ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.26 by root, Fri Nov 22 09:40:13 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_SHAREABLE = 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
62	#define F_ALLOW_STRINGREF 0x00000008UL //TODO	124	#define F_ALLOW_CYCLES 0x00000008UL
		125	#define F_ALLOW_WEAK_CYCLES 0x00000010UL
		126	#define F_FORBID_OBJECTS 0x00000020UL
		127	#define F_PACK_STRINGS 0x00000040UL
		128	#define F_TEXT_KEYS 0x00000080UL
		129	#define F_TEXT_STRINGS 0x00000100UL
		130	#define F_VALIDATE_UTF8 0x00000200UL
63		131
64	#define INIT_SIZE 32 // initial scalar size to be allocated	132	#define INIT_SIZE 32 // initial scalar size to be allocated
65		133
66	#define SB do {	134	#define SB do {
67	#define SE } while (0)	135	#define SE } while (0)
…		…
79	# define CBOR_SLOW 0	147	# define CBOR_SLOW 0
80	# define CBOR_STASH cbor_stash	148	# define CBOR_STASH cbor_stash
81	#endif	149	#endif
82		150
83	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::
84	static SV *types_true, *types_false, *types_error, *sv_cbor;	152	static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
85		153
86	typedef struct {	154	typedef struct {
87	U32 flags;	155	U32 flags;
88	U32 max_depth;	156	U32 max_depth;
89	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.
90	} CBOR;	164	} CBOR;
91		165
92	ecb_inline void	166	ecb_inline void
93	cbor_init (CBOR *cbor)	167	cbor_init (CBOR *cbor)
94	{	168	{
95	Zero (cbor, 1, CBOR);	169	Zero (cbor, 1, CBOR);
96	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);
97	}	178	}
98		179
99	/////////////////////////////////////////////////////////////////////////////	180	/////////////////////////////////////////////////////////////////////////////
100	// utility functions	181	// utility functions
101		182
…		…
124	#endif	205	#endif
125	}	206	}
126	}	207	}
127		208
128	// minimum length of a string to be registered for stringref	209	// minimum length of a string to be registered for stringref
129	ecb_inline int	210	ecb_inline STRLEN
130	minimum_string_length (UV idx)	211	minimum_string_length (UV idx)
131	{	212	{
132	return idx > 23	213	return idx <= 23 ? 3
133	? idx > 0xffU	214	: idx <= 0xffU ? 4
134	? idx > 0xffffU	215	: idx <= 0xffffU ? 5
135	? idx > 0xffffffffU	216	: idx <= 0xffffffffU ? 7
136	? 7	217	: 11;
137	: 6
138	: 5
139	: 4
140	: 3;
141	}	218	}
142		219
143	/////////////////////////////////////////////////////////////////////////////	220	/////////////////////////////////////////////////////////////////////////////
144	// encoder	221	// encoder
145		222
…		…
158	} enc_t;	235	} enc_t;
159		236
160	ecb_inline void	237	ecb_inline void
161	need (enc_t *enc, STRLEN len)	238	need (enc_t *enc, STRLEN len)
162	{	239	{
163	if (ecb_expect_false (enc->cur + len >= enc->end))	240	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
164	{	241	{
165	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	242	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
166	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	243	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
167	enc->cur = SvPVX (enc->sv) + cur;	244	enc->cur = SvPVX (enc->sv) + cur;
168	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	245	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
169	}	246	}
170	}	247	}
171		248
		249	static void encode_sv (enc_t *enc, SV *sv);
		250
172	ecb_inline void	251	ecb_inline void
173	encode_ch (enc_t *enc, char ch)	252	encode_ch (enc_t *enc, char ch)
174	{	253	{
175	need (enc, 1);	254	need (enc, 1);
176	*enc->cur++ = ch;	255	*enc->cur++ = ch;
177	}	256	}
178		257
		258	// used for tags, intregers, element counts and so on
179	static void	259	static void
180	encode_uint (enc_t *enc, int major, UV len)	260	encode_uint (enc_t *enc, int major, UV len)
181	{	261	{
182	need (enc, 9);	262	need (enc, 9);
183		263
184	if (len < 24)	264	if (ecb_expect_true (len < LENGTH_EXT1))
185	*enc->cur++ = major | len;	265	*enc->cur++ = major | len;
186	else if (len <= 0xff)	266	else if (ecb_expect_true (len <= 0xffU))
187	{	267	{
188	*enc->cur++ = major | 24;	268	*enc->cur++ = major | LENGTH_EXT1;
189	*enc->cur++ = len;	269	*enc->cur++ = len;
190	}	270	}
191	else if (len <= 0xffff)	271	else if (len <= 0xffffU)
192	{	272	{
193	*enc->cur++ = major | 25;	273	*enc->cur++ = major | LENGTH_EXT2;
194	*enc->cur++ = len >> 8;	274	*enc->cur++ = len >> 8;
195	*enc->cur++ = len;	275	*enc->cur++ = len;
196	}	276	}
197	else if (len <= 0xffffffff)	277	else if (len <= 0xffffffffU)
198	{	278	{
199	*enc->cur++ = major | 26;	279	*enc->cur++ = major | LENGTH_EXT4;
200	*enc->cur++ = len >> 24;	280	*enc->cur++ = len >> 24;
201	*enc->cur++ = len >> 16;	281	*enc->cur++ = len >> 16;
202	*enc->cur++ = len >> 8;	282	*enc->cur++ = len >> 8;
203	*enc->cur++ = len;	283	*enc->cur++ = len;
204	}	284	}
205	else	285	else
206	{	286	{
207	*enc->cur++ = major | 27;	287	*enc->cur++ = major | LENGTH_EXT8;
208	*enc->cur++ = len >> 56;	288	*enc->cur++ = len >> 56;
209	*enc->cur++ = len >> 48;	289	*enc->cur++ = len >> 48;
210	*enc->cur++ = len >> 40;	290	*enc->cur++ = len >> 40;
211	*enc->cur++ = len >> 32;	291	*enc->cur++ = len >> 32;
212	*enc->cur++ = len >> 24;	292	*enc->cur++ = len >> 24;
…		…
214	*enc->cur++ = len >> 8;	294	*enc->cur++ = len >> 8;
215	*enc->cur++ = len;	295	*enc->cur++ = len;
216	}	296	}
217	}	297	}
218		298
		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
219	ecb_inline void	311	ecb_inline void
220	encode_tag (enc_t *enc, UV tag)	312	encode_tag (enc_t *enc, UV tag)
221	{	313	{
222	encode_uint (enc, 0xc0, tag);	314	encode_uint (enc, MAJOR_TAG, tag);
223	}	315	}
224		316
225	static void	317	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		318	ecb_noinline static void
226	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	319	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
227	{	320	{
		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);
		351	need (enc, len);
		352	memcpy (enc->cur, str, len);
		353	enc->cur += len;
		354	}
		355
		356	ecb_inline void
		357	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		358	{
228	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF))	359	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
229	{	360	{
230	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	361	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
231		362
232	if (SvOK (*svp))	363	if (SvOK (*svp))
233	{	364	{
234	// already registered, use stringref	365	// already registered, use stringref
235	encode_tag (enc, CBOR_TAG_STRINGREF);	366	encode_tag (enc, CBOR_TAG_STRINGREF);
236	encode_uint (enc, 0x00, SvUV (*svp));	367	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
237	return;	368	return;
238	}	369	}
239	else if (len >= minimum_string_length (enc->stringref_idx))	370	else if (len >= minimum_string_length (enc->stringref_idx))
240	{	371	{
241	// register only	372	// register only
242	sv_setuv (*svp, enc->stringref_idx);	373	sv_setuv (*svp, enc->stringref_idx);
243	++enc->stringref_idx;	374	++enc->stringref_idx;
244	}	375	}
245	}	376	}
246		377
247	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	378	encode_str (enc, upgrade_utf8, utf8, str, len);
		379	}
		380
		381	ecb_inline void
		382	encode_float16 (enc_t *enc, NV nv)
		383	{
248	need (enc, len);	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
249	memcpy (enc->cur, str, len);	393	memcpy (enc->cur, &fp, 2);
250	enc->cur += len;	394	enc->cur += 2;
251	}	395	}
252		396
253	static void encode_sv (enc_t *enc, SV *sv);	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
		437	encode_array_as_map (enc_t *enc, SV *sv)
		438	{
		439	if (enc->depth >= enc->cbor.max_depth)
		440	croak (ERR_NESTING_EXCEEDED);
		441
		442	++enc->depth;
		443
		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	SvGETMAGIC (sv);	578	SvGETMAGIC (sv);
315
316	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
317	&& ecb_expect_false (SvREFCNT (sv) > 1))
318	{
319	if (!enc->shareable)
320	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
321
322	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
323
324	if (SvOK (*svp))
325	{
326	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
327	encode_uint (enc, 0x00, SvUV (*svp));
328	return;
329	}
330	else
331	{
332	sv_setuv (*svp, enc->shareable_idx);
333	++enc->shareable_idx;
334	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
335	}
336	}
337		579
338	svtype svt = SvTYPE (sv);	580	svtype svt = SvTYPE (sv);
339		581
340	if (ecb_expect_false (SvOBJECT (sv)))	582	if (ecb_expect_false (SvOBJECT (sv)))
341	{	583	{
…		…
348	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	590	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
349	? cbor_tagged_stash	591	? cbor_tagged_stash
350	: gv_stashpv ("CBOR::XS::Tagged" , 1);	592	: gv_stashpv ("CBOR::XS::Tagged" , 1);
351		593
352	HV *stash = SvSTASH (sv);	594	HV *stash = SvSTASH (sv);
353	GV *method;
354		595
355	if (stash == boolean_stash)	596	if (stash == boolean_stash)
356	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	597	{
		598	encode_bool (enc, SvIV (sv));
		599	return;
		600	}
357	else if (stash == error_stash)	601	else if (stash == error_stash)
358	encode_ch (enc, 0xe0 | 23);	602	{
		603	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		604	return;
		605	}
359	else if (stash == tagged_stash)	606	else if (stash == tagged_stash)
360	{	607	{
361	if (svt != SVt_PVAV)	608	if (svt != SVt_PVAV)
362	croak ("encountered CBOR::XS::Tagged object that isn't an array");	609	croak ("encountered CBOR::XS::Tagged object that isn't an array");
363		610
		611	switch (av_len ((AV *)sv))
		612	{
		613	case 2-1:
		614	// actually a tagged value
364	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	615	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
365	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))
366	}	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))));
367	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	662	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
368	{	663	{
369	dSP;	664	dSP;
370		665
371	ENTER; SAVETMPS; PUSHMARK (SP);	666	ENTER; SAVETMPS;
		667	PUSHMARK (SP);
372	// 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
373	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	669	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
374		670
375	PUTBACK;	671	PUTBACK;
376	// G_SCALAR ensures that return value is 1	672	// G_SCALAR ensures that return value is 1
…		…
389	}	685	}
390	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	686	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
391	{	687	{
392	dSP;	688	dSP;
393		689
394	ENTER; SAVETMPS; PUSHMARK (SP);	690	ENTER; SAVETMPS;
		691	PUSHMARK (SP);
395	EXTEND (SP, 2);	692	EXTEND (SP, 2);
396	// 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
397	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	694	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
398	PUSHs (sv_cbor);	695	PUSHs (sv_cbor);
399		696
…		…
404	// catch this surprisingly common error	701	// catch this surprisingly common error
405	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	702	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
406	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));
407		704
408	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	705	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
409	encode_uint (enc, 0x80, count + 1);	706	encode_uint (enc, MAJOR_ARRAY, count + 1);
410	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	707	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
411		708
412	while (count)	709	{
		710	int i;
		711
		712	for (i = 0; i < count; ++i)
413	encode_sv (enc, SP[1 - count--]);	713	encode_sv (enc, SP[i + 1 - count]);
		714
		715	SP -= count;
		716	}
414		717
415	PUTBACK;	718	PUTBACK;
416		719
417	FREETMPS; LEAVE;	720	FREETMPS; LEAVE;
418	}	721	}
…		…
436	{	739	{
437	double nv = SvNVX (sv);	740	double nv = SvNVX (sv);
438		741
439	need (enc, 9);	742	need (enc, 9);
440		743
441	if (ecb_expect_false (nv == (U32)nv))	744	if (ecb_expect_false (nv == (NV)(U32)nv))
442	encode_uint (enc, 0x00, (U32)nv);	745	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
443	//TODO: maybe I32?	746	//TODO: maybe I32?
444	else if (ecb_expect_false (nv == (float)nv))	747	else if (ecb_expect_false (nv == (float)nv))
445	{	748	encode_float32 (enc, nv);
446	uint32_t fp = ecb_float_to_binary32 (nv);
447
448	*enc->cur++ = 0xe0 | 26;
449
450	if (!ecb_big_endian ())
451	fp = ecb_bswap32 (fp);
452
453	memcpy (enc->cur, &fp, 4);
454	enc->cur += 4;
455	}
456	else	749	else
457	{	750	encode_float64 (enc, nv);
458	uint64_t fp = ecb_double_to_binary64 (nv);
459
460	*enc->cur++ = 0xe0 | 27;
461
462	if (!ecb_big_endian ())
463	fp = ecb_bswap64 (fp);
464
465	memcpy (enc->cur, &fp, 8);
466	enc->cur += 8;
467	}
468	}	751	}
469		752
470	static void	753	static void
471	encode_sv (enc_t *enc, SV *sv)	754	encode_sv (enc_t *enc, SV *sv)
472	{	755	{
…		…
474		757
475	if (SvPOKp (sv))	758	if (SvPOKp (sv))
476	{	759	{
477	STRLEN len;	760	STRLEN len;
478	char *str = SvPV (sv, len);	761	char *str = SvPV (sv, len);
479	encode_str (enc, SvUTF8 (sv), str, len);	762	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
480	}	763	}
481	else if (SvNOKp (sv))	764	else if (SvNOKp (sv))
482	encode_nv (enc, sv);	765	encode_nv (enc, sv);
483	else if (SvIOKp (sv))	766	else if (SvIOKp (sv))
484	{	767	encode_int (enc, sv);
485	if (SvIsUV (sv))
486	encode_uint (enc, 0x00, SvUVX (sv));
487	else if (SvIVX (sv) >= 0)
488	encode_uint (enc, 0x00, SvIVX (sv));
489	else
490	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
491	}
492	else if (SvROK (sv))	768	else if (SvROK (sv))
493	encode_rv (enc, SvRV (sv));	769	encode_rv (enc, SvRV (sv));
494	else if (!SvOK (sv))	770	else if (!SvOK (sv))
495	encode_ch (enc, 0xe0 | 22);	771	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
496	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	772	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
497	encode_ch (enc, 0xe0 | 23);	773	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
498	else	774	else
499	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",
500	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	776	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
501	}	777	}
502		778
503	static SV *	779	static SV *
504	encode_cbor (SV *scalar, CBOR *cbor)	780	encode_cbor (SV *scalar, CBOR *cbor)
505	{	781	{
506	enc_t enc = { };	782	enc_t enc = { 0 };
507		783
508	enc.cbor = *cbor;	784	enc.cbor = *cbor;
509	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	785	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
510	enc.cur = SvPVX (enc.sv);	786	enc.cur = SvPVX (enc.sv);
511	enc.end = SvEND (enc.sv);	787	enc.end = SvEND (enc.sv);
512		788
513	SvPOK_only (enc.sv);	789	SvPOK_only (enc.sv);
514		790
515	if (cbor->flags & F_ALLOW_STRINGREF)	791	if (cbor->flags & F_PACK_STRINGS)
516	{	792	{
517	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	793	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
518	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());	794	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
519	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());	795	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
520	}	796	}
…		…
542	CBOR cbor;	818	CBOR cbor;
543	U32 depth; // recursion depth	819	U32 depth; // recursion depth
544	U32 maxdepth; // recursion depth limit	820	U32 maxdepth; // recursion depth limit
545	AV *shareable;	821	AV *shareable;
546	AV *stringref;	822	AV *stringref;
		823	SV *decode_tagged;
		824	SV *err_sv; // optional sv for error, needs to be freed
547	} dec_t;	825	} dec_t;
548		826
549	#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);
550		834
551	#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;
552		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
553	#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)
554	#define DEC_DEC_DEPTH --dec->depth	864	#define DEC_DEC_DEPTH --dec->depth
555		865
556	static UV	866	static UV
557	decode_uint (dec_t *dec)	867	decode_uint (dec_t *dec)
558	{	868	{
559	switch (*dec->cur & 31)	869	U8 m = *dec->cur & MINOR_MASK;
560	{	870	++dec->cur;
561	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
562	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
563	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
564	return *dec->cur++ & 31;
565		871
566	case 24:	872	if (ecb_expect_true (m < LENGTH_EXT1))
		873	return m;
		874	else if (ecb_expect_true (m == LENGTH_EXT1))
		875	{
567	WANT (2);	876	WANT (1);
568	dec->cur += 2;	877	dec->cur += 1;
569	return dec->cur[-1];	878	return dec->cur[-1];
570		879	}
571	case 25:	880	else if (ecb_expect_true (m == LENGTH_EXT2))
		881	{
572	WANT (3);	882	WANT (2);
573	dec->cur += 3;	883	dec->cur += 2;
574	return (((UV)dec->cur[-2]) << 8)	884	return (((UV)dec->cur[-2]) << 8)
575	| ((UV)dec->cur[-1]);	885	| ((UV)dec->cur[-1]);
576		886	}
577	case 26:	887	else if (ecb_expect_true (m == LENGTH_EXT4))
		888	{
578	WANT (5);	889	WANT (4);
579	dec->cur += 5;	890	dec->cur += 4;
580	return (((UV)dec->cur[-4]) << 24)	891	return (((UV)dec->cur[-4]) << 24)
581	| (((UV)dec->cur[-3]) << 16)	892	| (((UV)dec->cur[-3]) << 16)
582	| (((UV)dec->cur[-2]) << 8)	893	| (((UV)dec->cur[-2]) << 8)
583	| ((UV)dec->cur[-1]);	894	| ((UV)dec->cur[-1]);
584		895	}
585	case 27:	896	else if (ecb_expect_true (m == LENGTH_EXT8))
		897	{
586	WANT (9);	898	WANT (8);
587	dec->cur += 9;	899	dec->cur += 8;
		900
		901	return
		902	#if UVSIZE < 8
		903	0
		904	#else
588	return (((UV)dec->cur[-8]) << 56)	905	(((UV)dec->cur[-8]) << 56)
589	| (((UV)dec->cur[-7]) << 48)	906	| (((UV)dec->cur[-7]) << 48)
590	| (((UV)dec->cur[-6]) << 40)	907	| (((UV)dec->cur[-6]) << 40)
591	| (((UV)dec->cur[-5]) << 32)	908	| (((UV)dec->cur[-5]) << 32)
		909	#endif
592	| (((UV)dec->cur[-4]) << 24)	910	| (((UV)dec->cur[-4]) << 24)
593	| (((UV)dec->cur[-3]) << 16)	911	| (((UV)dec->cur[-3]) << 16)
594	| (((UV)dec->cur[-2]) << 8)	912	| (((UV)dec->cur[-2]) << 8)
595	| ((UV)dec->cur[-1]);	913	| ((UV)dec->cur[-1]);
596		914	}
597	default:	915	else
598	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	916	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
599	}
600		917
601	fail:	918	fail:
602	return 0;	919	return 0;
603	}	920	}
604		921
…		…
609	{	926	{
610	AV *av = newAV ();	927	AV *av = newAV ();
611		928
612	DEC_INC_DEPTH;	929	DEC_INC_DEPTH;
613		930
614	if ((*dec->cur & 31) == 31)	931	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
615	{	932	{
616	++dec->cur;	933	++dec->cur;
617		934
618	for (;;)	935	for (;;)
619	{	936	{
620	WANT (1);	937	WANT (1);
621		938
622	if (*dec->cur == (0xe0 | 31))	939	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
623	{	940	{
624	++dec->cur;	941	++dec->cur;
625	break;	942	break;
626	}	943	}
627		944
628	av_push (av, decode_sv (dec));	945	av_push (av, decode_sv (dec));
629	}	946	}
630	}	947	}
631	else	948	else
632	{	949	{
633	int i, len = decode_uint (dec);	950	UV i, len = decode_uint (dec);
634		951
		952	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
635	av_fill (av, len - 1);	953	av_fill (av, len - 1);
636		954
637	for (i = 0; i < len; ++i)	955	for (i = 0; i < len; ++i)
638	AvARRAY (av)[i] = decode_sv (dec);	956	AvARRAY (av)[i] = decode_sv (dec);
639	}	957	}
640		958
641	DEC_DEC_DEPTH;	959	DEC_DEC_DEPTH;
642	return newRV_noinc ((SV *)av);	960	return newRV_noinc ((SV *)av);
643		961
644	fail:	962	fail:
645	SvREFCNT_dec (av);	963	SvREFCNT_dec_NN (av);
646	DEC_DEC_DEPTH;	964	DEC_DEC_DEPTH;
647	return &PL_sv_undef;	965	return &PL_sv_undef;
648	}	966	}
649		967
650	static void	968	static void
…		…
652	{	970	{
653	// for speed reasons, we specialcase single-string	971	// for speed reasons, we specialcase single-string
654	// byte or utf-8 strings as keys, but only when !stringref	972	// byte or utf-8 strings as keys, but only when !stringref
655		973
656	if (ecb_expect_true (!dec->stringref))	974	if (ecb_expect_true (!dec->stringref))
657	if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)	975	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
658	{	976	{
659	I32 len = decode_uint (dec);	977	STRLEN len = decode_uint (dec);
660	char *key = (char *)dec->cur;	978	char *key = (char *)dec->cur;
661		979
		980	WANT (len);
662	dec->cur += len;	981	dec->cur += len;
663		982
664	if (ecb_expect_false (dec->stringref))
665	av_push (dec->stringref, newSVpvn (key, len));
666
667	hv_store (hv, key, len, decode_sv (dec), 0);	983	hv_store (hv, key, len, decode_sv (dec), 0);
668		984
669	return;	985	return;
670	}	986	}
671	else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)	987	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
672	{	988	{
673	I32 len = decode_uint (dec);	989	STRLEN len = decode_uint (dec);
674	char *key = (char *)dec->cur;	990	char *key = (char *)dec->cur;
675		991
		992	WANT (len);
676	dec->cur += len;	993	dec->cur += len;
677		994
678	if (ecb_expect_false (dec->stringref))	995	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
679	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	996	if (!cbor_is_utf8_string ((U8 *)key, len))
		997	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
680		998
681	hv_store (hv, key, -len, decode_sv (dec), 0);	999	hv_store (hv, key, -len, decode_sv (dec), 0);
682		1000
683	return;	1001	return;
684	}	1002	}
685		1003
686	SV *k = decode_sv (dec);	1004	SV *k = decode_sv (dec);
687	SV *v = decode_sv (dec);	1005	SV *v = decode_sv (dec);
688		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
689	hv_store_ent (hv, k, v, 0);	1035	hv_store_ent (hv, k, v, 0);
690	SvREFCNT_dec (k);	1036	SvREFCNT_dec_NN (k);
		1037
		1038	fail:
		1039	;
691	}	1040	}
692		1041
693	static SV *	1042	static SV *
694	decode_hv (dec_t *dec)	1043	decode_hv (dec_t *dec)
695	{	1044	{
696	HV *hv = newHV ();	1045	HV *hv = newHV ();
697		1046
698	DEC_INC_DEPTH;	1047	DEC_INC_DEPTH;
699		1048
700	if ((*dec->cur & 31) == 31)	1049	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
701	{	1050	{
702	++dec->cur;	1051	++dec->cur;
703		1052
704	for (;;)	1053	for (;;)
705	{	1054	{
706	WANT (1);	1055	WANT (1);
707		1056
708	if (*dec->cur == (0xe0 | 31))	1057	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
709	{	1058	{
710	++dec->cur;	1059	++dec->cur;
711	break;	1060	break;
712	}	1061	}
713		1062
714	decode_he (dec, hv);	1063	decode_he (dec, hv);
715	}	1064	}
716	}	1065	}
717	else	1066	else
718	{	1067	{
719	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
720		1071
721	while (pairs--)	1072	while (pairs--)
722	decode_he (dec, hv);	1073	decode_he (dec, hv);
723	}	1074	}
724		1075
725	DEC_DEC_DEPTH;	1076	DEC_DEC_DEPTH;
726	return newRV_noinc ((SV *)hv);	1077	return newRV_noinc ((SV *)hv);
727		1078
728	fail:	1079	fail:
729	SvREFCNT_dec (hv);	1080	SvREFCNT_dec_NN (hv);
730	DEC_DEC_DEPTH;	1081	DEC_DEC_DEPTH;
731	return &PL_sv_undef;	1082	return &PL_sv_undef;
732	}	1083	}
733		1084
734	static SV *	1085	static SV *
735	decode_str (dec_t *dec, int utf8)	1086	decode_str (dec_t *dec, int utf8)
736	{	1087	{
737	SV *sv = 0;	1088	SV *sv = 0;
738		1089
739	if ((*dec->cur & 31) == 31)	1090	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
740	{	1091	{
		1092	// indefinite length strings
741	++dec->cur;	1093	++dec->cur;
742		1094
		1095	U8 major = *dec->cur & MAJOR_MISC;
		1096
743	sv = newSVpvn ("", 0);	1097	sv = newSVpvn ("", 0);
744		1098
745	// not very fast, and certainly not robust against illegal input
746	for (;;)	1099	for (;;)
747	{	1100	{
748	WANT (1);	1101	WANT (1);
749		1102
750	if (*dec->cur == (0xe0 | 31))	1103	if ((*dec->cur - major) > LENGTH_EXT8)
		1104	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
751	{	1105	{
752	++dec->cur;	1106	++dec->cur;
753	break;	1107	break;
754	}	1108	}
		1109	else
		1110	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
755		1111
756	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;
757	}	1117	}
758	}	1118	}
759	else	1119	else
760	{	1120	{
761	STRLEN len = decode_uint (dec);	1121	STRLEN len = decode_uint (dec);
…		…
768	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	1128	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
769	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	1129	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770	}	1130	}
771		1131
772	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
773	SvUTF8_on (sv);	1138	SvUTF8_on (sv);
		1139	}
774		1140
775	return sv;	1141	return sv;
776		1142
777	fail:	1143	fail:
778	SvREFCNT_dec (sv);	1144	SvREFCNT_dec (sv);
…		…
797	sv = newRV_noinc (decode_sv (dec));	1163	sv = newRV_noinc (decode_sv (dec));
798	break;	1164	break;
799		1165
800	case CBOR_TAG_STRINGREF_NAMESPACE:	1166	case CBOR_TAG_STRINGREF_NAMESPACE:
801	{	1167	{
		1168	// do not use SAVETMPS/FREETMPS, as these will
		1169	// erase mortalised caches, e.g. "shareable"
802	ENTER; SAVETMPS;	1170	ENTER;
803		1171
804	SAVESPTR (dec->stringref);	1172	SAVESPTR (dec->stringref);
805	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());	1173	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
806		1174
807	sv = decode_sv (dec);	1175	sv = decode_sv (dec);
808		1176
809	FREETMPS; LEAVE;	1177	LEAVE;
810	}	1178	}
811	break;	1179	break;
812		1180
813	case CBOR_TAG_STRINGREF:	1181	case CBOR_TAG_STRINGREF:
814	{	1182	{
815	if ((*dec->cur >> 5) != 0)	1183	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
816	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1184	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
817		1185
818	UV idx = decode_uint (dec);	1186	UV idx = decode_uint (dec);
819		1187
820	if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))	1188	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
821	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");	1189	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
822		1190
823	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1191	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
824	}	1192	}
825	break;	1193	break;
…		…
827	case CBOR_TAG_VALUE_SHAREABLE:	1195	case CBOR_TAG_VALUE_SHAREABLE:
828	{	1196	{
829	if (ecb_expect_false (!dec->shareable))	1197	if (ecb_expect_false (!dec->shareable))
830	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());	1198	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
831		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;
832	sv = newSV (0);	1206	AV *av = newAV ();
833	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1207	av_push (dec->shareable, (SV *)av);
		1208	idx = AvFILLp (dec->shareable);
834		1209
835	SV *osv = decode_sv (dec);	1210	sv = decode_sv (dec);
836	sv_setsv (sv, osv);	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
837	SvREFCNT_dec_NN (osv);	1223	SvREFCNT_dec_NN ((SV *)av);
		1224	AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
		1225	}
		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	}
838	}	1233	}
839	break;	1234	break;
840		1235
841	case CBOR_TAG_VALUE_SHAREDREF:	1236	case CBOR_TAG_VALUE_SHAREDREF:
842	{	1237	{
843	if ((*dec->cur >> 5) != 0)	1238	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
844	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1239	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
845		1240
846	UV idx = decode_uint (dec);	1241	UV idx = decode_uint (dec);
847		1242
848	if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))	1243	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
849	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1244	ERR ("corrupted CBOR data (sharedref index out of bounds)");
850		1245
851	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1246	sv = AvARRAY (dec->shareable)[idx];
		1247
		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));
		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);
852	}	1260	}
853	break;	1261	break;
854		1262
855	case CBOR_TAG_PERL_OBJECT:	1263	case CBOR_TAG_PERL_OBJECT:
856	{	1264	{
		1265	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1266	goto filter;
		1267
857	sv = decode_sv (dec);	1268	sv = decode_sv (dec);
858		1269
859	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1270	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
860	ERR ("corrupted CBOR data (non-array perl object)");	1271	ERR ("corrupted CBOR data (non-array perl object)");
861		1272
…		…
871	if (!method)	1282	if (!method)
872	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)");
873		1284
874	dSP;	1285	dSP;
875		1286
876	ENTER; SAVETMPS; PUSHMARK (SP);	1287	ENTER; SAVETMPS;
		1288	PUSHMARK (SP);
877	EXTEND (SP, len + 1);	1289	EXTEND (SP, len + 1);
878	// 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
879	PUSHs (*av_fetch (av, 0, 1));	1291	PUSHs (*av_fetch (av, 0, 1));
880	PUSHs (sv_cbor);	1292	PUSHs (sv_cbor);
881		1293
…		…
889	SPAGAIN;	1301	SPAGAIN;
890		1302
891	if (SvTRUE (ERRSV))	1303	if (SvTRUE (ERRSV))
892	{	1304	{
893	FREETMPS; LEAVE;	1305	FREETMPS; LEAVE;
894	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1306	ERR_ERRSV;
895	}	1307	}
896		1308
897	SvREFCNT_dec (sv);	1309	SvREFCNT_dec_NN (sv);
898	sv = SvREFCNT_inc (POPs);	1310	sv = SvREFCNT_inc (POPs);
899		1311
900	PUTBACK;	1312	PUTBACK;
901		1313
902	FREETMPS; LEAVE;	1314	FREETMPS; LEAVE;
903	}	1315	}
904	break;	1316	break;
905		1317
906	default:	1318	default:
		1319	filter:
907	{	1320	{
		1321	SV *tag_sv = newSVuv (tag);
		1322
908	sv = decode_sv (dec);	1323	sv = decode_sv (dec);
909		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))
		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	{
910	AV *av = newAV ();	1352	AV *av = newAV ();
911	av_push (av, newSVuv (tag));	1353	av_push (av, tag_sv);
912	av_push (av, sv);	1354	av_push (av, sv);
913		1355
914	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1356	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
915	? cbor_tagged_stash	1357	? cbor_tagged_stash
916	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1358	: gv_stashpv ("CBOR::XS::Tagged" , 1);
917
918	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);	1359	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
		1360	}
		1361
		1362	PUTBACK;
		1363
		1364	FREETMPS; LEAVE;
919	}	1365	}
920	break;	1366	break;
921	}	1367	}
922		1368
923	return sv;	1369	return sv;
…		…
930	static SV *	1376	static SV *
931	decode_sv (dec_t *dec)	1377	decode_sv (dec_t *dec)
932	{	1378	{
933	WANT (1);	1379	WANT (1);
934		1380
935	switch (*dec->cur >> 5)	1381	switch (*dec->cur >> MAJOR_SHIFT)
936	{	1382	{
937	case 0: // unsigned int	1383	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
938	return newSVuv (decode_uint (dec));	1384	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
939	case 1: // negative int	1385	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
940	return newSViv (-1 - (IV)decode_uint (dec));	1386	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
941	case 2: // octet string	1387	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
942	return decode_str (dec, 0);	1388	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
943	case 3: // utf-8 string	1389	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
944	return decode_str (dec, 1);	1390
945	case 4: // array	1391	case MAJOR_MISC >> MAJOR_SHIFT:
946	return decode_av (dec);
947	case 5: // map
948	return decode_hv (dec);
949	case 6: // tag
950	return decode_tagged (dec);
951	case 7: // misc
952	switch (*dec->cur++ & 31)	1392	switch (*dec->cur++ & MINOR_MASK)
953	{	1393	{
954	case 20:	1394	case SIMPLE_FALSE:
955	#if CBOR_SLOW	1395	#if CBOR_SLOW
956	types_false = get_bool ("Types::Serialiser::false");	1396	types_false = get_bool ("Types::Serialiser::false");
957	#endif	1397	#endif
958	return newSVsv (types_false);	1398	return newSVsv (types_false);
959	case 21:	1399	case SIMPLE_TRUE:
960	#if CBOR_SLOW	1400	#if CBOR_SLOW
961	types_true = get_bool ("Types::Serialiser::true");	1401	types_true = get_bool ("Types::Serialiser::true");
962	#endif	1402	#endif
963	return newSVsv (types_true);	1403	return newSVsv (types_true);
964	case 22:	1404	case SIMPLE_NULL:
965	return newSVsv (&PL_sv_undef);	1405	return newSVsv (&PL_sv_undef);
966	case 23:	1406	case SIMPLE_UNDEF:
967	#if CBOR_SLOW	1407	#if CBOR_SLOW
968	types_error = get_bool ("Types::Serialiser::error");	1408	types_error = get_bool ("Types::Serialiser::error");
969	#endif	1409	#endif
970	return newSVsv (types_error);	1410	return newSVsv (types_error);
971		1411
972	case 25:	1412	case MISC_FLOAT16:
973	{	1413	{
974	WANT (2);	1414	WANT (2);
975		1415
976	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1416	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
977	dec->cur += 2;	1417	dec->cur += 2;
978		1418
979	return newSVnv (ecb_binary16_to_float (fp));	1419	return newSVnv (ecb_binary16_to_float (fp));
980	}	1420	}
981		1421
982	case 26:	1422	case MISC_FLOAT32:
983	{	1423	{
984	uint32_t fp;	1424	uint32_t fp;
985	WANT (4);	1425	WANT (4);
986	memcpy (&fp, dec->cur, 4);	1426	memcpy (&fp, dec->cur, 4);
987	dec->cur += 4;	1427	dec->cur += 4;
…		…
990	fp = ecb_bswap32 (fp);	1430	fp = ecb_bswap32 (fp);
991		1431
992	return newSVnv (ecb_binary32_to_float (fp));	1432	return newSVnv (ecb_binary32_to_float (fp));
993	}	1433	}
994		1434
995	case 27:	1435	case MISC_FLOAT64:
996	{	1436	{
997	uint64_t fp;	1437	uint64_t fp;
998	WANT (8);	1438	WANT (8);
999	memcpy (&fp, dec->cur, 8);	1439	memcpy (&fp, dec->cur, 8);
1000	dec->cur += 8;	1440	dec->cur += 8;
…		…
1003	fp = ecb_bswap64 (fp);	1443	fp = ecb_bswap64 (fp);
1004		1444
1005	return newSVnv (ecb_binary64_to_double (fp));	1445	return newSVnv (ecb_binary64_to_double (fp));
1006	}	1446	}
1007		1447
1008	// 0..19 unassigned	1448	// 0..19 unassigned simple
1009	// 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
1010	default:	1452	default:
1011	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1453	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1012	}	1454	}
1013		1455
1014	break;	1456	break;
1015	}	1457	}
1016		1458
…		…
1019	}	1461	}
1020		1462
1021	static SV *	1463	static SV *
1022	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1464	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1023	{	1465	{
1024	dec_t dec = { };	1466	dec_t dec = { 0 };
1025	SV *sv;	1467	SV *sv;
1026	STRLEN len;	1468	STRLEN len;
1027	char *data = SvPVbyte (string, len);	1469	char *data = SvPVbyte (string, len);
1028		1470
1029	if (len > cbor->max_size && cbor->max_size)	1471	if (len > cbor->max_size && cbor->max_size)
…		…
1043	if (dec.cur != dec.end && !dec.err)	1485	if (dec.cur != dec.end && !dec.err)
1044	dec.err = "garbage after CBOR object";	1486	dec.err = "garbage after CBOR object";
1045		1487
1046	if (dec.err)	1488	if (dec.err)
1047	{	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
1048	SvREFCNT_dec (sv);	1501	SvREFCNT_dec_NN (sv);
		1502
		1503	if (dec.err_sv)
		1504	sv_2mortal (dec.err_sv);
		1505
1049	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);
1050	}	1507	}
1051		1508
1052	sv = sv_2mortal (sv);	1509	sv = sv_2mortal (sv);
1053		1510
1054	return sv;	1511	return sv;
1055	}	1512	}
1056		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
1057	/////////////////////////////////////////////////////////////////////////////	1639	/////////////////////////////////////////////////////////////////////////////
1058	// XS interface functions	1640	// XS interface functions
1059		1641
1060	MODULE = CBOR::XS PACKAGE = CBOR::XS	1642	MODULE = CBOR::XS PACKAGE = CBOR::XS
1061		1643
…		…
1069		1651
1070	types_true = get_bool ("Types::Serialiser::true" );	1652	types_true = get_bool ("Types::Serialiser::true" );
1071	types_false = get_bool ("Types::Serialiser::false");	1653	types_false = get_bool ("Types::Serialiser::false");
1072	types_error = get_bool ("Types::Serialiser::error");	1654	types_error = get_bool ("Types::Serialiser::error");
1073		1655
		1656	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
		1657
1074	sv_cbor = newSVpv ("CBOR", 0);	1658	sv_cbor = newSVpv ("CBOR", 0);
1075	SvREADONLY_on (sv_cbor);	1659	SvREADONLY_on (sv_cbor);
		1660
		1661	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1076	}	1662	}
1077		1663
1078	PROTOTYPES: DISABLE	1664	PROTOTYPES: DISABLE
1079		1665
1080	void CLONE (...)	1666	void CLONE (...)
…		…
1096	)));	1682	)));
1097	}	1683	}
1098		1684
1099	void shrink (CBOR *self, int enable = 1)	1685	void shrink (CBOR *self, int enable = 1)
1100	ALIAS:	1686	ALIAS:
1101	shrink = F_SHRINK	1687	shrink = F_SHRINK
1102	allow_unknown = F_ALLOW_UNKNOWN	1688	allow_unknown = F_ALLOW_UNKNOWN
1103	allow_sharing = F_ALLOW_SHARING	1689	allow_sharing = F_ALLOW_SHARING
1104	allow_stringref = F_ALLOW_STRINGREF	1690	allow_cycles = F_ALLOW_CYCLES
		1691	allow_weak_cycles = F_ALLOW_WEAK_CYCLES
		1692	forbid_objects = F_FORBID_OBJECTS
		1693	pack_strings = F_PACK_STRINGS
		1694	text_keys = F_TEXT_KEYS
		1695	text_strings = F_TEXT_STRINGS
		1696	validate_utf8 = F_VALIDATE_UTF8
1105	PPCODE:	1697	PPCODE:
1106	{	1698	{
1107	if (enable)	1699	if (enable)
1108	self->flags |= ix;	1700	self->flags |= ix;
1109	else	1701	else
…		…
1112	XPUSHs (ST (0));	1704	XPUSHs (ST (0));
1113	}	1705	}
1114		1706
1115	void get_shrink (CBOR *self)	1707	void get_shrink (CBOR *self)
1116	ALIAS:	1708	ALIAS:
1117	get_shrink = F_SHRINK	1709	get_shrink = F_SHRINK
1118	get_allow_unknown = F_ALLOW_UNKNOWN	1710	get_allow_unknown = F_ALLOW_UNKNOWN
1119	get_allow_sharing = F_ALLOW_SHARING	1711	get_allow_sharing = F_ALLOW_SHARING
1120	get_allow_stringref = F_ALLOW_STRINGREF	1712	get_allow_cycles = F_ALLOW_CYCLES
		1713	get_allow_weak_cycles = F_ALLOW_WEAK_CYCLES
		1714	get_forbid_objects = F_FORBID_OBJECTS
		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
1121	PPCODE:	1719	PPCODE:
1122	XPUSHs (boolSV (self->flags & ix));	1720	XPUSHs (boolSV (self->flags & ix));
1123		1721
1124	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1722	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1125	PPCODE:	1723	PPCODE:
…		…
1138	XPUSHs (ST (0));	1736	XPUSHs (ST (0));
1139		1737
1140	int get_max_size (CBOR *self)	1738	int get_max_size (CBOR *self)
1141	CODE:	1739	CODE:
1142	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);
1143	OUTPUT:	1753	OUTPUT:
1144	RETVAL	1754	RETVAL
1145		1755
1146	void encode (CBOR *self, SV *scalar)	1756	void encode (CBOR *self, SV *scalar)
1147	PPCODE:	1757	PPCODE:
…		…
1162	EXTEND (SP, 2);	1772	EXTEND (SP, 2);
1163	PUSHs (sv);	1773	PUSHs (sv);
1164	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1774	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1165	}	1775	}
1166		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
1167	PROTOTYPES: ENABLE	1833	PROTOTYPES: ENABLE
1168		1834
1169	void encode_cbor (SV *scalar)	1835	void encode_cbor (SV *scalar)
		1836	ALIAS:
		1837	encode_cbor = 0
		1838	encode_cbor_sharing = F_ALLOW_SHARING
1170	PPCODE:	1839	PPCODE:
1171	{	1840	{
1172	CBOR cbor;	1841	CBOR cbor;
1173	cbor_init (&cbor);	1842	cbor_init (&cbor);
		1843	cbor.flags |= ix;
1174	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1844	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1175	XPUSHs (scalar);	1845	XPUSHs (scalar);
1176	}	1846	}
1177		1847
1178	void decode_cbor (SV *cborstr)	1848	void decode_cbor (SV *cborstr)
…		…
1182	cbor_init (&cbor);	1852	cbor_init (&cbor);
1183	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1853	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1184	XPUSHs (cborstr);	1854	XPUSHs (cborstr);
1185	}	1855	}
1186		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