ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.20 by root, Wed Nov 20 11:06:42 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
		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
…		…
122	SvPV_shrink_to_cur (sv);	202	SvPV_shrink_to_cur (sv);
123	#elif defined (SvPV_renew)	203	#elif defined (SvPV_renew)
124	SvPV_renew (sv, SvCUR (sv) + 1);	204	SvPV_renew (sv, SvCUR (sv) + 1);
125	#endif	205	#endif
126	}	206	}
		207	}
		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;
127	}	218	}
128		219
129	/////////////////////////////////////////////////////////////////////////////	220	/////////////////////////////////////////////////////////////////////////////
130	// encoder	221	// encoder
131		222
…		…
144	} enc_t;	235	} enc_t;
145		236
146	ecb_inline void	237	ecb_inline void
147	need (enc_t *enc, STRLEN len)	238	need (enc_t *enc, STRLEN len)
148	{	239	{
149	if (ecb_expect_false (enc->cur + len >= enc->end))	240	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
150	{	241	{
151	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	242	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
152	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	243	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
153	enc->cur = SvPVX (enc->sv) + cur;	244	enc->cur = SvPVX (enc->sv) + cur;
154	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	245	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
155	}	246	}
156	}	247	}
157		248
		249	static void encode_sv (enc_t *enc, SV *sv);
		250
158	ecb_inline void	251	ecb_inline void
159	encode_ch (enc_t *enc, char ch)	252	encode_ch (enc_t *enc, char ch)
160	{	253	{
161	need (enc, 1);	254	need (enc, 1);
162	*enc->cur++ = ch;	255	*enc->cur++ = ch;
163	}	256	}
164		257
		258	// used for tags, intregers, element counts and so on
165	static void	259	static void
166	encode_uint (enc_t *enc, int major, UV len)	260	encode_uint (enc_t *enc, int major, UV len)
167	{	261	{
168	need (enc, 9);	262	need (enc, 9);
169		263
170	if (len < 24)	264	if (ecb_expect_true (len < LENGTH_EXT1))
171	*enc->cur++ = major | len;	265	*enc->cur++ = major | len;
172	else if (len <= 0xff)	266	else if (ecb_expect_true (len <= 0xffU))
173	{	267	{
174	*enc->cur++ = major | 24;	268	*enc->cur++ = major | LENGTH_EXT1;
175	*enc->cur++ = len;	269	*enc->cur++ = len;
176	}	270	}
177	else if (len <= 0xffff)	271	else if (len <= 0xffffU)
178	{	272	{
179	*enc->cur++ = major | 25;	273	*enc->cur++ = major | LENGTH_EXT2;
180	*enc->cur++ = len >> 8;	274	*enc->cur++ = len >> 8;
181	*enc->cur++ = len;	275	*enc->cur++ = len;
182	}	276	}
183	else if (len <= 0xffffffff)	277	else if (len <= 0xffffffffU)
184	{	278	{
185	*enc->cur++ = major | 26;	279	*enc->cur++ = major | LENGTH_EXT4;
186	*enc->cur++ = len >> 24;	280	*enc->cur++ = len >> 24;
187	*enc->cur++ = len >> 16;	281	*enc->cur++ = len >> 16;
188	*enc->cur++ = len >> 8;	282	*enc->cur++ = len >> 8;
189	*enc->cur++ = len;	283	*enc->cur++ = len;
190	}	284	}
191	else	285	else
192	{	286	{
193	*enc->cur++ = major | 27;	287	*enc->cur++ = major | LENGTH_EXT8;
194	*enc->cur++ = len >> 56;	288	*enc->cur++ = len >> 56;
195	*enc->cur++ = len >> 48;	289	*enc->cur++ = len >> 48;
196	*enc->cur++ = len >> 40;	290	*enc->cur++ = len >> 40;
197	*enc->cur++ = len >> 32;	291	*enc->cur++ = len >> 32;
198	*enc->cur++ = len >> 24;	292	*enc->cur++ = len >> 24;
…		…
200	*enc->cur++ = len >> 8;	294	*enc->cur++ = len >> 8;
201	*enc->cur++ = len;	295	*enc->cur++ = len;
202	}	296	}
203	}	297	}
204		298
205	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
206	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	319	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
207	{	320	{
208	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);
209	need (enc, len);	351	need (enc, len);
210	memcpy (enc->cur, str, len);	352	memcpy (enc->cur, str, len);
211	enc->cur += len;	353	enc->cur += len;
212	}	354	}
213		355
214	ecb_inline void	356	ecb_inline void
		357	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		358	{
		359	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
		360	{
		361	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
		362
		363	if (SvOK (*svp))
		364	{
		365	// already registered, use stringref
		366	encode_tag (enc, CBOR_TAG_STRINGREF);
		367	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		368	return;
		369	}
		370	else if (len >= minimum_string_length (enc->stringref_idx))
		371	{
		372	// register only
		373	sv_setuv (*svp, enc->stringref_idx);
		374	++enc->stringref_idx;
		375	}
		376	}
		377
		378	encode_str (enc, upgrade_utf8, utf8, str, len);
		379	}
		380
		381	ecb_inline void
215	encode_tag (enc_t *enc, UV tag)	382	encode_float16 (enc_t *enc, NV nv)
216	{	383	{
217	encode_uint (enc, 0xc0, tag);	384	need (enc, 1+2);
218	}
219		385
220	static void encode_sv (enc_t *enc, SV *sv);	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
		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	}
221		511
222	static void	512	static void
223	encode_av (enc_t *enc, AV *av)	513	encode_av (enc_t *enc, AV *av)
224	{	514	{
225	int i, len = av_len (av);	515	int i, len = av_len (av);
…		…
227	if (enc->depth >= enc->cbor.max_depth)	517	if (enc->depth >= enc->cbor.max_depth)
228	croak (ERR_NESTING_EXCEEDED);	518	croak (ERR_NESTING_EXCEEDED);
229		519
230	++enc->depth;	520	++enc->depth;
231		521
232	encode_uint (enc, 0x80, len + 1);	522	encode_uint (enc, MAJOR_ARRAY, len + 1);
233		523
		524	if (ecb_expect_false (SvMAGICAL (av)))
234	for (i = 0; i <= len; ++i)	525	for (i = 0; i <= len; ++i)
235	{	526	{
236	SV **svp = av_fetch (av, i, 0);	527	SV **svp = av_fetch (av, i, 0);
237	encode_sv (enc, svp ? *svp : &PL_sv_undef);	528	encode_sv (enc, svp ? *svp : &PL_sv_undef);
238	}	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	}
239		536
240	--enc->depth;	537	--enc->depth;
241	}
242
243	ecb_inline void
244	encode_he (enc_t *enc, HE *he)
245	{
246	if (HeKLEN (he) == HEf_SVKEY)
247	encode_sv (enc, HeSVKEY (he));
248	else
249	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
250	}	538	}
251		539
252	static void	540	static void
253	encode_hv (enc_t *enc, HV *hv)	541	encode_hv (enc_t *enc, HV *hv)
254	{	542	{
…		…
260	++enc->depth;	548	++enc->depth;
261		549
262	int pairs = hv_iterinit (hv);	550	int pairs = hv_iterinit (hv);
263	int mg = SvMAGICAL (hv);	551	int mg = SvMAGICAL (hv);
264		552
265	if (mg)	553	if (ecb_expect_false (mg))
266	encode_ch (enc, 0xa0 | 31);	554	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
267	else	555	else
268	encode_uint (enc, 0xa0, pairs);	556	encode_uint (enc, MAJOR_MAP, pairs);
269		557
270	while ((he = hv_iternext (hv)))	558	while ((he = hv_iternext (hv)))
271	{	559	{
272	if (ecb_expect_false (enc->cbor.flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)))
273	{
274	SV **svp;
275
276	if (HeKLEN (he) == HEf_SVKEY)	560	if (HeKLEN (he) == HEf_SVKEY)
277	svp = hv_fetch_ent (enc->stringref[!! SvUTF8 (HeSVKEY (he))], HeSVKEY (he) , 1, 0);//TODO return HE :/	561	encode_sv (enc, HeSVKEY (he));
278	else
279	svp = hv_fetch (enc->stringref[!! HeKUTF8 (he) ], HeKEY (he), HeKLEN (he), 1);
280
281	if (SvOK (*svp))
282	{
283	encode_tag (enc, CBOR_TAG_STRINGREF);
284	encode_uint (enc, 0x00, SvUV (*svp));
285	}
286	else
287	{
288	sv_setuv (*svp, enc->stringref_idx);
289	++enc->stringref_idx;
290	encode_he (enc, he);
291	}
292	}
293	else	562	else
294	encode_he (enc, he);	563	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
295		564
296	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));
297	}	566	}
298		567
299	if (mg)	568	if (ecb_expect_false (mg))
300	encode_ch (enc, 0xe0 | 31);	569	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
301		570
302	--enc->depth;	571	--enc->depth;
303	}	572	}
304		573
305	// encode objects, arrays and special \0=false and \1=true values.	574	// encode objects, arrays and special \0=false and \1=true values.
306	static void	575	static void
307	encode_rv (enc_t *enc, SV *sv)	576	encode_rv (enc_t *enc, SV *sv)
308	{	577	{
309	SvGETMAGIC (sv);	578	SvGETMAGIC (sv);
310
311	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
312	&& ecb_expect_false (SvREFCNT (sv) > 1))
313	{
314	if (!enc->shareable)
315	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
316
317	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
318
319	if (SvOK (*svp))
320	{
321	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
322	encode_uint (enc, 0x00, SvUV (*svp));
323	return;
324	}
325	else
326	{
327	sv_setuv (*svp, enc->shareable_idx);
328	++enc->shareable_idx;
329	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
330	}
331	}
332		579
333	svtype svt = SvTYPE (sv);	580	svtype svt = SvTYPE (sv);
334		581
335	if (ecb_expect_false (SvOBJECT (sv)))	582	if (ecb_expect_false (SvOBJECT (sv)))
336	{	583	{
…		…
343	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	590	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
344	? cbor_tagged_stash	591	? cbor_tagged_stash
345	: gv_stashpv ("CBOR::XS::Tagged" , 1);	592	: gv_stashpv ("CBOR::XS::Tagged" , 1);
346		593
347	HV *stash = SvSTASH (sv);	594	HV *stash = SvSTASH (sv);
348	GV *method;
349		595
350	if (stash == boolean_stash)	596	if (stash == boolean_stash)
351	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	597	{
		598	encode_bool (enc, SvIV (sv));
		599	return;
		600	}
352	else if (stash == error_stash)	601	else if (stash == error_stash)
353	encode_ch (enc, 0xe0 | 23);	602	{
		603	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		604	return;
		605	}
354	else if (stash == tagged_stash)	606	else if (stash == tagged_stash)
355	{	607	{
356	if (svt != SVt_PVAV)	608	if (svt != SVt_PVAV)
357	croak ("encountered CBOR::XS::Tagged object that isn't an array");	609	croak ("encountered CBOR::XS::Tagged object that isn't an array");
358		610
		611	switch (av_len ((AV *)sv))
		612	{
		613	case 2-1:
		614	// actually a tagged value
359	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	615	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
360	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))
361	}	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))));
362	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	662	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
363	{	663	{
364	dSP;	664	dSP;
365		665
366	ENTER; SAVETMPS; PUSHMARK (SP);	666	ENTER; SAVETMPS;
		667	PUSHMARK (SP);
367	// 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
368	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	669	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
369		670
370	PUTBACK;	671	PUTBACK;
371	// G_SCALAR ensures that return value is 1	672	// G_SCALAR ensures that return value is 1
…		…
384	}	685	}
385	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	686	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
386	{	687	{
387	dSP;	688	dSP;
388		689
389	ENTER; SAVETMPS; PUSHMARK (SP);	690	ENTER; SAVETMPS;
		691	PUSHMARK (SP);
390	EXTEND (SP, 2);	692	EXTEND (SP, 2);
391	// 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
392	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	694	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
393	PUSHs (sv_cbor);	695	PUSHs (sv_cbor);
394		696
…		…
399	// catch this surprisingly common error	701	// catch this surprisingly common error
400	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	702	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
401	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));
402		704
403	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	705	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
404	encode_uint (enc, 0x80, count + 1);	706	encode_uint (enc, MAJOR_ARRAY, count + 1);
405	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	707	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
406		708
407	while (count)	709	{
		710	int i;
		711
		712	for (i = 0; i < count; ++i)
408	encode_sv (enc, SP[1 - count--]);	713	encode_sv (enc, SP[i + 1 - count]);
		714
		715	SP -= count;
		716	}
409		717
410	PUTBACK;	718	PUTBACK;
411		719
412	FREETMPS; LEAVE;	720	FREETMPS; LEAVE;
413	}	721	}
…		…
431	{	739	{
432	double nv = SvNVX (sv);	740	double nv = SvNVX (sv);
433		741
434	need (enc, 9);	742	need (enc, 9);
435		743
436	if (ecb_expect_false (nv == (U32)nv))	744	if (ecb_expect_false (nv == (NV)(U32)nv))
437	encode_uint (enc, 0x00, (U32)nv);	745	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
438	//TODO: maybe I32?	746	//TODO: maybe I32?
439	else if (ecb_expect_false (nv == (float)nv))	747	else if (ecb_expect_false (nv == (float)nv))
440	{	748	encode_float32 (enc, nv);
441	uint32_t fp = ecb_float_to_binary32 (nv);
442
443	*enc->cur++ = 0xe0 | 26;
444
445	if (!ecb_big_endian ())
446	fp = ecb_bswap32 (fp);
447
448	memcpy (enc->cur, &fp, 4);
449	enc->cur += 4;
450	}
451	else	749	else
452	{	750	encode_float64 (enc, nv);
453	uint64_t fp = ecb_double_to_binary64 (nv);
454
455	*enc->cur++ = 0xe0 | 27;
456
457	if (!ecb_big_endian ())
458	fp = ecb_bswap64 (fp);
459
460	memcpy (enc->cur, &fp, 8);
461	enc->cur += 8;
462	}
463	}	751	}
464		752
465	static void	753	static void
466	encode_sv (enc_t *enc, SV *sv)	754	encode_sv (enc_t *enc, SV *sv)
467	{	755	{
…		…
469		757
470	if (SvPOKp (sv))	758	if (SvPOKp (sv))
471	{	759	{
472	STRLEN len;	760	STRLEN len;
473	char *str = SvPV (sv, len);	761	char *str = SvPV (sv, len);
474	encode_str (enc, SvUTF8 (sv), str, len);	762	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
475	}	763	}
476	else if (SvNOKp (sv))	764	else if (SvNOKp (sv))
477	encode_nv (enc, sv);	765	encode_nv (enc, sv);
478	else if (SvIOKp (sv))	766	else if (SvIOKp (sv))
479	{	767	encode_int (enc, sv);
480	if (SvIsUV (sv))
481	encode_uint (enc, 0x00, SvUVX (sv));
482	else if (SvIVX (sv) >= 0)
483	encode_uint (enc, 0x00, SvIVX (sv));
484	else
485	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
486	}
487	else if (SvROK (sv))	768	else if (SvROK (sv))
488	encode_rv (enc, SvRV (sv));	769	encode_rv (enc, SvRV (sv));
489	else if (!SvOK (sv))	770	else if (!SvOK (sv))
490	encode_ch (enc, 0xe0 | 22);	771	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
491	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	772	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
492	encode_ch (enc, 0xe0 | 23);	773	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
493	else	774	else
494	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",
495	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	776	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
496	}	777	}
497		778
498	static SV *	779	static SV *
499	encode_cbor (SV *scalar, CBOR *cbor)	780	encode_cbor (SV *scalar, CBOR *cbor)
500	{	781	{
501	enc_t enc = { };	782	enc_t enc = { 0 };
502		783
503	enc.cbor = *cbor;	784	enc.cbor = *cbor;
504	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	785	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
505	enc.cur = SvPVX (enc.sv);	786	enc.cur = SvPVX (enc.sv);
506	enc.end = SvEND (enc.sv);	787	enc.end = SvEND (enc.sv);
507		788
508	SvPOK_only (enc.sv);	789	SvPOK_only (enc.sv);
509		790
510	if (cbor->flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS))	791	if (cbor->flags & F_PACK_STRINGS)
511	{	792	{
512	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	793	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
513	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());	794	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
514	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());	795	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
515	}	796	}
…		…
537	CBOR cbor;	818	CBOR cbor;
538	U32 depth; // recursion depth	819	U32 depth; // recursion depth
539	U32 maxdepth; // recursion depth limit	820	U32 maxdepth; // recursion depth limit
540	AV *shareable;	821	AV *shareable;
541	AV *stringref;	822	AV *stringref;
		823	SV *decode_tagged;
		824	SV *err_sv; // optional sv for error, needs to be freed
542	} dec_t;	825	} dec_t;
543		826
544	#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);
545		834
546	#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;
547		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
548	#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)
549	#define DEC_DEC_DEPTH --dec->depth	864	#define DEC_DEC_DEPTH --dec->depth
550		865
551	static UV	866	static UV
552	decode_uint (dec_t *dec)	867	decode_uint (dec_t *dec)
553	{	868	{
554	switch (*dec->cur & 31)	869	U8 m = *dec->cur & MINOR_MASK;
555	{	870	++dec->cur;
556	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
557	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
558	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
559	return *dec->cur++ & 31;
560		871
561	case 24:	872	if (ecb_expect_true (m < LENGTH_EXT1))
		873	return m;
		874	else if (ecb_expect_true (m == LENGTH_EXT1))
		875	{
562	WANT (2);	876	WANT (1);
563	dec->cur += 2;	877	dec->cur += 1;
564	return dec->cur[-1];	878	return dec->cur[-1];
565		879	}
566	case 25:	880	else if (ecb_expect_true (m == LENGTH_EXT2))
		881	{
567	WANT (3);	882	WANT (2);
568	dec->cur += 3;	883	dec->cur += 2;
569	return (((UV)dec->cur[-2]) << 8)	884	return (((UV)dec->cur[-2]) << 8)
570	| ((UV)dec->cur[-1]);	885	| ((UV)dec->cur[-1]);
571		886	}
572	case 26:	887	else if (ecb_expect_true (m == LENGTH_EXT4))
		888	{
573	WANT (5);	889	WANT (4);
574	dec->cur += 5;	890	dec->cur += 4;
575	return (((UV)dec->cur[-4]) << 24)	891	return (((UV)dec->cur[-4]) << 24)
576	| (((UV)dec->cur[-3]) << 16)	892	| (((UV)dec->cur[-3]) << 16)
577	| (((UV)dec->cur[-2]) << 8)	893	| (((UV)dec->cur[-2]) << 8)
578	| ((UV)dec->cur[-1]);	894	| ((UV)dec->cur[-1]);
579		895	}
580	case 27:	896	else if (ecb_expect_true (m == LENGTH_EXT8))
		897	{
581	WANT (9);	898	WANT (8);
582	dec->cur += 9;	899	dec->cur += 8;
		900
		901	return
		902	#if UVSIZE < 8
		903	0
		904	#else
583	return (((UV)dec->cur[-8]) << 56)	905	(((UV)dec->cur[-8]) << 56)
584	| (((UV)dec->cur[-7]) << 48)	906	| (((UV)dec->cur[-7]) << 48)
585	| (((UV)dec->cur[-6]) << 40)	907	| (((UV)dec->cur[-6]) << 40)
586	| (((UV)dec->cur[-5]) << 32)	908	| (((UV)dec->cur[-5]) << 32)
		909	#endif
587	| (((UV)dec->cur[-4]) << 24)	910	| (((UV)dec->cur[-4]) << 24)
588	| (((UV)dec->cur[-3]) << 16)	911	| (((UV)dec->cur[-3]) << 16)
589	| (((UV)dec->cur[-2]) << 8)	912	| (((UV)dec->cur[-2]) << 8)
590	| ((UV)dec->cur[-1]);	913	| ((UV)dec->cur[-1]);
591		914	}
592	default:	915	else
593	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	916	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
594	}
595		917
596	fail:	918	fail:
597	return 0;	919	return 0;
598	}	920	}
599		921
…		…
604	{	926	{
605	AV *av = newAV ();	927	AV *av = newAV ();
606		928
607	DEC_INC_DEPTH;	929	DEC_INC_DEPTH;
608		930
609	if ((*dec->cur & 31) == 31)	931	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
610	{	932	{
611	++dec->cur;	933	++dec->cur;
612		934
613	for (;;)	935	for (;;)
614	{	936	{
615	WANT (1);	937	WANT (1);
616		938
617	if (*dec->cur == (0xe0 | 31))	939	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
618	{	940	{
619	++dec->cur;	941	++dec->cur;
620	break;	942	break;
621	}	943	}
622		944
623	av_push (av, decode_sv (dec));	945	av_push (av, decode_sv (dec));
624	}	946	}
625	}	947	}
626	else	948	else
627	{	949	{
628	int i, len = decode_uint (dec);	950	UV i, len = decode_uint (dec);
629		951
		952	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
630	av_fill (av, len - 1);	953	av_fill (av, len - 1);
631		954
632	for (i = 0; i < len; ++i)	955	for (i = 0; i < len; ++i)
633	AvARRAY (av)[i] = decode_sv (dec);	956	AvARRAY (av)[i] = decode_sv (dec);
634	}	957	}
635		958
636	DEC_DEC_DEPTH;	959	DEC_DEC_DEPTH;
637	return newRV_noinc ((SV *)av);	960	return newRV_noinc ((SV *)av);
638		961
639	fail:	962	fail:
640	SvREFCNT_dec (av);	963	SvREFCNT_dec_NN (av);
641	DEC_DEC_DEPTH;	964	DEC_DEC_DEPTH;
642	return &PL_sv_undef;	965	return &PL_sv_undef;
643	}	966	}
644		967
645	static void	968	static void
646	decode_he (dec_t *dec, HV *hv)	969	decode_he (dec_t *dec, HV *hv)
647	{	970	{
648	// for speed reasons, we specialcase single-string	971	// for speed reasons, we specialcase single-string
649	// byte or utf-8 strings as keys.	972	// byte or utf-8 strings as keys, but only when !stringref
650		973
651	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))
652	{	976	{
653	I32 len = decode_uint (dec);	977	STRLEN len = decode_uint (dec);
654	char *key = (char *)dec->cur;	978	char *key = (char *)dec->cur;
655		979
		980	WANT (len);
656	dec->cur += len;	981	dec->cur += len;
657		982
658	if (ecb_expect_false (dec->stringref))
659	av_push (dec->stringref, newSVpvn (key, len));
660
661	hv_store (hv, key, len, decode_sv (dec), 0);	983	hv_store (hv, key, len, decode_sv (dec), 0);
		984
		985	return;
662	}	986	}
663	else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)	987	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
664	{	988	{
665	I32 len = decode_uint (dec);	989	STRLEN len = decode_uint (dec);
666	char *key = (char *)dec->cur;	990	char *key = (char *)dec->cur;
667		991
		992	WANT (len);
668	dec->cur += len;	993	dec->cur += len;
669		994
670	if (ecb_expect_false (dec->stringref))	995	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
671	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)");
672		998
673	hv_store (hv, key, -len, decode_sv (dec), 0);	999	hv_store (hv, key, -len, decode_sv (dec), 0);
		1000
		1001	return;
674	}	1002	}
675	else	1003
676	{
677	SV *k = decode_sv (dec);	1004	SV *k = decode_sv (dec);
678	SV *v = decode_sv (dec);	1005	SV *v = decode_sv (dec);
679		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
680	hv_store_ent (hv, k, v, 0);	1035	hv_store_ent (hv, k, v, 0);
681	SvREFCNT_dec (k);	1036	SvREFCNT_dec_NN (k);
682	}	1037
		1038	fail:
		1039	;
683	}	1040	}
684		1041
685	static SV *	1042	static SV *
686	decode_hv (dec_t *dec)	1043	decode_hv (dec_t *dec)
687	{	1044	{
688	HV *hv = newHV ();	1045	HV *hv = newHV ();
689		1046
690	DEC_INC_DEPTH;	1047	DEC_INC_DEPTH;
691		1048
692	if ((*dec->cur & 31) == 31)	1049	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
693	{	1050	{
694	++dec->cur;	1051	++dec->cur;
695		1052
696	for (;;)	1053	for (;;)
697	{	1054	{
698	WANT (1);	1055	WANT (1);
699		1056
700	if (*dec->cur == (0xe0 | 31))	1057	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
701	{	1058	{
702	++dec->cur;	1059	++dec->cur;
703	break;	1060	break;
704	}	1061	}
705		1062
706	decode_he (dec, hv);	1063	decode_he (dec, hv);
707	}	1064	}
708	}	1065	}
709	else	1066	else
710	{	1067	{
711	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
712		1071
713	while (pairs--)	1072	while (pairs--)
714	decode_he (dec, hv);	1073	decode_he (dec, hv);
715	}	1074	}
716		1075
717	DEC_DEC_DEPTH;	1076	DEC_DEC_DEPTH;
718	return newRV_noinc ((SV *)hv);	1077	return newRV_noinc ((SV *)hv);
719		1078
720	fail:	1079	fail:
721	SvREFCNT_dec (hv);	1080	SvREFCNT_dec_NN (hv);
722	DEC_DEC_DEPTH;	1081	DEC_DEC_DEPTH;
723	return &PL_sv_undef;	1082	return &PL_sv_undef;
724	}	1083	}
725		1084
726	static SV *	1085	static SV *
727	decode_str (dec_t *dec, int utf8)	1086	decode_str (dec_t *dec, int utf8)
728	{	1087	{
729	SV *sv = 0;	1088	SV *sv = 0;
730		1089
731	if ((*dec->cur & 31) == 31)	1090	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
732	{	1091	{
		1092	// indefinite length strings
733	++dec->cur;	1093	++dec->cur;
734		1094
		1095	U8 major = *dec->cur & MAJOR_MISC;
		1096
735	sv = newSVpvn ("", 0);	1097	sv = newSVpvn ("", 0);
736		1098
737	// not very fast, and certainly not robust against illegal input
738	for (;;)	1099	for (;;)
739	{	1100	{
740	WANT (1);	1101	WANT (1);
741		1102
742	if (*dec->cur == (0xe0 | 31))	1103	if ((*dec->cur - major) > LENGTH_EXT8)
		1104	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
743	{	1105	{
744	++dec->cur;	1106	++dec->cur;
745	break;	1107	break;
746	}	1108	}
		1109	else
		1110	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747		1111
748	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;
749	}	1117	}
750	}	1118	}
751	else	1119	else
752	{	1120	{
753	STRLEN len = decode_uint (dec);	1121	STRLEN len = decode_uint (dec);
754		1122
755	WANT (len);	1123	WANT (len);
756	sv = newSVpvn (dec->cur, len);	1124	sv = newSVpvn (dec->cur, len);
757	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));
758	}	1130	}
759		1131
760	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
761	SvUTF8_on (sv);	1138	SvUTF8_on (sv);
762		1139	}
763	if (ecb_expect_false (dec->stringref))
764	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
765		1140
766	return sv;	1141	return sv;
767		1142
768	fail:	1143	fail:
769	SvREFCNT_dec (sv);	1144	SvREFCNT_dec (sv);
…		…
788	sv = newRV_noinc (decode_sv (dec));	1163	sv = newRV_noinc (decode_sv (dec));
789	break;	1164	break;
790		1165
791	case CBOR_TAG_STRINGREF_NAMESPACE:	1166	case CBOR_TAG_STRINGREF_NAMESPACE:
792	{	1167	{
		1168	// do not use SAVETMPS/FREETMPS, as these will
		1169	// erase mortalised caches, e.g. "shareable"
793	ENTER; SAVETMPS;	1170	ENTER;
794		1171
795	SAVESPTR (dec->stringref);	1172	SAVESPTR (dec->stringref);
796	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());	1173	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
797		1174
798	sv = decode_sv (dec);	1175	sv = decode_sv (dec);
799		1176
800	FREETMPS; LEAVE;	1177	LEAVE;
801	}	1178	}
802	break;	1179	break;
803		1180
804	case CBOR_TAG_STRINGREF:	1181	case CBOR_TAG_STRINGREF:
805	{	1182	{
806	if ((*dec->cur >> 5) != 0)	1183	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
807	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1184	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
808		1185
809	UV idx = decode_uint (dec);	1186	UV idx = decode_uint (dec);
810		1187
811	if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))	1188	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
812	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)");
813		1190
814	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1191	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
815	}	1192	}
816	break;	1193	break;
…		…
818	case CBOR_TAG_VALUE_SHAREABLE:	1195	case CBOR_TAG_VALUE_SHAREABLE:
819	{	1196	{
820	if (ecb_expect_false (!dec->shareable))	1197	if (ecb_expect_false (!dec->shareable))
821	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());	1198	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
822		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;
823	sv = newSV (0);	1206	AV *av = newAV ();
824	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1207	av_push (dec->shareable, (SV *)av);
		1208	idx = AvFILLp (dec->shareable);
825		1209
826	SV *osv = decode_sv (dec);	1210	sv = decode_sv (dec);
827	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
828	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	}
829	}	1233	}
830	break;	1234	break;
831		1235
832	case CBOR_TAG_VALUE_SHAREDREF:	1236	case CBOR_TAG_VALUE_SHAREDREF:
833	{	1237	{
834	if ((*dec->cur >> 5) != 0)	1238	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
835	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1239	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
836		1240
837	UV idx = decode_uint (dec);	1241	UV idx = decode_uint (dec);
838		1242
839	if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))	1243	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
840	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1244	ERR ("corrupted CBOR data (sharedref index out of bounds)");
841		1245
842	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);
843	}	1260	}
844	break;	1261	break;
845		1262
846	case CBOR_TAG_PERL_OBJECT:	1263	case CBOR_TAG_PERL_OBJECT:
847	{	1264	{
		1265	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1266	goto filter;
		1267
848	sv = decode_sv (dec);	1268	sv = decode_sv (dec);
849		1269
850	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1270	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
851	ERR ("corrupted CBOR data (non-array perl object)");	1271	ERR ("corrupted CBOR data (non-array perl object)");
852		1272
…		…
862	if (!method)	1282	if (!method)
863	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)");
864		1284
865	dSP;	1285	dSP;
866		1286
867	ENTER; SAVETMPS; PUSHMARK (SP);	1287	ENTER; SAVETMPS;
		1288	PUSHMARK (SP);
868	EXTEND (SP, len + 1);	1289	EXTEND (SP, len + 1);
869	// 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
870	PUSHs (*av_fetch (av, 0, 1));	1291	PUSHs (*av_fetch (av, 0, 1));
871	PUSHs (sv_cbor);	1292	PUSHs (sv_cbor);
872		1293
…		…
880	SPAGAIN;	1301	SPAGAIN;
881		1302
882	if (SvTRUE (ERRSV))	1303	if (SvTRUE (ERRSV))
883	{	1304	{
884	FREETMPS; LEAVE;	1305	FREETMPS; LEAVE;
885	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1306	ERR_ERRSV;
886	}	1307	}
887		1308
888	SvREFCNT_dec (sv);	1309	SvREFCNT_dec_NN (sv);
889	sv = SvREFCNT_inc (POPs);	1310	sv = SvREFCNT_inc (POPs);
890		1311
891	PUTBACK;	1312	PUTBACK;
892		1313
893	FREETMPS; LEAVE;	1314	FREETMPS; LEAVE;
894	}	1315	}
895	break;	1316	break;
896		1317
897	default:	1318	default:
		1319	filter:
898	{	1320	{
		1321	SV *tag_sv = newSVuv (tag);
		1322
899	sv = decode_sv (dec);	1323	sv = decode_sv (dec);
900		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	{
901	AV *av = newAV ();	1352	AV *av = newAV ();
902	av_push (av, newSVuv (tag));	1353	av_push (av, tag_sv);
903	av_push (av, sv);	1354	av_push (av, sv);
904		1355
905	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1356	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
906	? cbor_tagged_stash	1357	? cbor_tagged_stash
907	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1358	: gv_stashpv ("CBOR::XS::Tagged" , 1);
908
909	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;
910	}	1365	}
911	break;	1366	break;
912	}	1367	}
913		1368
914	return sv;	1369	return sv;
…		…
921	static SV *	1376	static SV *
922	decode_sv (dec_t *dec)	1377	decode_sv (dec_t *dec)
923	{	1378	{
924	WANT (1);	1379	WANT (1);
925		1380
926	switch (*dec->cur >> 5)	1381	switch (*dec->cur >> MAJOR_SHIFT)
927	{	1382	{
928	case 0: // unsigned int	1383	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
929	return newSVuv (decode_uint (dec));	1384	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
930	case 1: // negative int	1385	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
931	return newSViv (-1 - (IV)decode_uint (dec));	1386	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
932	case 2: // octet string	1387	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
933	return decode_str (dec, 0);	1388	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
934	case 3: // utf-8 string	1389	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
935	return decode_str (dec, 1);	1390
936	case 4: // array	1391	case MAJOR_MISC >> MAJOR_SHIFT:
937	return decode_av (dec);
938	case 5: // map
939	return decode_hv (dec);
940	case 6: // tag
941	return decode_tagged (dec);
942	case 7: // misc
943	switch (*dec->cur++ & 31)	1392	switch (*dec->cur++ & MINOR_MASK)
944	{	1393	{
945	case 20:	1394	case SIMPLE_FALSE:
946	#if CBOR_SLOW	1395	#if CBOR_SLOW
947	types_false = get_bool ("Types::Serialiser::false");	1396	types_false = get_bool ("Types::Serialiser::false");
948	#endif	1397	#endif
949	return newSVsv (types_false);	1398	return newSVsv (types_false);
950	case 21:	1399	case SIMPLE_TRUE:
951	#if CBOR_SLOW	1400	#if CBOR_SLOW
952	types_true = get_bool ("Types::Serialiser::true");	1401	types_true = get_bool ("Types::Serialiser::true");
953	#endif	1402	#endif
954	return newSVsv (types_true);	1403	return newSVsv (types_true);
955	case 22:	1404	case SIMPLE_NULL:
956	return newSVsv (&PL_sv_undef);	1405	return newSVsv (&PL_sv_undef);
957	case 23:	1406	case SIMPLE_UNDEF:
958	#if CBOR_SLOW	1407	#if CBOR_SLOW
959	types_error = get_bool ("Types::Serialiser::error");	1408	types_error = get_bool ("Types::Serialiser::error");
960	#endif	1409	#endif
961	return newSVsv (types_error);	1410	return newSVsv (types_error);
962		1411
963	case 25:	1412	case MISC_FLOAT16:
964	{	1413	{
965	WANT (2);	1414	WANT (2);
966		1415
967	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1416	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
968	dec->cur += 2;	1417	dec->cur += 2;
969		1418
970	return newSVnv (ecb_binary16_to_float (fp));	1419	return newSVnv (ecb_binary16_to_float (fp));
971	}	1420	}
972		1421
973	case 26:	1422	case MISC_FLOAT32:
974	{	1423	{
975	uint32_t fp;	1424	uint32_t fp;
976	WANT (4);	1425	WANT (4);
977	memcpy (&fp, dec->cur, 4);	1426	memcpy (&fp, dec->cur, 4);
978	dec->cur += 4;	1427	dec->cur += 4;
…		…
981	fp = ecb_bswap32 (fp);	1430	fp = ecb_bswap32 (fp);
982		1431
983	return newSVnv (ecb_binary32_to_float (fp));	1432	return newSVnv (ecb_binary32_to_float (fp));
984	}	1433	}
985		1434
986	case 27:	1435	case MISC_FLOAT64:
987	{	1436	{
988	uint64_t fp;	1437	uint64_t fp;
989	WANT (8);	1438	WANT (8);
990	memcpy (&fp, dec->cur, 8);	1439	memcpy (&fp, dec->cur, 8);
991	dec->cur += 8;	1440	dec->cur += 8;
…		…
994	fp = ecb_bswap64 (fp);	1443	fp = ecb_bswap64 (fp);
995		1444
996	return newSVnv (ecb_binary64_to_double (fp));	1445	return newSVnv (ecb_binary64_to_double (fp));
997	}	1446	}
998		1447
999	// 0..19 unassigned	1448	// 0..19 unassigned simple
1000	// 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
1001	default:	1452	default:
1002	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1453	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1003	}	1454	}
1004		1455
1005	break;	1456	break;
1006	}	1457	}
1007		1458
…		…
1010	}	1461	}
1011		1462
1012	static SV *	1463	static SV *
1013	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1464	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1014	{	1465	{
1015	dec_t dec = { };	1466	dec_t dec = { 0 };
1016	SV *sv;	1467	SV *sv;
1017	STRLEN len;	1468	STRLEN len;
1018	char *data = SvPVbyte (string, len);	1469	char *data = SvPVbyte (string, len);
1019		1470
1020	if (len > cbor->max_size && cbor->max_size)	1471	if (len > cbor->max_size && cbor->max_size)
…		…
1034	if (dec.cur != dec.end && !dec.err)	1485	if (dec.cur != dec.end && !dec.err)
1035	dec.err = "garbage after CBOR object";	1486	dec.err = "garbage after CBOR object";
1036		1487
1037	if (dec.err)	1488	if (dec.err)
1038	{	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
1039	SvREFCNT_dec (sv);	1501	SvREFCNT_dec_NN (sv);
		1502
		1503	if (dec.err_sv)
		1504	sv_2mortal (dec.err_sv);
		1505
1040	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);
1041	}	1507	}
1042		1508
1043	sv = sv_2mortal (sv);	1509	sv = sv_2mortal (sv);
1044		1510
1045	return sv;	1511	return sv;
1046	}	1512	}
1047		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
1048	/////////////////////////////////////////////////////////////////////////////	1639	/////////////////////////////////////////////////////////////////////////////
1049	// XS interface functions	1640	// XS interface functions
1050		1641
1051	MODULE = CBOR::XS PACKAGE = CBOR::XS	1642	MODULE = CBOR::XS PACKAGE = CBOR::XS
1052		1643
…		…
1060		1651
1061	types_true = get_bool ("Types::Serialiser::true" );	1652	types_true = get_bool ("Types::Serialiser::true" );
1062	types_false = get_bool ("Types::Serialiser::false");	1653	types_false = get_bool ("Types::Serialiser::false");
1063	types_error = get_bool ("Types::Serialiser::error");	1654	types_error = get_bool ("Types::Serialiser::error");
1064		1655
		1656	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
		1657
1065	sv_cbor = newSVpv ("CBOR", 0);	1658	sv_cbor = newSVpv ("CBOR", 0);
1066	SvREADONLY_on (sv_cbor);	1659	SvREADONLY_on (sv_cbor);
		1660
		1661	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1067	}	1662	}
1068		1663
1069	PROTOTYPES: DISABLE	1664	PROTOTYPES: DISABLE
1070		1665
1071	void CLONE (...)	1666	void CLONE (...)
…		…
1087	)));	1682	)));
1088	}	1683	}
1089		1684
1090	void shrink (CBOR *self, int enable = 1)	1685	void shrink (CBOR *self, int enable = 1)
1091	ALIAS:	1686	ALIAS:
1092	shrink = F_SHRINK	1687	shrink = F_SHRINK
1093	allow_unknown = F_ALLOW_UNKNOWN	1688	allow_unknown = F_ALLOW_UNKNOWN
1094	allow_sharing = F_ALLOW_SHARING	1689	allow_sharing = F_ALLOW_SHARING
1095	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
1096	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
1097	PPCODE:	1697	PPCODE:
1098	{	1698	{
1099	if (enable)	1699	if (enable)
1100	self->flags |= ix;	1700	self->flags |= ix;
1101	else	1701	else
…		…
1104	XPUSHs (ST (0));	1704	XPUSHs (ST (0));
1105	}	1705	}
1106		1706
1107	void get_shrink (CBOR *self)	1707	void get_shrink (CBOR *self)
1108	ALIAS:	1708	ALIAS:
1109	get_shrink = F_SHRINK	1709	get_shrink = F_SHRINK
1110	get_allow_unknown = F_ALLOW_UNKNOWN	1710	get_allow_unknown = F_ALLOW_UNKNOWN
1111	get_allow_sharing = F_ALLOW_SHARING	1711	get_allow_sharing = F_ALLOW_SHARING
1112	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
1113	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
1114	PPCODE:	1719	PPCODE:
1115	XPUSHs (boolSV (self->flags & ix));	1720	XPUSHs (boolSV (self->flags & ix));
1116		1721
1117	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1722	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1118	PPCODE:	1723	PPCODE:
…		…
1131	XPUSHs (ST (0));	1736	XPUSHs (ST (0));
1132		1737
1133	int get_max_size (CBOR *self)	1738	int get_max_size (CBOR *self)
1134	CODE:	1739	CODE:
1135	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);
1136	OUTPUT:	1753	OUTPUT:
1137	RETVAL	1754	RETVAL
1138		1755
1139	void encode (CBOR *self, SV *scalar)	1756	void encode (CBOR *self, SV *scalar)
1140	PPCODE:	1757	PPCODE:
…		…
1155	EXTEND (SP, 2);	1772	EXTEND (SP, 2);
1156	PUSHs (sv);	1773	PUSHs (sv);
1157	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1774	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1158	}	1775	}
1159		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
1160	PROTOTYPES: ENABLE	1833	PROTOTYPES: ENABLE
1161		1834
1162	void encode_cbor (SV *scalar)	1835	void encode_cbor (SV *scalar)
		1836	ALIAS:
		1837	encode_cbor = 0
		1838	encode_cbor_sharing = F_ALLOW_SHARING
1163	PPCODE:	1839	PPCODE:
1164	{	1840	{
1165	CBOR cbor;	1841	CBOR cbor;
1166	cbor_init (&cbor);	1842	cbor_init (&cbor);
		1843	cbor.flags |= ix;
1167	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1844	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1168	XPUSHs (scalar);	1845	XPUSHs (scalar);
1169	}	1846	}
1170		1847
1171	void decode_cbor (SV *cborstr)	1848	void decode_cbor (SV *cborstr)
…		…
1175	cbor_init (&cbor);	1852	cbor_init (&cbor);
1176	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1853	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1177	XPUSHs (cborstr);	1854	XPUSHs (cborstr);
1178	}	1855	}
1179		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