ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.15 by root, Tue Oct 29 18:37:31 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
29	CBOR_TAG_PERL_OBJECT = 256,	77	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
30	CBOR_TAG_GENERIC_OBJECT = 257,	78	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
		79	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
		80	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
		81	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
		82	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
		83	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
31		84
32	// rfc7049	85	// rfc7049
33	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8	86	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any	87	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35	CBOR_TAG_POS_BIGNUM = 2, // byte string	88	CBOR_TAG_POS_BIGNUM = 2, // byte string
36	CBOR_TAG_NEG_BIGNUM = 3, // byte string	89	CBOR_TAG_NEG_BIGNUM = 3, // byte string
37	CBOR_TAG_DECIMAL = 4, // decimal fraction, array	90	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
38	CBOR_TAG_BIGFLOAT = 5, // array	91	CBOR_TAG_BIGFLOAT = 5, // array
39		92
40	CBOR_TAG_CONV_B64U = 21, // base64url, any	93	CBOR_TAG_CONV_B64U = 21, // base64url, any
41	CBOR_TAG_CONV_B64 = 22, // base64, any	94	CBOR_TAG_CONV_B64 = 22, // base64, any
42	CBOR_TAG_CONV_HEX = 23, // base16, any	95	CBOR_TAG_CONV_HEX = 23, // base16, any
43	CBOR_TAG_CBOR = 24, // embedded cbor, byte string	96	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
44		97
45	CBOR_TAG_URI = 32, // URI rfc3986, utf-8	98	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
46	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8	99	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
47	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8	100	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
48	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8	101	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
49	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	102	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50		103
51	CBOR_TAG_MAGIC = 55799 // self-describe cbor	104	CBOR_TAG_MAGIC = 55799, // self-describe cbor
52	};	105	};
53		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
54	#define F_SHRINK 0x00000200UL	121	#define F_SHRINK 0x00000001UL
55	#define F_ALLOW_UNKNOWN 0x00002000UL	122	#define F_ALLOW_UNKNOWN 0x00000002UL
		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
		127	#define F_PACK_STRINGS 0x00000040UL
		128	#define F_TEXT_KEYS 0x00000080UL
		129	#define F_TEXT_STRINGS 0x00000100UL
		130	#define F_VALIDATE_UTF8 0x00000200UL
56		131
57	#define INIT_SIZE 32 // initial scalar size to be allocated	132	#define INIT_SIZE 32 // initial scalar size to be allocated
58		133
59	#define SB do {	134	#define SB do {
60	#define SE } while (0)	135	#define SE } while (0)
…		…
72	# define CBOR_SLOW 0	147	# define CBOR_SLOW 0
73	# define CBOR_STASH cbor_stash	148	# define CBOR_STASH cbor_stash
74	#endif	149	#endif
75		150
76	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::
77	static SV *types_true, *types_false, *types_error, *sv_cbor;	152	static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
78		153
79	typedef struct {	154	typedef struct {
80	U32 flags;	155	U32 flags;
81	U32 max_depth;	156	U32 max_depth;
82	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.
83	} CBOR;	164	} CBOR;
84		165
85	ecb_inline void	166	ecb_inline void
86	cbor_init (CBOR *cbor)	167	cbor_init (CBOR *cbor)
87	{	168	{
88	Zero (cbor, 1, CBOR);	169	Zero (cbor, 1, CBOR);
89	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);
90	}	178	}
91		179
92	/////////////////////////////////////////////////////////////////////////////	180	/////////////////////////////////////////////////////////////////////////////
93	// utility functions	181	// utility functions
94		182
…		…
116	SvPV_renew (sv, SvCUR (sv) + 1);	204	SvPV_renew (sv, SvCUR (sv) + 1);
117	#endif	205	#endif
118	}	206	}
119	}	207	}
120		208
121	/////////////////////////////////////////////////////////////////////////////	209	// minimum length of a string to be registered for stringref
122	// fp hell	210	ecb_inline STRLEN
123		211	minimum_string_length (UV idx)
124	//TODO	212	{
		213	return idx <= 23 ? 3
		214	: idx <= 0xffU ? 4
		215	: idx <= 0xffffU ? 5
		216	: idx <= 0xffffffffU ? 7
		217	: 11;
		218	}
125		219
126	/////////////////////////////////////////////////////////////////////////////	220	/////////////////////////////////////////////////////////////////////////////
127	// encoder	221	// encoder
128		222
129	// structure used for encoding CBOR	223	// structure used for encoding CBOR
…		…
132	char *cur; // SvPVX (sv) + current output position	226	char *cur; // SvPVX (sv) + current output position
133	char *end; // SvEND (sv)	227	char *end; // SvEND (sv)
134	SV *sv; // result scalar	228	SV *sv; // result scalar
135	CBOR cbor;	229	CBOR cbor;
136	U32 depth; // recursion level	230	U32 depth; // recursion level
		231	HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
		232	UV stringref_idx;
		233	HV *shareable; // ptr => index, or 0
		234	UV shareable_idx;
137	} enc_t;	235	} enc_t;
138		236
139	ecb_inline void	237	ecb_inline void
140	need (enc_t *enc, STRLEN len)	238	need (enc_t *enc, STRLEN len)
141	{	239	{
142	if (ecb_expect_false (enc->cur + len >= enc->end))	240	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
143	{	241	{
144	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	242	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
145	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	243	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
146	enc->cur = SvPVX (enc->sv) + cur;	244	enc->cur = SvPVX (enc->sv) + cur;
147	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	245	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
148	}	246	}
149	}	247	}
150		248
		249	static void encode_sv (enc_t *enc, SV *sv);
		250
151	ecb_inline void	251	ecb_inline void
152	encode_ch (enc_t *enc, char ch)	252	encode_ch (enc_t *enc, char ch)
153	{	253	{
154	need (enc, 1);	254	need (enc, 1);
155	*enc->cur++ = ch;	255	*enc->cur++ = ch;
156	}	256	}
157		257
		258	// used for tags, intregers, element counts and so on
158	static void	259	static void
159	encode_uint (enc_t *enc, int major, UV len)	260	encode_uint (enc_t *enc, int major, UV len)
160	{	261	{
161	need (enc, 9);	262	need (enc, 9);
162		263
163	if (len < 24)	264	if (ecb_expect_true (len < LENGTH_EXT1))
164	*enc->cur++ = major | len;	265	*enc->cur++ = major | len;
165	else if (len <= 0xff)	266	else if (ecb_expect_true (len <= 0xffU))
166	{	267	{
167	*enc->cur++ = major | 24;	268	*enc->cur++ = major | LENGTH_EXT1;
168	*enc->cur++ = len;	269	*enc->cur++ = len;
169	}	270	}
170	else if (len <= 0xffff)	271	else if (len <= 0xffffU)
171	{	272	{
172	*enc->cur++ = major | 25;	273	*enc->cur++ = major | LENGTH_EXT2;
173	*enc->cur++ = len >> 8;	274	*enc->cur++ = len >> 8;
174	*enc->cur++ = len;	275	*enc->cur++ = len;
175	}	276	}
176	else if (len <= 0xffffffff)	277	else if (len <= 0xffffffffU)
177	{	278	{
178	*enc->cur++ = major | 26;	279	*enc->cur++ = major | LENGTH_EXT4;
179	*enc->cur++ = len >> 24;	280	*enc->cur++ = len >> 24;
180	*enc->cur++ = len >> 16;	281	*enc->cur++ = len >> 16;
181	*enc->cur++ = len >> 8;	282	*enc->cur++ = len >> 8;
182	*enc->cur++ = len;	283	*enc->cur++ = len;
183	}	284	}
184	else	285	else
185	{	286	{
186	*enc->cur++ = major | 27;	287	*enc->cur++ = major | LENGTH_EXT8;
187	*enc->cur++ = len >> 56;	288	*enc->cur++ = len >> 56;
188	*enc->cur++ = len >> 48;	289	*enc->cur++ = len >> 48;
189	*enc->cur++ = len >> 40;	290	*enc->cur++ = len >> 40;
190	*enc->cur++ = len >> 32;	291	*enc->cur++ = len >> 32;
191	*enc->cur++ = len >> 24;	292	*enc->cur++ = len >> 24;
…		…
193	*enc->cur++ = len >> 8;	294	*enc->cur++ = len >> 8;
194	*enc->cur++ = len;	295	*enc->cur++ = len;
195	}	296	}
196	}	297	}
197		298
198	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
199	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	319	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
200	{	320	{
201	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);
202	need (enc, len);	351	need (enc, len);
203	memcpy (enc->cur, str, len);	352	memcpy (enc->cur, str, len);
204	enc->cur += len;	353	enc->cur += len;
205	}	354	}
206		355
207	static void encode_sv (enc_t *enc, SV *sv);	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
		382	encode_float16 (enc_t *enc, NV nv)
		383	{
		384	need (enc, 1+2);
		385
		386	*enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
		387
		388	uint16_t fp = ecb_float_to_binary16 (nv);
		389
		390	if (!ecb_big_endian ())
		391	fp = ecb_bswap16 (fp);
		392
		393	memcpy (enc->cur, &fp, 2);
		394	enc->cur += 2;
		395	}
		396
		397	ecb_inline void
		398	encode_float32 (enc_t *enc, NV nv)
		399	{
		400	need (enc, 1+4);
		401
		402	*enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
		403
		404	uint32_t fp = ecb_float_to_binary32 (nv);
		405
		406	if (!ecb_big_endian ())
		407	fp = ecb_bswap32 (fp);
		408
		409	memcpy (enc->cur, &fp, 4);
		410	enc->cur += 4;
		411	}
		412
		413	ecb_inline void
		414	encode_float64 (enc_t *enc, NV nv)
		415	{
		416	need (enc, 1+8);
		417
		418	*enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
		419
		420	uint64_t fp = ecb_double_to_binary64 (nv);
		421
		422	if (!ecb_big_endian ())
		423	fp = ecb_bswap64 (fp);
		424
		425	memcpy (enc->cur, &fp, 8);
		426	enc->cur += 8;
		427	}
		428
		429	ecb_inline void
		430	encode_bool (enc_t *enc, int istrue)
		431	{
		432	encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
		433	}
		434
		435	// encodes an arrayref containing key-value pairs as CBOR map
		436	ecb_inline void
		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	}
208		511
209	static void	512	static void
210	encode_av (enc_t *enc, AV *av)	513	encode_av (enc_t *enc, AV *av)
211	{	514	{
212	int i, len = av_len (av);	515	int i, len = av_len (av);
…		…
214	if (enc->depth >= enc->cbor.max_depth)	517	if (enc->depth >= enc->cbor.max_depth)
215	croak (ERR_NESTING_EXCEEDED);	518	croak (ERR_NESTING_EXCEEDED);
216		519
217	++enc->depth;	520	++enc->depth;
218		521
219	encode_uint (enc, 0x80, len + 1);	522	encode_uint (enc, MAJOR_ARRAY, len + 1);
220		523
		524	if (ecb_expect_false (SvMAGICAL (av)))
221	for (i = 0; i <= len; ++i)	525	for (i = 0; i <= len; ++i)
222	{	526	{
223	SV **svp = av_fetch (av, i, 0);	527	SV **svp = av_fetch (av, i, 0);
224	encode_sv (enc, svp ? *svp : &PL_sv_undef);	528	encode_sv (enc, svp ? *svp : &PL_sv_undef);
225	}	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	}
226		536
227	--enc->depth;	537	--enc->depth;
228	}	538	}
229		539
230	static void	540	static void
…		…
238	++enc->depth;	548	++enc->depth;
239		549
240	int pairs = hv_iterinit (hv);	550	int pairs = hv_iterinit (hv);
241	int mg = SvMAGICAL (hv);	551	int mg = SvMAGICAL (hv);
242		552
243	if (mg)	553	if (ecb_expect_false (mg))
244	encode_ch (enc, 0xa0 | 31);	554	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
245	else	555	else
246	encode_uint (enc, 0xa0, pairs);	556	encode_uint (enc, MAJOR_MAP, pairs);
247		557
248	while ((he = hv_iternext (hv)))	558	while ((he = hv_iternext (hv)))
249	{	559	{
250	if (HeKLEN (he) == HEf_SVKEY)	560	if (HeKLEN (he) == HEf_SVKEY)
251	encode_sv (enc, HeSVKEY (he));	561	encode_sv (enc, HeSVKEY (he));
252	else	562	else
253	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));
254		564
255	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));
256	}	566	}
257		567
258	if (mg)	568	if (ecb_expect_false (mg))
259	encode_ch (enc, 0xe0 | 31);	569	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
260		570
261	--enc->depth;	571	--enc->depth;
262	}	572	}
263		573
264	// encode objects, arrays and special \0=false and \1=true values.	574	// encode objects, arrays and special \0=false and \1=true values.
265	static void	575	static void
266	encode_rv (enc_t *enc, SV *sv)	576	encode_rv (enc_t *enc, SV *sv)
267	{	577	{
268	svtype svt;
269
270	SvGETMAGIC (sv);	578	SvGETMAGIC (sv);
		579
271	svt = SvTYPE (sv);	580	svtype svt = SvTYPE (sv);
272		581
273	if (ecb_expect_false (SvOBJECT (sv)))	582	if (ecb_expect_false (SvOBJECT (sv)))
274	{	583	{
275	HV *boolean_stash = !CBOR_SLOW || types_boolean_stash	584	HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
276	? types_boolean_stash	585	? types_boolean_stash
…		…
281	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	590	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
282	? cbor_tagged_stash	591	? cbor_tagged_stash
283	: gv_stashpv ("CBOR::XS::Tagged" , 1);	592	: gv_stashpv ("CBOR::XS::Tagged" , 1);
284		593
285	HV *stash = SvSTASH (sv);	594	HV *stash = SvSTASH (sv);
286	GV *method;
287		595
288	if (stash == boolean_stash)	596	if (stash == boolean_stash)
289	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	597	{
		598	encode_bool (enc, SvIV (sv));
		599	return;
		600	}
290	else if (stash == error_stash)	601	else if (stash == error_stash)
291	encode_ch (enc, 0xe0 | 23);	602	{
		603	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		604	return;
		605	}
292	else if (stash == tagged_stash)	606	else if (stash == tagged_stash)
293	{	607	{
294	if (svt != SVt_PVAV)	608	if (svt != SVt_PVAV)
295	croak ("encountered CBOR::XS::Tagged object that isn't an array");	609	croak ("encountered CBOR::XS::Tagged object that isn't an array");
296		610
		611	switch (av_len ((AV *)sv))
		612	{
		613	case 2-1:
		614	// actually a tagged value
297	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	615	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
298	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))
299	}	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))));
300	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	662	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301	{	663	{
302	dSP;	664	dSP;
303		665
304	ENTER; SAVETMPS; PUSHMARK (SP);	666	ENTER; SAVETMPS;
		667	PUSHMARK (SP);
305	// 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
306	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	669	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
307		670
308	PUTBACK;	671	PUTBACK;
309	// G_SCALAR ensures that return value is 1	672	// G_SCALAR ensures that return value is 1
…		…
322	}	685	}
323	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	686	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
324	{	687	{
325	dSP;	688	dSP;
326		689
327	ENTER; SAVETMPS; PUSHMARK (SP);	690	ENTER; SAVETMPS;
		691	PUSHMARK (SP);
328	EXTEND (SP, 2);	692	EXTEND (SP, 2);
329	// 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
330	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	694	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
331	PUSHs (sv_cbor);	695	PUSHs (sv_cbor);
332		696
…		…
336		700
337	// catch this surprisingly common error	701	// catch this surprisingly common error
338	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	702	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
339	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));
340		704
341	encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT);	705	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
342	encode_uint (enc, 0x80, count + 1);	706	encode_uint (enc, MAJOR_ARRAY, count + 1);
343	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	707	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344		708
345	while (count)	709	{
		710	int i;
		711
		712	for (i = 0; i < count; ++i)
346	encode_sv (enc, SP[1 - count--]);	713	encode_sv (enc, SP[i + 1 - count]);
		714
		715	SP -= count;
		716	}
347		717
348	PUTBACK;	718	PUTBACK;
349		719
350	FREETMPS; LEAVE;	720	FREETMPS; LEAVE;
351	}	721	}
…		…
355	}	725	}
356	else if (svt == SVt_PVHV)	726	else if (svt == SVt_PVHV)
357	encode_hv (enc, (HV *)sv);	727	encode_hv (enc, (HV *)sv);
358	else if (svt == SVt_PVAV)	728	else if (svt == SVt_PVAV)
359	encode_av (enc, (AV *)sv);	729	encode_av (enc, (AV *)sv);
360	else if (svt < SVt_PVAV)
361	{
362	STRLEN len = 0;
363	char *pv = svt ? SvPV (sv, len) : 0;
364
365	if (len == 1 && *pv == '1')
366	encode_ch (enc, 0xe0 | 21);
367	else if (len == 1 && *pv == '0')
368	encode_ch (enc, 0xe0 | 20);
369	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
370	encode_ch (enc, 0xe0 | 23);
371	else
372	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
373	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374	}
375	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
376	encode_ch (enc, 0xe0 | 23);
377	else	730	else
378	croak ("encountered %s, but CBOR can only represent references to arrays or hashes",	731	{
379	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	732	encode_tag (enc, CBOR_TAG_INDIRECTION);
		733	encode_sv (enc, sv);
		734	}
380	}	735	}
381		736
382	static void	737	static void
383	encode_nv (enc_t *enc, SV *sv)	738	encode_nv (enc_t *enc, SV *sv)
384	{	739	{
385	double nv = SvNVX (sv);	740	double nv = SvNVX (sv);
386		741
387	need (enc, 9);	742	need (enc, 9);
388		743
389	if (ecb_expect_false (nv == (U32)nv))	744	if (ecb_expect_false (nv == (NV)(U32)nv))
390	encode_uint (enc, 0x00, (U32)nv);	745	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
391	//TODO: maybe I32?	746	//TODO: maybe I32?
392	else if (ecb_expect_false (nv == (float)nv))	747	else if (ecb_expect_false (nv == (float)nv))
393	{	748	encode_float32 (enc, nv);
394	uint32_t fp = ecb_float_to_binary32 (nv);
395
396	*enc->cur++ = 0xe0 | 26;
397
398	if (!ecb_big_endian ())
399	fp = ecb_bswap32 (fp);
400
401	memcpy (enc->cur, &fp, 4);
402	enc->cur += 4;
403	}
404	else	749	else
405	{	750	encode_float64 (enc, nv);
406	uint64_t fp = ecb_double_to_binary64 (nv);
407
408	*enc->cur++ = 0xe0 | 27;
409
410	if (!ecb_big_endian ())
411	fp = ecb_bswap64 (fp);
412
413	memcpy (enc->cur, &fp, 8);
414	enc->cur += 8;
415	}
416	}	751	}
417		752
418	static void	753	static void
419	encode_sv (enc_t *enc, SV *sv)	754	encode_sv (enc_t *enc, SV *sv)
420	{	755	{
…		…
422		757
423	if (SvPOKp (sv))	758	if (SvPOKp (sv))
424	{	759	{
425	STRLEN len;	760	STRLEN len;
426	char *str = SvPV (sv, len);	761	char *str = SvPV (sv, len);
427	encode_str (enc, SvUTF8 (sv), str, len);	762	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
428	}	763	}
429	else if (SvNOKp (sv))	764	else if (SvNOKp (sv))
430	encode_nv (enc, sv);	765	encode_nv (enc, sv);
431	else if (SvIOKp (sv))	766	else if (SvIOKp (sv))
432	{	767	encode_int (enc, sv);
433	if (SvIsUV (sv))
434	encode_uint (enc, 0x00, SvUVX (sv));
435	else if (SvIVX (sv) >= 0)
436	encode_uint (enc, 0x00, SvIVX (sv));
437	else
438	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
439	}
440	else if (SvROK (sv))	768	else if (SvROK (sv))
441	encode_rv (enc, SvRV (sv));	769	encode_rv (enc, SvRV (sv));
442	else if (!SvOK (sv))	770	else if (!SvOK (sv))
443	encode_ch (enc, 0xe0 | 22);	771	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
444	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	772	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
445	encode_ch (enc, 0xe0 | 23);	773	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
446	else	774	else
447	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",
448	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	776	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
449	}	777	}
450		778
451	static SV *	779	static SV *
452	encode_cbor (SV *scalar, CBOR *cbor)	780	encode_cbor (SV *scalar, CBOR *cbor)
453	{	781	{
454	enc_t enc;	782	enc_t enc = { 0 };
455		783
456	enc.cbor = *cbor;	784	enc.cbor = *cbor;
457	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	785	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458	enc.cur = SvPVX (enc.sv);	786	enc.cur = SvPVX (enc.sv);
459	enc.end = SvEND (enc.sv);	787	enc.end = SvEND (enc.sv);
460	enc.depth = 0;
461		788
462	SvPOK_only (enc.sv);	789	SvPOK_only (enc.sv);
		790
		791	if (cbor->flags & F_PACK_STRINGS)
		792	{
		793	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
		794	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
		795	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
		796	}
		797
463	encode_sv (&enc, scalar);	798	encode_sv (&enc, scalar);
464		799
465	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	800	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
466	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	801	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
467		802
…		…
481	U8 *end; // end of input string	816	U8 *end; // end of input string
482	const char *err; // parse error, if != 0	817	const char *err; // parse error, if != 0
483	CBOR cbor;	818	CBOR cbor;
484	U32 depth; // recursion depth	819	U32 depth; // recursion depth
485	U32 maxdepth; // recursion depth limit	820	U32 maxdepth; // recursion depth limit
		821	AV *shareable;
		822	AV *stringref;
		823	SV *decode_tagged;
		824	SV *err_sv; // optional sv for error, needs to be freed
486	} dec_t;	825	} dec_t;
487		826
488	#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);
489		834
490	#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;
491		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
492	#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)
493	#define DEC_DEC_DEPTH --dec->depth	864	#define DEC_DEC_DEPTH --dec->depth
494		865
495	static UV	866	static UV
496	decode_uint (dec_t *dec)	867	decode_uint (dec_t *dec)
497	{	868	{
498	switch (*dec->cur & 31)	869	U8 m = *dec->cur & MINOR_MASK;
499	{	870	++dec->cur;
500	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
501	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
502	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
503	return *dec->cur++ & 31;
504		871
505	case 24:	872	if (ecb_expect_true (m < LENGTH_EXT1))
		873	return m;
		874	else if (ecb_expect_true (m == LENGTH_EXT1))
		875	{
506	WANT (2);	876	WANT (1);
507	dec->cur += 2;	877	dec->cur += 1;
508	return dec->cur[-1];	878	return dec->cur[-1];
509		879	}
510	case 25:	880	else if (ecb_expect_true (m == LENGTH_EXT2))
		881	{
511	WANT (3);	882	WANT (2);
512	dec->cur += 3;	883	dec->cur += 2;
513	return (((UV)dec->cur[-2]) << 8)	884	return (((UV)dec->cur[-2]) << 8)
514	| ((UV)dec->cur[-1]);	885	| ((UV)dec->cur[-1]);
515		886	}
516	case 26:	887	else if (ecb_expect_true (m == LENGTH_EXT4))
		888	{
517	WANT (5);	889	WANT (4);
518	dec->cur += 5;	890	dec->cur += 4;
519	return (((UV)dec->cur[-4]) << 24)	891	return (((UV)dec->cur[-4]) << 24)
520	| (((UV)dec->cur[-3]) << 16)	892	| (((UV)dec->cur[-3]) << 16)
521	| (((UV)dec->cur[-2]) << 8)	893	| (((UV)dec->cur[-2]) << 8)
522	| ((UV)dec->cur[-1]);	894	| ((UV)dec->cur[-1]);
523		895	}
524	case 27:	896	else if (ecb_expect_true (m == LENGTH_EXT8))
		897	{
525	WANT (9);	898	WANT (8);
526	dec->cur += 9;	899	dec->cur += 8;
		900
		901	return
		902	#if UVSIZE < 8
		903	0
		904	#else
527	return (((UV)dec->cur[-8]) << 56)	905	(((UV)dec->cur[-8]) << 56)
528	| (((UV)dec->cur[-7]) << 48)	906	| (((UV)dec->cur[-7]) << 48)
529	| (((UV)dec->cur[-6]) << 40)	907	| (((UV)dec->cur[-6]) << 40)
530	| (((UV)dec->cur[-5]) << 32)	908	| (((UV)dec->cur[-5]) << 32)
		909	#endif
531	| (((UV)dec->cur[-4]) << 24)	910	| (((UV)dec->cur[-4]) << 24)
532	| (((UV)dec->cur[-3]) << 16)	911	| (((UV)dec->cur[-3]) << 16)
533	| (((UV)dec->cur[-2]) << 8)	912	| (((UV)dec->cur[-2]) << 8)
534	| ((UV)dec->cur[-1]);	913	| ((UV)dec->cur[-1]);
535		914	}
536	default:	915	else
537	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	916	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
538	}
539		917
540	fail:	918	fail:
541	return 0;	919	return 0;
542	}	920	}
543		921
…		…
548	{	926	{
549	AV *av = newAV ();	927	AV *av = newAV ();
550		928
551	DEC_INC_DEPTH;	929	DEC_INC_DEPTH;
552		930
553	if ((*dec->cur & 31) == 31)	931	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
554	{	932	{
555	++dec->cur;	933	++dec->cur;
556		934
557	for (;;)	935	for (;;)
558	{	936	{
559	WANT (1);	937	WANT (1);
560		938
561	if (*dec->cur == (0xe0 | 31))	939	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
562	{	940	{
563	++dec->cur;	941	++dec->cur;
564	break;	942	break;
565	}	943	}
566		944
567	av_push (av, decode_sv (dec));	945	av_push (av, decode_sv (dec));
568	}	946	}
569	}	947	}
570	else	948	else
571	{	949	{
572	int i, len = decode_uint (dec);	950	UV i, len = decode_uint (dec);
573		951
		952	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
574	av_fill (av, len - 1);	953	av_fill (av, len - 1);
575		954
576	for (i = 0; i < len; ++i)	955	for (i = 0; i < len; ++i)
577	AvARRAY (av)[i] = decode_sv (dec);	956	AvARRAY (av)[i] = decode_sv (dec);
578	}	957	}
579		958
580	DEC_DEC_DEPTH;	959	DEC_DEC_DEPTH;
581	return newRV_noinc ((SV *)av);	960	return newRV_noinc ((SV *)av);
582		961
583	fail:	962	fail:
584	SvREFCNT_dec (av);	963	SvREFCNT_dec_NN (av);
585	DEC_DEC_DEPTH;	964	DEC_DEC_DEPTH;
586	return &PL_sv_undef;	965	return &PL_sv_undef;
587	}	966	}
588		967
		968	static void
		969	decode_he (dec_t *dec, HV *hv)
		970	{
		971	// for speed reasons, we specialcase single-string
		972	// byte or utf-8 strings as keys, but only when !stringref
		973
		974	if (ecb_expect_true (!dec->stringref))
		975	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
		976	{
		977	STRLEN len = decode_uint (dec);
		978	char *key = (char *)dec->cur;
		979
		980	WANT (len);
		981	dec->cur += len;
		982
		983	hv_store (hv, key, len, decode_sv (dec), 0);
		984
		985	return;
		986	}
		987	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
		988	{
		989	STRLEN len = decode_uint (dec);
		990	char *key = (char *)dec->cur;
		991
		992	WANT (len);
		993	dec->cur += len;
		994
		995	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		996	if (!cbor_is_utf8_string ((U8 *)key, len))
		997	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
		998
		999	hv_store (hv, key, -len, decode_sv (dec), 0);
		1000
		1001	return;
		1002	}
		1003
		1004	SV *k = decode_sv (dec);
		1005	SV *v = decode_sv (dec);
		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
		1035	hv_store_ent (hv, k, v, 0);
		1036	SvREFCNT_dec_NN (k);
		1037
		1038	fail:
		1039	;
		1040	}
		1041
589	static SV *	1042	static SV *
590	decode_hv (dec_t *dec)	1043	decode_hv (dec_t *dec)
591	{	1044	{
592	HV *hv = newHV ();	1045	HV *hv = newHV ();
593		1046
594	DEC_INC_DEPTH;	1047	DEC_INC_DEPTH;
595		1048
596	if ((*dec->cur & 31) == 31)	1049	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
597	{	1050	{
598	++dec->cur;	1051	++dec->cur;
599		1052
600	for (;;)	1053	for (;;)
601	{	1054	{
602	WANT (1);	1055	WANT (1);
603		1056
604	if (*dec->cur == (0xe0 | 31))	1057	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
605	{	1058	{
606	++dec->cur;	1059	++dec->cur;
607	break;	1060	break;
608	}	1061	}
609		1062
610	SV *k = decode_sv (dec);	1063	decode_he (dec, hv);
611	SV *v = decode_sv (dec);
612
613	hv_store_ent (hv, k, v, 0);
614	SvREFCNT_dec (k);
615	}	1064	}
616	}	1065	}
617	else	1066	else
618	{	1067	{
619	int len = decode_uint (dec);	1068	UV pairs = decode_uint (dec);
620		1069
		1070	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
		1071
621	while (len--)	1072	while (pairs--)
622	{	1073	decode_he (dec, hv);
623	SV *k = decode_sv (dec);
624	SV *v = decode_sv (dec);
625
626	hv_store_ent (hv, k, v, 0);
627	SvREFCNT_dec (k);
628	}
629	}	1074	}
630		1075
631	DEC_DEC_DEPTH;	1076	DEC_DEC_DEPTH;
632	return newRV_noinc ((SV *)hv);	1077	return newRV_noinc ((SV *)hv);
633		1078
634	fail:	1079	fail:
635	SvREFCNT_dec (hv);	1080	SvREFCNT_dec_NN (hv);
636	DEC_DEC_DEPTH;	1081	DEC_DEC_DEPTH;
637	return &PL_sv_undef;	1082	return &PL_sv_undef;
638	}	1083	}
639		1084
640	static SV *	1085	static SV *
641	decode_str (dec_t *dec, int utf8)	1086	decode_str (dec_t *dec, int utf8)
642	{	1087	{
643	SV *sv = 0;	1088	SV *sv = 0;
644		1089
645	if ((*dec->cur & 31) == 31)	1090	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
646	{	1091	{
		1092	// indefinite length strings
647	++dec->cur;	1093	++dec->cur;
648		1094
		1095	U8 major = *dec->cur & MAJOR_MISC;
		1096
649	sv = newSVpvn ("", 0);	1097	sv = newSVpvn ("", 0);
650		1098
651	// not very fast, and certainly not robust against illegal input
652	for (;;)	1099	for (;;)
653	{	1100	{
654	WANT (1);	1101	WANT (1);
655		1102
656	if (*dec->cur == (0xe0 | 31))	1103	if ((*dec->cur - major) > LENGTH_EXT8)
		1104	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
657	{	1105	{
658	++dec->cur;	1106	++dec->cur;
659	break;	1107	break;
660	}	1108	}
		1109	else
		1110	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
661		1111
662	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;
663	}	1117	}
664	}	1118	}
665	else	1119	else
666	{	1120	{
667	STRLEN len = decode_uint (dec);	1121	STRLEN len = decode_uint (dec);
668		1122
669	WANT (len);	1123	WANT (len);
670	sv = newSVpvn (dec->cur, len);	1124	sv = newSVpvn (dec->cur, len);
671	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));
672	}	1130	}
673		1131
674	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
675	SvUTF8_on (sv);	1138	SvUTF8_on (sv);
		1139	}
676		1140
677	return sv;	1141	return sv;
678		1142
679	fail:	1143	fail:
680	SvREFCNT_dec (sv);	1144	SvREFCNT_dec (sv);
…		…
682	}	1146	}
683		1147
684	static SV *	1148	static SV *
685	decode_tagged (dec_t *dec)	1149	decode_tagged (dec_t *dec)
686	{	1150	{
		1151	SV *sv = 0;
687	UV tag = decode_uint (dec);	1152	UV tag = decode_uint (dec);
		1153
		1154	WANT (1);
		1155
		1156	switch (tag)
		1157	{
		1158	case CBOR_TAG_MAGIC:
688	SV *sv = decode_sv (dec);	1159	sv = decode_sv (dec);
		1160	break;
689		1161
690	if (tag == CBOR_TAG_MAGIC)	1162	case CBOR_TAG_INDIRECTION:
691	return sv;	1163	sv = newRV_noinc (decode_sv (dec));
692	else if (tag == CBOR_TAG_PERL_OBJECT)	1164	break;
693	{	1165
		1166	case CBOR_TAG_STRINGREF_NAMESPACE:
		1167	{
		1168	// do not use SAVETMPS/FREETMPS, as these will
		1169	// erase mortalised caches, e.g. "shareable"
		1170	ENTER;
		1171
		1172	SAVESPTR (dec->stringref);
		1173	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
		1174
		1175	sv = decode_sv (dec);
		1176
		1177	LEAVE;
		1178	}
		1179	break;
		1180
		1181	case CBOR_TAG_STRINGREF:
		1182	{
		1183	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
		1184	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
		1185
		1186	UV idx = decode_uint (dec);
		1187
		1188	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
		1189	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
		1190
		1191	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
		1192	}
		1193	break;
		1194
		1195	case CBOR_TAG_VALUE_SHAREABLE:
		1196	{
		1197	if (ecb_expect_false (!dec->shareable))
		1198	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
		1199
		1200	if (ecb_expect_false (dec->cbor.flags & (F_ALLOW_CYCLES | F_ALLOW_WEAK_CYCLES)))
		1201	{
		1202	// if cycles are allowed, then we store an AV as value
		1203	// while it is being decoded, and gather unresolved
		1204	// references in it, to be re4solved after decoding.
		1205	int idx, i;
		1206	AV *av = newAV ();
		1207	av_push (dec->shareable, (SV *)av);
		1208	idx = AvFILLp (dec->shareable);
		1209
		1210	sv = decode_sv (dec);
		1211
		1212	// the AV now contains \undef for all unresolved references,
		1213	// so we fix them up here.
		1214	for (i = 0; i <= AvFILLp (av); ++i)
		1215	SvRV_set (AvARRAY (av)[i], SvREFCNT_inc_NN (SvRV (sv)));
		1216
		1217	// weaken all recursive references
		1218	if (dec->cbor.flags & F_ALLOW_WEAK_CYCLES)
		1219	for (i = 0; i <= AvFILLp (av); ++i)
		1220	sv_rvweaken (AvARRAY (av)[i]);
		1221
		1222	// now replace the AV by a reference to the completed value
		1223	SvREFCNT_dec_NN ((SV *)av);
		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	}
		1233	}
		1234	break;
		1235
		1236	case CBOR_TAG_VALUE_SHAREDREF:
		1237	{
		1238	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
		1239	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
		1240
		1241	UV idx = decode_uint (dec);
		1242
		1243	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
		1244	ERR ("corrupted CBOR data (sharedref index out of bounds)");
		1245
		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);
		1260	}
		1261	break;
		1262
		1263	case CBOR_TAG_PERL_OBJECT:
		1264	{
		1265	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1266	goto filter;
		1267
		1268	sv = decode_sv (dec);
		1269
694	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1270	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
695	ERR ("corrupted CBOR data (non-array perl object)");	1271	ERR ("corrupted CBOR data (non-array perl object)");
696		1272
697	AV *av = (AV *)SvRV (sv);	1273	AV *av = (AV *)SvRV (sv);
698	int len = av_len (av) + 1;	1274	int len = av_len (av) + 1;
699	HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);	1275	HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
700		1276
701	if (!stash)	1277	if (!stash)
702	ERR ("cannot decode perl-object (package does not exist)");	1278	ERR ("cannot decode perl-object (package does not exist)");
703		1279
704	GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);	1280	GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
705		1281
706	if (!method)	1282	if (!method)
707	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)");
708		1284
709	dSP;	1285	dSP;
710		1286
711	ENTER; SAVETMPS; PUSHMARK (SP);	1287	ENTER; SAVETMPS;
		1288	PUSHMARK (SP);
712	EXTEND (SP, len + 1);	1289	EXTEND (SP, len + 1);
713	// 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
714	PUSHs (*av_fetch (av, 0, 1));	1291	PUSHs (*av_fetch (av, 0, 1));
715	PUSHs (sv_cbor);	1292	PUSHs (sv_cbor);
716		1293
717	int i;	1294	int i;
718		1295
719	for (i = 1; i < len; ++i)	1296	for (i = 1; i < len; ++i)
720	PUSHs (*av_fetch (av, i, 1));	1297	PUSHs (*av_fetch (av, i, 1));
721		1298
722	PUTBACK;	1299	PUTBACK;
723	call_sv ((SV *)GvCV (method), G_SCALAR);	1300	call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
724	SPAGAIN;	1301	SPAGAIN;
725		1302
		1303	if (SvTRUE (ERRSV))
		1304	{
		1305	FREETMPS; LEAVE;
		1306	ERR_ERRSV;
		1307	}
		1308
726	SvREFCNT_dec (sv);	1309	SvREFCNT_dec_NN (sv);
727	sv = SvREFCNT_inc (POPs);	1310	sv = SvREFCNT_inc (POPs);
728		1311
729	PUTBACK;	1312	PUTBACK;
730		1313
731	FREETMPS; LEAVE;	1314	FREETMPS; LEAVE;
		1315	}
		1316	break;
732		1317
733	return sv;	1318	default:
734	}	1319	filter:
735	else	1320	{
736	{	1321	SV *tag_sv = newSVuv (tag);
		1322
		1323	sv = decode_sv (dec);
		1324
		1325	dSP;
		1326	ENTER; SAVETMPS;
		1327	PUSHMARK (SP);
		1328	EXTEND (SP, 2);
		1329	PUSHs (tag_sv);
		1330	PUSHs (sv);
		1331
		1332	PUTBACK;
		1333	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
		1334	SPAGAIN;
		1335
		1336	if (SvTRUE (ERRSV))
		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	{
737	AV *av = newAV ();	1352	AV *av = newAV ();
738	av_push (av, newSVuv (tag));	1353	av_push (av, tag_sv);
739	av_push (av, sv);	1354	av_push (av, sv);
740		1355
741	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1356	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
742	? cbor_tagged_stash	1357	? cbor_tagged_stash
743	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1358	: gv_stashpv ("CBOR::XS::Tagged" , 1);
744
745	return 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;
		1365	}
		1366	break;
746	}	1367	}
		1368
		1369	return sv;
747		1370
748	fail:	1371	fail:
749	SvREFCNT_dec (sv);	1372	SvREFCNT_dec (sv);
750	return &PL_sv_undef;	1373	return &PL_sv_undef;
751	}	1374	}
…		…
753	static SV *	1376	static SV *
754	decode_sv (dec_t *dec)	1377	decode_sv (dec_t *dec)
755	{	1378	{
756	WANT (1);	1379	WANT (1);
757		1380
758	switch (*dec->cur >> 5)	1381	switch (*dec->cur >> MAJOR_SHIFT)
759	{	1382	{
760	case 0: // unsigned int	1383	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
761	return newSVuv (decode_uint (dec));	1384	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
762	case 1: // negative int	1385	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
763	return newSViv (-1 - (IV)decode_uint (dec));	1386	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
764	case 2: // octet string	1387	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
765	return decode_str (dec, 0);	1388	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
766	case 3: // utf-8 string	1389	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
767	return decode_str (dec, 1);	1390
768	case 4: // array	1391	case MAJOR_MISC >> MAJOR_SHIFT:
769	return decode_av (dec);
770	case 5: // map
771	return decode_hv (dec);
772	case 6: // tag
773	return decode_tagged (dec);
774	case 7: // misc
775	switch (*dec->cur++ & 31)	1392	switch (*dec->cur++ & MINOR_MASK)
776	{	1393	{
777	case 20:	1394	case SIMPLE_FALSE:
778	#if CBOR_SLOW	1395	#if CBOR_SLOW
779	types_false = get_bool ("Types::Serialiser::false");	1396	types_false = get_bool ("Types::Serialiser::false");
780	#endif	1397	#endif
781	return newSVsv (types_false);	1398	return newSVsv (types_false);
782	case 21:	1399	case SIMPLE_TRUE:
783	#if CBOR_SLOW	1400	#if CBOR_SLOW
784	types_true = get_bool ("Types::Serialiser::true");	1401	types_true = get_bool ("Types::Serialiser::true");
785	#endif	1402	#endif
786	return newSVsv (types_true);	1403	return newSVsv (types_true);
787	case 22:	1404	case SIMPLE_NULL:
788	return newSVsv (&PL_sv_undef);	1405	return newSVsv (&PL_sv_undef);
789	case 23:	1406	case SIMPLE_UNDEF:
790	#if CBOR_SLOW	1407	#if CBOR_SLOW
791	types_error = get_bool ("Types::Serialiser::error");	1408	types_error = get_bool ("Types::Serialiser::error");
792	#endif	1409	#endif
793	return newSVsv (types_error);	1410	return newSVsv (types_error);
794		1411
795	case 25:	1412	case MISC_FLOAT16:
796	{	1413	{
797	WANT (2);	1414	WANT (2);
798		1415
799	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1416	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
800	dec->cur += 2;	1417	dec->cur += 2;
801		1418
802	return newSVnv (ecb_binary16_to_float (fp));	1419	return newSVnv (ecb_binary16_to_float (fp));
803	}	1420	}
804		1421
805	case 26:	1422	case MISC_FLOAT32:
806	{	1423	{
807	uint32_t fp;	1424	uint32_t fp;
808	WANT (4);	1425	WANT (4);
809	memcpy (&fp, dec->cur, 4);	1426	memcpy (&fp, dec->cur, 4);
810	dec->cur += 4;	1427	dec->cur += 4;
…		…
813	fp = ecb_bswap32 (fp);	1430	fp = ecb_bswap32 (fp);
814		1431
815	return newSVnv (ecb_binary32_to_float (fp));	1432	return newSVnv (ecb_binary32_to_float (fp));
816	}	1433	}
817		1434
818	case 27:	1435	case MISC_FLOAT64:
819	{	1436	{
820	uint64_t fp;	1437	uint64_t fp;
821	WANT (8);	1438	WANT (8);
822	memcpy (&fp, dec->cur, 8);	1439	memcpy (&fp, dec->cur, 8);
823	dec->cur += 8;	1440	dec->cur += 8;
…		…
826	fp = ecb_bswap64 (fp);	1443	fp = ecb_bswap64 (fp);
827		1444
828	return newSVnv (ecb_binary64_to_double (fp));	1445	return newSVnv (ecb_binary64_to_double (fp));
829	}	1446	}
830		1447
831	// 0..19 unassigned	1448	// 0..19 unassigned simple
832	// 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
833	default:	1452	default:
834	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1453	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
835	}	1454	}
836		1455
837	break;	1456	break;
838	}	1457	}
839		1458
…		…
842	}	1461	}
843		1462
844	static SV *	1463	static SV *
845	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1464	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
846	{	1465	{
847	dec_t dec;	1466	dec_t dec = { 0 };
848	SV *sv;	1467	SV *sv;
		1468	STRLEN len;
		1469	char *data = SvPVbyte (string, len);
849		1470
850	/* work around bugs in 5.10 where manipulating magic values
851	* makes perl ignore the magic in subsequent accesses.
852	* also make a copy of non-PV values, to get them into a clean
853	* state (SvPV should do that, but it's buggy, see below).
854	*/
855	/*SvGETMAGIC (string);*/
856	if (SvMAGICAL (string) || !SvPOK (string))
857	string = sv_2mortal (newSVsv (string));
858
859	SvUPGRADE (string, SVt_PV);
860
861	/* work around a bug in perl 5.10, which causes SvCUR to fail an
862	* assertion with -DDEBUGGING, although SvCUR is documented to
863	* return the xpv_cur field which certainly exists after upgrading.
864	* according to nicholas clark, calling SvPOK fixes this.
865	* But it doesn't fix it, so try another workaround, call SvPV_nolen
866	* and hope for the best.
867	* Damnit, SvPV_nolen still trips over yet another assertion. This
868	* assertion business is seriously broken, try yet another workaround
869	* for the broken -DDEBUGGING.
870	*/
871	{
872	#ifdef DEBUGGING
873	STRLEN offset = SvOK (string) ? sv_len (string) : 0;
874	#else
875	STRLEN offset = SvCUR (string);
876	#endif
877
878	if (offset > cbor->max_size && cbor->max_size)	1471	if (len > cbor->max_size && cbor->max_size)
879	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",	1472	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
880	(unsigned long)SvCUR (string), (unsigned long)cbor->max_size);	1473	(unsigned long)len, (unsigned long)cbor->max_size);
881	}
882
883	sv_utf8_downgrade (string, 0);
884		1474
885	dec.cbor = *cbor;	1475	dec.cbor = *cbor;
886	dec.cur = (U8 *)SvPVX (string);	1476	dec.cur = (U8 *)data;
887	dec.end = (U8 *)SvEND (string);	1477	dec.end = (U8 *)data + len;
888	dec.err = 0;
889	dec.depth = 0;
890		1478
891	sv = decode_sv (&dec);	1479	sv = decode_sv (&dec);
892		1480
893	if (offset_return)	1481	if (offset_return)
894	*offset_return = dec.cur;	1482	*offset_return = dec.cur;
…		…
897	if (dec.cur != dec.end && !dec.err)	1485	if (dec.cur != dec.end && !dec.err)
898	dec.err = "garbage after CBOR object";	1486	dec.err = "garbage after CBOR object";
899		1487
900	if (dec.err)	1488	if (dec.err)
901	{	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
902	SvREFCNT_dec (sv);	1501	SvREFCNT_dec_NN (sv);
		1502
		1503	if (dec.err_sv)
		1504	sv_2mortal (dec.err_sv);
		1505
903	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (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);
904	}	1507	}
905		1508
906	sv = sv_2mortal (sv);	1509	sv = sv_2mortal (sv);
907		1510
908	return sv;	1511	return sv;
909	}	1512	}
910		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
911	/////////////////////////////////////////////////////////////////////////////	1639	/////////////////////////////////////////////////////////////////////////////
912	// XS interface functions	1640	// XS interface functions
913		1641
914	MODULE = CBOR::XS PACKAGE = CBOR::XS	1642	MODULE = CBOR::XS PACKAGE = CBOR::XS
915		1643
…		…
923		1651
924	types_true = get_bool ("Types::Serialiser::true" );	1652	types_true = get_bool ("Types::Serialiser::true" );
925	types_false = get_bool ("Types::Serialiser::false");	1653	types_false = get_bool ("Types::Serialiser::false");
926	types_error = get_bool ("Types::Serialiser::error");	1654	types_error = get_bool ("Types::Serialiser::error");
927		1655
		1656	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
		1657
928	sv_cbor = newSVpv ("CBOR", 0);	1658	sv_cbor = newSVpv ("CBOR", 0);
929	SvREADONLY_on (sv_cbor);	1659	SvREADONLY_on (sv_cbor);
		1660
		1661	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
930	}	1662	}
931		1663
932	PROTOTYPES: DISABLE	1664	PROTOTYPES: DISABLE
933		1665
934	void CLONE (...)	1666	void CLONE (...)
…		…
950	)));	1682	)));
951	}	1683	}
952		1684
953	void shrink (CBOR *self, int enable = 1)	1685	void shrink (CBOR *self, int enable = 1)
954	ALIAS:	1686	ALIAS:
955	shrink = F_SHRINK	1687	shrink = F_SHRINK
956	allow_unknown = F_ALLOW_UNKNOWN	1688	allow_unknown = F_ALLOW_UNKNOWN
		1689	allow_sharing = F_ALLOW_SHARING
		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
957	PPCODE:	1697	PPCODE:
958	{	1698	{
959	if (enable)	1699	if (enable)
960	self->flags |= ix;	1700	self->flags |= ix;
961	else	1701	else
…		…
964	XPUSHs (ST (0));	1704	XPUSHs (ST (0));
965	}	1705	}
966		1706
967	void get_shrink (CBOR *self)	1707	void get_shrink (CBOR *self)
968	ALIAS:	1708	ALIAS:
969	get_shrink = F_SHRINK	1709	get_shrink = F_SHRINK
970	get_allow_unknown = F_ALLOW_UNKNOWN	1710	get_allow_unknown = F_ALLOW_UNKNOWN
		1711	get_allow_sharing = F_ALLOW_SHARING
		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
971	PPCODE:	1719	PPCODE:
972	XPUSHs (boolSV (self->flags & ix));	1720	XPUSHs (boolSV (self->flags & ix));
973		1721
974	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1722	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
975	PPCODE:	1723	PPCODE:
…		…
988	XPUSHs (ST (0));	1736	XPUSHs (ST (0));
989		1737
990	int get_max_size (CBOR *self)	1738	int get_max_size (CBOR *self)
991	CODE:	1739	CODE:
992	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);
993	OUTPUT:	1753	OUTPUT:
994	RETVAL	1754	RETVAL
995		1755
996	void encode (CBOR *self, SV *scalar)	1756	void encode (CBOR *self, SV *scalar)
997	PPCODE:	1757	PPCODE:
…		…
1012	EXTEND (SP, 2);	1772	EXTEND (SP, 2);
1013	PUSHs (sv);	1773	PUSHs (sv);
1014	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1774	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1015	}	1775	}
1016		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
1017	PROTOTYPES: ENABLE	1833	PROTOTYPES: ENABLE
1018		1834
1019	void encode_cbor (SV *scalar)	1835	void encode_cbor (SV *scalar)
		1836	ALIAS:
		1837	encode_cbor = 0
		1838	encode_cbor_sharing = F_ALLOW_SHARING
1020	PPCODE:	1839	PPCODE:
1021	{	1840	{
1022	CBOR cbor;	1841	CBOR cbor;
1023	cbor_init (&cbor);	1842	cbor_init (&cbor);
		1843	cbor.flags |= ix;
1024	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1844	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1025	XPUSHs (scalar);	1845	XPUSHs (scalar);
1026	}	1846	}
1027		1847
1028	void decode_cbor (SV *cborstr)	1848	void decode_cbor (SV *cborstr)
…		…
1032	cbor_init (&cbor);	1852	cbor_init (&cbor);
1033	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1853	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1034	XPUSHs (cborstr);	1854	XPUSHs (cborstr);
1035	}	1855	}
1036		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