F_LATIN1
…		…
1823	shrink = F_SHRINK	2129	shrink = F_SHRINK
1824	allow_blessed = F_ALLOW_BLESSED	2130	allow_blessed = F_ALLOW_BLESSED
1825	convert_blessed = F_CONV_BLESSED	2131	convert_blessed = F_CONV_BLESSED
1826	relaxed = F_RELAXED	2132	relaxed = F_RELAXED
1827	allow_unknown = F_ALLOW_UNKNOWN	2133	allow_unknown = F_ALLOW_UNKNOWN
		2134	allow_tags = F_ALLOW_TAGS
1828	PPCODE:	2135	PPCODE:
1829	{	2136	{
1830	if (enable)	2137	if (enable)
1831	self->flags \|= ix;	2138	self->flags \|= ix;
1832	else	2139	else
…		…
1848	get_shrink = F_SHRINK	2155	get_shrink = F_SHRINK
1849	get_allow_blessed = F_ALLOW_BLESSED	2156	get_allow_blessed = F_ALLOW_BLESSED
1850	get_convert_blessed = F_CONV_BLESSED	2157	get_convert_blessed = F_CONV_BLESSED
1851	get_relaxed = F_RELAXED	2158	get_relaxed = F_RELAXED
1852	get_allow_unknown = F_ALLOW_UNKNOWN	2159	get_allow_unknown = F_ALLOW_UNKNOWN
		2160	get_allow_tags = F_ALLOW_TAGS
1853	PPCODE:	2161	PPCODE:
1854	XPUSHs (boolSV (self->flags & ix));	2162	XPUSHs (boolSV (self->flags & ix));
1855		2163
1856	void max_depth (JSON *self, U32 max_depth = 0x80000000UL)	2164	void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1857	PPCODE:	2165	PPCODE:
…		…
1885	}	2193	}
1886		2194
1887	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)
1888	PPCODE:	2196	PPCODE:
1889	{	2197	{
1890	if (!self->cb_sk_object)	2198	if (!self->cb_sk_object)
1891	self->cb_sk_object = newHV ();	2199	self->cb_sk_object = newHV ();
1892		2200
1893	if (SvOK (cb))	2201	if (SvOK (cb))
1894	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);	2202	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1895	else	2203	else
…		…
1906	XPUSHs (ST (0));	2214	XPUSHs (ST (0));
1907	}	2215	}
1908		2216
1909	void encode (JSON self, SV scalar)	2217	void encode (JSON self, SV scalar)
1910	PPCODE:	2218	PPCODE:
1911	XPUSHs (encode_json (scalar, self));	2219	PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
		2220	XPUSHs (scalar);
1912		2221
1913	void decode (JSON self, SV jsonstr)	2222	void decode (JSON self, SV jsonstr)
1914	PPCODE:	2223	PPCODE:
1915	XPUSHs (decode_json (jsonstr, self, 0));	2224	PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
		2225	XPUSHs (jsonstr);
1916		2226
1917	void decode_prefix (JSON self, SV jsonstr)	2227	void decode_prefix (JSON self, SV jsonstr)
1918	PPCODE:	2228	PPCODE:
1919	{	2229	{
1920	char *offset;	2230	SV *sv;
		2231	STRLEN offset;
		2232	PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1921	EXTEND (SP, 2);	2233	EXTEND (SP, 2);
1922	PUSHs (decode_json (jsonstr, self, &offset));	2234	PUSHs (sv);
1923	PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));	2235	PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
1924	}	2236	}
1925		2237
1926	void incr_parse (JSON self, SV jsonstr = 0)	2238	void incr_parse (JSON self, SV jsonstr = 0)
1927	PPCODE:	2239	PPCODE:
1928	{	2240	{
1929	if (!self->incr_text)	2241	if (!self->incr_text)
1930	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	}
1931		2262
1932	// append data, if any	2263	// append data, if any
1933	if (jsonstr)	2264	if (jsonstr)
1934	{	2265	{
		2266	/* make sure both strings have same encoding */
		2267	if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1935	if (SvUTF8 (jsonstr))	2268	if (SvUTF8 (jsonstr))
		2269	sv_utf8_downgrade (jsonstr, 0);
1936	{	2270	else
1937	if (!SvUTF8 (self->incr_text))
1938	{
1939	/* utf-8-ness differs, need to upgrade */
1940	sv_utf8_upgrade (self->incr_text);
1941
1942	if (self->incr_pos)
1943	self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1944	- (U8 *)SvPVX (self->incr_text);
1945	}
1946	}
1947	else if (SvUTF8 (self->incr_text))
1948	sv_utf8_upgrade (jsonstr);	2271	sv_utf8_upgrade (jsonstr);
1949		2272
		2273	/* and then just blindly append */
1950	{	2274	{
1951	STRLEN len;	2275	STRLEN len;
1952	const char *str = SvPV (jsonstr, len);	2276	const char *str = SvPV (jsonstr, len);
1953	STRLEN cur = SvCUR (self->incr_text);	2277	STRLEN cur = SvCUR (self->incr_text);
1954		2278
1955	if (SvLEN (self->incr_text) <= cur + len)	2279	if (SvLEN (self->incr_text) - cur <= len)
1956	SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	2280	json_sv_grow (self->incr_text, cur, len);
1957		2281
1958	Move (str, SvEND (self->incr_text), len, char);	2282	Move (str, SvEND (self->incr_text), len, char);
1959	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);	2283	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1960	*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
1961	}	2285	}
1962	}	2286	}
1963		2287
1964	if (GIMME_V != G_VOID)	2288	if (GIMME_V != G_VOID)
1965	do	2289	do
1966	{	2290	{
		2291	SV *sv;
1967	char *offset;	2292	STRLEN offset;
1968		2293
1969	if (!INCR_DONE (self))	2294	if (!INCR_DONE (self))
1970	{	2295	{
1971	incr_parse (self);	2296	incr_parse (self);
1972		2297
1973	if (self->incr_pos > self->max_size && self->max_size)	2298	if (self->incr_pos > self->max_size && self->max_size)
1974	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",
1975	(unsigned long)self->incr_pos, (unsigned long)self->max_size);	2300	(unsigned long)self->incr_pos, (unsigned long)self->max_size);
1976		2301
1977	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
1978	break;	2311	break;
		2312	}
1979	}	2313	}
1980		2314
1981	XPUSHs (decode_json (self->incr_text, self, &offset));	2315	PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
		2316	XPUSHs (sv);
1982		2317
1983	self->incr_pos -= offset - SvPVX (self->incr_text);	2318	self->incr_pos -= offset;
1984	self->incr_nest = 0;	2319	self->incr_nest = 0;
1985	self->incr_mode = 0;	2320	self->incr_mode = 0;
1986		2321
1987	sv_chop (self->incr_text, offset);	2322	sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
1988	}	2323	}
1989	while (GIMME_V == G_ARRAY);	2324	while (GIMME_V == G_ARRAY);
1990	}	2325	}
1991		2326
1992	SV incr_text (JSON self)	2327	SV incr_text (JSON self)
…		…
2023	self->incr_mode = 0;	2358	self->incr_mode = 0;
2024	}	2359	}
2025		2360
2026	void DESTROY (JSON *self)	2361	void DESTROY (JSON *self)
2027	CODE:	2362	CODE:
		2363	SvREFCNT_dec (self->v_false);
		2364	SvREFCNT_dec (self->v_true);
2028	SvREFCNT_dec (self->cb_sk_object);	2365	SvREFCNT_dec (self->cb_sk_object);
2029	SvREFCNT_dec (self->cb_object);	2366	SvREFCNT_dec (self->cb_object);
2030	SvREFCNT_dec (self->incr_text);	2367	SvREFCNT_dec (self->incr_text);
2031		2368
2032	PROTOTYPES: ENABLE	2369	PROTOTYPES: ENABLE
2033		2370
2034	void encode_json (SV *scalar)	2371	void encode_json (SV *scalar)
2035	ALIAS:
2036	to_json_ = 0
2037	encode_json = F_UTF8
2038	PPCODE:	2372	PPCODE:
2039	{	2373	{
2040	JSON json;	2374	JSON json;
2041	json_init (&json);	2375	json_init (&json);
2042	json.flags \|= ix;	2376	json.flags \|= F_UTF8;
2043	XPUSHs (encode_json (scalar, &json));	2377	PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
		2378	XPUSHs (scalar);
2044	}	2379	}
2045		2380
2046	void decode_json (SV *jsonstr)	2381	void decode_json (SV *jsonstr)
2047	ALIAS:
2048	from_json_ = 0
2049	decode_json = F_UTF8
2050	PPCODE:	2382	PPCODE:
2051	{	2383	{
2052	JSON json;	2384	JSON json;
2053	json_init (&json);	2385	json_init (&json);
2054	json.flags \|= ix;	2386	json.flags \|= F_UTF8;
2055	XPUSHs (decode_json (jsonstr, &json, 0));	2387	PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
		2388	XPUSHs (jsonstr);
2056	}	2389	}
2057		2390

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.106 by root, Tue Jan 19 01:36:34 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 1.106 by root, Tue Jan 19 01:36:34 2010 UTC vs.
Revision 1.143 by root, Mon Jun 21 01:04:30 2021 UTC