ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.3 by root, Sat Oct 26 11:08:34 2013 UTC vs.
Revision 1.59 by root, Fri Nov 25 23:37:27 2016 UTC

…		…
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		11
		12	#define ECB_NO_THREADS 1
12	#include "ecb.h"	13	#include "ecb.h"
13		14
		15	// compatibility with perl <5.18
		16	#ifndef HvNAMELEN_get
		17	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
		18	#endif
		19	#ifndef HvNAMELEN
		20	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
		21	#endif
		22	#ifndef HvNAMEUTF8
		23	# define HvNAMEUTF8(hv) 0
		24	#endif
		25	#ifndef SvREFCNT_dec_NN
		26	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
		27	#endif
		28
		29	// known major and minor types
		30	enum cbor_type
		31	{
		32	MAJOR_SHIFT = 5,
		33	MINOR_MASK = 0x1f,
		34
		35	MAJOR_POS_INT = 0 << MAJOR_SHIFT,
		36	MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
		37	MAJOR_BYTES = 2 << MAJOR_SHIFT,
		38	MAJOR_TEXT = 3 << MAJOR_SHIFT,
		39	MAJOR_ARRAY = 4 << MAJOR_SHIFT,
		40	MAJOR_MAP = 5 << MAJOR_SHIFT,
		41	MAJOR_TAG = 6 << MAJOR_SHIFT,
		42	MAJOR_MISC = 7 << MAJOR_SHIFT,
		43
		44	// INT/STRING/ARRAY/MAP subtypes
		45	LENGTH_EXT1 = 24,
		46	LENGTH_EXT2 = 25,
		47	LENGTH_EXT4 = 26,
		48	LENGTH_EXT8 = 27,
		49
		50	// SIMPLE types (effectively MISC subtypes)
		51	SIMPLE_FALSE = 20,
		52	SIMPLE_TRUE = 21,
		53	SIMPLE_NULL = 22,
		54	SIMPLE_UNDEF = 23,
		55
		56	// MISC subtype (unused)
		57	MISC_EXT1 = 24,
		58	MISC_FLOAT16 = 25,
		59	MISC_FLOAT32 = 26,
		60	MISC_FLOAT64 = 27,
		61
		62	// BYTES/TEXT/ARRAY/MAP
		63	MINOR_INDEF = 31,
		64	};
		65
		66	// known tags
		67	enum cbor_tag
		68	{
		69	// extensions
		70	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
		71	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
		72	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
		73	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
		74	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
		75	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
		76	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
		77
		78	// rfc7049
		79	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
		80	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
		81	CBOR_TAG_POS_BIGNUM = 2, // byte string
		82	CBOR_TAG_NEG_BIGNUM = 3, // byte string
		83	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
		84	CBOR_TAG_BIGFLOAT = 5, // array
		85
		86	CBOR_TAG_CONV_B64U = 21, // base64url, any
		87	CBOR_TAG_CONV_B64 = 22, // base64, any
		88	CBOR_TAG_CONV_HEX = 23, // base16, any
		89	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
		90
		91	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
		92	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
		93	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
		94	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
		95	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
		96
		97	CBOR_TAG_MAGIC = 55799, // self-describe cbor
		98	};
		99
14	#define F_SHRINK 0x00000200UL	100	#define F_SHRINK 0x00000001UL
15	#define F_ALLOW_UNKNOWN 0x00002000UL	101	#define F_ALLOW_UNKNOWN 0x00000002UL
		102	#define F_ALLOW_SHARING 0x00000004UL
		103	#define F_ALLOW_CYCLES 0x00000008UL
		104	#define F_PACK_STRINGS 0x00000010UL
		105	#define F_TEXT_KEYS 0x00000020UL
		106	#define F_TEXT_STRINGS 0x00000040UL
		107	#define F_VALIDATE_UTF8 0x00000080UL
16		108
17	#define INIT_SIZE 32 // initial scalar size to be allocated	109	#define INIT_SIZE 32 // initial scalar size to be allocated
18		110
19	#define SB do {	111	#define SB do {
20	#define SE } while (0)	112	#define SE } while (0)
21
22	#if __GNUC__ >= 3
23	# define expect(expr,value) __builtin_expect ((expr), (value))
24	# define INLINE static inline
25	#else
26	# define expect(expr,value) (expr)
27	# define INLINE static
28	#endif
29
30	#define expect_false(expr) expect ((expr) != 0, 0)
31	#define expect_true(expr) expect ((expr) != 0, 1)
32		113
33	#define IN_RANGE_INC(type,val,beg,end) \	114	#define IN_RANGE_INC(type,val,beg,end) \
34	((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \	115	((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
35	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))	116	<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
36		117
…		…
42	#else	123	#else
43	# define CBOR_SLOW 0	124	# define CBOR_SLOW 0
44	# define CBOR_STASH cbor_stash	125	# define CBOR_STASH cbor_stash
45	#endif	126	#endif
46		127
47	static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS::	128	static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
48	static SV *cbor_true, *cbor_false;	129	static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
49		130
50	typedef struct {	131	typedef struct {
51	U32 flags;	132	U32 flags;
52	U32 max_depth;	133	U32 max_depth;
53	STRLEN max_size;	134	STRLEN max_size;
		135	SV *filter;
54		136
55	SV *cb_object;	137	// for the incremental parser
56	HV *cb_sk_object;	138	STRLEN incr_pos; // the current offset into the text
		139	STRLEN incr_need; // minimum bytes needed to decode
		140	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
57	} CBOR;	141	} CBOR;
58		142
59	INLINE void	143	ecb_inline void
60	cbor_init (CBOR *cbor)	144	cbor_init (CBOR *cbor)
61	{	145	{
62	Zero (cbor, 1, CBOR);	146	Zero (cbor, 1, CBOR);
63	cbor->max_depth = 512;	147	cbor->max_depth = 512;
64	}	148	}
65		149
		150	ecb_inline void
		151	cbor_free (CBOR *cbor)
		152	{
		153	SvREFCNT_dec (cbor->filter);
		154	SvREFCNT_dec (cbor->incr_count);
		155	}
		156
66	/////////////////////////////////////////////////////////////////////////////	157	/////////////////////////////////////////////////////////////////////////////
67	// utility functions	158	// utility functions
68		159
69	INLINE SV *	160	ecb_inline SV *
70	get_bool (const char *name)	161	get_bool (const char *name)
71	{	162	{
72	SV *sv = get_sv (name, 1);	163	SV *sv = get_sv (name, 1);
73		164
74	SvREADONLY_on (sv);	165	SvREADONLY_on (sv);
75	SvREADONLY_on (SvRV (sv));	166	SvREADONLY_on (SvRV (sv));
76		167
77	return sv;	168	return sv;
78	}	169	}
79		170
80	INLINE void	171	ecb_inline void
81	shrink (SV *sv)	172	shrink (SV *sv)
82	{	173	{
83	sv_utf8_downgrade (sv, 1);	174	sv_utf8_downgrade (sv, 1);
84		175
85	if (SvLEN (sv) > SvCUR (sv) + 1)	176	if (SvLEN (sv) > SvCUR (sv) + 1)
…		…
90	SvPV_renew (sv, SvCUR (sv) + 1);	181	SvPV_renew (sv, SvCUR (sv) + 1);
91	#endif	182	#endif
92	}	183	}
93	}	184	}
94		185
95	/////////////////////////////////////////////////////////////////////////////	186	// minimum length of a string to be registered for stringref
96	// fp hell	187	ecb_inline int
97		188	minimum_string_length (UV idx)
98	//TODO	189	{
		190	return idx > 23
		191	? idx > 0xffU
		192	? idx > 0xffffU
		193	? idx > 0xffffffffU
		194	? 11
		195	: 7
		196	: 5
		197	: 4
		198	: 3;
		199	}
99		200
100	/////////////////////////////////////////////////////////////////////////////	201	/////////////////////////////////////////////////////////////////////////////
101	// encoder	202	// encoder
102		203
103	// structure used for encoding CBOR	204	// structure used for encoding CBOR
…		…
106	char *cur; // SvPVX (sv) + current output position	207	char *cur; // SvPVX (sv) + current output position
107	char *end; // SvEND (sv)	208	char *end; // SvEND (sv)
108	SV *sv; // result scalar	209	SV *sv; // result scalar
109	CBOR cbor;	210	CBOR cbor;
110	U32 depth; // recursion level	211	U32 depth; // recursion level
		212	HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
		213	UV stringref_idx;
		214	HV *shareable; // ptr => index, or 0
		215	UV shareable_idx;
111	} enc_t;	216	} enc_t;
112		217
113	INLINE void	218	ecb_inline void
114	need (enc_t *enc, STRLEN len)	219	need (enc_t *enc, STRLEN len)
115	{	220	{
116	if (expect_false (enc->cur + len >= enc->end))	221	if (ecb_expect_false (enc->cur + len >= enc->end))
117	{	222	{
118	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	223	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
119	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	224	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
120	enc->cur = SvPVX (enc->sv) + cur;	225	enc->cur = SvPVX (enc->sv) + cur;
121	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	226	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
122	}	227	}
123	}	228	}
124		229
125	INLINE void	230	ecb_inline void
126	encode_ch (enc_t *enc, char ch)	231	encode_ch (enc_t *enc, char ch)
127	{	232	{
128	need (enc, 1);	233	need (enc, 1);
129	*enc->cur++ = ch;	234	*enc->cur++ = ch;
130	}	235	}
…		…
132	static void	237	static void
133	encode_uint (enc_t *enc, int major, UV len)	238	encode_uint (enc_t *enc, int major, UV len)
134	{	239	{
135	need (enc, 9);	240	need (enc, 9);
136		241
137	if (len < 24)	242	if (ecb_expect_true (len < LENGTH_EXT1))
138	*enc->cur++ = major | len;	243	*enc->cur++ = major | len;
139	else if (len < 0x100)	244	else if (ecb_expect_true (len <= 0xffU))
140	{	245	{
141	*enc->cur++ = major | 24;	246	*enc->cur++ = major | LENGTH_EXT1;
142	*enc->cur++ = len;	247	*enc->cur++ = len;
143	}	248	}
144	else if (len < 0x10000)	249	else if (len <= 0xffffU)
145	{	250	{
146	*enc->cur++ = major | 25;	251	*enc->cur++ = major | LENGTH_EXT2;
147	*enc->cur++ = len >> 8;	252	*enc->cur++ = len >> 8;
148	*enc->cur++ = len;	253	*enc->cur++ = len;
149	}	254	}
150	else if (len < 0x100000000)	255	else if (len <= 0xffffffffU)
151	{	256	{
152	*enc->cur++ = major | 26;	257	*enc->cur++ = major | LENGTH_EXT4;
153	*enc->cur++ = len >> 24;	258	*enc->cur++ = len >> 24;
154	*enc->cur++ = len >> 16;	259	*enc->cur++ = len >> 16;
155	*enc->cur++ = len >> 8;	260	*enc->cur++ = len >> 8;
156	*enc->cur++ = len;	261	*enc->cur++ = len;
157	}	262	}
158	else if (len)	263	else
159	{	264	{
160	*enc->cur++ = major | 27;	265	*enc->cur++ = major | LENGTH_EXT8;
161	*enc->cur++ = len >> 56;	266	*enc->cur++ = len >> 56;
162	*enc->cur++ = len >> 48;	267	*enc->cur++ = len >> 48;
163	*enc->cur++ = len >> 40;	268	*enc->cur++ = len >> 40;
164	*enc->cur++ = len >> 32;	269	*enc->cur++ = len >> 32;
165	*enc->cur++ = len >> 24;	270	*enc->cur++ = len >> 24;
…		…
167	*enc->cur++ = len >> 8;	272	*enc->cur++ = len >> 8;
168	*enc->cur++ = len;	273	*enc->cur++ = len;
169	}	274	}
170	}	275	}
171		276
172	static void	277	ecb_inline void
		278	encode_tag (enc_t *enc, UV tag)
		279	{
		280	encode_uint (enc, MAJOR_TAG, tag);
		281	}
		282
		283	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		284	ecb_noinline static void
173	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	285	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
174	{	286	{
175	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	287	STRLEN ulen = len;
		288	U8 *p, *pend = (U8 *)str + len;
		289
		290	for (p = (U8 *)str; p < pend; ++p)
		291	ulen += *p >> 7; // count set high bits
		292
		293	encode_uint (enc, MAJOR_TEXT, ulen);
		294
		295	need (enc, ulen);
		296	for (p = (U8 *)str; p < pend; ++p)
		297	if (*p < 0x80)
		298	*enc->cur++ = *p;
		299	else
		300	{
		301	*enc->cur++ = 0xc0 + (*p >> 6);
		302	*enc->cur++ = 0x80 + (*p & 63);
		303	}
		304	}
		305
		306	ecb_inline void
		307	encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		308	{
		309	if (ecb_expect_false (upgrade_utf8))
		310	if (!utf8)
		311	{
		312	encode_str_utf8 (enc, utf8, str, len);
		313	return;
		314	}
		315
		316	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
176	need (enc, len);	317	need (enc, len);
177	memcpy (enc->cur, str, len);	318	memcpy (enc->cur, str, len);
178	enc->cur += len;	319	enc->cur += len;
179	}	320	}
180		321
		322	ecb_inline void
		323	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		324	{
		325	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
		326	{
		327	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
		328
		329	if (SvOK (*svp))
		330	{
		331	// already registered, use stringref
		332	encode_tag (enc, CBOR_TAG_STRINGREF);
		333	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		334	return;
		335	}
		336	else if (len >= minimum_string_length (enc->stringref_idx))
		337	{
		338	// register only
		339	sv_setuv (*svp, enc->stringref_idx);
		340	++enc->stringref_idx;
		341	}
		342	}
		343
		344	encode_str (enc, upgrade_utf8, utf8, str, len);
		345	}
		346
181	static void encode_sv (enc_t *enc, SV *sv);	347	static void encode_sv (enc_t *enc, SV *sv);
182		348
183	static void	349	static void
184	encode_av (enc_t *enc, AV *av)	350	encode_av (enc_t *enc, AV *av)
185	{	351	{
…		…
188	if (enc->depth >= enc->cbor.max_depth)	354	if (enc->depth >= enc->cbor.max_depth)
189	croak (ERR_NESTING_EXCEEDED);	355	croak (ERR_NESTING_EXCEEDED);
190		356
191	++enc->depth;	357	++enc->depth;
192		358
193	encode_uint (enc, 0x80, len + 1);	359	encode_uint (enc, MAJOR_ARRAY, len + 1);
194		360
		361	if (SvMAGICAL (av))
195	for (i = 0; i <= len; ++i)	362	for (i = 0; i <= len; ++i)
196	{	363	{
197	SV **svp = av_fetch (av, i, 0);	364	SV **svp = av_fetch (av, i, 0);
198	encode_sv (enc, svp ? *svp : &PL_sv_undef);	365	encode_sv (enc, svp ? *svp : &PL_sv_undef);
199	}	366	}
		367	else
		368	for (i = 0; i <= len; ++i)
		369	{
		370	SV *sv = AvARRAY (av)[i];
		371	encode_sv (enc, sv ? sv : &PL_sv_undef);
		372	}
200		373
201	--enc->depth;	374	--enc->depth;
202	}	375	}
203		376
204	static void	377	static void
…		…
213		386
214	int pairs = hv_iterinit (hv);	387	int pairs = hv_iterinit (hv);
215	int mg = SvMAGICAL (hv);	388	int mg = SvMAGICAL (hv);
216		389
217	if (mg)	390	if (mg)
218	encode_ch (enc, 0xa0 | 31);	391	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
219	else	392	else
220	encode_uint (enc, 0xa0, pairs);	393	encode_uint (enc, MAJOR_MAP, pairs);
221		394
222	while ((he = hv_iternext (hv)))	395	while ((he = hv_iternext (hv)))
223	{	396	{
224	if (HeKLEN (he) == HEf_SVKEY)	397	if (HeKLEN (he) == HEf_SVKEY)
225	encode_sv (enc, HeSVKEY (he));	398	encode_sv (enc, HeSVKEY (he));
226	else	399	else
227	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	400	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
228		401
229	encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	402	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
230	}	403	}
231		404
232	if (mg)	405	if (mg)
233	encode_ch (enc, 0xe0 | 31);	406	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
234		407
235	--enc->depth;	408	--enc->depth;
236	}	409	}
237		410
238	// encode objects, arrays and special \0=false and \1=true values.	411	// encode objects, arrays and special \0=false and \1=true values.
239	static void	412	static void
240	encode_rv (enc_t *enc, SV *sv)	413	encode_rv (enc_t *enc, SV *sv)
241	{	414	{
242	svtype svt;
243
244	SvGETMAGIC (sv);	415	SvGETMAGIC (sv);
		416
245	svt = SvTYPE (sv);	417	svtype svt = SvTYPE (sv);
246		418
247	if (expect_false (SvOBJECT (sv)))	419	if (ecb_expect_false (SvOBJECT (sv)))
248	{	420	{
249	HV *stash = !CBOR_SLOW || cbor_boolean_stash	421	HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
250	? cbor_boolean_stash	422	? types_boolean_stash
		423	: gv_stashpv ("Types::Serialiser::Boolean", 1);
		424	HV *error_stash = !CBOR_SLOW || types_error_stash
		425	? types_error_stash
		426	: gv_stashpv ("Types::Serialiser::Error", 1);
		427	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
		428	? cbor_tagged_stash
251	: gv_stashpv ("CBOR::XS::Boolean", 1);	429	: gv_stashpv ("CBOR::XS::Tagged" , 1);
252		430
253	if (SvSTASH (sv) == stash)	431	HV *stash = SvSTASH (sv);
254	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	432
		433	if (stash == boolean_stash)
		434	{
		435	encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
		436	return;
		437	}
		438	else if (stash == error_stash)
		439	{
		440	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		441	return;
		442	}
		443	else if (stash == tagged_stash)
		444	{
		445	if (svt != SVt_PVAV)
		446	croak ("encountered CBOR::XS::Tagged object that isn't an array");
		447
		448	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
		449	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
		450
		451	return;
		452	}
		453	}
		454
		455	if (ecb_expect_false (SvREFCNT (sv) > 1)
		456	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		457	{
		458	if (!enc->shareable)
		459	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
		460
		461	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
		462
		463	if (SvOK (*svp))
		464	{
		465	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		466	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		467	return;
		468	}
255	else	469	else
256	{	470	{
257	#if 0 //TODO	471	sv_setuv (*svp, enc->shareable_idx);
258	if (enc->cbor.flags & F_CONV_BLESSED)	472	++enc->shareable_idx;
		473	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		474	}
		475	}
		476
		477	if (ecb_expect_false (SvOBJECT (sv)))
		478	{
		479	HV *stash = SvSTASH (sv);
		480	GV *method;
		481
		482	if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
259	{	483	{
		484	dSP;
		485
		486	ENTER; SAVETMPS;
		487	PUSHMARK (SP);
260	// we re-bless the reference to get overload and other niceties right	488	// we re-bless the reference to get overload and other niceties right
261	GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
262
263	if (to_cbor)
264	{
265	dSP;
266
267	ENTER; SAVETMPS; PUSHMARK (SP);
268	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));	489	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
269		490
270	// calling with G_SCALAR ensures that we always get a 1 return value
271	PUTBACK;	491	PUTBACK;
		492	// G_SCALAR ensures that return value is 1
272	call_sv ((SV *)GvCV (to_cbor), G_SCALAR);	493	call_sv ((SV *)GvCV (method), G_SCALAR);
273	SPAGAIN;	494	SPAGAIN;
274		495
275	// catch this surprisingly common error	496	// catch this surprisingly common error
276	if (SvROK (TOPs) && SvRV (TOPs) == sv)	497	if (SvROK (TOPs) && SvRV (TOPs) == sv)
277	croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));	498	croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
278		499
279	sv = POPs;
280	PUTBACK;
281
282	encode_sv (enc, sv);	500	encode_sv (enc, POPs);
283		501
		502	PUTBACK;
		503
284	FREETMPS; LEAVE;	504	FREETMPS; LEAVE;
285	}
286	else if (enc->cbor.flags & F_ALLOW_BLESSED)
287	encode_str (enc, "null", 4, 0);
288	else
289	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",
290	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
291	}	505	}
292	else if (enc->cbor.flags & F_ALLOW_BLESSED)	506	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
293	encode_str (enc, "null", 4, 0);	507	{
		508	dSP;
		509
		510	ENTER; SAVETMPS;
		511	SAVESTACK_POS ();
		512	PUSHMARK (SP);
		513	EXTEND (SP, 2);
		514	// we re-bless the reference to get overload and other niceties right
		515	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
		516	PUSHs (sv_cbor);
		517
		518	PUTBACK;
		519	int count = call_sv ((SV *)GvCV (method), G_ARRAY);
		520	SPAGAIN;
		521
		522	// catch this surprisingly common error
		523	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
		524	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
		525
		526	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
		527	encode_uint (enc, MAJOR_ARRAY, count + 1);
		528	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
		529
		530	while (count)
		531	encode_sv (enc, SP[1 - count--]);
		532
		533	PUTBACK;
		534
		535	FREETMPS; LEAVE;
		536	}
294	else	537	else
295	croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",	538	croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
296	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	539	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
297	#endif
298	}
299	}	540	}
300	else if (svt == SVt_PVHV)	541	else if (svt == SVt_PVHV)
301	encode_hv (enc, (HV *)sv);	542	encode_hv (enc, (HV *)sv);
302	else if (svt == SVt_PVAV)	543	else if (svt == SVt_PVAV)
303	encode_av (enc, (AV *)sv);	544	encode_av (enc, (AV *)sv);
304	else if (svt < SVt_PVAV)
305	{
306	STRLEN len = 0;
307	char *pv = svt ? SvPV (sv, len) : 0;
308
309	if (len == 1 && *pv == '1')
310	encode_ch (enc, 0xe0 | 21);
311	else if (len == 1 && *pv == '0')
312	encode_ch (enc, 0xe0 | 20);
313	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
314	encode_ch (enc, 0xe0 | 23);
315	else
316	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
317	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
318	}
319	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
320	encode_ch (enc, 0xe0 | 23);
321	else	545	else
322	croak ("encountered %s, but CBOR can only represent references to arrays or hashes",	546	{
323	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	547	encode_tag (enc, CBOR_TAG_INDIRECTION);
		548	encode_sv (enc, sv);
		549	}
324	}	550	}
325		551
326	static void	552	static void
327	encode_nv (enc_t *enc, SV *sv)	553	encode_nv (enc_t *enc, SV *sv)
328	{	554	{
329	double nv = SvNVX (sv);	555	double nv = SvNVX (sv);
330		556
331	need (enc, 9);	557	need (enc, 9);
332		558
333	if (expect_false (nv == (U32)nv))	559	if (ecb_expect_false (nv == (NV)(U32)nv))
334	encode_uint (enc, 0x00, (U32)nv);	560	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
335	//TODO: maybe I32?	561	//TODO: maybe I32?
336	else if (expect_false (nv == (float)nv))	562	else if (ecb_expect_false (nv == (float)nv))
337	{	563	{
338	uint32_t fp = ecb_float_to_binary32 (nv);	564	uint32_t fp = ecb_float_to_binary32 (nv);
339		565
340	*enc->cur++ = 0xe0 | 26;	566	*enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
341		567
342	if (!ecb_big_endian ())	568	if (!ecb_big_endian ())
343	fp = ecb_bswap32 (fp);	569	fp = ecb_bswap32 (fp);
344		570
345	memcpy (enc->cur, &fp, 4);	571	memcpy (enc->cur, &fp, 4);
…		…
347	}	573	}
348	else	574	else
349	{	575	{
350	uint64_t fp = ecb_double_to_binary64 (nv);	576	uint64_t fp = ecb_double_to_binary64 (nv);
351		577
352	*enc->cur++ = 0xe0 | 27;	578	*enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
353		579
354	if (!ecb_big_endian ())	580	if (!ecb_big_endian ())
355	fp = ecb_bswap64 (fp);	581	fp = ecb_bswap64 (fp);
356		582
357	memcpy (enc->cur, &fp, 8);	583	memcpy (enc->cur, &fp, 8);
…		…
366		592
367	if (SvPOKp (sv))	593	if (SvPOKp (sv))
368	{	594	{
369	STRLEN len;	595	STRLEN len;
370	char *str = SvPV (sv, len);	596	char *str = SvPV (sv, len);
371	encode_str (enc, SvUTF8 (sv), str, len);	597	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
372	}	598	}
373	else if (SvNOKp (sv))	599	else if (SvNOKp (sv))
374	encode_nv (enc, sv);	600	encode_nv (enc, sv);
375	else if (SvIOKp (sv))	601	else if (SvIOKp (sv))
376	{	602	{
377	if (SvIsUV (sv))	603	if (SvIsUV (sv))
378	encode_uint (enc, 0x00, SvUVX (sv));	604	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
379	else if (SvIVX (sv) >= 0)	605	else if (SvIVX (sv) >= 0)
380	encode_uint (enc, 0x00, SvIVX (sv));	606	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
381	else	607	else
382	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));	608	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
383	}	609	}
384	else if (SvROK (sv))	610	else if (SvROK (sv))
385	encode_rv (enc, SvRV (sv));	611	encode_rv (enc, SvRV (sv));
386	else if (!SvOK (sv))	612	else if (!SvOK (sv))
387	encode_ch (enc, 0xe0 | 22);	613	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
388	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	614	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
389	encode_ch (enc, 0xe0 | 23);	615	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
390	else	616	else
391	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	617	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
392	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	618	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
393	}	619	}
394		620
395	static SV *	621	static SV *
396	encode_cbor (SV *scalar, CBOR *cbor)	622	encode_cbor (SV *scalar, CBOR *cbor)
397	{	623	{
398	enc_t enc;	624	enc_t enc = { 0 };
399		625
400	enc.cbor = *cbor;	626	enc.cbor = *cbor;
401	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	627	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
402	enc.cur = SvPVX (enc.sv);	628	enc.cur = SvPVX (enc.sv);
403	enc.end = SvEND (enc.sv);	629	enc.end = SvEND (enc.sv);
404	enc.depth = 0;
405		630
406	SvPOK_only (enc.sv);	631	SvPOK_only (enc.sv);
		632
		633	if (cbor->flags & F_PACK_STRINGS)
		634	{
		635	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
		636	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
		637	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
		638	}
		639
407	encode_sv (&enc, scalar);	640	encode_sv (&enc, scalar);
408		641
409	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	642	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
410	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	643	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
411		644
…		…
425	U8 *end; // end of input string	658	U8 *end; // end of input string
426	const char *err; // parse error, if != 0	659	const char *err; // parse error, if != 0
427	CBOR cbor;	660	CBOR cbor;
428	U32 depth; // recursion depth	661	U32 depth; // recursion depth
429	U32 maxdepth; // recursion depth limit	662	U32 maxdepth; // recursion depth limit
		663	AV *shareable;
		664	AV *stringref;
		665	SV *decode_tagged;
		666	SV *err_sv; // optional sv for error, needs to be freed
430	} dec_t;	667	} dec_t;
431		668
432	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	669	// set dec->err to ERRSV
		670	ecb_cold static void
		671	err_errsv (dec_t *dec)
		672	{
		673	if (!dec->err)
		674	{
		675	dec->err_sv = newSVsv (ERRSV);
433		676
434	#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data");	677	// chop off the trailing \n
		678	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		679	*SvEND (dec->err_sv) = 0;
435		680
		681	dec->err = SvPVutf8_nolen (dec->err_sv);
		682	}
		683	}
		684
		685	// the following functions are used to reduce code size and help the compiler to optimise
		686	ecb_cold static void
		687	err_set (dec_t *dec, const char *reason)
		688	{
		689	if (!dec->err)
		690	dec->err = reason;
		691	}
		692
		693	ecb_cold static void
		694	err_unexpected_end (dec_t *dec)
		695	{
		696	err_set (dec, "unexpected end of CBOR data");
		697	}
		698
		699	ecb_cold static void
		700	err_nesting_exceeded (dec_t *dec)
		701	{
		702	err_set (dec, ERR_NESTING_EXCEEDED);
		703	}
		704
		705	#define ERR_DO(do) SB do; goto fail; SE
		706	#define ERR(reason) ERR_DO (err_set (dec, reason))
		707	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		708
		709	#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec))
		710
436	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	711	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
437	#define DEC_DEC_DEPTH --dec->depth	712	#define DEC_DEC_DEPTH --dec->depth
438		713
439	static UV	714	static UV
440	decode_uint (dec_t *dec)	715	decode_uint (dec_t *dec)
441	{	716	{
442	switch (*dec->cur & 31)	717	U8 m = *dec->cur & MINOR_MASK;
443	{	718	++dec->cur;
444	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
445	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
446	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
447	return *dec->cur++ & 31;
448		719
449	case 24:	720	if (ecb_expect_true (m < LENGTH_EXT1))
		721	return m;
		722	else if (ecb_expect_true (m == LENGTH_EXT1))
		723	{
450	WANT (2);	724	WANT (1);
451	dec->cur += 2;	725	dec->cur += 1;
452	return dec->cur[-1];	726	return dec->cur[-1];
453		727	}
454	case 25:	728	else if (ecb_expect_true (m == LENGTH_EXT2))
		729	{
455	WANT (3);	730	WANT (2);
456	dec->cur += 3;	731	dec->cur += 2;
457	return (((UV)dec->cur[-2]) << 8)	732	return (((UV)dec->cur[-2]) << 8)
458	| ((UV)dec->cur[-1]);	733	| ((UV)dec->cur[-1]);
459		734	}
460	case 26:	735	else if (ecb_expect_true (m == LENGTH_EXT4))
		736	{
461	WANT (5);	737	WANT (4);
462	dec->cur += 5;	738	dec->cur += 4;
463	return (((UV)dec->cur[-4]) << 24)	739	return (((UV)dec->cur[-4]) << 24)
464	| (((UV)dec->cur[-3]) << 16)	740	| (((UV)dec->cur[-3]) << 16)
465	| (((UV)dec->cur[-2]) << 8)	741	| (((UV)dec->cur[-2]) << 8)
466	| ((UV)dec->cur[-1]);	742	| ((UV)dec->cur[-1]);
467		743	}
468	case 27:	744	else if (ecb_expect_true (m == LENGTH_EXT8))
		745	{
469	WANT (9);	746	WANT (8);
470	dec->cur += 9;	747	dec->cur += 8;
		748
		749	return
		750	#if UVSIZE < 8
		751	0
		752	#else
471	return (((UV)dec->cur[-8]) << 56)	753	(((UV)dec->cur[-8]) << 56)
472	| (((UV)dec->cur[-7]) << 48)	754	| (((UV)dec->cur[-7]) << 48)
473	| (((UV)dec->cur[-6]) << 40)	755	| (((UV)dec->cur[-6]) << 40)
474	| (((UV)dec->cur[-5]) << 32)	756	| (((UV)dec->cur[-5]) << 32)
		757	#endif
475	| (((UV)dec->cur[-4]) << 24)	758	| (((UV)dec->cur[-4]) << 24)
476	| (((UV)dec->cur[-3]) << 16)	759	| (((UV)dec->cur[-3]) << 16)
477	| (((UV)dec->cur[-2]) << 8)	760	| (((UV)dec->cur[-2]) << 8)
478	| ((UV)dec->cur[-1]);	761	| ((UV)dec->cur[-1]);
479		762	}
480	default:	763	else
481	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	764	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
482	}
483		765
484	fail:	766	fail:
485	return 0;	767	return 0;
486	}	768	}
487		769
…		…
492	{	774	{
493	AV *av = newAV ();	775	AV *av = newAV ();
494		776
495	DEC_INC_DEPTH;	777	DEC_INC_DEPTH;
496		778
497	if ((*dec->cur & 31) == 31)	779	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
498	{	780	{
499	++dec->cur;	781	++dec->cur;
500		782
501	for (;;)	783	for (;;)
502	{	784	{
503	WANT (1);	785	WANT (1);
504		786
505	if (*dec->cur == (0xe0 | 31))	787	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
506	{	788	{
507	++dec->cur;	789	++dec->cur;
508	break;	790	break;
509	}	791	}
510		792
511	av_push (av, decode_sv (dec));	793	av_push (av, decode_sv (dec));
512	}	794	}
513	}	795	}
514	else	796	else
515	{	797	{
516	int i, len = decode_uint (dec);	798	UV i, len = decode_uint (dec);
517		799
		800	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
518	av_fill (av, len - 1);	801	av_fill (av, len - 1);
519		802
520	for (i = 0; i < len; ++i)	803	for (i = 0; i < len; ++i)
521	AvARRAY (av)[i] = decode_sv (dec);	804	AvARRAY (av)[i] = decode_sv (dec);
522	}	805	}
…		…
528	SvREFCNT_dec (av);	811	SvREFCNT_dec (av);
529	DEC_DEC_DEPTH;	812	DEC_DEC_DEPTH;
530	return &PL_sv_undef;	813	return &PL_sv_undef;
531	}	814	}
532		815
		816	static void
		817	decode_he (dec_t *dec, HV *hv)
		818	{
		819	// for speed reasons, we specialcase single-string
		820	// byte or utf-8 strings as keys, but only when !stringref
		821
		822	if (ecb_expect_true (!dec->stringref))
		823	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
		824	{
		825	STRLEN len = decode_uint (dec);
		826	char *key = (char *)dec->cur;
		827
		828	WANT (len);
		829	dec->cur += len;
		830
		831	hv_store (hv, key, len, decode_sv (dec), 0);
		832
		833	return;
		834	}
		835	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
		836	{
		837	STRLEN len = decode_uint (dec);
		838	char *key = (char *)dec->cur;
		839
		840	WANT (len);
		841	dec->cur += len;
		842
		843	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		844	if (!is_utf8_string (key, len))
		845	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
		846
		847	hv_store (hv, key, -len, decode_sv (dec), 0);
		848
		849	return;
		850	}
		851
		852	SV *k = decode_sv (dec);
		853	SV *v = decode_sv (dec);
		854
		855	// we leak memory if uncaught exceptions are thrown by random magical
		856	// methods, and this is hopefully the only place where it can happen,
		857	// so if there is a chance of an exception, take the very slow path.
		858	// since catching exceptions is "undocumented/internal/forbidden" by
		859	// the new p5p powers, we need to call out to a perl function :/
		860	if (ecb_expect_false (SvAMAGIC (k)))
		861	{
		862	dSP;
		863
		864	ENTER; SAVETMPS;
		865	PUSHMARK (SP);
		866	EXTEND (SP, 3);
		867	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		868	PUSHs (sv_2mortal (k));
		869	PUSHs (sv_2mortal (v));
		870
		871	PUTBACK;
		872	call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
		873	SPAGAIN;
		874
		875	FREETMPS; LEAVE;
		876
		877	if (SvTRUE (ERRSV))
		878	ERR_ERRSV;
		879
		880	return;
		881	}
		882
		883	hv_store_ent (hv, k, v, 0);
		884	SvREFCNT_dec (k);
		885
		886	fail:
		887	;
		888	}
		889
533	static SV *	890	static SV *
534	decode_hv (dec_t *dec)	891	decode_hv (dec_t *dec)
535	{	892	{
536	HV *hv = newHV ();	893	HV *hv = newHV ();
537		894
538	DEC_INC_DEPTH;	895	DEC_INC_DEPTH;
539		896
540	if ((*dec->cur & 31) == 31)	897	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
541	{	898	{
542	++dec->cur;	899	++dec->cur;
543		900
544	for (;;)	901	for (;;)
545	{	902	{
546	WANT (1);	903	WANT (1);
547		904
548	if (*dec->cur == (0xe0 | 31))	905	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
549	{	906	{
550	++dec->cur;	907	++dec->cur;
551	break;	908	break;
552	}	909	}
553		910
554	SV *k = decode_sv (dec);	911	decode_he (dec, hv);
555	SV *v = decode_sv (dec);
556
557	hv_store_ent (hv, k, v, 0);
558	}	912	}
559	}	913	}
560	else	914	else
561	{	915	{
562	int len = decode_uint (dec);	916	UV pairs = decode_uint (dec);
563		917
		918	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
		919
564	while (len--)	920	while (pairs--)
565	{	921	decode_he (dec, hv);
566	SV *k = decode_sv (dec);
567	SV *v = decode_sv (dec);
568
569	hv_store_ent (hv, k, v, 0);
570	}
571	}	922	}
572		923
573	DEC_DEC_DEPTH;	924	DEC_DEC_DEPTH;
574	return newRV_noinc ((SV *)hv);	925	return newRV_noinc ((SV *)hv);
575
576	#if 0
577	SV *sv;
578	HV *hv = newHV ();
579
580	DEC_INC_DEPTH;
581	decode_ws (dec);
582
583	for (;;)
584	{
585	// heuristic: assume that
586	// a) decode_str + hv_store_ent are abysmally slow.
587	// b) most hash keys are short, simple ascii text.
588	// => try to "fast-match" such strings to avoid
589	// the overhead of decode_str + hv_store_ent.
590	{
591	SV *value;
592	char *p = dec->cur;
593	char *e = p + 24; // only try up to 24 bytes
594
595	for (;;)
596	{
597	// the >= 0x80 is false on most architectures
598	if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
599	{
600	// slow path, back up and use decode_str
601	SV *key = decode_str (dec);
602	if (!key)
603	goto fail;
604
605	decode_ws (dec); EXPECT_CH (':');
606
607	decode_ws (dec);
608	value = decode_sv (dec);
609	if (!value)
610	{
611	SvREFCNT_dec (key);
612	goto fail;
613	}
614
615	hv_store_ent (hv, key, value, 0);
616	SvREFCNT_dec (key);
617
618	break;
619	}
620	else if (*p == '"')
621	{
622	// fast path, got a simple key
623	char *key = dec->cur;
624	int len = p - key;
625	dec->cur = p + 1;
626
627	decode_ws (dec); EXPECT_CH (':');
628
629	decode_ws (dec);
630	value = decode_sv (dec);
631	if (!value)
632	goto fail;
633
634	hv_store (hv, key, len, value, 0);
635
636	break;
637	}
638
639	++p;
640	}
641	}
642
643	decode_ws (dec);
644
645	if (*dec->cur == '}')
646	{
647	++dec->cur;
648	break;
649	}
650
651	if (*dec->cur != ',')
652	ERR (", or } expected while parsing object/hash");
653
654	++dec->cur;
655
656	decode_ws (dec);
657
658	if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
659	{
660	++dec->cur;
661	break;
662	}
663	}
664
665	DEC_DEC_DEPTH;
666	sv = newRV_noinc ((SV *)hv);
667
668	// check filter callbacks
669	if (dec->cbor.flags & F_HOOK)
670	{
671	if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
672	{
673	HE *cb, *he;
674
675	hv_iterinit (hv);
676	he = hv_iternext (hv);
677	hv_iterinit (hv);
678
679	// the next line creates a mortal sv each time its called.
680	// might want to optimise this for common cases.
681	cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
682
683	if (cb)
684	{
685	dSP;
686	int count;
687
688	ENTER; SAVETMPS; PUSHMARK (SP);
689	XPUSHs (HeVAL (he));
690	sv_2mortal (sv);
691
692	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
693
694	if (count == 1)
695	{
696	sv = newSVsv (POPs);
697	FREETMPS; LEAVE;
698	return sv;
699	}
700
701	SvREFCNT_inc (sv);
702	FREETMPS; LEAVE;
703	}
704	}
705
706	if (dec->cbor.cb_object)
707	{
708	dSP;
709	int count;
710
711	ENTER; SAVETMPS; PUSHMARK (SP);
712	XPUSHs (sv_2mortal (sv));
713
714	PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
715
716	if (count == 1)
717	{
718	sv = newSVsv (POPs);
719	FREETMPS; LEAVE;
720	return sv;
721	}
722
723	SvREFCNT_inc (sv);
724	FREETMPS; LEAVE;
725	}
726	}
727
728	return sv;
729	#endif
730		926
731	fail:	927	fail:
732	SvREFCNT_dec (hv);	928	SvREFCNT_dec (hv);
733	DEC_DEC_DEPTH;	929	DEC_DEC_DEPTH;
734	return &PL_sv_undef;	930	return &PL_sv_undef;
735	}	931	}
736		932
737	static SV *	933	static SV *
738	decode_str (dec_t *dec, int utf8)	934	decode_str (dec_t *dec, int utf8)
739	{	935	{
740	SV *sv;	936	SV *sv = 0;
741		937
742	if ((*dec->cur & 31) == 31)	938	if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
743	{	939	{
		940	// indefinite length strings
744	++dec->cur;	941	++dec->cur;
745		942
		943	U8 major = *dec->cur & MAJOR_MISC;
		944
746	sv = newSVpvn ("", 0);	945	sv = newSVpvn ("", 0);
747		946
748	// not very fast, and certainly not robust against illegal input
749	for (;;)	947	for (;;)
750	{	948	{
751	WANT (1);	949	WANT (1);
752		950
753	if (*dec->cur == (0xe0 | 31))	951	if ((*dec->cur - major) > LENGTH_EXT8)
		952	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
754	{	953	{
755	++dec->cur;	954	++dec->cur;
756	break;	955	break;
757	}	956	}
		957	else
		958	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
758		959
759	SV *sv2 = decode_sv (dec);	960	STRLEN len = decode_uint (dec);
760	sv_catsv (sv, sv2);	961
		962	WANT (len);
		963	sv_catpvn (sv, dec->cur, len);
		964	dec->cur += len;
761	}	965	}
762	}	966	}
763	else	967	else
764	{	968	{
765	STRLEN len = decode_uint (dec);	969	STRLEN len = decode_uint (dec);
766		970
767	WANT (len);	971	WANT (len);
768	sv = newSVpvn (dec->cur, len);	972	sv = newSVpvn (dec->cur, len);
769	dec->cur += len;	973	dec->cur += len;
		974
		975	if (ecb_expect_false (dec->stringref)
		976	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
		977	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770	}	978	}
771		979
772	if (utf8)	980	if (utf8)
		981	{
		982	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		983	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		984	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		985
773	SvUTF8_on (sv);	986	SvUTF8_on (sv);
		987	}
774		988
775	return sv;	989	return sv;
776		990
777	fail:	991	fail:
		992	SvREFCNT_dec (sv);
778	return &PL_sv_undef;	993	return &PL_sv_undef;
779	}	994	}
780		995
781	static SV *	996	static SV *
782	decode_tagged (dec_t *dec)	997	decode_tagged (dec_t *dec)
783	{	998	{
		999	SV *sv = 0;
784	UV tag = decode_uint (dec);	1000	UV tag = decode_uint (dec);
		1001
		1002	WANT (1);
		1003
		1004	switch (tag)
		1005	{
		1006	case CBOR_TAG_MAGIC:
785	SV *sv = decode_sv (dec);	1007	sv = decode_sv (dec);
		1008	break;
786		1009
787	if (tag == 55799) // 2.4.5 Self-Describe CBOR	1010	case CBOR_TAG_INDIRECTION:
		1011	sv = newRV_noinc (decode_sv (dec));
		1012	break;
		1013
		1014	case CBOR_TAG_STRINGREF_NAMESPACE:
		1015	{
		1016	// do nmot use SAVETMPS/FREETMPS, as these will
		1017	// erase mortalised caches, e.g. "shareable"
		1018	ENTER;
		1019
		1020	SAVESPTR (dec->stringref);
		1021	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
		1022
		1023	sv = decode_sv (dec);
		1024
		1025	LEAVE;
		1026	}
		1027	break;
		1028
		1029	case CBOR_TAG_STRINGREF:
		1030	{
		1031	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
		1032	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
		1033
		1034	UV idx = decode_uint (dec);
		1035
		1036	if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
		1037	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
		1038
		1039	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
		1040	}
		1041	break;
		1042
		1043	case CBOR_TAG_VALUE_SHAREABLE:
		1044	{
		1045	if (ecb_expect_false (!dec->shareable))
		1046	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
		1047
		1048	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1049	{
		1050	sv = newSV (0);
		1051	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
		1052
		1053	SV *osv = decode_sv (dec);
		1054	sv_setsv (sv, osv);
		1055	SvREFCNT_dec_NN (osv);
		1056	}
		1057	else
		1058	{
		1059	av_push (dec->shareable, &PL_sv_undef);
		1060	int idx = AvFILLp (dec->shareable);
		1061	sv = decode_sv (dec);
		1062	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		1063	}
		1064	}
		1065	break;
		1066
		1067	case CBOR_TAG_VALUE_SHAREDREF:
		1068	{
		1069	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
		1070	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
		1071
		1072	UV idx = decode_uint (dec);
		1073
		1074	if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
		1075	ERR ("corrupted CBOR data (sharedref index out of bounds)");
		1076
		1077	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		1078
		1079	if (sv == &PL_sv_undef)
		1080	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
		1081	}
		1082	break;
		1083
		1084	case CBOR_TAG_PERL_OBJECT:
		1085	{
		1086	sv = decode_sv (dec);
		1087
		1088	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
		1089	ERR ("corrupted CBOR data (non-array perl object)");
		1090
		1091	AV *av = (AV *)SvRV (sv);
		1092	int len = av_len (av) + 1;
		1093	HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
		1094
		1095	if (!stash)
		1096	ERR ("cannot decode perl-object (package does not exist)");
		1097
		1098	GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
		1099
		1100	if (!method)
		1101	ERR ("cannot decode perl-object (package does not have a THAW method)");
		1102
		1103	dSP;
		1104
		1105	ENTER; SAVETMPS;
		1106	PUSHMARK (SP);
		1107	EXTEND (SP, len + 1);
		1108	// we re-bless the reference to get overload and other niceties right
		1109	PUSHs (*av_fetch (av, 0, 1));
		1110	PUSHs (sv_cbor);
		1111
		1112	int i;
		1113
		1114	for (i = 1; i < len; ++i)
		1115	PUSHs (*av_fetch (av, i, 1));
		1116
		1117	PUTBACK;
		1118	call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
		1119	SPAGAIN;
		1120
		1121	if (SvTRUE (ERRSV))
		1122	{
		1123	FREETMPS; LEAVE;
		1124	ERR_ERRSV;
		1125	}
		1126
		1127	SvREFCNT_dec (sv);
		1128	sv = SvREFCNT_inc (POPs);
		1129
		1130	PUTBACK;
		1131
		1132	FREETMPS; LEAVE;
		1133	}
		1134	break;
		1135
		1136	default:
		1137	{
		1138	SV *tag_sv = newSVuv (tag);
		1139
		1140	sv = decode_sv (dec);
		1141
		1142	dSP;
		1143	ENTER; SAVETMPS;
		1144	SAVESTACK_POS ();
		1145	PUSHMARK (SP);
		1146	EXTEND (SP, 2);
		1147	PUSHs (tag_sv);
		1148	PUSHs (sv);
		1149
		1150	PUTBACK;
		1151	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
		1152	SPAGAIN;
		1153
		1154	if (SvTRUE (ERRSV))
		1155	{
		1156	SvREFCNT_dec (tag_sv);
		1157	FREETMPS; LEAVE;
		1158	ERR_ERRSV;
		1159	}
		1160
		1161	if (count)
		1162	{
		1163	SvREFCNT_dec (tag_sv);
		1164	SvREFCNT_dec (sv);
		1165	sv = SvREFCNT_inc (POPs);
		1166	}
		1167	else
		1168	{
		1169	AV *av = newAV ();
		1170	av_push (av, tag_sv);
		1171	av_push (av, sv);
		1172
		1173	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
		1174	? cbor_tagged_stash
		1175	: gv_stashpv ("CBOR::XS::Tagged" , 1);
		1176	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
		1177	}
		1178
		1179	PUTBACK;
		1180
		1181	FREETMPS; LEAVE;
		1182	}
		1183	break;
		1184	}
		1185
788	return sv;	1186	return sv;
789		1187
790	AV *av = newAV ();	1188	fail:
791	av_push (av, newSVuv (tag));	1189	SvREFCNT_dec (sv);
792	av_push (av, sv);	1190	return &PL_sv_undef;
793	return newRV_noinc ((SV *)av);
794	}	1191	}
795		1192
796	static SV *	1193	static SV *
797	decode_sv (dec_t *dec)	1194	decode_sv (dec_t *dec)
798	{	1195	{
799	WANT (1);	1196	WANT (1);
800		1197
801	switch (*dec->cur >> 5)	1198	switch (*dec->cur >> MAJOR_SHIFT)
802	{	1199	{
803	case 0: // unsigned int	1200	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
804	//TODO: 64 bit values on 3 2bit perls	1201	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
805	return newSVuv (decode_uint (dec));	1202	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
806	case 1: // negative int	1203	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
807	return newSViv (-1 - (IV)decode_uint (dec));	1204	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
808	case 2: // octet string	1205	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
809	return decode_str (dec, 0);	1206	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
810	case 3: // utf-8 string	1207
811	return decode_str (dec, 1);	1208	case MAJOR_MISC >> MAJOR_SHIFT:
812	case 4: // array
813	return decode_av (dec);
814	case 5: // map
815	return decode_hv (dec);
816	case 6: // tag
817	return decode_tagged (dec);
818	case 7: // misc
819	switch (*dec->cur++ & 31)	1209	switch (*dec->cur++ & MINOR_MASK)
820	{	1210	{
821	case 20:	1211	case SIMPLE_FALSE:
822	#if CBOR_SLOW	1212	#if CBOR_SLOW
823	cbor_false = get_bool ("CBOR::XS::false");	1213	types_false = get_bool ("Types::Serialiser::false");
824	#endif	1214	#endif
825	return newSVsv (cbor_false);	1215	return newSVsv (types_false);
826	case 21:	1216	case SIMPLE_TRUE:
827	#if CBOR_SLOW	1217	#if CBOR_SLOW
828	cbor_true = get_bool ("CBOR::XS::true");	1218	types_true = get_bool ("Types::Serialiser::true");
829	#endif	1219	#endif
830	return newSVsv (cbor_true);	1220	return newSVsv (types_true);
831	case 22:	1221	case SIMPLE_NULL:
832	return newSVsv (&PL_sv_undef);	1222	return newSVsv (&PL_sv_undef);
		1223	case SIMPLE_UNDEF:
		1224	#if CBOR_SLOW
		1225	types_error = get_bool ("Types::Serialiser::error");
		1226	#endif
		1227	return newSVsv (types_error);
833		1228
834	case 25:	1229	case MISC_FLOAT16:
835	{	1230	{
836	WANT (2);	1231	WANT (2);
837		1232
838	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1233	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
839	dec->cur += 2;	1234	dec->cur += 2;
840		1235
841	return newSVnv (ecb_binary16_to_float (fp));	1236	return newSVnv (ecb_binary16_to_float (fp));
842	}	1237	}
843		1238
844	case 26:	1239	case MISC_FLOAT32:
845	{	1240	{
846	uint32_t fp;	1241	uint32_t fp;
847	WANT (4);	1242	WANT (4);
848	memcpy (&fp, dec->cur, 4);	1243	memcpy (&fp, dec->cur, 4);
849	dec->cur += 4;	1244	dec->cur += 4;
…		…
852	fp = ecb_bswap32 (fp);	1247	fp = ecb_bswap32 (fp);
853		1248
854	return newSVnv (ecb_binary32_to_float (fp));	1249	return newSVnv (ecb_binary32_to_float (fp));
855	}	1250	}
856		1251
857	case 27:	1252	case MISC_FLOAT64:
858	{	1253	{
859	uint64_t fp;	1254	uint64_t fp;
860	WANT (8);	1255	WANT (8);
861	memcpy (&fp, dec->cur, 8);	1256	memcpy (&fp, dec->cur, 8);
862	dec->cur += 8;	1257	dec->cur += 8;
…		…
865	fp = ecb_bswap64 (fp);	1260	fp = ecb_bswap64 (fp);
866		1261
867	return newSVnv (ecb_binary64_to_double (fp));	1262	return newSVnv (ecb_binary64_to_double (fp));
868	}	1263	}
869		1264
870	// 0..19 unassigned	1265	// 0..19 unassigned simple
871	// 24 reserved + unassigned (reserved values are not encodable)	1266	// 24 reserved + unassigned simple (reserved values are not encodable)
		1267	// 28-30 unassigned misc
		1268	// 31 break code
872	default:	1269	default:
873	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1270	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
874	}	1271	}
875		1272
876	break;	1273	break;
877	}	1274	}
878	#if 0
879	switch (*dec->cur)
880	{
881	//case '"': ++dec->cur; return decode_str (dec);
882	case '[': ++dec->cur; return decode_av (dec);
883	case '{': ++dec->cur; return decode_hv (dec);
884
885	case '-':
886	case '0': case '1': case '2': case '3': case '4':
887	case '5': case '6': case '7': case '8': case '9':
888	//TODO return decode_num (dec);
889
890	case 't':
891	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
892	{
893	dec->cur += 4;
894	#if CBOR_SLOW
895	cbor_true = get_bool ("CBOR::XS::true");
896	#endif
897	return newSVsv (cbor_true);
898	}
899	else
900	ERR ("'true' expected");
901
902	break;
903
904	case 'f':
905	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
906	{
907	dec->cur += 5;
908	#if CBOR_SLOW
909	cbor_false = get_bool ("CBOR::XS::false");
910	#endif
911	return newSVsv (cbor_false);
912	}
913	else
914	ERR ("'false' expected");
915
916	break;
917
918	case 'n':
919	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
920	{
921	dec->cur += 4;
922	return newSVsv (&PL_sv_undef);
923	}
924	else
925	ERR ("'null' expected");
926
927	break;
928
929	default:
930	ERR ("malformed CBOR string, neither array, object, number, string or atom");
931	break;
932	}
933	#endif
934		1275
935	fail:	1276	fail:
936	return &PL_sv_undef;	1277	return &PL_sv_undef;
937	}	1278	}
938		1279
939	static SV *	1280	static SV *
940	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1281	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
941	{	1282	{
942	dec_t dec;	1283	dec_t dec = { 0 };
943	SV *sv;	1284	SV *sv;
		1285	STRLEN len;
		1286	char *data = SvPVbyte (string, len);
944		1287
945	/* work around bugs in 5.10 where manipulating magic values
946	* makes perl ignore the magic in subsequent accesses.
947	* also make a copy of non-PV values, to get them into a clean
948	* state (SvPV should do that, but it's buggy, see below).
949	*/
950	/*SvGETMAGIC (string);*/
951	if (SvMAGICAL (string) || !SvPOK (string))
952	string = sv_2mortal (newSVsv (string));
953
954	SvUPGRADE (string, SVt_PV);
955
956	/* work around a bug in perl 5.10, which causes SvCUR to fail an
957	* assertion with -DDEBUGGING, although SvCUR is documented to
958	* return the xpv_cur field which certainly exists after upgrading.
959	* according to nicholas clark, calling SvPOK fixes this.
960	* But it doesn't fix it, so try another workaround, call SvPV_nolen
961	* and hope for the best.
962	* Damnit, SvPV_nolen still trips over yet another assertion. This
963	* assertion business is seriously broken, try yet another workaround
964	* for the broken -DDEBUGGING.
965	*/
966	{
967	#ifdef DEBUGGING
968	STRLEN offset = SvOK (string) ? sv_len (string) : 0;
969	#else
970	STRLEN offset = SvCUR (string);
971	#endif
972
973	if (offset > cbor->max_size && cbor->max_size)	1288	if (len > cbor->max_size && cbor->max_size)
974	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",	1289	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
975	(unsigned long)SvCUR (string), (unsigned long)cbor->max_size);	1290	(unsigned long)len, (unsigned long)cbor->max_size);
976	}
977
978	sv_utf8_downgrade (string, 0);
979		1291
980	dec.cbor = *cbor;	1292	dec.cbor = *cbor;
981	dec.cur = (U8 *)SvPVX (string);	1293	dec.cur = (U8 *)data;
982	dec.end = (U8 *)SvEND (string);	1294	dec.end = (U8 *)data + len;
983	dec.err = 0;
984	dec.depth = 0;
985
986	if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
987	;//TODO dec.cbor.flags |= F_HOOK;
988		1295
989	sv = decode_sv (&dec);	1296	sv = decode_sv (&dec);
990		1297
991	if (offset_return)	1298	if (offset_return)
992	*offset_return = dec.cur;	1299	*offset_return = dec.cur;
…		…
995	if (dec.cur != dec.end && !dec.err)	1302	if (dec.cur != dec.end && !dec.err)
996	dec.err = "garbage after CBOR object";	1303	dec.err = "garbage after CBOR object";
997		1304
998	if (dec.err)	1305	if (dec.err)
999	{	1306	{
		1307	if (dec.shareable)
		1308	{
		1309	// need to break cyclic links, which would all be in shareable
		1310	int i;
		1311	SV **svp;
		1312
		1313	for (i = av_len (dec.shareable) + 1; i--; )
		1314	if ((svp = av_fetch (dec.shareable, i, 0)))
		1315	sv_setsv (*svp, &PL_sv_undef);
		1316	}
		1317
1000	SvREFCNT_dec (sv);	1318	SvREFCNT_dec (sv);
		1319
		1320	if (dec.err_sv)
		1321	sv_2mortal (dec.err_sv);
		1322
1001	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur);	1323	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1002	}	1324	}
1003		1325
1004	sv = sv_2mortal (sv);	1326	sv = sv_2mortal (sv);
1005		1327
1006	return sv;	1328	return sv;
1007	}	1329	}
1008		1330
		1331	/////////////////////////////////////////////////////////////////////////////
		1332	// incremental parser
		1333
		1334	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1335
		1336	// returns 0 for notyet, 1 for success or error
		1337	static int
		1338	incr_parse (CBOR *self, SV *cborstr)
		1339	{
		1340	STRLEN cur;
		1341	SvPV (cborstr, cur);
		1342
		1343	while (ecb_expect_true (self->incr_need <= cur))
		1344	{
		1345	// table of integer count bytes
		1346	static I8 incr_len[MINOR_MASK + 1] = {
		1347	0, 0, 0, 0, 0, 0, 0, 0,
		1348	0, 0, 0, 0, 0, 0, 0, 0,
		1349	0, 0, 0, 0, 0, 0, 0, 0,
		1350	1, 2, 4, 8,-1,-1,-1,-2
		1351	};
		1352
		1353	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1354	U8 m = *p & MINOR_MASK;
		1355	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1356	I8 ilen = incr_len[m];
		1357
		1358	self->incr_need = self->incr_pos + 1;
		1359
		1360	if (ecb_expect_false (ilen < 0))
		1361	{
		1362	if (m != MINOR_INDEF)
		1363	return 1; // error
		1364
		1365	if (*p == (MAJOR_MISC | MINOR_INDEF))
		1366	{
		1367	if (count >= 0)
		1368	return 1; // error
		1369
		1370	count = 1;
		1371	}
		1372	else
		1373	{
		1374	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1375	count = -1;
		1376	}
		1377	}
		1378	else
		1379	{
		1380	self->incr_need += ilen;
		1381	if (ecb_expect_false (self->incr_need > cur))
		1382	return 0;
		1383
		1384	int major = *p >> MAJOR_SHIFT;
		1385
		1386	switch (major)
		1387	{
		1388	case MAJOR_TAG >> MAJOR_SHIFT:
		1389	++count; // tags merely prefix another value
		1390	break;
		1391
		1392	case MAJOR_BYTES >> MAJOR_SHIFT:
		1393	case MAJOR_TEXT >> MAJOR_SHIFT:
		1394	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1395	case MAJOR_MAP >> MAJOR_SHIFT:
		1396	{
		1397	UV len;
		1398
		1399	if (ecb_expect_false (ilen))
		1400	{
		1401	len = 0;
		1402
		1403	do {
		1404	len = (len << 8) | *++p;
		1405	} while (--ilen);
		1406	}
		1407	else
		1408	len = m;
		1409
		1410	switch (major)
		1411	{
		1412	case MAJOR_BYTES >> MAJOR_SHIFT:
		1413	case MAJOR_TEXT >> MAJOR_SHIFT:
		1414	self->incr_need += len;
		1415	if (ecb_expect_false (self->incr_need > cur))
		1416	return 0;
		1417
		1418	break;
		1419
		1420	case MAJOR_MAP >> MAJOR_SHIFT:
		1421	len <<= 1;
		1422	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1423	if (len)
		1424	{
		1425	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1426	count = len + 1;
		1427	}
		1428	break;
		1429	}
		1430	}
		1431	}
		1432	}
		1433
		1434	self->incr_pos = self->incr_need;
		1435
		1436	if (count > 0)
		1437	{
		1438	while (!--count)
		1439	{
		1440	if (!AvFILLp (self->incr_count))
		1441	return 1; // done
		1442
		1443	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1444	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1445	}
		1446
		1447	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1448	}
		1449	}
		1450
		1451	return 0;
		1452	}
		1453
		1454
1009	/////////////////////////////////////////////////////////////////////////////	1455	/////////////////////////////////////////////////////////////////////////////
1010	// XS interface functions	1456	// XS interface functions
1011		1457
1012	MODULE = CBOR::XS PACKAGE = CBOR::XS	1458	MODULE = CBOR::XS PACKAGE = CBOR::XS
1013		1459
1014	BOOT:	1460	BOOT:
1015	{	1461	{
1016	cbor_stash = gv_stashpv ("CBOR::XS" , 1);	1462	cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1017	cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);	1463	cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1018		1464
1019	cbor_true = get_bool ("CBOR::XS::true");	1465	types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1020	cbor_false = get_bool ("CBOR::XS::false");	1466	types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
		1467
		1468	types_true = get_bool ("Types::Serialiser::true" );
		1469	types_false = get_bool ("Types::Serialiser::false");
		1470	types_error = get_bool ("Types::Serialiser::error");
		1471
		1472	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
		1473
		1474	sv_cbor = newSVpv ("CBOR", 0);
		1475	SvREADONLY_on (sv_cbor);
		1476
		1477	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1021	}	1478	}
1022		1479
1023	PROTOTYPES: DISABLE	1480	PROTOTYPES: DISABLE
1024		1481
1025	void CLONE (...)	1482	void CLONE (...)
1026	CODE:	1483	CODE:
1027	cbor_stash = 0;	1484	cbor_stash = 0;
		1485	cbor_tagged_stash = 0;
		1486	types_error_stash = 0;
1028	cbor_boolean_stash = 0;	1487	types_boolean_stash = 0;
1029		1488
1030	void new (char *klass)	1489	void new (char *klass)
1031	PPCODE:	1490	PPCODE:
1032	{	1491	{
1033	SV *pv = NEWSV (0, sizeof (CBOR));	1492	SV *pv = NEWSV (0, sizeof (CBOR));
…		…
1041		1500
1042	void shrink (CBOR *self, int enable = 1)	1501	void shrink (CBOR *self, int enable = 1)
1043	ALIAS:	1502	ALIAS:
1044	shrink = F_SHRINK	1503	shrink = F_SHRINK
1045	allow_unknown = F_ALLOW_UNKNOWN	1504	allow_unknown = F_ALLOW_UNKNOWN
		1505	allow_sharing = F_ALLOW_SHARING
		1506	allow_cycles = F_ALLOW_CYCLES
		1507	pack_strings = F_PACK_STRINGS
		1508	text_keys = F_TEXT_KEYS
		1509	text_strings = F_TEXT_STRINGS
		1510	validate_utf8 = F_VALIDATE_UTF8
1046	PPCODE:	1511	PPCODE:
1047	{	1512	{
1048	if (enable)	1513	if (enable)
1049	self->flags |= ix;	1514	self->flags |= ix;
1050	else	1515	else
…		…
1055		1520
1056	void get_shrink (CBOR *self)	1521	void get_shrink (CBOR *self)
1057	ALIAS:	1522	ALIAS:
1058	get_shrink = F_SHRINK	1523	get_shrink = F_SHRINK
1059	get_allow_unknown = F_ALLOW_UNKNOWN	1524	get_allow_unknown = F_ALLOW_UNKNOWN
		1525	get_allow_sharing = F_ALLOW_SHARING
		1526	get_allow_cycles = F_ALLOW_CYCLES
		1527	get_pack_strings = F_PACK_STRINGS
		1528	get_text_keys = F_TEXT_KEYS
		1529	get_text_strings = F_TEXT_STRINGS
		1530	get_validate_utf8 = F_VALIDATE_UTF8
1060	PPCODE:	1531	PPCODE:
1061	XPUSHs (boolSV (self->flags & ix));	1532	XPUSHs (boolSV (self->flags & ix));
1062		1533
1063	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1534	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1064	PPCODE:	1535	PPCODE:
…		…
1080	CODE:	1551	CODE:
1081	RETVAL = self->max_size;	1552	RETVAL = self->max_size;
1082	OUTPUT:	1553	OUTPUT:
1083	RETVAL	1554	RETVAL
1084		1555
1085	#if 0 //TODO	1556	void filter (CBOR *self, SV *filter = 0)
1086
1087	void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1088	PPCODE:	1557	PPCODE:
1089	{
1090	SvREFCNT_dec (self->cb_object);	1558	SvREFCNT_dec (self->filter);
1091	self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;	1559	self->filter = filter ? newSVsv (filter) : filter;
1092
1093	XPUSHs (ST (0));	1560	XPUSHs (ST (0));
1094	}
1095		1561
1096	void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef)	1562	SV *get_filter (CBOR *self)
1097	PPCODE:	1563	CODE:
1098	{	1564	RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1099	if (!self->cb_sk_object)	1565	OUTPUT:
1100	self->cb_sk_object = newHV ();	1566	RETVAL
1101
1102	if (SvOK (cb))
1103	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1104	else
1105	{
1106	hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1107
1108	if (!HvKEYS (self->cb_sk_object))
1109	{
1110	SvREFCNT_dec (self->cb_sk_object);
1111	self->cb_sk_object = 0;
1112	}
1113	}
1114
1115	XPUSHs (ST (0));
1116	}
1117
1118	#endif
1119		1567
1120	void encode (CBOR *self, SV *scalar)	1568	void encode (CBOR *self, SV *scalar)
1121	PPCODE:	1569	PPCODE:
1122	PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;	1570	PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1123	XPUSHs (scalar);	1571	XPUSHs (scalar);
…		…
1136	EXTEND (SP, 2);	1584	EXTEND (SP, 2);
1137	PUSHs (sv);	1585	PUSHs (sv);
1138	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1586	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1139	}	1587	}
1140		1588
		1589	void incr_parse (CBOR *self, SV *cborstr)
		1590	ALIAS:
		1591	incr_parse_multiple = 1
		1592	PPCODE:
		1593	{
		1594	if (SvUTF8 (cborstr))
		1595	sv_utf8_downgrade (cborstr, 0);
		1596
		1597	if (!self->incr_count)
		1598	{
		1599	self->incr_count = newAV ();
		1600	self->incr_pos = 0;
		1601	self->incr_need = 1;
		1602
		1603	av_push (self->incr_count, newSViv (1));
		1604	}
		1605
		1606	do
		1607	{
		1608	if (!incr_parse (self, cborstr))
		1609	{
		1610	if (self->incr_need > self->max_size && self->max_size)
		1611	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1612	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1613
		1614	break;
		1615	}
		1616
		1617	SV *sv;
		1618	char *offset;
		1619
		1620	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1621	XPUSHs (sv);
		1622
		1623	sv_chop (cborstr, offset);
		1624
		1625	av_clear (self->incr_count);
		1626	av_push (self->incr_count, newSViv (1));
		1627
		1628	self->incr_pos = 0;
		1629	self->incr_need = self->incr_pos + 1;
		1630	}
		1631	while (ix);
		1632	}
		1633
		1634	void incr_reset (CBOR *self)
		1635	CODE:
		1636	{
		1637	SvREFCNT_dec (self->incr_count);
		1638	self->incr_count = 0;
		1639	}
		1640
1141	void DESTROY (CBOR *self)	1641	void DESTROY (CBOR *self)
1142	CODE:	1642	PPCODE:
1143	SvREFCNT_dec (self->cb_sk_object);	1643	cbor_free (self);
1144	SvREFCNT_dec (self->cb_object);
1145		1644
1146	PROTOTYPES: ENABLE	1645	PROTOTYPES: ENABLE
1147		1646
1148	void encode_cbor (SV *scalar)	1647	void encode_cbor (SV *scalar)
		1648	ALIAS:
		1649	encode_cbor = 0
		1650	encode_cbor_sharing = F_ALLOW_SHARING
1149	PPCODE:	1651	PPCODE:
1150	{	1652	{
1151	CBOR cbor;	1653	CBOR cbor;
1152	cbor_init (&cbor);	1654	cbor_init (&cbor);
		1655	cbor.flags |= ix;
1153	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1656	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1154	XPUSHs (scalar);	1657	XPUSHs (scalar);
1155	}	1658	}
1156		1659
1157	void decode_cbor (SV *cborstr)	1660	void decode_cbor (SV *cborstr)

Diff Legend

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