ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.18 by root, Wed Nov 20 01:09:46 2013 UTC vs.
Revision 1.80 by root, Fri Sep 8 20:03:06 2023 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
		13	#define ECB_NO_THREADS 1
12	#include "ecb.h"	14	#include "ecb.h"
13		15
14	// compatibility with perl <5.18	16	// compatibility with perl <5.18
15	#ifndef HvNAMELEN_get	17	#ifndef HvNAMELEN_get
16	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))	18	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
…		…
19	# define HvNAMELEN(hv) HvNAMELEN_get (hv)	21	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20	#endif	22	#endif
21	#ifndef HvNAMEUTF8	23	#ifndef HvNAMEUTF8
22	# define HvNAMEUTF8(hv) 0	24	# define HvNAMEUTF8(hv) 0
23	#endif	25	#endif
		26	#ifndef SvREFCNT_inc_NN
		27	# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
		28	#endif
		29	#ifndef SvREFCNT_dec_NN
		30	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
		31	#endif
		32
		33	// perl's is_utf8_string interprets len=0 as "calculate len", but we want it to mean 0
		34	#define cbor_is_utf8_string(str,len) (!(len) || is_utf8_string ((str), (len)))
		35
		36	// known major and minor types
		37	enum cbor_type
		38	{
		39	MAJOR_SHIFT = 5,
		40	MINOR_MASK = 0x1f,
		41
		42	MAJOR_POS_INT = 0 << MAJOR_SHIFT,
		43	MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
		44	MAJOR_BYTES = 2 << MAJOR_SHIFT,
		45	MAJOR_TEXT = 3 << MAJOR_SHIFT,
		46	MAJOR_ARRAY = 4 << MAJOR_SHIFT,
		47	MAJOR_MAP = 5 << MAJOR_SHIFT,
		48	MAJOR_TAG = 6 << MAJOR_SHIFT,
		49	MAJOR_MISC = 7 << MAJOR_SHIFT,
		50
		51	// INT/STRING/ARRAY/MAP subtypes
		52	LENGTH_EXT1 = 24,
		53	LENGTH_EXT2 = 25,
		54	LENGTH_EXT4 = 26,
		55	LENGTH_EXT8 = 27,
		56
		57	// SIMPLE types (effectively MISC subtypes)
		58	SIMPLE_FALSE = 20,
		59	SIMPLE_TRUE = 21,
		60	SIMPLE_NULL = 22,
		61	SIMPLE_UNDEF = 23,
		62
		63	// MISC subtype (unused)
		64	MISC_EXT1 = 24,
		65	MISC_FLOAT16 = 25,
		66	MISC_FLOAT32 = 26,
		67	MISC_FLOAT64 = 27,
		68
		69	// BYTES/TEXT/ARRAY/MAP
		70	MINOR_INDEF = 31,
		71	};
24		72
25	// known tags	73	// known tags
26	enum cbor_tag	74	enum cbor_tag
27	{	75	{
28	// inofficial extensions (pending iana registration)	76	// extensions
		77	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29	CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object	78	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30	CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object	79	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31	CBOR_TAG_VALUE_SHARABLE = 26, // http://cbor.schmorp.de/value-sharing	80	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32	CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing	81	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33	CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref	82	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34	CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection	83	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36		84
37	// rfc7049	85	// rfc7049
38	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8	86	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any	87	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40	CBOR_TAG_POS_BIGNUM = 2, // byte string	88	CBOR_TAG_POS_BIGNUM = 2, // byte string
41	CBOR_TAG_NEG_BIGNUM = 3, // byte string	89	CBOR_TAG_NEG_BIGNUM = 3, // byte string
42	CBOR_TAG_DECIMAL = 4, // decimal fraction, array	90	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43	CBOR_TAG_BIGFLOAT = 5, // array	91	CBOR_TAG_BIGFLOAT = 5, // array
44		92
45	CBOR_TAG_CONV_B64U = 21, // base64url, any	93	CBOR_TAG_CONV_B64U = 21, // base64url, any
46	CBOR_TAG_CONV_B64 = 22, // base64, any	94	CBOR_TAG_CONV_B64 = 22, // base64, any
47	CBOR_TAG_CONV_HEX = 23, // base16, any	95	CBOR_TAG_CONV_HEX = 23, // base16, any
48	CBOR_TAG_CBOR = 24, // embedded cbor, byte string	96	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49		97
50	CBOR_TAG_URI = 32, // URI rfc3986, utf-8	98	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8	99	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8	100	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8	101	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	102	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55		103
56	CBOR_TAG_MAGIC = 55799 // self-describe cbor	104	CBOR_TAG_MAGIC = 55799, // self-describe cbor
57	};	105	};
58		106
		107	// known forced types, also hardcoded in CBOR.pm
		108	enum
		109	{
		110	AS_CBOR = 0,
		111	AS_INT = 1,
		112	AS_BYTES = 2,
		113	AS_TEXT = 3,
		114	AS_FLOAT16 = 4,
		115	AS_FLOAT32 = 5,
		116	AS_FLOAT64 = 6,
		117	AS_MAP = 7,
		118	// possibly future enhancements: (generic) float, (generic) string
		119	};
		120
59	#define F_SHRINK 0x00000001UL	121	#define F_SHRINK 0x00000001UL
60	#define F_ALLOW_UNKNOWN 0x00000002UL	122	#define F_ALLOW_UNKNOWN 0x00000002UL
61	#define F_ALLOW_SHARING 0x00000004UL //TODO	123	#define F_ALLOW_SHARING 0x00000004UL
		124	#define F_ALLOW_CYCLES 0x00000008UL
		125	#define F_ALLOW_WEAK_CYCLES 0x00000010UL
		126	#define F_FORBID_OBJECTS 0x00000020UL
62	#define F_DEDUP_STRINGS 0x00000008UL //TODO	127	#define F_PACK_STRINGS 0x00000040UL
63	#define F_DEDUP_KEYS 0x00000010UL //TODO	128	#define F_TEXT_KEYS 0x00000080UL
		129	#define F_TEXT_STRINGS 0x00000100UL
		130	#define F_VALIDATE_UTF8 0x00000200UL
64		131
65	#define INIT_SIZE 32 // initial scalar size to be allocated	132	#define INIT_SIZE 32 // initial scalar size to be allocated
66		133
67	#define SB do {	134	#define SB do {
68	#define SE } while (0)	135	#define SE } while (0)
…		…
80	# define CBOR_SLOW 0	147	# define CBOR_SLOW 0
81	# define CBOR_STASH cbor_stash	148	# define CBOR_STASH cbor_stash
82	#endif	149	#endif
83		150
84	static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::	151	static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85	static SV *types_true, *types_false, *types_error, *sv_cbor;	152	static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86		153
87	typedef struct {	154	typedef struct {
88	U32 flags;	155	U32 flags;
89	U32 max_depth;	156	U32 max_depth;
90	STRLEN max_size;	157	STRLEN max_size;
		158	SV *filter;
		159
		160	// for the incremental parser
		161	STRLEN incr_pos; // the current offset into the text
		162	STRLEN incr_need; // minimum bytes needed to decode
		163	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91	} CBOR;	164	} CBOR;
92		165
93	ecb_inline void	166	ecb_inline void
94	cbor_init (CBOR *cbor)	167	cbor_init (CBOR *cbor)
95	{	168	{
96	Zero (cbor, 1, CBOR);	169	Zero (cbor, 1, CBOR);
97	cbor->max_depth = 512;	170	cbor->max_depth = 512;
		171	}
		172
		173	ecb_inline void
		174	cbor_free (CBOR *cbor)
		175	{
		176	SvREFCNT_dec (cbor->filter);
		177	SvREFCNT_dec (cbor->incr_count);
98	}	178	}
99		179
100	/////////////////////////////////////////////////////////////////////////////	180	/////////////////////////////////////////////////////////////////////////////
101	// utility functions	181	// utility functions
102		182
…		…
124	SvPV_renew (sv, SvCUR (sv) + 1);	204	SvPV_renew (sv, SvCUR (sv) + 1);
125	#endif	205	#endif
126	}	206	}
127	}	207	}
128		208
		209	// minimum length of a string to be registered for stringref
		210	ecb_inline STRLEN
		211	minimum_string_length (UV idx)
		212	{
		213	return idx <= 23 ? 3
		214	: idx <= 0xffU ? 4
		215	: idx <= 0xffffU ? 5
		216	: idx <= 0xffffffffU ? 7
		217	: 11;
		218	}
		219
129	/////////////////////////////////////////////////////////////////////////////	220	/////////////////////////////////////////////////////////////////////////////
130	// encoder	221	// encoder
131		222
132	// structure used for encoding CBOR	223	// structure used for encoding CBOR
133	typedef struct	224	typedef struct
…		…
135	char *cur; // SvPVX (sv) + current output position	226	char *cur; // SvPVX (sv) + current output position
136	char *end; // SvEND (sv)	227	char *end; // SvEND (sv)
137	SV *sv; // result scalar	228	SV *sv; // result scalar
138	CBOR cbor;	229	CBOR cbor;
139	U32 depth; // recursion level	230	U32 depth; // recursion level
140	HV *stringref; // string => index, or 0	231	HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
		232	UV stringref_idx;
141	HV *sharable; // ptr => index, or 0	233	HV *shareable; // ptr => index, or 0
142	HV *sharable_idx;	234	UV shareable_idx;
143	} enc_t;	235	} enc_t;
144		236
145	ecb_inline void	237	ecb_inline void
146	need (enc_t *enc, STRLEN len)	238	need (enc_t *enc, STRLEN len)
147	{	239	{
148	if (ecb_expect_false (enc->cur + len >= enc->end))	240	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
149	{	241	{
150	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	242	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
151	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	243	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
152	enc->cur = SvPVX (enc->sv) + cur;	244	enc->cur = SvPVX (enc->sv) + cur;
153	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	245	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
154	}	246	}
155	}	247	}
156		248
		249	static void encode_sv (enc_t *enc, SV *sv);
		250
157	ecb_inline void	251	ecb_inline void
158	encode_ch (enc_t *enc, char ch)	252	encode_ch (enc_t *enc, char ch)
159	{	253	{
160	need (enc, 1);	254	need (enc, 1);
161	*enc->cur++ = ch;	255	*enc->cur++ = ch;
162	}	256	}
163		257
		258	// used for tags, intregers, element counts and so on
164	static void	259	static void
165	encode_uint (enc_t *enc, int major, UV len)	260	encode_uint (enc_t *enc, int major, UV len)
166	{	261	{
167	need (enc, 9);	262	need (enc, 9);
168		263
169	if (len < 24)	264	if (ecb_expect_true (len < LENGTH_EXT1))
170	*enc->cur++ = major | len;	265	*enc->cur++ = major | len;
171	else if (len <= 0xff)	266	else if (ecb_expect_true (len <= 0xffU))
172	{	267	{
173	*enc->cur++ = major | 24;	268	*enc->cur++ = major | LENGTH_EXT1;
174	*enc->cur++ = len;	269	*enc->cur++ = len;
175	}	270	}
176	else if (len <= 0xffff)	271	else if (len <= 0xffffU)
177	{	272	{
178	*enc->cur++ = major | 25;	273	*enc->cur++ = major | LENGTH_EXT2;
179	*enc->cur++ = len >> 8;	274	*enc->cur++ = len >> 8;
180	*enc->cur++ = len;	275	*enc->cur++ = len;
181	}	276	}
182	else if (len <= 0xffffffff)	277	else if (len <= 0xffffffffU)
183	{	278	{
184	*enc->cur++ = major | 26;	279	*enc->cur++ = major | LENGTH_EXT4;
185	*enc->cur++ = len >> 24;	280	*enc->cur++ = len >> 24;
186	*enc->cur++ = len >> 16;	281	*enc->cur++ = len >> 16;
187	*enc->cur++ = len >> 8;	282	*enc->cur++ = len >> 8;
188	*enc->cur++ = len;	283	*enc->cur++ = len;
189	}	284	}
190	else	285	else
191	{	286	{
192	*enc->cur++ = major | 27;	287	*enc->cur++ = major | LENGTH_EXT8;
193	*enc->cur++ = len >> 56;	288	*enc->cur++ = len >> 56;
194	*enc->cur++ = len >> 48;	289	*enc->cur++ = len >> 48;
195	*enc->cur++ = len >> 40;	290	*enc->cur++ = len >> 40;
196	*enc->cur++ = len >> 32;	291	*enc->cur++ = len >> 32;
197	*enc->cur++ = len >> 24;	292	*enc->cur++ = len >> 24;
…		…
199	*enc->cur++ = len >> 8;	294	*enc->cur++ = len >> 8;
200	*enc->cur++ = len;	295	*enc->cur++ = len;
201	}	296	}
202	}	297	}
203		298
204	static void	299	// encodes a perl value into a CBOR integer
		300	ecb_inline void
		301	encode_int (enc_t *enc, SV *sv)
		302	{
		303	if (SvIsUV (sv))
		304	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
		305	else if (SvIVX (sv) >= 0)
		306	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
		307	else
		308	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
		309	}
		310
		311	ecb_inline void
		312	encode_tag (enc_t *enc, UV tag)
		313	{
		314	encode_uint (enc, MAJOR_TAG, tag);
		315	}
		316
		317	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		318	ecb_noinline static void
205	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	319	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
206	{	320	{
207	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	321	STRLEN ulen = len;
		322	U8 *p, *pend = (U8 *)str + len;
		323
		324	for (p = (U8 *)str; p < pend; ++p)
		325	ulen += *p >> 7; // count set high bits
		326
		327	encode_uint (enc, MAJOR_TEXT, ulen);
		328
		329	need (enc, ulen);
		330	for (p = (U8 *)str; p < pend; ++p)
		331	if (*p < 0x80)
		332	*enc->cur++ = *p;
		333	else
		334	{
		335	*enc->cur++ = 0xc0 + (*p >> 6);
		336	*enc->cur++ = 0x80 + (*p & 63);
		337	}
		338	}
		339
		340	ecb_inline void
		341	encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		342	{
		343	if (ecb_expect_false (upgrade_utf8))
		344	if (!utf8)
		345	{
		346	encode_str_utf8 (enc, utf8, str, len);
		347	return;
		348	}
		349
		350	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
208	need (enc, len);	351	need (enc, len);
209	memcpy (enc->cur, str, len);	352	memcpy (enc->cur, str, len);
210	enc->cur += len;	353	enc->cur += len;
211	}	354	}
212		355
213	ecb_inline void	356	ecb_inline void
214	encode_tag (enc_t *enc, UV tag)	357	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
215	{	358	{
216	encode_uint (enc, 0xc0, tag);	359	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
217	}	360	{
		361	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
218		362
219	static int
220	encode_sharable2 (enc_t *enc, SV *sv)
221	{
222	if (!enc->sharable)
223	enc->sharable = (HV *)sv_2mortal ((SV *)newHV ());
224
225	SV **svp = hv_fetch (enc->sharable, &sv, sizeof (sv), 1);
226
227	if (SvOK (*svp))	363	if (SvOK (*svp))
228	{	364	{
		365	// already registered, use stringref
229	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);	366	encode_tag (enc, CBOR_TAG_STRINGREF);
230	encode_uint (enc, 0x00, SvUV (*svp));	367	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
231
232	return 1;	368	return;
233	}	369	}
234	else	370	else if (len >= minimum_string_length (enc->stringref_idx))
235	{	371	{
		372	// register only
236	sv_setuv (*svp, enc->sharable_idx++);	373	sv_setuv (*svp, enc->stringref_idx);
237	encode_tag (enc, CBOR_TAG_VALUE_SHARABLE);	374	++enc->stringref_idx;
238		375	}
239	return 0;
240	}	376	}
241	}
242		377
		378	encode_str (enc, upgrade_utf8, utf8, str, len);
		379	}
		380
243	ecb_inline int	381	ecb_inline void
		382	encode_float16 (enc_t *enc, NV nv)
		383	{
		384	need (enc, 1+2);
		385
		386	*enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
		387
		388	uint16_t fp = ecb_float_to_binary16 (nv);
		389
		390	if (!ecb_big_endian ())
		391	fp = ecb_bswap16 (fp);
		392
		393	memcpy (enc->cur, &fp, 2);
		394	enc->cur += 2;
		395	}
		396
		397	ecb_inline void
		398	encode_float32 (enc_t *enc, NV nv)
		399	{
		400	need (enc, 1+4);
		401
		402	*enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
		403
		404	uint32_t fp = ecb_float_to_binary32 (nv);
		405
		406	if (!ecb_big_endian ())
		407	fp = ecb_bswap32 (fp);
		408
		409	memcpy (enc->cur, &fp, 4);
		410	enc->cur += 4;
		411	}
		412
		413	ecb_inline void
		414	encode_float64 (enc_t *enc, NV nv)
		415	{
		416	need (enc, 1+8);
		417
		418	*enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
		419
		420	uint64_t fp = ecb_double_to_binary64 (nv);
		421
		422	if (!ecb_big_endian ())
		423	fp = ecb_bswap64 (fp);
		424
		425	memcpy (enc->cur, &fp, 8);
		426	enc->cur += 8;
		427	}
		428
		429	ecb_inline void
		430	encode_bool (enc_t *enc, int istrue)
		431	{
		432	encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
		433	}
		434
		435	// encodes an arrayref containing key-value pairs as CBOR map
		436	ecb_inline void
244	encode_sharable (enc_t *enc, SV *sv)	437	encode_array_as_map (enc_t *enc, SV *sv)
245	{	438	{
246	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)	439	if (enc->depth >= enc->cbor.max_depth)
247	&& ecb_expect_false (SvREFCNT (sv) > 1))	440	croak (ERR_NESTING_EXCEEDED);
248	return encode_sharable2 (enc, sv);
249		441
250	return 0;	442	++enc->depth;
251	}
252		443
253	static void encode_sv (enc_t *enc, SV *sv);	444	// as_map does error checking for us, but we re-check in case
		445	// things have changed.
		446
		447	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
		448	croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
		449
		450	AV *av = (AV *)SvRV (sv);
		451	int i, len = av_len (av);
		452
		453	if (!(len & 1))
		454	croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
		455
		456	encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
		457
		458	for (i = 0; i <= len; ++i)
		459	{
		460	SV **svp = av_fetch (av, i, 0);
		461	encode_sv (enc, svp ? *svp : &PL_sv_undef);
		462	}
		463
		464	--enc->depth;
		465	}
		466
		467	ecb_inline void
		468	encode_forced (enc_t *enc, UV type, SV *sv)
		469	{
		470	switch (type)
		471	{
		472	case AS_CBOR:
		473	{
		474	STRLEN len;
		475	char *str = SvPVbyte (sv, len);
		476
		477	need (enc, len);
		478	memcpy (enc->cur, str, len);
		479	enc->cur += len;
		480	}
		481	break;
		482
		483	case AS_BYTES:
		484	{
		485	STRLEN len;
		486	char *str = SvPVbyte (sv, len);
		487	encode_strref (enc, 0, 0, str, len);
		488	}
		489	break;
		490
		491	case AS_TEXT:
		492	{
		493	STRLEN len;
		494	char *str = SvPVutf8 (sv, len);
		495	encode_strref (enc, 1, 1, str, len);
		496	}
		497	break;
		498
		499	case AS_INT: encode_int (enc, sv); break;
		500
		501	case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
		502	case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
		503	case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
		504
		505	case AS_MAP: encode_array_as_map (enc, sv); break;
		506
		507	default:
		508	croak ("encountered malformed CBOR::XS::Tagged object");
		509	}
		510	}
254		511
255	static void	512	static void
256	encode_av (enc_t *enc, AV *av)	513	encode_av (enc_t *enc, AV *av)
257	{	514	{
258	int i, len = av_len (av);	515	int i, len = av_len (av);
…		…
260	if (enc->depth >= enc->cbor.max_depth)	517	if (enc->depth >= enc->cbor.max_depth)
261	croak (ERR_NESTING_EXCEEDED);	518	croak (ERR_NESTING_EXCEEDED);
262		519
263	++enc->depth;	520	++enc->depth;
264		521
265	encode_uint (enc, 0x80, len + 1);	522	encode_uint (enc, MAJOR_ARRAY, len + 1);
266		523
		524	if (ecb_expect_false (SvMAGICAL (av)))
267	for (i = 0; i <= len; ++i)	525	for (i = 0; i <= len; ++i)
268	{	526	{
269	SV **svp = av_fetch (av, i, 0);	527	SV **svp = av_fetch (av, i, 0);
270	encode_sv (enc, svp ? *svp : &PL_sv_undef);	528	encode_sv (enc, svp ? *svp : &PL_sv_undef);
271	}	529	}
		530	else
		531	for (i = 0; i <= len; ++i)
		532	{
		533	SV *sv = AvARRAY (av)[i];
		534	encode_sv (enc, sv ? sv : &PL_sv_undef);
		535	}
272		536
273	--enc->depth;	537	--enc->depth;
274	}	538	}
275		539
276	static void	540	static void
…		…
284	++enc->depth;	548	++enc->depth;
285		549
286	int pairs = hv_iterinit (hv);	550	int pairs = hv_iterinit (hv);
287	int mg = SvMAGICAL (hv);	551	int mg = SvMAGICAL (hv);
288		552
289	if (mg)	553	if (ecb_expect_false (mg))
290	encode_ch (enc, 0xa0 | 31);	554	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291	else	555	else
292	encode_uint (enc, 0xa0, pairs);	556	encode_uint (enc, MAJOR_MAP, pairs);
293		557
294	while ((he = hv_iternext (hv)))	558	while ((he = hv_iternext (hv)))
295	{	559	{
296	if (HeKLEN (he) == HEf_SVKEY)	560	if (HeKLEN (he) == HEf_SVKEY)
297	encode_sv (enc, HeSVKEY (he));	561	encode_sv (enc, HeSVKEY (he));
298	else	562	else
299	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	563	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300		564
301	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	565	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302	}	566	}
303		567
304	if (mg)	568	if (ecb_expect_false (mg))
305	encode_ch (enc, 0xe0 | 31);	569	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306		570
307	--enc->depth;	571	--enc->depth;
308	}	572	}
309		573
310	// encode objects, arrays and special \0=false and \1=true values.	574	// encode objects, arrays and special \0=false and \1=true values.
311	static void	575	static void
312	encode_rv (enc_t *enc, SV *sv)	576	encode_rv (enc_t *enc, SV *sv)
313	{	577	{
314	svtype svt;
315
316	SvGETMAGIC (sv);	578	SvGETMAGIC (sv);
		579
317	svt = SvTYPE (sv);	580	svtype svt = SvTYPE (sv);
318
319	if (encode_sharable (enc, sv))
320	return;
321		581
322	if (ecb_expect_false (SvOBJECT (sv)))	582	if (ecb_expect_false (SvOBJECT (sv)))
323	{	583	{
324	HV *boolean_stash = !CBOR_SLOW || types_boolean_stash	584	HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
325	? types_boolean_stash	585	? types_boolean_stash
…		…
330	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	590	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
331	? cbor_tagged_stash	591	? cbor_tagged_stash
332	: gv_stashpv ("CBOR::XS::Tagged" , 1);	592	: gv_stashpv ("CBOR::XS::Tagged" , 1);
333		593
334	HV *stash = SvSTASH (sv);	594	HV *stash = SvSTASH (sv);
335	GV *method;
336		595
337	if (stash == boolean_stash)	596	if (stash == boolean_stash)
338	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	597	{
		598	encode_bool (enc, SvIV (sv));
		599	return;
		600	}
339	else if (stash == error_stash)	601	else if (stash == error_stash)
340	encode_ch (enc, 0xe0 | 23);	602	{
		603	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		604	return;
		605	}
341	else if (stash == tagged_stash)	606	else if (stash == tagged_stash)
342	{	607	{
343	if (svt != SVt_PVAV)	608	if (svt != SVt_PVAV)
344	croak ("encountered CBOR::XS::Tagged object that isn't an array");	609	croak ("encountered CBOR::XS::Tagged object that isn't an array");
345		610
		611	switch (av_len ((AV *)sv))
		612	{
		613	case 2-1:
		614	// actually a tagged value
346	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	615	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
347	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));	616	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
		617	break;
		618
		619	case 3-1:
		620	// a forced type [value, type, undef]
		621	encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
		622	break;
		623
		624	default:
		625	croak ("encountered malformed CBOR::XS::Tagged object");
		626	}
		627
		628	return;
		629	}
		630	}
		631
		632	if (ecb_expect_false (SvREFCNT (sv) > 1)
		633	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		634	{
		635	if (ecb_expect_false (!enc->shareable))
		636	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
		637
		638	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
		639
		640	if (SvOK (*svp))
348	}	641	{
		642	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		643	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		644	return;
		645	}
		646	else
		647	{
		648	sv_setuv (*svp, enc->shareable_idx);
		649	++enc->shareable_idx;
		650	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		651	}
		652	}
		653
		654	if (ecb_expect_false (SvOBJECT (sv)))
		655	{
		656	HV *stash = SvSTASH (sv);
		657	GV *method;
		658
		659	if (enc->cbor.flags & F_FORBID_OBJECTS)
		660	croak ("encountered object '%s', but forbid_objects is enabled",
		661	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
349	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	662	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
350	{	663	{
351	dSP;	664	dSP;
352		665
353	ENTER; SAVETMPS; PUSHMARK (SP);	666	ENTER; SAVETMPS;
		667	PUSHMARK (SP);
354	// we re-bless the reference to get overload and other niceties right	668	// we re-bless the reference to get overload and other niceties right
355	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	669	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
356		670
357	PUTBACK;	671	PUTBACK;
358	// G_SCALAR ensures that return value is 1	672	// G_SCALAR ensures that return value is 1
…		…
371	}	685	}
372	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	686	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
373	{	687	{
374	dSP;	688	dSP;
375		689
376	ENTER; SAVETMPS; PUSHMARK (SP);	690	ENTER; SAVETMPS;
		691	PUSHMARK (SP);
377	EXTEND (SP, 2);	692	EXTEND (SP, 2);
378	// we re-bless the reference to get overload and other niceties right	693	// we re-bless the reference to get overload and other niceties right
379	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	694	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
380	PUSHs (sv_cbor);	695	PUSHs (sv_cbor);
381		696
…		…
386	// catch this surprisingly common error	701	// catch this surprisingly common error
387	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	702	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
388	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	703	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
389		704
390	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	705	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
391	encode_uint (enc, 0x80, count + 1);	706	encode_uint (enc, MAJOR_ARRAY, count + 1);
392	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	707	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
393		708
394	while (count)	709	{
		710	int i;
		711
		712	for (i = 0; i < count; ++i)
395	encode_sv (enc, SP[1 - count--]);	713	encode_sv (enc, SP[i + 1 - count]);
		714
		715	SP -= count;
		716	}
396		717
397	PUTBACK;	718	PUTBACK;
398		719
399	FREETMPS; LEAVE;	720	FREETMPS; LEAVE;
400	}	721	}
…		…
418	{	739	{
419	double nv = SvNVX (sv);	740	double nv = SvNVX (sv);
420		741
421	need (enc, 9);	742	need (enc, 9);
422		743
423	if (ecb_expect_false (nv == (U32)nv))	744	if (ecb_expect_false (nv == (NV)(U32)nv))
424	encode_uint (enc, 0x00, (U32)nv);	745	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
425	//TODO: maybe I32?	746	//TODO: maybe I32?
426	else if (ecb_expect_false (nv == (float)nv))	747	else if (ecb_expect_false (nv == (float)nv))
427	{	748	encode_float32 (enc, nv);
428	uint32_t fp = ecb_float_to_binary32 (nv);
429
430	*enc->cur++ = 0xe0 | 26;
431
432	if (!ecb_big_endian ())
433	fp = ecb_bswap32 (fp);
434
435	memcpy (enc->cur, &fp, 4);
436	enc->cur += 4;
437	}
438	else	749	else
439	{	750	encode_float64 (enc, nv);
440	uint64_t fp = ecb_double_to_binary64 (nv);
441
442	*enc->cur++ = 0xe0 | 27;
443
444	if (!ecb_big_endian ())
445	fp = ecb_bswap64 (fp);
446
447	memcpy (enc->cur, &fp, 8);
448	enc->cur += 8;
449	}
450	}	751	}
451		752
452	static void	753	static void
453	encode_sv (enc_t *enc, SV *sv)	754	encode_sv (enc_t *enc, SV *sv)
454	{	755	{
455	SvGETMAGIC (sv);	756	SvGETMAGIC (sv);
456		757
457	if (encode_sharable (enc, sv))
458	return;
459
460	if (SvPOKp (sv))	758	if (SvPOKp (sv))
461	{	759	{
462	STRLEN len;	760	STRLEN len;
463	char *str = SvPV (sv, len);	761	char *str = SvPV (sv, len);
464	encode_str (enc, SvUTF8 (sv), str, len);	762	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
465	}	763	}
466	else if (SvNOKp (sv))	764	else if (SvNOKp (sv))
467	encode_nv (enc, sv);	765	encode_nv (enc, sv);
468	else if (SvIOKp (sv))	766	else if (SvIOKp (sv))
469	{	767	encode_int (enc, sv);
470	if (SvIsUV (sv))
471	encode_uint (enc, 0x00, SvUVX (sv));
472	else if (SvIVX (sv) >= 0)
473	encode_uint (enc, 0x00, SvIVX (sv));
474	else
475	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
476	}
477	else if (SvROK (sv))	768	else if (SvROK (sv))
478	encode_rv (enc, SvRV (sv));	769	encode_rv (enc, SvRV (sv));
479	else if (!SvOK (sv))	770	else if (!SvOK (sv))
480	encode_ch (enc, 0xe0 | 22);	771	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
481	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	772	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
482	encode_ch (enc, 0xe0 | 23);	773	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
483	else	774	else
484	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	775	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
485	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	776	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
486	}	777	}
487		778
488	static SV *	779	static SV *
489	encode_cbor (SV *scalar, CBOR *cbor)	780	encode_cbor (SV *scalar, CBOR *cbor)
490	{	781	{
491	enc_t enc = { };	782	enc_t enc = { 0 };
492		783
493	enc.cbor = *cbor;	784	enc.cbor = *cbor;
494	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	785	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
495	enc.cur = SvPVX (enc.sv);	786	enc.cur = SvPVX (enc.sv);
496	enc.end = SvEND (enc.sv);	787	enc.end = SvEND (enc.sv);
497		788
498	SvPOK_only (enc.sv);	789	SvPOK_only (enc.sv);
		790
		791	if (cbor->flags & F_PACK_STRINGS)
		792	{
		793	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
		794	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
		795	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
		796	}
		797
499	encode_sv (&enc, scalar);	798	encode_sv (&enc, scalar);
500		799
501	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	800	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
502	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	801	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
503		802
…		…
517	U8 *end; // end of input string	816	U8 *end; // end of input string
518	const char *err; // parse error, if != 0	817	const char *err; // parse error, if != 0
519	CBOR cbor;	818	CBOR cbor;
520	U32 depth; // recursion depth	819	U32 depth; // recursion depth
521	U32 maxdepth; // recursion depth limit	820	U32 maxdepth; // recursion depth limit
522	AV *sharable;	821	AV *shareable;
		822	AV *stringref;
		823	SV *decode_tagged;
		824	SV *err_sv; // optional sv for error, needs to be freed
523	} dec_t;	825	} dec_t;
524		826
525	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	827	// set dec->err to ERRSV
		828	ecb_cold static void
		829	err_errsv (dec_t *dec)
		830	{
		831	if (!dec->err)
		832	{
		833	dec->err_sv = newSVsv (ERRSV);
526		834
527	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	835	// chop off the trailing \n
		836	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		837	*SvEND (dec->err_sv) = 0;
528		838
		839	dec->err = SvPVutf8_nolen (dec->err_sv);
		840	}
		841	}
		842
		843	// the following functions are used to reduce code size and help the compiler to optimise
		844	ecb_cold static void
		845	err_set (dec_t *dec, const char *reason)
		846	{
		847	if (!dec->err)
		848	dec->err = reason;
		849	}
		850
		851	ecb_cold static void
		852	err_unexpected_end (dec_t *dec)
		853	{
		854	err_set (dec, "unexpected end of CBOR data");
		855	}
		856
		857	#define ERR_DO(do) SB do; goto fail; SE
		858	#define ERR(reason) ERR_DO (err_set (dec, reason))
		859	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		860
		861	#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
		862
529	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	863	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
530	#define DEC_DEC_DEPTH --dec->depth	864	#define DEC_DEC_DEPTH --dec->depth
531		865
532	static UV	866	static UV
533	decode_uint (dec_t *dec)	867	decode_uint (dec_t *dec)
534	{	868	{
535	switch (*dec->cur & 31)	869	U8 m = *dec->cur & MINOR_MASK;
536	{	870	++dec->cur;
537	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
538	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
539	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
540	return *dec->cur++ & 31;
541		871
542	case 24:	872	if (ecb_expect_true (m < LENGTH_EXT1))
		873	return m;
		874	else if (ecb_expect_true (m == LENGTH_EXT1))
		875	{
543	WANT (2);	876	WANT (1);
544	dec->cur += 2;	877	dec->cur += 1;
545	return dec->cur[-1];	878	return dec->cur[-1];
546		879	}
547	case 25:	880	else if (ecb_expect_true (m == LENGTH_EXT2))
		881	{
548	WANT (3);	882	WANT (2);
549	dec->cur += 3;	883	dec->cur += 2;
550	return (((UV)dec->cur[-2]) << 8)	884	return (((UV)dec->cur[-2]) << 8)
551	| ((UV)dec->cur[-1]);	885	| ((UV)dec->cur[-1]);
552		886	}
553	case 26:	887	else if (ecb_expect_true (m == LENGTH_EXT4))
		888	{
554	WANT (5);	889	WANT (4);
555	dec->cur += 5;	890	dec->cur += 4;
556	return (((UV)dec->cur[-4]) << 24)	891	return (((UV)dec->cur[-4]) << 24)
557	| (((UV)dec->cur[-3]) << 16)	892	| (((UV)dec->cur[-3]) << 16)
558	| (((UV)dec->cur[-2]) << 8)	893	| (((UV)dec->cur[-2]) << 8)
559	| ((UV)dec->cur[-1]);	894	| ((UV)dec->cur[-1]);
560		895	}
561	case 27:	896	else if (ecb_expect_true (m == LENGTH_EXT8))
		897	{
562	WANT (9);	898	WANT (8);
563	dec->cur += 9;	899	dec->cur += 8;
		900
		901	return
		902	#if UVSIZE < 8
		903	0
		904	#else
564	return (((UV)dec->cur[-8]) << 56)	905	(((UV)dec->cur[-8]) << 56)
565	| (((UV)dec->cur[-7]) << 48)	906	| (((UV)dec->cur[-7]) << 48)
566	| (((UV)dec->cur[-6]) << 40)	907	| (((UV)dec->cur[-6]) << 40)
567	| (((UV)dec->cur[-5]) << 32)	908	| (((UV)dec->cur[-5]) << 32)
		909	#endif
568	| (((UV)dec->cur[-4]) << 24)	910	| (((UV)dec->cur[-4]) << 24)
569	| (((UV)dec->cur[-3]) << 16)	911	| (((UV)dec->cur[-3]) << 16)
570	| (((UV)dec->cur[-2]) << 8)	912	| (((UV)dec->cur[-2]) << 8)
571	| ((UV)dec->cur[-1]);	913	| ((UV)dec->cur[-1]);
572		914	}
573	default:	915	else
574	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	916	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
575	}
576		917
577	fail:	918	fail:
578	return 0;	919	return 0;
579	}	920	}
580		921
…		…
585	{	926	{
586	AV *av = newAV ();	927	AV *av = newAV ();
587		928
588	DEC_INC_DEPTH;	929	DEC_INC_DEPTH;
589		930
590	if ((*dec->cur & 31) == 31)	931	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
591	{	932	{
592	++dec->cur;	933	++dec->cur;
593		934
594	for (;;)	935	for (;;)
595	{	936	{
596	WANT (1);	937	WANT (1);
597		938
598	if (*dec->cur == (0xe0 | 31))	939	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
599	{	940	{
600	++dec->cur;	941	++dec->cur;
601	break;	942	break;
602	}	943	}
603		944
604	av_push (av, decode_sv (dec));	945	av_push (av, decode_sv (dec));
605	}	946	}
606	}	947	}
607	else	948	else
608	{	949	{
609	int i, len = decode_uint (dec);	950	UV i, len = decode_uint (dec);
610		951
		952	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
611	av_fill (av, len - 1);	953	av_fill (av, len - 1);
612		954
613	for (i = 0; i < len; ++i)	955	for (i = 0; i < len; ++i)
614	AvARRAY (av)[i] = decode_sv (dec);	956	AvARRAY (av)[i] = decode_sv (dec);
615	}	957	}
616		958
617	DEC_DEC_DEPTH;	959	DEC_DEC_DEPTH;
618	return newRV_noinc ((SV *)av);	960	return newRV_noinc ((SV *)av);
619		961
620	fail:	962	fail:
621	SvREFCNT_dec (av);	963	SvREFCNT_dec_NN (av);
622	DEC_DEC_DEPTH;	964	DEC_DEC_DEPTH;
623	return &PL_sv_undef;	965	return &PL_sv_undef;
624	}	966	}
625		967
626	static void	968	static void
627	decode_he (dec_t *dec, HV *hv)	969	decode_he (dec_t *dec, HV *hv)
628	{	970	{
629	// for speed reasons, we specialcase single-string	971	// for speed reasons, we specialcase single-string
630	// byte or utf-8 strings as keys.	972	// byte or utf-8 strings as keys, but only when !stringref
631		973
632	if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)	974	if (ecb_expect_true (!dec->stringref))
		975	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
633	{	976	{
634	I32 len = decode_uint (dec);	977	STRLEN len = decode_uint (dec);
635	char *key = (char *)dec->cur;	978	char *key = (char *)dec->cur;
636		979
		980	WANT (len);
637	dec->cur += len;	981	dec->cur += len;
638		982
639	hv_store (hv, key, len, decode_sv (dec), 0);	983	hv_store (hv, key, len, decode_sv (dec), 0);
		984
		985	return;
640	}	986	}
641	else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)	987	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
642	{	988	{
643	I32 len = decode_uint (dec);	989	STRLEN len = decode_uint (dec);
644	char *key = (char *)dec->cur;	990	char *key = (char *)dec->cur;
645		991
		992	WANT (len);
646	dec->cur += len;	993	dec->cur += len;
647		994
		995	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		996	if (!cbor_is_utf8_string ((U8 *)key, len))
		997	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
		998
648	hv_store (hv, key, -len, decode_sv (dec), 0);	999	hv_store (hv, key, -len, decode_sv (dec), 0);
		1000
		1001	return;
649	}	1002	}
650	else	1003
651	{
652	SV *k = decode_sv (dec);	1004	SV *k = decode_sv (dec);
653	SV *v = decode_sv (dec);	1005	SV *v = decode_sv (dec);
654		1006
		1007	// we leak memory if uncaught exceptions are thrown by random magical
		1008	// methods, and this is hopefully the only place where it can happen,
		1009	// so if there is a chance of an exception, take the very slow path.
		1010	// since catching exceptions is "undocumented/internal/forbidden" by
		1011	// the new p5p powers, we need to call out to a perl function :/
		1012	if (ecb_expect_false (SvAMAGIC (k)))
		1013	{
		1014	dSP;
		1015
		1016	ENTER; SAVETMPS;
		1017	PUSHMARK (SP);
		1018	EXTEND (SP, 3);
		1019	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		1020	PUSHs (sv_2mortal (k));
		1021	PUSHs (sv_2mortal (v));
		1022
		1023	PUTBACK;
		1024	call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
		1025	SPAGAIN;
		1026
		1027	FREETMPS; LEAVE;
		1028
		1029	if (SvTRUE (ERRSV))
		1030	ERR_ERRSV;
		1031
		1032	return;
		1033	}
		1034
655	hv_store_ent (hv, k, v, 0);	1035	hv_store_ent (hv, k, v, 0);
656	SvREFCNT_dec (k);	1036	SvREFCNT_dec_NN (k);
657	}	1037
		1038	fail:
		1039	;
658	}	1040	}
659		1041
660	static SV *	1042	static SV *
661	decode_hv (dec_t *dec)	1043	decode_hv (dec_t *dec)
662	{	1044	{
663	HV *hv = newHV ();	1045	HV *hv = newHV ();
664		1046
665	DEC_INC_DEPTH;	1047	DEC_INC_DEPTH;
666		1048
667	if ((*dec->cur & 31) == 31)	1049	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
668	{	1050	{
669	++dec->cur;	1051	++dec->cur;
670		1052
671	for (;;)	1053	for (;;)
672	{	1054	{
673	WANT (1);	1055	WANT (1);
674		1056
675	if (*dec->cur == (0xe0 | 31))	1057	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
676	{	1058	{
677	++dec->cur;	1059	++dec->cur;
678	break;	1060	break;
679	}	1061	}
680		1062
681	decode_he (dec, hv);	1063	decode_he (dec, hv);
682	}	1064	}
683	}	1065	}
684	else	1066	else
685	{	1067	{
686	int pairs = decode_uint (dec);	1068	UV pairs = decode_uint (dec);
		1069
		1070	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
687		1071
688	while (pairs--)	1072	while (pairs--)
689	decode_he (dec, hv);	1073	decode_he (dec, hv);
690	}	1074	}
691		1075
692	DEC_DEC_DEPTH;	1076	DEC_DEC_DEPTH;
693	return newRV_noinc ((SV *)hv);	1077	return newRV_noinc ((SV *)hv);
694		1078
695	fail:	1079	fail:
696	SvREFCNT_dec (hv);	1080	SvREFCNT_dec_NN (hv);
697	DEC_DEC_DEPTH;	1081	DEC_DEC_DEPTH;
698	return &PL_sv_undef;	1082	return &PL_sv_undef;
699	}	1083	}
700		1084
701	static SV *	1085	static SV *
702	decode_str (dec_t *dec, int utf8)	1086	decode_str (dec_t *dec, int utf8)
703	{	1087	{
704	SV *sv = 0;	1088	SV *sv = 0;
705		1089
706	if ((*dec->cur & 31) == 31)	1090	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
707	{	1091	{
		1092	// indefinite length strings
708	++dec->cur;	1093	++dec->cur;
709		1094
		1095	U8 major = *dec->cur & MAJOR_MISC;
		1096
710	sv = newSVpvn ("", 0);	1097	sv = newSVpvn ("", 0);
711		1098
712	// not very fast, and certainly not robust against illegal input
713	for (;;)	1099	for (;;)
714	{	1100	{
715	WANT (1);	1101	WANT (1);
716		1102
717	if (*dec->cur == (0xe0 | 31))	1103	if ((*dec->cur - major) > LENGTH_EXT8)
		1104	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
718	{	1105	{
719	++dec->cur;	1106	++dec->cur;
720	break;	1107	break;
721	}	1108	}
		1109	else
		1110	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
722		1111
723	sv_catsv (sv, decode_sv (dec));	1112	STRLEN len = decode_uint (dec);
		1113
		1114	WANT (len);
		1115	sv_catpvn (sv, dec->cur, len);
		1116	dec->cur += len;
724	}	1117	}
725	}	1118	}
726	else	1119	else
727	{	1120	{
728	STRLEN len = decode_uint (dec);	1121	STRLEN len = decode_uint (dec);
729		1122
730	WANT (len);	1123	WANT (len);
731	sv = newSVpvn (dec->cur, len);	1124	sv = newSVpvn (dec->cur, len);
732	dec->cur += len;	1125	dec->cur += len;
		1126
		1127	if (ecb_expect_false (dec->stringref)
		1128	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
		1129	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
733	}	1130	}
734		1131
735	if (utf8)	1132	if (utf8)
		1133	{
		1134	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		1135	if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv)))
		1136	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		1137
736	SvUTF8_on (sv);	1138	SvUTF8_on (sv);
		1139	}
737		1140
738	return sv;	1141	return sv;
739		1142
740	fail:	1143	fail:
741	SvREFCNT_dec (sv);	1144	SvREFCNT_dec (sv);
…		…
743	}	1146	}
744		1147
745	static SV *	1148	static SV *
746	decode_tagged (dec_t *dec)	1149	decode_tagged (dec_t *dec)
747	{	1150	{
		1151	SV *sv = 0;
748	UV tag = decode_uint (dec);	1152	UV tag = decode_uint (dec);
749	SV *sv = decode_sv (dec);	1153
		1154	WANT (1);
750		1155
751	switch (tag)	1156	switch (tag)
752	{	1157	{
753	case CBOR_TAG_MAGIC:	1158	case CBOR_TAG_MAGIC:
754	return sv;	1159	sv = decode_sv (dec);
		1160	break;
755		1161
756	case CBOR_TAG_INDIRECTION:	1162	case CBOR_TAG_INDIRECTION:
757	return newRV_noinc (sv);	1163	sv = newRV_noinc (decode_sv (dec));
		1164	break;
758		1165
		1166	case CBOR_TAG_STRINGREF_NAMESPACE:
		1167	{
		1168	// do not use SAVETMPS/FREETMPS, as these will
		1169	// erase mortalised caches, e.g. "shareable"
		1170	ENTER;
		1171
		1172	SAVESPTR (dec->stringref);
		1173	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
		1174
		1175	sv = decode_sv (dec);
		1176
		1177	LEAVE;
		1178	}
		1179	break;
		1180
		1181	case CBOR_TAG_STRINGREF:
		1182	{
		1183	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
		1184	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
		1185
		1186	UV idx = decode_uint (dec);
		1187
		1188	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
		1189	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
		1190
		1191	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
		1192	}
		1193	break;
		1194
759	case CBOR_TAG_VALUE_SHARABLE:	1195	case CBOR_TAG_VALUE_SHAREABLE:
		1196	{
760	if (ecb_expect_false (!dec->sharable))	1197	if (ecb_expect_false (!dec->shareable))
761	dec->sharable = (AV *)sv_2mortal ((SV *)newAV ());	1198	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
762		1199
		1200	if (ecb_expect_false (dec->cbor.flags & (F_ALLOW_CYCLES | F_ALLOW_WEAK_CYCLES)))
		1201	{
		1202	// if cycles are allowed, then we store an AV as value
		1203	// while it is being decoded, and gather unresolved
		1204	// references in it, to be re4solved after decoding.
		1205	int idx, i;
		1206	AV *av = newAV ();
		1207	av_push (dec->shareable, (SV *)av);
		1208	idx = AvFILLp (dec->shareable);
		1209
		1210	sv = decode_sv (dec);
		1211
		1212	// the AV now contains \undef for all unresolved references,
		1213	// so we fix them up here.
		1214	for (i = 0; i <= AvFILLp (av); ++i)
		1215	SvRV_set (AvARRAY (av)[i], SvREFCNT_inc_NN (SvRV (sv)));
		1216
		1217	// weaken all recursive references
		1218	if (dec->cbor.flags & F_ALLOW_WEAK_CYCLES)
		1219	for (i = 0; i <= AvFILLp (av); ++i)
		1220	sv_rvweaken (AvARRAY (av)[i]);
		1221
		1222	// now replace the AV by a reference to the completed value
		1223	SvREFCNT_dec_NN ((SV *)av);
763	av_push (dec->sharable, SvREFCNT_inc_NN (sv));	1224	AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
764		1225	}
765	return sv;	1226	else
		1227	{
		1228	av_push (dec->shareable, &PL_sv_undef);
		1229	int idx = AvFILLp (dec->shareable);
		1230	sv = decode_sv (dec);
		1231	AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
		1232	}
		1233	}
		1234	break;
766		1235
767	case CBOR_TAG_VALUE_SHAREDREF:	1236	case CBOR_TAG_VALUE_SHAREDREF:
768	{	1237	{
769	// TODO: should verify that the sv atcually was a CBOR unsigned integer	1238	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
770	UV idx = SvUV (sv);	1239	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
771		1240
		1241	UV idx = decode_uint (dec);
		1242
772	if (!dec->sharable || idx > AvFILLp (dec->sharable))	1243	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
773	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1244	ERR ("corrupted CBOR data (sharedref index out of bounds)");
774		1245
775	SvREFCNT_dec (sv);	1246	sv = AvARRAY (dec->shareable)[idx];
776		1247
777	return SvREFCNT_inc_NN (AvARRAY (dec->sharable)[idx]);	1248	// reference to cycle, we create a new \undef and use that, and also
		1249	// registerr it in the AV for later fixing
		1250	if (ecb_expect_false (SvTYPE (sv) == SVt_PVAV))
		1251	{
		1252	AV *av = (AV *)sv;
		1253	sv = newRV_noinc (&PL_sv_undef);
		1254	av_push (av, SvREFCNT_inc_NN (sv));
778	}	1255	}
		1256	else if (ecb_expect_false (sv == &PL_sv_undef)) // not yet decoded, but cycles not allowed
		1257	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
		1258	else // we decoded the object earlier, no cycle
		1259	sv = newSVsv (sv);
		1260	}
		1261	break;
779		1262
780	case CBOR_TAG_PERL_OBJECT:	1263	case CBOR_TAG_PERL_OBJECT:
781	{	1264	{
		1265	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1266	goto filter;
		1267
		1268	sv = decode_sv (dec);
		1269
782	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1270	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
783	ERR ("corrupted CBOR data (non-array perl object)");	1271	ERR ("corrupted CBOR data (non-array perl object)");
784		1272
785	AV *av = (AV *)SvRV (sv);	1273	AV *av = (AV *)SvRV (sv);
786	int len = av_len (av) + 1;	1274	int len = av_len (av) + 1;
…		…
794	if (!method)	1282	if (!method)
795	ERR ("cannot decode perl-object (package does not have a THAW method)");	1283	ERR ("cannot decode perl-object (package does not have a THAW method)");
796		1284
797	dSP;	1285	dSP;
798		1286
799	ENTER; SAVETMPS; PUSHMARK (SP);	1287	ENTER; SAVETMPS;
		1288	PUSHMARK (SP);
800	EXTEND (SP, len + 1);	1289	EXTEND (SP, len + 1);
801	// we re-bless the reference to get overload and other niceties right	1290	// we re-bless the reference to get overload and other niceties right
802	PUSHs (*av_fetch (av, 0, 1));	1291	PUSHs (*av_fetch (av, 0, 1));
803	PUSHs (sv_cbor);	1292	PUSHs (sv_cbor);
804		1293
…		…
812	SPAGAIN;	1301	SPAGAIN;
813		1302
814	if (SvTRUE (ERRSV))	1303	if (SvTRUE (ERRSV))
815	{	1304	{
816	FREETMPS; LEAVE;	1305	FREETMPS; LEAVE;
817	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1306	ERR_ERRSV;
818	}	1307	}
819		1308
820	SvREFCNT_dec (sv);	1309	SvREFCNT_dec_NN (sv);
821	sv = SvREFCNT_inc (POPs);	1310	sv = SvREFCNT_inc (POPs);
822		1311
823	PUTBACK;	1312	PUTBACK;
824		1313
825	FREETMPS; LEAVE;	1314	FREETMPS; LEAVE;
826
827	return sv;
828	}	1315	}
		1316	break;
829		1317
830	default:	1318	default:
		1319	filter:
		1320	{
		1321	SV *tag_sv = newSVuv (tag);
		1322
		1323	sv = decode_sv (dec);
		1324
		1325	dSP;
		1326	ENTER; SAVETMPS;
		1327	PUSHMARK (SP);
		1328	EXTEND (SP, 2);
		1329	PUSHs (tag_sv);
		1330	PUSHs (sv);
		1331
		1332	PUTBACK;
		1333	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
		1334	SPAGAIN;
		1335
		1336	if (SvTRUE (ERRSV))
831	{	1337	{
		1338	SvREFCNT_dec_NN (tag_sv);
		1339	FREETMPS; LEAVE;
		1340	ERR_ERRSV;
		1341	}
		1342
		1343	if (count)
		1344	{
		1345	SvREFCNT_dec_NN (tag_sv);
		1346	SvREFCNT_dec_NN (sv);
		1347	sv = SvREFCNT_inc_NN (TOPs);
		1348	SP -= count;
		1349	}
		1350	else
		1351	{
832	AV *av = newAV ();	1352	AV *av = newAV ();
833	av_push (av, newSVuv (tag));	1353	av_push (av, tag_sv);
834	av_push (av, sv);	1354	av_push (av, sv);
835		1355
836	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1356	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
837	? cbor_tagged_stash	1357	? cbor_tagged_stash
838	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1358	: gv_stashpv ("CBOR::XS::Tagged" , 1);
839
840	return sv_bless (newRV_noinc ((SV *)av), tagged_stash);	1359	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
841	}	1360	}
		1361
		1362	PUTBACK;
		1363
		1364	FREETMPS; LEAVE;
		1365	}
		1366	break;
842	}	1367	}
		1368
		1369	return sv;
843		1370
844	fail:	1371	fail:
845	SvREFCNT_dec (sv);	1372	SvREFCNT_dec (sv);
846	return &PL_sv_undef;	1373	return &PL_sv_undef;
847	}	1374	}
…		…
849	static SV *	1376	static SV *
850	decode_sv (dec_t *dec)	1377	decode_sv (dec_t *dec)
851	{	1378	{
852	WANT (1);	1379	WANT (1);
853		1380
854	switch (*dec->cur >> 5)	1381	switch (*dec->cur >> MAJOR_SHIFT)
855	{	1382	{
856	case 0: // unsigned int	1383	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
857	return newSVuv (decode_uint (dec));	1384	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
858	case 1: // negative int	1385	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
859	return newSViv (-1 - (IV)decode_uint (dec));	1386	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
860	case 2: // octet string	1387	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
861	return decode_str (dec, 0);	1388	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
862	case 3: // utf-8 string	1389	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
863	return decode_str (dec, 1);	1390
864	case 4: // array	1391	case MAJOR_MISC >> MAJOR_SHIFT:
865	return decode_av (dec);
866	case 5: // map
867	return decode_hv (dec);
868	case 6: // tag
869	return decode_tagged (dec);
870	case 7: // misc
871	switch (*dec->cur++ & 31)	1392	switch (*dec->cur++ & MINOR_MASK)
872	{	1393	{
873	case 20:	1394	case SIMPLE_FALSE:
874	#if CBOR_SLOW	1395	#if CBOR_SLOW
875	types_false = get_bool ("Types::Serialiser::false");	1396	types_false = get_bool ("Types::Serialiser::false");
876	#endif	1397	#endif
877	return newSVsv (types_false);	1398	return newSVsv (types_false);
878	case 21:	1399	case SIMPLE_TRUE:
879	#if CBOR_SLOW	1400	#if CBOR_SLOW
880	types_true = get_bool ("Types::Serialiser::true");	1401	types_true = get_bool ("Types::Serialiser::true");
881	#endif	1402	#endif
882	return newSVsv (types_true);	1403	return newSVsv (types_true);
883	case 22:	1404	case SIMPLE_NULL:
884	return newSVsv (&PL_sv_undef);	1405	return newSVsv (&PL_sv_undef);
885	case 23:	1406	case SIMPLE_UNDEF:
886	#if CBOR_SLOW	1407	#if CBOR_SLOW
887	types_error = get_bool ("Types::Serialiser::error");	1408	types_error = get_bool ("Types::Serialiser::error");
888	#endif	1409	#endif
889	return newSVsv (types_error);	1410	return newSVsv (types_error);
890		1411
891	case 25:	1412	case MISC_FLOAT16:
892	{	1413	{
893	WANT (2);	1414	WANT (2);
894		1415
895	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1416	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
896	dec->cur += 2;	1417	dec->cur += 2;
897		1418
898	return newSVnv (ecb_binary16_to_float (fp));	1419	return newSVnv (ecb_binary16_to_float (fp));
899	}	1420	}
900		1421
901	case 26:	1422	case MISC_FLOAT32:
902	{	1423	{
903	uint32_t fp;	1424	uint32_t fp;
904	WANT (4);	1425	WANT (4);
905	memcpy (&fp, dec->cur, 4);	1426	memcpy (&fp, dec->cur, 4);
906	dec->cur += 4;	1427	dec->cur += 4;
…		…
909	fp = ecb_bswap32 (fp);	1430	fp = ecb_bswap32 (fp);
910		1431
911	return newSVnv (ecb_binary32_to_float (fp));	1432	return newSVnv (ecb_binary32_to_float (fp));
912	}	1433	}
913		1434
914	case 27:	1435	case MISC_FLOAT64:
915	{	1436	{
916	uint64_t fp;	1437	uint64_t fp;
917	WANT (8);	1438	WANT (8);
918	memcpy (&fp, dec->cur, 8);	1439	memcpy (&fp, dec->cur, 8);
919	dec->cur += 8;	1440	dec->cur += 8;
…		…
922	fp = ecb_bswap64 (fp);	1443	fp = ecb_bswap64 (fp);
923		1444
924	return newSVnv (ecb_binary64_to_double (fp));	1445	return newSVnv (ecb_binary64_to_double (fp));
925	}	1446	}
926		1447
927	// 0..19 unassigned	1448	// 0..19 unassigned simple
928	// 24 reserved + unassigned (reserved values are not encodable)	1449	// 24 reserved + unassigned simple (reserved values are not encodable)
		1450	// 28-30 unassigned misc
		1451	// 31 break code
929	default:	1452	default:
930	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1453	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
931	}	1454	}
932		1455
933	break;	1456	break;
934	}	1457	}
935		1458
…		…
938	}	1461	}
939		1462
940	static SV *	1463	static SV *
941	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1464	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
942	{	1465	{
943	dec_t dec = { };	1466	dec_t dec = { 0 };
944	SV *sv;	1467	SV *sv;
945	STRLEN len;	1468	STRLEN len;
946	char *data = SvPVbyte (string, len);	1469	char *data = SvPVbyte (string, len);
947		1470
948	if (len > cbor->max_size && cbor->max_size)	1471	if (len > cbor->max_size && cbor->max_size)
…		…
962	if (dec.cur != dec.end && !dec.err)	1485	if (dec.cur != dec.end && !dec.err)
963	dec.err = "garbage after CBOR object";	1486	dec.err = "garbage after CBOR object";
964		1487
965	if (dec.err)	1488	if (dec.err)
966	{	1489	{
		1490	if (dec.shareable)
		1491	{
		1492	// need to break cyclic links, which would all be in shareable
		1493	int i;
		1494	SV **svp;
		1495
		1496	for (i = av_len (dec.shareable) + 1; i--; )
		1497	if ((svp = av_fetch (dec.shareable, i, 0)))
		1498	sv_setsv (*svp, &PL_sv_undef);
		1499	}
		1500
967	SvREFCNT_dec (sv);	1501	SvREFCNT_dec_NN (sv);
		1502
		1503	if (dec.err_sv)
		1504	sv_2mortal (dec.err_sv);
		1505
968	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);	1506	croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur);
969	}	1507	}
970		1508
971	sv = sv_2mortal (sv);	1509	sv = sv_2mortal (sv);
972		1510
973	return sv;	1511	return sv;
974	}	1512	}
975		1513
		1514	/////////////////////////////////////////////////////////////////////////////
		1515	// incremental parser
		1516
		1517	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1518
		1519	// returns 0 for notyet, 1 for success or error
		1520	static int
		1521	incr_parse (CBOR *self, SV *cborstr)
		1522	{
		1523	STRLEN cur;
		1524	SvPV (cborstr, cur);
		1525
		1526	while (ecb_expect_true (self->incr_need <= cur))
		1527	{
		1528	// table of integer count bytes
		1529	static I8 incr_len[MINOR_MASK + 1] = {
		1530	0, 0, 0, 0, 0, 0, 0, 0,
		1531	0, 0, 0, 0, 0, 0, 0, 0,
		1532	0, 0, 0, 0, 0, 0, 0, 0,
		1533	1, 2, 4, 8,-1,-1,-1,-2
		1534	};
		1535
		1536	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1537	U8 m = *p & MINOR_MASK;
		1538	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1539	I8 ilen = incr_len[m];
		1540
		1541	self->incr_need = self->incr_pos + 1;
		1542
		1543	if (ecb_expect_false (ilen < 0))
		1544	{
		1545	if (m != MINOR_INDEF)
		1546	return 1; // error
		1547
		1548	if (*p == (MAJOR_MISC | MINOR_INDEF))
		1549	{
		1550	if (count >= 0)
		1551	return 1; // error
		1552
		1553	count = 1;
		1554	}
		1555	else
		1556	{
		1557	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1558	count = -1;
		1559	}
		1560	}
		1561	else
		1562	{
		1563	self->incr_need += ilen;
		1564	if (ecb_expect_false (self->incr_need > cur))
		1565	return 0;
		1566
		1567	int major = *p >> MAJOR_SHIFT;
		1568
		1569	switch (major)
		1570	{
		1571	case MAJOR_TAG >> MAJOR_SHIFT:
		1572	++count; // tags merely prefix another value
		1573	break;
		1574
		1575	case MAJOR_BYTES >> MAJOR_SHIFT:
		1576	case MAJOR_TEXT >> MAJOR_SHIFT:
		1577	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1578	case MAJOR_MAP >> MAJOR_SHIFT:
		1579	{
		1580	UV len;
		1581
		1582	if (ecb_expect_false (ilen))
		1583	{
		1584	len = 0;
		1585
		1586	do {
		1587	len = (len << 8) | *++p;
		1588	} while (--ilen);
		1589	}
		1590	else
		1591	len = m;
		1592
		1593	switch (major)
		1594	{
		1595	case MAJOR_BYTES >> MAJOR_SHIFT:
		1596	case MAJOR_TEXT >> MAJOR_SHIFT:
		1597	self->incr_need += len;
		1598	if (ecb_expect_false (self->incr_need > cur))
		1599	return 0;
		1600
		1601	break;
		1602
		1603	case MAJOR_MAP >> MAJOR_SHIFT:
		1604	len <<= 1;
		1605	/* FALLTHROUGH */
		1606	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1607	if (len)
		1608	{
		1609	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1610	count = len + 1;
		1611	}
		1612	break;
		1613	}
		1614	}
		1615	}
		1616	}
		1617
		1618	self->incr_pos = self->incr_need;
		1619
		1620	if (count > 0)
		1621	{
		1622	while (!--count)
		1623	{
		1624	if (!AvFILLp (self->incr_count))
		1625	return 1; // done
		1626
		1627	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1628	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1629	}
		1630
		1631	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1632	}
		1633	}
		1634
		1635	return 0;
		1636	}
		1637
		1638
976	/////////////////////////////////////////////////////////////////////////////	1639	/////////////////////////////////////////////////////////////////////////////
977	// XS interface functions	1640	// XS interface functions
978		1641
979	MODULE = CBOR::XS PACKAGE = CBOR::XS	1642	MODULE = CBOR::XS PACKAGE = CBOR::XS
980		1643
…		…
988		1651
989	types_true = get_bool ("Types::Serialiser::true" );	1652	types_true = get_bool ("Types::Serialiser::true" );
990	types_false = get_bool ("Types::Serialiser::false");	1653	types_false = get_bool ("Types::Serialiser::false");
991	types_error = get_bool ("Types::Serialiser::error");	1654	types_error = get_bool ("Types::Serialiser::error");
992		1655
		1656	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
		1657
993	sv_cbor = newSVpv ("CBOR", 0);	1658	sv_cbor = newSVpv ("CBOR", 0);
994	SvREADONLY_on (sv_cbor);	1659	SvREADONLY_on (sv_cbor);
		1660
		1661	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
995	}	1662	}
996		1663
997	PROTOTYPES: DISABLE	1664	PROTOTYPES: DISABLE
998		1665
999	void CLONE (...)	1666	void CLONE (...)
…		…
1015	)));	1682	)));
1016	}	1683	}
1017		1684
1018	void shrink (CBOR *self, int enable = 1)	1685	void shrink (CBOR *self, int enable = 1)
1019	ALIAS:	1686	ALIAS:
1020	shrink = F_SHRINK	1687	shrink = F_SHRINK
1021	allow_unknown = F_ALLOW_UNKNOWN	1688	allow_unknown = F_ALLOW_UNKNOWN
1022	allow_sharing = F_ALLOW_SHARING	1689	allow_sharing = F_ALLOW_SHARING
1023	dedup_keys = F_DEDUP_KEYS	1690	allow_cycles = F_ALLOW_CYCLES
		1691	allow_weak_cycles = F_ALLOW_WEAK_CYCLES
		1692	forbid_objects = F_FORBID_OBJECTS
1024	dedup_strings = F_DEDUP_STRINGS	1693	pack_strings = F_PACK_STRINGS
		1694	text_keys = F_TEXT_KEYS
		1695	text_strings = F_TEXT_STRINGS
		1696	validate_utf8 = F_VALIDATE_UTF8
1025	PPCODE:	1697	PPCODE:
1026	{	1698	{
1027	if (enable)	1699	if (enable)
1028	self->flags |= ix;	1700	self->flags |= ix;
1029	else	1701	else
…		…
1032	XPUSHs (ST (0));	1704	XPUSHs (ST (0));
1033	}	1705	}
1034		1706
1035	void get_shrink (CBOR *self)	1707	void get_shrink (CBOR *self)
1036	ALIAS:	1708	ALIAS:
1037	get_shrink = F_SHRINK	1709	get_shrink = F_SHRINK
1038	get_allow_unknown = F_ALLOW_UNKNOWN	1710	get_allow_unknown = F_ALLOW_UNKNOWN
1039	get_allow_sharing = F_ALLOW_SHARING	1711	get_allow_sharing = F_ALLOW_SHARING
1040	get_dedup_keys = F_DEDUP_KEYS	1712	get_allow_cycles = F_ALLOW_CYCLES
		1713	get_allow_weak_cycles = F_ALLOW_WEAK_CYCLES
		1714	get_forbid_objects = F_FORBID_OBJECTS
1041	get_dedup_strings = F_DEDUP_STRINGS	1715	get_pack_strings = F_PACK_STRINGS
		1716	get_text_keys = F_TEXT_KEYS
		1717	get_text_strings = F_TEXT_STRINGS
		1718	get_validate_utf8 = F_VALIDATE_UTF8
1042	PPCODE:	1719	PPCODE:
1043	XPUSHs (boolSV (self->flags & ix));	1720	XPUSHs (boolSV (self->flags & ix));
1044		1721
1045	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1722	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1046	PPCODE:	1723	PPCODE:
…		…
1059	XPUSHs (ST (0));	1736	XPUSHs (ST (0));
1060		1737
1061	int get_max_size (CBOR *self)	1738	int get_max_size (CBOR *self)
1062	CODE:	1739	CODE:
1063	RETVAL = self->max_size;	1740	RETVAL = self->max_size;
		1741	OUTPUT:
		1742	RETVAL
		1743
		1744	void filter (CBOR *self, SV *filter = 0)
		1745	PPCODE:
		1746	SvREFCNT_dec (self->filter);
		1747	self->filter = filter ? newSVsv (filter) : filter;
		1748	XPUSHs (ST (0));
		1749
		1750	SV *get_filter (CBOR *self)
		1751	CODE:
		1752	RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1064	OUTPUT:	1753	OUTPUT:
1065	RETVAL	1754	RETVAL
1066		1755
1067	void encode (CBOR *self, SV *scalar)	1756	void encode (CBOR *self, SV *scalar)
1068	PPCODE:	1757	PPCODE:
…		…
1083	EXTEND (SP, 2);	1772	EXTEND (SP, 2);
1084	PUSHs (sv);	1773	PUSHs (sv);
1085	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1774	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1086	}	1775	}
1087		1776
		1777	void incr_parse (CBOR *self, SV *cborstr)
		1778	ALIAS:
		1779	incr_parse_multiple = 1
		1780	PPCODE:
		1781	{
		1782	if (SvUTF8 (cborstr))
		1783	sv_utf8_downgrade (cborstr, 0);
		1784
		1785	if (!self->incr_count)
		1786	{
		1787	self->incr_count = newAV ();
		1788	self->incr_pos = 0;
		1789	self->incr_need = 1;
		1790
		1791	av_push (self->incr_count, newSViv (1));
		1792	}
		1793
		1794	do
		1795	{
		1796	if (!incr_parse (self, cborstr))
		1797	{
		1798	if (self->incr_need > self->max_size && self->max_size)
		1799	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1800	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1801
		1802	break;
		1803	}
		1804
		1805	SV *sv;
		1806	char *offset;
		1807
		1808	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1809	XPUSHs (sv);
		1810
		1811	sv_chop (cborstr, offset);
		1812
		1813	av_clear (self->incr_count);
		1814	av_push (self->incr_count, newSViv (1));
		1815
		1816	self->incr_pos = 0;
		1817	self->incr_need = self->incr_pos + 1;
		1818	}
		1819	while (ix);
		1820	}
		1821
		1822	void incr_reset (CBOR *self)
		1823	CODE:
		1824	{
		1825	SvREFCNT_dec (self->incr_count);
		1826	self->incr_count = 0;
		1827	}
		1828
		1829	void DESTROY (CBOR *self)
		1830	PPCODE:
		1831	cbor_free (self);
		1832
1088	PROTOTYPES: ENABLE	1833	PROTOTYPES: ENABLE
1089		1834
1090	void encode_cbor (SV *scalar)	1835	void encode_cbor (SV *scalar)
		1836	ALIAS:
		1837	encode_cbor = 0
		1838	encode_cbor_sharing = F_ALLOW_SHARING
1091	PPCODE:	1839	PPCODE:
1092	{	1840	{
1093	CBOR cbor;	1841	CBOR cbor;
1094	cbor_init (&cbor);	1842	cbor_init (&cbor);
		1843	cbor.flags |= ix;
1095	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1844	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1096	XPUSHs (scalar);	1845	XPUSHs (scalar);
1097	}	1846	}
1098		1847
1099	void decode_cbor (SV *cborstr)	1848	void decode_cbor (SV *cborstr)
…		…
1103	cbor_init (&cbor);	1852	cbor_init (&cbor);
1104	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1853	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1105	XPUSHs (cborstr);	1854	XPUSHs (cborstr);
1106	}	1855	}
1107		1856
		1857	#ifdef __AFL_COMPILER
		1858
		1859	void
		1860	afl_init ()
		1861	CODE:
		1862	__AFL_INIT ();
		1863
		1864	int
		1865	afl_loop (unsigned int count = 10000)
		1866	CODE:
		1867	RETVAL = __AFL_LOOP (count);
		1868	OUTPUT:
		1869	RETVAL
		1870
		1871	#endif
		1872

Diff Legend

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