ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.29 by root, Thu Nov 28 09:13:12 2013 UTC vs.
Revision 1.78 by root, Fri Sep 8 06:01:12 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
24	#ifndef SvREFCNT_dec_NN	29	#ifndef SvREFCNT_dec_NN
25	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)	30	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26	#endif	31	#endif
27		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	};
		72
28	// known tags	73	// known tags
29	enum cbor_tag	74	enum cbor_tag
30	{	75	{
31	// extensions	76	// extensions
32	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref	77	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
33	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object	78	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
34	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object	79	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
35	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing	80	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
36	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing	81	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
37	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref	82	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
38	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection	83	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39		84
40	// rfc7049	85	// rfc7049
41	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8	86	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
42	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any	87	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
43	CBOR_TAG_POS_BIGNUM = 2, // byte string	88	CBOR_TAG_POS_BIGNUM = 2, // byte string
44	CBOR_TAG_NEG_BIGNUM = 3, // byte string	89	CBOR_TAG_NEG_BIGNUM = 3, // byte string
45	CBOR_TAG_DECIMAL = 4, // decimal fraction, array	90	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
46	CBOR_TAG_BIGFLOAT = 5, // array	91	CBOR_TAG_BIGFLOAT = 5, // array
47		92
48	CBOR_TAG_CONV_B64U = 21, // base64url, any	93	CBOR_TAG_CONV_B64U = 21, // base64url, any
49	CBOR_TAG_CONV_B64 = 22, // base64, any	94	CBOR_TAG_CONV_B64 = 22, // base64, any
50	CBOR_TAG_CONV_HEX = 23, // base16, any	95	CBOR_TAG_CONV_HEX = 23, // base16, any
51	CBOR_TAG_CBOR = 24, // embedded cbor, byte string	96	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
52		97
53	CBOR_TAG_URI = 32, // URI rfc3986, utf-8	98	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
54	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8	99	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
55	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8	100	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
56	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8	101	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
57	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	102	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
58		103
59	CBOR_TAG_MAGIC = 55799 // self-describe cbor	104	CBOR_TAG_MAGIC = 55799, // self-describe cbor
		105	};
		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
60	};	119	};
61		120
62	#define F_SHRINK 0x00000001UL	121	#define F_SHRINK 0x00000001UL
63	#define F_ALLOW_UNKNOWN 0x00000002UL	122	#define F_ALLOW_UNKNOWN 0x00000002UL
64	#define F_ALLOW_SHARING 0x00000004UL //TODO	123	#define F_ALLOW_SHARING 0x00000004UL
65	#define F_ALLOW_STRINGREF 0x00000008UL //TODO	124	#define F_ALLOW_CYCLES 0x00000008UL
		125	#define F_FORBID_OBJECTS 0x00000010UL
		126	#define F_PACK_STRINGS 0x00000020UL
		127	#define F_TEXT_KEYS 0x00000040UL
		128	#define F_TEXT_STRINGS 0x00000080UL
		129	#define F_VALIDATE_UTF8 0x00000100UL
66		130
67	#define INIT_SIZE 32 // initial scalar size to be allocated	131	#define INIT_SIZE 32 // initial scalar size to be allocated
68		132
69	#define SB do {	133	#define SB do {
70	#define SE } while (0)	134	#define SE } while (0)
…		…
89	typedef struct {	153	typedef struct {
90	U32 flags;	154	U32 flags;
91	U32 max_depth;	155	U32 max_depth;
92	STRLEN max_size;	156	STRLEN max_size;
93	SV *filter;	157	SV *filter;
		158
		159	// for the incremental parser
		160	STRLEN incr_pos; // the current offset into the text
		161	STRLEN incr_need; // minimum bytes needed to decode
		162	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
94	} CBOR;	163	} CBOR;
95		164
96	ecb_inline void	165	ecb_inline void
97	cbor_init (CBOR *cbor)	166	cbor_init (CBOR *cbor)
98	{	167	{
…		…
102		171
103	ecb_inline void	172	ecb_inline void
104	cbor_free (CBOR *cbor)	173	cbor_free (CBOR *cbor)
105	{	174	{
106	SvREFCNT_dec (cbor->filter);	175	SvREFCNT_dec (cbor->filter);
		176	SvREFCNT_dec (cbor->incr_count);
107	}	177	}
108		178
109	/////////////////////////////////////////////////////////////////////////////	179	/////////////////////////////////////////////////////////////////////////////
110	// utility functions	180	// utility functions
111		181
…		…
134	#endif	204	#endif
135	}	205	}
136	}	206	}
137		207
138	// minimum length of a string to be registered for stringref	208	// minimum length of a string to be registered for stringref
139	ecb_inline int	209	ecb_inline STRLEN
140	minimum_string_length (UV idx)	210	minimum_string_length (UV idx)
141	{	211	{
142	return idx > 23	212	return idx <= 23 ? 3
143	? idx > 0xffU	213	: idx <= 0xffU ? 4
144	? idx > 0xffffU	214	: idx <= 0xffffU ? 5
145	? idx > 0xffffffffU	215	: idx <= 0xffffffffU ? 7
146	? 11	216	: 11;
147	: 7
148	: 5
149	: 4
150	: 3;
151	}	217	}
152		218
153	/////////////////////////////////////////////////////////////////////////////	219	/////////////////////////////////////////////////////////////////////////////
154	// encoder	220	// encoder
155		221
…		…
168	} enc_t;	234	} enc_t;
169		235
170	ecb_inline void	236	ecb_inline void
171	need (enc_t *enc, STRLEN len)	237	need (enc_t *enc, STRLEN len)
172	{	238	{
173	if (ecb_expect_false (enc->cur + len >= enc->end))	239	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
174	{	240	{
175	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	241	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
176	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	242	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
177	enc->cur = SvPVX (enc->sv) + cur;	243	enc->cur = SvPVX (enc->sv) + cur;
178	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	244	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
179	}	245	}
180	}	246	}
181		247
		248	static void encode_sv (enc_t *enc, SV *sv);
		249
182	ecb_inline void	250	ecb_inline void
183	encode_ch (enc_t *enc, char ch)	251	encode_ch (enc_t *enc, char ch)
184	{	252	{
185	need (enc, 1);	253	need (enc, 1);
186	*enc->cur++ = ch;	254	*enc->cur++ = ch;
187	}	255	}
188		256
		257	// used for tags, intregers, element counts and so on
189	static void	258	static void
190	encode_uint (enc_t *enc, int major, UV len)	259	encode_uint (enc_t *enc, int major, UV len)
191	{	260	{
192	need (enc, 9);	261	need (enc, 9);
193		262
194	if (len < 24)	263	if (ecb_expect_true (len < LENGTH_EXT1))
195	*enc->cur++ = major | len;	264	*enc->cur++ = major | len;
196	else if (len <= 0xff)	265	else if (ecb_expect_true (len <= 0xffU))
197	{	266	{
198	*enc->cur++ = major | 24;	267	*enc->cur++ = major | LENGTH_EXT1;
199	*enc->cur++ = len;	268	*enc->cur++ = len;
200	}	269	}
201	else if (len <= 0xffff)	270	else if (len <= 0xffffU)
202	{	271	{
203	*enc->cur++ = major | 25;	272	*enc->cur++ = major | LENGTH_EXT2;
204	*enc->cur++ = len >> 8;	273	*enc->cur++ = len >> 8;
205	*enc->cur++ = len;	274	*enc->cur++ = len;
206	}	275	}
207	else if (len <= 0xffffffff)	276	else if (len <= 0xffffffffU)
208	{	277	{
209	*enc->cur++ = major | 26;	278	*enc->cur++ = major | LENGTH_EXT4;
210	*enc->cur++ = len >> 24;	279	*enc->cur++ = len >> 24;
211	*enc->cur++ = len >> 16;	280	*enc->cur++ = len >> 16;
212	*enc->cur++ = len >> 8;	281	*enc->cur++ = len >> 8;
213	*enc->cur++ = len;	282	*enc->cur++ = len;
214	}	283	}
215	else	284	else
216	{	285	{
217	*enc->cur++ = major | 27;	286	*enc->cur++ = major | LENGTH_EXT8;
218	*enc->cur++ = len >> 56;	287	*enc->cur++ = len >> 56;
219	*enc->cur++ = len >> 48;	288	*enc->cur++ = len >> 48;
220	*enc->cur++ = len >> 40;	289	*enc->cur++ = len >> 40;
221	*enc->cur++ = len >> 32;	290	*enc->cur++ = len >> 32;
222	*enc->cur++ = len >> 24;	291	*enc->cur++ = len >> 24;
…		…
224	*enc->cur++ = len >> 8;	293	*enc->cur++ = len >> 8;
225	*enc->cur++ = len;	294	*enc->cur++ = len;
226	}	295	}
227	}	296	}
228		297
		298	// encodes a perl value into a CBOR integer
		299	ecb_inline void
		300	encode_int (enc_t *enc, SV *sv)
		301	{
		302	if (SvIsUV (sv))
		303	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
		304	else if (SvIVX (sv) >= 0)
		305	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
		306	else
		307	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
		308	}
		309
229	ecb_inline void	310	ecb_inline void
230	encode_tag (enc_t *enc, UV tag)	311	encode_tag (enc_t *enc, UV tag)
231	{	312	{
232	encode_uint (enc, 0xc0, tag);	313	encode_uint (enc, MAJOR_TAG, tag);
233	}	314	}
234		315
235	static void	316	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		317	ecb_noinline static void
236	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	318	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
237	{	319	{
		320	STRLEN ulen = len;
		321	U8 *p, *pend = (U8 *)str + len;
		322
		323	for (p = (U8 *)str; p < pend; ++p)
		324	ulen += *p >> 7; // count set high bits
		325
		326	encode_uint (enc, MAJOR_TEXT, ulen);
		327
		328	need (enc, ulen);
		329	for (p = (U8 *)str; p < pend; ++p)
		330	if (*p < 0x80)
		331	*enc->cur++ = *p;
		332	else
		333	{
		334	*enc->cur++ = 0xc0 + (*p >> 6);
		335	*enc->cur++ = 0x80 + (*p & 63);
		336	}
		337	}
		338
		339	ecb_inline void
		340	encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		341	{
		342	if (ecb_expect_false (upgrade_utf8))
		343	if (!utf8)
		344	{
		345	encode_str_utf8 (enc, utf8, str, len);
		346	return;
		347	}
		348
		349	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
		350	need (enc, len);
		351	memcpy (enc->cur, str, len);
		352	enc->cur += len;
		353	}
		354
		355	ecb_inline void
		356	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		357	{
238	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF))	358	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
239	{	359	{
240	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	360	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
241		361
242	if (SvOK (*svp))	362	if (SvOK (*svp))
243	{	363	{
244	// already registered, use stringref	364	// already registered, use stringref
245	encode_tag (enc, CBOR_TAG_STRINGREF);	365	encode_tag (enc, CBOR_TAG_STRINGREF);
246	encode_uint (enc, 0x00, SvUV (*svp));	366	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
247	return;	367	return;
248	}	368	}
249	else if (len >= minimum_string_length (enc->stringref_idx))	369	else if (len >= minimum_string_length (enc->stringref_idx))
250	{	370	{
251	// register only	371	// register only
252	sv_setuv (*svp, enc->stringref_idx);	372	sv_setuv (*svp, enc->stringref_idx);
253	++enc->stringref_idx;	373	++enc->stringref_idx;
254	}	374	}
255	}	375	}
256		376
257	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	377	encode_str (enc, upgrade_utf8, utf8, str, len);
		378	}
		379
		380	ecb_inline void
		381	encode_float16 (enc_t *enc, NV nv)
		382	{
258	need (enc, len);	383	need (enc, 1+2);
		384
		385	*enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
		386
		387	uint16_t fp = ecb_float_to_binary16 (nv);
		388
		389	if (!ecb_big_endian ())
		390	fp = ecb_bswap16 (fp);
		391
259	memcpy (enc->cur, str, len);	392	memcpy (enc->cur, &fp, 2);
260	enc->cur += len;	393	enc->cur += 2;
261	}	394	}
262		395
263	static void encode_sv (enc_t *enc, SV *sv);	396	ecb_inline void
		397	encode_float32 (enc_t *enc, NV nv)
		398	{
		399	need (enc, 1+4);
		400
		401	*enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
		402
		403	uint32_t fp = ecb_float_to_binary32 (nv);
		404
		405	if (!ecb_big_endian ())
		406	fp = ecb_bswap32 (fp);
		407
		408	memcpy (enc->cur, &fp, 4);
		409	enc->cur += 4;
		410	}
		411
		412	ecb_inline void
		413	encode_float64 (enc_t *enc, NV nv)
		414	{
		415	need (enc, 1+8);
		416
		417	*enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
		418
		419	uint64_t fp = ecb_double_to_binary64 (nv);
		420
		421	if (!ecb_big_endian ())
		422	fp = ecb_bswap64 (fp);
		423
		424	memcpy (enc->cur, &fp, 8);
		425	enc->cur += 8;
		426	}
		427
		428	ecb_inline void
		429	encode_bool (enc_t *enc, int istrue)
		430	{
		431	encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
		432	}
		433
		434	// encodes an arrayref containing key-value pairs as CBOR map
		435	ecb_inline void
		436	encode_array_as_map (enc_t *enc, SV *sv)
		437	{
		438	if (enc->depth >= enc->cbor.max_depth)
		439	croak (ERR_NESTING_EXCEEDED);
		440
		441	++enc->depth;
		442
		443	// as_map does error checking for us, but we re-check in case
		444	// things have changed.
		445
		446	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
		447	croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
		448
		449	AV *av = (AV *)SvRV (sv);
		450	int i, len = av_len (av);
		451
		452	if (!(len & 1))
		453	croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
		454
		455	encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
		456
		457	for (i = 0; i <= len; ++i)
		458	{
		459	SV **svp = av_fetch (av, i, 0);
		460	encode_sv (enc, svp ? *svp : &PL_sv_undef);
		461	}
		462
		463	--enc->depth;
		464	}
		465
		466	ecb_inline void
		467	encode_forced (enc_t *enc, UV type, SV *sv)
		468	{
		469	switch (type)
		470	{
		471	case AS_CBOR:
		472	{
		473	STRLEN len;
		474	char *str = SvPVbyte (sv, len);
		475
		476	need (enc, len);
		477	memcpy (enc->cur, str, len);
		478	enc->cur += len;
		479	}
		480	break;
		481
		482	case AS_BYTES:
		483	{
		484	STRLEN len;
		485	char *str = SvPVbyte (sv, len);
		486	encode_strref (enc, 0, 0, str, len);
		487	}
		488	break;
		489
		490	case AS_TEXT:
		491	{
		492	STRLEN len;
		493	char *str = SvPVutf8 (sv, len);
		494	encode_strref (enc, 1, 1, str, len);
		495	}
		496	break;
		497
		498	case AS_INT: encode_int (enc, sv); break;
		499
		500	case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
		501	case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
		502	case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
		503
		504	case AS_MAP: encode_array_as_map (enc, sv); break;
		505
		506	default:
		507	croak ("encountered malformed CBOR::XS::Tagged object");
		508	}
		509	}
264		510
265	static void	511	static void
266	encode_av (enc_t *enc, AV *av)	512	encode_av (enc_t *enc, AV *av)
267	{	513	{
268	int i, len = av_len (av);	514	int i, len = av_len (av);
…		…
270	if (enc->depth >= enc->cbor.max_depth)	516	if (enc->depth >= enc->cbor.max_depth)
271	croak (ERR_NESTING_EXCEEDED);	517	croak (ERR_NESTING_EXCEEDED);
272		518
273	++enc->depth;	519	++enc->depth;
274		520
275	encode_uint (enc, 0x80, len + 1);	521	encode_uint (enc, MAJOR_ARRAY, len + 1);
276		522
		523	if (ecb_expect_false (SvMAGICAL (av)))
277	for (i = 0; i <= len; ++i)	524	for (i = 0; i <= len; ++i)
278	{	525	{
279	SV **svp = av_fetch (av, i, 0);	526	SV **svp = av_fetch (av, i, 0);
280	encode_sv (enc, svp ? *svp : &PL_sv_undef);	527	encode_sv (enc, svp ? *svp : &PL_sv_undef);
281	}	528	}
		529	else
		530	for (i = 0; i <= len; ++i)
		531	{
		532	SV *sv = AvARRAY (av)[i];
		533	encode_sv (enc, sv ? sv : &PL_sv_undef);
		534	}
282		535
283	--enc->depth;	536	--enc->depth;
284	}	537	}
285		538
286	static void	539	static void
…		…
294	++enc->depth;	547	++enc->depth;
295		548
296	int pairs = hv_iterinit (hv);	549	int pairs = hv_iterinit (hv);
297	int mg = SvMAGICAL (hv);	550	int mg = SvMAGICAL (hv);
298		551
299	if (mg)	552	if (ecb_expect_false (mg))
300	encode_ch (enc, 0xa0 | 31);	553	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
301	else	554	else
302	encode_uint (enc, 0xa0, pairs);	555	encode_uint (enc, MAJOR_MAP, pairs);
303		556
304	while ((he = hv_iternext (hv)))	557	while ((he = hv_iternext (hv)))
305	{	558	{
306	if (HeKLEN (he) == HEf_SVKEY)	559	if (HeKLEN (he) == HEf_SVKEY)
307	encode_sv (enc, HeSVKEY (he));	560	encode_sv (enc, HeSVKEY (he));
308	else	561	else
309	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	562	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
310		563
311	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	564	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
312	}	565	}
313		566
314	if (mg)	567	if (ecb_expect_false (mg))
315	encode_ch (enc, 0xe0 | 31);	568	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
316		569
317	--enc->depth;	570	--enc->depth;
318	}	571	}
319		572
320	// encode objects, arrays and special \0=false and \1=true values.	573	// encode objects, arrays and special \0=false and \1=true values.
321	static void	574	static void
322	encode_rv (enc_t *enc, SV *sv)	575	encode_rv (enc_t *enc, SV *sv)
323	{	576	{
324	SvGETMAGIC (sv);	577	SvGETMAGIC (sv);
325
326	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
327	&& ecb_expect_false (SvREFCNT (sv) > 1))
328	{
329	if (!enc->shareable)
330	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
331
332	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
333
334	if (SvOK (*svp))
335	{
336	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
337	encode_uint (enc, 0x00, SvUV (*svp));
338	return;
339	}
340	else
341	{
342	sv_setuv (*svp, enc->shareable_idx);
343	++enc->shareable_idx;
344	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
345	}
346	}
347		578
348	svtype svt = SvTYPE (sv);	579	svtype svt = SvTYPE (sv);
349		580
350	if (ecb_expect_false (SvOBJECT (sv)))	581	if (ecb_expect_false (SvOBJECT (sv)))
351	{	582	{
…		…
358	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	589	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
359	? cbor_tagged_stash	590	? cbor_tagged_stash
360	: gv_stashpv ("CBOR::XS::Tagged" , 1);	591	: gv_stashpv ("CBOR::XS::Tagged" , 1);
361		592
362	HV *stash = SvSTASH (sv);	593	HV *stash = SvSTASH (sv);
363	GV *method;
364		594
365	if (stash == boolean_stash)	595	if (stash == boolean_stash)
366	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	596	{
		597	encode_bool (enc, SvIV (sv));
		598	return;
		599	}
367	else if (stash == error_stash)	600	else if (stash == error_stash)
368	encode_ch (enc, 0xe0 | 23);	601	{
		602	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		603	return;
		604	}
369	else if (stash == tagged_stash)	605	else if (stash == tagged_stash)
370	{	606	{
371	if (svt != SVt_PVAV)	607	if (svt != SVt_PVAV)
372	croak ("encountered CBOR::XS::Tagged object that isn't an array");	608	croak ("encountered CBOR::XS::Tagged object that isn't an array");
373		609
		610	switch (av_len ((AV *)sv))
		611	{
		612	case 2-1:
		613	// actually a tagged value
374	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	614	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
375	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));	615	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
		616	break;
		617
		618	case 3-1:
		619	// a forced type [value, type, undef]
		620	encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
		621	break;
		622
		623	default:
		624	croak ("encountered malformed CBOR::XS::Tagged object");
		625	}
		626
		627	return;
		628	}
		629	}
		630
		631	if (ecb_expect_false (SvREFCNT (sv) > 1)
		632	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		633	{
		634	if (ecb_expect_false (!enc->shareable))
		635	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
		636
		637	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
		638
		639	if (SvOK (*svp))
376	}	640	{
		641	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		642	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		643	return;
		644	}
		645	else
		646	{
		647	sv_setuv (*svp, enc->shareable_idx);
		648	++enc->shareable_idx;
		649	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		650	}
		651	}
		652
		653	if (ecb_expect_false (SvOBJECT (sv)))
		654	{
		655	HV *stash = SvSTASH (sv);
		656	GV *method;
		657
		658	if (enc->cbor.flags & F_FORBID_OBJECTS)
		659	croak ("encountered object '%s', but forbid_objects is enabled",
		660	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
377	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	661	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
378	{	662	{
379	dSP;	663	dSP;
380		664
381	ENTER; SAVETMPS; PUSHMARK (SP);	665	ENTER; SAVETMPS;
		666	PUSHMARK (SP);
382	// we re-bless the reference to get overload and other niceties right	667	// we re-bless the reference to get overload and other niceties right
383	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	668	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
384		669
385	PUTBACK;	670	PUTBACK;
386	// G_SCALAR ensures that return value is 1	671	// G_SCALAR ensures that return value is 1
…		…
399	}	684	}
400	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	685	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
401	{	686	{
402	dSP;	687	dSP;
403		688
404	ENTER; SAVETMPS; PUSHMARK (SP);	689	ENTER; SAVETMPS;
		690	PUSHMARK (SP);
405	EXTEND (SP, 2);	691	EXTEND (SP, 2);
406	// we re-bless the reference to get overload and other niceties right	692	// we re-bless the reference to get overload and other niceties right
407	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	693	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
408	PUSHs (sv_cbor);	694	PUSHs (sv_cbor);
409		695
…		…
414	// catch this surprisingly common error	700	// catch this surprisingly common error
415	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	701	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
416	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	702	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
417		703
418	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	704	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
419	encode_uint (enc, 0x80, count + 1);	705	encode_uint (enc, MAJOR_ARRAY, count + 1);
420	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	706	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
421		707
422	while (count)	708	{
		709	int i;
		710
		711	for (i = 0; i < count; ++i)
423	encode_sv (enc, SP[1 - count--]);	712	encode_sv (enc, SP[i + 1 - count]);
		713
		714	SP -= count;
		715	}
424		716
425	PUTBACK;	717	PUTBACK;
426		718
427	FREETMPS; LEAVE;	719	FREETMPS; LEAVE;
428	}	720	}
…		…
446	{	738	{
447	double nv = SvNVX (sv);	739	double nv = SvNVX (sv);
448		740
449	need (enc, 9);	741	need (enc, 9);
450		742
451	if (ecb_expect_false (nv == (U32)nv))	743	if (ecb_expect_false (nv == (NV)(U32)nv))
452	encode_uint (enc, 0x00, (U32)nv);	744	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
453	//TODO: maybe I32?	745	//TODO: maybe I32?
454	else if (ecb_expect_false (nv == (float)nv))	746	else if (ecb_expect_false (nv == (float)nv))
455	{	747	encode_float32 (enc, nv);
456	uint32_t fp = ecb_float_to_binary32 (nv);
457
458	*enc->cur++ = 0xe0 | 26;
459
460	if (!ecb_big_endian ())
461	fp = ecb_bswap32 (fp);
462
463	memcpy (enc->cur, &fp, 4);
464	enc->cur += 4;
465	}
466	else	748	else
467	{	749	encode_float64 (enc, nv);
468	uint64_t fp = ecb_double_to_binary64 (nv);
469
470	*enc->cur++ = 0xe0 | 27;
471
472	if (!ecb_big_endian ())
473	fp = ecb_bswap64 (fp);
474
475	memcpy (enc->cur, &fp, 8);
476	enc->cur += 8;
477	}
478	}	750	}
479		751
480	static void	752	static void
481	encode_sv (enc_t *enc, SV *sv)	753	encode_sv (enc_t *enc, SV *sv)
482	{	754	{
…		…
484		756
485	if (SvPOKp (sv))	757	if (SvPOKp (sv))
486	{	758	{
487	STRLEN len;	759	STRLEN len;
488	char *str = SvPV (sv, len);	760	char *str = SvPV (sv, len);
489	encode_str (enc, SvUTF8 (sv), str, len);	761	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
490	}	762	}
491	else if (SvNOKp (sv))	763	else if (SvNOKp (sv))
492	encode_nv (enc, sv);	764	encode_nv (enc, sv);
493	else if (SvIOKp (sv))	765	else if (SvIOKp (sv))
494	{	766	encode_int (enc, sv);
495	if (SvIsUV (sv))
496	encode_uint (enc, 0x00, SvUVX (sv));
497	else if (SvIVX (sv) >= 0)
498	encode_uint (enc, 0x00, SvIVX (sv));
499	else
500	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
501	}
502	else if (SvROK (sv))	767	else if (SvROK (sv))
503	encode_rv (enc, SvRV (sv));	768	encode_rv (enc, SvRV (sv));
504	else if (!SvOK (sv))	769	else if (!SvOK (sv))
505	encode_ch (enc, 0xe0 | 22);	770	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
506	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	771	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
507	encode_ch (enc, 0xe0 | 23);	772	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
508	else	773	else
509	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	774	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
510	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	775	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
511	}	776	}
512		777
513	static SV *	778	static SV *
514	encode_cbor (SV *scalar, CBOR *cbor)	779	encode_cbor (SV *scalar, CBOR *cbor)
515	{	780	{
516	enc_t enc = { };	781	enc_t enc = { 0 };
517		782
518	enc.cbor = *cbor;	783	enc.cbor = *cbor;
519	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	784	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
520	enc.cur = SvPVX (enc.sv);	785	enc.cur = SvPVX (enc.sv);
521	enc.end = SvEND (enc.sv);	786	enc.end = SvEND (enc.sv);
522		787
523	SvPOK_only (enc.sv);	788	SvPOK_only (enc.sv);
524		789
525	if (cbor->flags & F_ALLOW_STRINGREF)	790	if (cbor->flags & F_PACK_STRINGS)
526	{	791	{
527	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	792	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
528	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());	793	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
529	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());	794	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
530	}	795	}
…		…
553	U32 depth; // recursion depth	818	U32 depth; // recursion depth
554	U32 maxdepth; // recursion depth limit	819	U32 maxdepth; // recursion depth limit
555	AV *shareable;	820	AV *shareable;
556	AV *stringref;	821	AV *stringref;
557	SV *decode_tagged;	822	SV *decode_tagged;
		823	SV *err_sv; // optional sv for error, needs to be freed
558	} dec_t;	824	} dec_t;
559		825
560	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	826	// set dec->err to ERRSV
		827	ecb_cold static void
		828	err_errsv (dec_t *dec)
		829	{
		830	if (!dec->err)
		831	{
		832	dec->err_sv = newSVsv (ERRSV);
561		833
562	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	834	// chop off the trailing \n
		835	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		836	*SvEND (dec->err_sv) = 0;
563		837
		838	dec->err = SvPVutf8_nolen (dec->err_sv);
		839	}
		840	}
		841
		842	// the following functions are used to reduce code size and help the compiler to optimise
		843	ecb_cold static void
		844	err_set (dec_t *dec, const char *reason)
		845	{
		846	if (!dec->err)
		847	dec->err = reason;
		848	}
		849
		850	ecb_cold static void
		851	err_unexpected_end (dec_t *dec)
		852	{
		853	err_set (dec, "unexpected end of CBOR data");
		854	}
		855
		856	#define ERR_DO(do) SB do; goto fail; SE
		857	#define ERR(reason) ERR_DO (err_set (dec, reason))
		858	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		859
		860	#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
		861
564	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	862	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
565	#define DEC_DEC_DEPTH --dec->depth	863	#define DEC_DEC_DEPTH --dec->depth
566		864
567	static UV	865	static UV
568	decode_uint (dec_t *dec)	866	decode_uint (dec_t *dec)
569	{	867	{
570	switch (*dec->cur & 31)	868	U8 m = *dec->cur & MINOR_MASK;
571	{	869	++dec->cur;
572	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
573	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
574	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
575	return *dec->cur++ & 31;
576		870
577	case 24:	871	if (ecb_expect_true (m < LENGTH_EXT1))
		872	return m;
		873	else if (ecb_expect_true (m == LENGTH_EXT1))
		874	{
578	WANT (2);	875	WANT (1);
579	dec->cur += 2;	876	dec->cur += 1;
580	return dec->cur[-1];	877	return dec->cur[-1];
581		878	}
582	case 25:	879	else if (ecb_expect_true (m == LENGTH_EXT2))
		880	{
583	WANT (3);	881	WANT (2);
584	dec->cur += 3;	882	dec->cur += 2;
585	return (((UV)dec->cur[-2]) << 8)	883	return (((UV)dec->cur[-2]) << 8)
586	| ((UV)dec->cur[-1]);	884	| ((UV)dec->cur[-1]);
587		885	}
588	case 26:	886	else if (ecb_expect_true (m == LENGTH_EXT4))
		887	{
589	WANT (5);	888	WANT (4);
590	dec->cur += 5;	889	dec->cur += 4;
591	return (((UV)dec->cur[-4]) << 24)	890	return (((UV)dec->cur[-4]) << 24)
592	| (((UV)dec->cur[-3]) << 16)	891	| (((UV)dec->cur[-3]) << 16)
593	| (((UV)dec->cur[-2]) << 8)	892	| (((UV)dec->cur[-2]) << 8)
594	| ((UV)dec->cur[-1]);	893	| ((UV)dec->cur[-1]);
595		894	}
596	case 27:	895	else if (ecb_expect_true (m == LENGTH_EXT8))
		896	{
597	WANT (9);	897	WANT (8);
598	dec->cur += 9;	898	dec->cur += 8;
		899
		900	return
		901	#if UVSIZE < 8
		902	0
		903	#else
599	return (((UV)dec->cur[-8]) << 56)	904	(((UV)dec->cur[-8]) << 56)
600	| (((UV)dec->cur[-7]) << 48)	905	| (((UV)dec->cur[-7]) << 48)
601	| (((UV)dec->cur[-6]) << 40)	906	| (((UV)dec->cur[-6]) << 40)
602	| (((UV)dec->cur[-5]) << 32)	907	| (((UV)dec->cur[-5]) << 32)
		908	#endif
603	| (((UV)dec->cur[-4]) << 24)	909	| (((UV)dec->cur[-4]) << 24)
604	| (((UV)dec->cur[-3]) << 16)	910	| (((UV)dec->cur[-3]) << 16)
605	| (((UV)dec->cur[-2]) << 8)	911	| (((UV)dec->cur[-2]) << 8)
606	| ((UV)dec->cur[-1]);	912	| ((UV)dec->cur[-1]);
607		913	}
608	default:	914	else
609	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	915	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
610	}
611		916
612	fail:	917	fail:
613	return 0;	918	return 0;
614	}	919	}
615		920
…		…
620	{	925	{
621	AV *av = newAV ();	926	AV *av = newAV ();
622		927
623	DEC_INC_DEPTH;	928	DEC_INC_DEPTH;
624		929
625	if ((*dec->cur & 31) == 31)	930	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
626	{	931	{
627	++dec->cur;	932	++dec->cur;
628		933
629	for (;;)	934	for (;;)
630	{	935	{
631	WANT (1);	936	WANT (1);
632		937
633	if (*dec->cur == (0xe0 | 31))	938	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
634	{	939	{
635	++dec->cur;	940	++dec->cur;
636	break;	941	break;
637	}	942	}
638		943
639	av_push (av, decode_sv (dec));	944	av_push (av, decode_sv (dec));
640	}	945	}
641	}	946	}
642	else	947	else
643	{	948	{
644	int i, len = decode_uint (dec);	949	UV i, len = decode_uint (dec);
645		950
		951	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
646	av_fill (av, len - 1);	952	av_fill (av, len - 1);
647		953
648	for (i = 0; i < len; ++i)	954	for (i = 0; i < len; ++i)
649	AvARRAY (av)[i] = decode_sv (dec);	955	AvARRAY (av)[i] = decode_sv (dec);
650	}	956	}
651		957
652	DEC_DEC_DEPTH;	958	DEC_DEC_DEPTH;
653	return newRV_noinc ((SV *)av);	959	return newRV_noinc ((SV *)av);
654		960
655	fail:	961	fail:
656	SvREFCNT_dec (av);	962	SvREFCNT_dec_NN (av);
657	DEC_DEC_DEPTH;	963	DEC_DEC_DEPTH;
658	return &PL_sv_undef;	964	return &PL_sv_undef;
659	}	965	}
660		966
661	static void	967	static void
…		…
663	{	969	{
664	// for speed reasons, we specialcase single-string	970	// for speed reasons, we specialcase single-string
665	// byte or utf-8 strings as keys, but only when !stringref	971	// byte or utf-8 strings as keys, but only when !stringref
666		972
667	if (ecb_expect_true (!dec->stringref))	973	if (ecb_expect_true (!dec->stringref))
668	if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)	974	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
669	{	975	{
670	I32 len = decode_uint (dec);	976	STRLEN len = decode_uint (dec);
671	char *key = (char *)dec->cur;	977	char *key = (char *)dec->cur;
672		978
		979	WANT (len);
673	dec->cur += len;	980	dec->cur += len;
674		981
675	if (ecb_expect_false (dec->stringref))
676	av_push (dec->stringref, newSVpvn (key, len));
677
678	hv_store (hv, key, len, decode_sv (dec), 0);	982	hv_store (hv, key, len, decode_sv (dec), 0);
679		983
680	return;	984	return;
681	}	985	}
682	else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)	986	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
683	{	987	{
684	I32 len = decode_uint (dec);	988	STRLEN len = decode_uint (dec);
685	char *key = (char *)dec->cur;	989	char *key = (char *)dec->cur;
686		990
		991	WANT (len);
687	dec->cur += len;	992	dec->cur += len;
688		993
689	if (ecb_expect_false (dec->stringref))	994	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
690	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	995	if (!cbor_is_utf8_string ((U8 *)key, len))
		996	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
691		997
692	hv_store (hv, key, -len, decode_sv (dec), 0);	998	hv_store (hv, key, -len, decode_sv (dec), 0);
693		999
694	return;	1000	return;
695	}	1001	}
696		1002
697	SV *k = decode_sv (dec);	1003	SV *k = decode_sv (dec);
698	SV *v = decode_sv (dec);	1004	SV *v = decode_sv (dec);
699		1005
		1006	// we leak memory if uncaught exceptions are thrown by random magical
		1007	// methods, and this is hopefully the only place where it can happen,
		1008	// so if there is a chance of an exception, take the very slow path.
		1009	// since catching exceptions is "undocumented/internal/forbidden" by
		1010	// the new p5p powers, we need to call out to a perl function :/
		1011	if (ecb_expect_false (SvAMAGIC (k)))
		1012	{
		1013	dSP;
		1014
		1015	ENTER; SAVETMPS;
		1016	PUSHMARK (SP);
		1017	EXTEND (SP, 3);
		1018	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		1019	PUSHs (sv_2mortal (k));
		1020	PUSHs (sv_2mortal (v));
		1021
		1022	PUTBACK;
		1023	call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
		1024	SPAGAIN;
		1025
		1026	FREETMPS; LEAVE;
		1027
		1028	if (SvTRUE (ERRSV))
		1029	ERR_ERRSV;
		1030
		1031	return;
		1032	}
		1033
700	hv_store_ent (hv, k, v, 0);	1034	hv_store_ent (hv, k, v, 0);
701	SvREFCNT_dec (k);	1035	SvREFCNT_dec_NN (k);
		1036
		1037	fail:
		1038	;
702	}	1039	}
703		1040
704	static SV *	1041	static SV *
705	decode_hv (dec_t *dec)	1042	decode_hv (dec_t *dec)
706	{	1043	{
707	HV *hv = newHV ();	1044	HV *hv = newHV ();
708		1045
709	DEC_INC_DEPTH;	1046	DEC_INC_DEPTH;
710		1047
711	if ((*dec->cur & 31) == 31)	1048	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
712	{	1049	{
713	++dec->cur;	1050	++dec->cur;
714		1051
715	for (;;)	1052	for (;;)
716	{	1053	{
717	WANT (1);	1054	WANT (1);
718		1055
719	if (*dec->cur == (0xe0 | 31))	1056	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
720	{	1057	{
721	++dec->cur;	1058	++dec->cur;
722	break;	1059	break;
723	}	1060	}
724		1061
725	decode_he (dec, hv);	1062	decode_he (dec, hv);
726	}	1063	}
727	}	1064	}
728	else	1065	else
729	{	1066	{
730	int pairs = decode_uint (dec);	1067	UV pairs = decode_uint (dec);
		1068
		1069	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
731		1070
732	while (pairs--)	1071	while (pairs--)
733	decode_he (dec, hv);	1072	decode_he (dec, hv);
734	}	1073	}
735		1074
736	DEC_DEC_DEPTH;	1075	DEC_DEC_DEPTH;
737	return newRV_noinc ((SV *)hv);	1076	return newRV_noinc ((SV *)hv);
738		1077
739	fail:	1078	fail:
740	SvREFCNT_dec (hv);	1079	SvREFCNT_dec_NN (hv);
741	DEC_DEC_DEPTH;	1080	DEC_DEC_DEPTH;
742	return &PL_sv_undef;	1081	return &PL_sv_undef;
743	}	1082	}
744		1083
745	static SV *	1084	static SV *
746	decode_str (dec_t *dec, int utf8)	1085	decode_str (dec_t *dec, int utf8)
747	{	1086	{
748	SV *sv = 0;	1087	SV *sv = 0;
749		1088
750	if ((*dec->cur & 31) == 31)	1089	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
751	{	1090	{
		1091	// indefinite length strings
752	++dec->cur;	1092	++dec->cur;
753		1093
		1094	U8 major = *dec->cur & MAJOR_MISC;
		1095
754	sv = newSVpvn ("", 0);	1096	sv = newSVpvn ("", 0);
755		1097
756	// not very fast, and certainly not robust against illegal input
757	for (;;)	1098	for (;;)
758	{	1099	{
759	WANT (1);	1100	WANT (1);
760		1101
761	if (*dec->cur == (0xe0 | 31))	1102	if ((*dec->cur - major) > LENGTH_EXT8)
		1103	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
762	{	1104	{
763	++dec->cur;	1105	++dec->cur;
764	break;	1106	break;
765	}	1107	}
		1108	else
		1109	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
766		1110
767	sv_catsv (sv, decode_sv (dec));	1111	STRLEN len = decode_uint (dec);
		1112
		1113	WANT (len);
		1114	sv_catpvn (sv, dec->cur, len);
		1115	dec->cur += len;
768	}	1116	}
769	}	1117	}
770	else	1118	else
771	{	1119	{
772	STRLEN len = decode_uint (dec);	1120	STRLEN len = decode_uint (dec);
…		…
779	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	1127	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
780	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	1128	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
781	}	1129	}
782		1130
783	if (utf8)	1131	if (utf8)
		1132	{
		1133	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		1134	if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv)))
		1135	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		1136
784	SvUTF8_on (sv);	1137	SvUTF8_on (sv);
		1138	}
785		1139
786	return sv;	1140	return sv;
787		1141
788	fail:	1142	fail:
789	SvREFCNT_dec (sv);	1143	SvREFCNT_dec (sv);
…		…
808	sv = newRV_noinc (decode_sv (dec));	1162	sv = newRV_noinc (decode_sv (dec));
809	break;	1163	break;
810		1164
811	case CBOR_TAG_STRINGREF_NAMESPACE:	1165	case CBOR_TAG_STRINGREF_NAMESPACE:
812	{	1166	{
		1167	// do not use SAVETMPS/FREETMPS, as these will
		1168	// erase mortalised caches, e.g. "shareable"
813	ENTER; SAVETMPS;	1169	ENTER;
814		1170
815	SAVESPTR (dec->stringref);	1171	SAVESPTR (dec->stringref);
816	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());	1172	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
817		1173
818	sv = decode_sv (dec);	1174	sv = decode_sv (dec);
819		1175
820	FREETMPS; LEAVE;	1176	LEAVE;
821	}	1177	}
822	break;	1178	break;
823		1179
824	case CBOR_TAG_STRINGREF:	1180	case CBOR_TAG_STRINGREF:
825	{	1181	{
826	if ((*dec->cur >> 5) != 0)	1182	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
827	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1183	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
828		1184
829	UV idx = decode_uint (dec);	1185	UV idx = decode_uint (dec);
830		1186
831	if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))	1187	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
832	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");	1188	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
833		1189
834	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1190	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
835	}	1191	}
836	break;	1192	break;
…		…
838	case CBOR_TAG_VALUE_SHAREABLE:	1194	case CBOR_TAG_VALUE_SHAREABLE:
839	{	1195	{
840	if (ecb_expect_false (!dec->shareable))	1196	if (ecb_expect_false (!dec->shareable))
841	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());	1197	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
842		1198
		1199	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1200	{
		1201	// if cycles are allowed, then we store an AV as value
		1202	// while it is being decoded, and gather unresolved
		1203	// references in it, to be re4solved after decoding.
		1204	int idx, i;
843	sv = newSV (0);	1205	AV *av = newAV ();
844	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1206	av_push (dec->shareable, (SV *)av);
		1207	idx = AvFILLp (dec->shareable);
845		1208
846	SV *osv = decode_sv (dec);	1209	sv = decode_sv (dec);
847	sv_setsv (sv, osv);	1210
		1211	// the AV now contains \undef for all unresolved references,
		1212	// so we fix them up here.
		1213	for (i = 0; i <= AvFILLp (av); ++i)
		1214	SvRV_set (AvARRAY (av)[i], SvREFCNT_inc_NN (SvRV (sv)));
		1215
		1216	// now replace the AV by a reference to the completed value
848	SvREFCNT_dec_NN (osv);	1217	SvREFCNT_dec_NN ((SV *)av);
		1218	AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
		1219	}
		1220	else
		1221	{
		1222	av_push (dec->shareable, &PL_sv_undef);
		1223	int idx = AvFILLp (dec->shareable);
		1224	sv = decode_sv (dec);
		1225	AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
		1226	}
849	}	1227	}
850	break;	1228	break;
851		1229
852	case CBOR_TAG_VALUE_SHAREDREF:	1230	case CBOR_TAG_VALUE_SHAREDREF:
853	{	1231	{
854	if ((*dec->cur >> 5) != 0)	1232	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
855	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1233	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
856		1234
857	UV idx = decode_uint (dec);	1235	UV idx = decode_uint (dec);
858		1236
859	if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))	1237	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
860	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1238	ERR ("corrupted CBOR data (sharedref index out of bounds)");
861		1239
862	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1240	sv = AvARRAY (dec->shareable)[idx];
		1241
		1242	// reference to cycle, we create a new \undef and use that, and also
		1243	// registerr it in the AV for later fixing
		1244	if (SvTYPE (sv) == SVt_PVAV)
		1245	{
		1246	AV *av = (AV *)sv;
		1247	sv = newRV_noinc (&PL_sv_undef);
		1248	av_push (av, sv);
		1249	}
		1250	else if (sv == &PL_sv_undef) // not yet decoded, but cycles not allowed
		1251	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
		1252	else // we decoded the object earlier, no cycle
		1253	sv = newSVsv (sv);
863	}	1254	}
864	break;	1255	break;
865		1256
866	case CBOR_TAG_PERL_OBJECT:	1257	case CBOR_TAG_PERL_OBJECT:
867	{	1258	{
		1259	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1260	goto filter;
		1261
868	sv = decode_sv (dec);	1262	sv = decode_sv (dec);
869		1263
870	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1264	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
871	ERR ("corrupted CBOR data (non-array perl object)");	1265	ERR ("corrupted CBOR data (non-array perl object)");
872		1266
…		…
882	if (!method)	1276	if (!method)
883	ERR ("cannot decode perl-object (package does not have a THAW method)");	1277	ERR ("cannot decode perl-object (package does not have a THAW method)");
884		1278
885	dSP;	1279	dSP;
886		1280
887	ENTER; SAVETMPS; PUSHMARK (SP);	1281	ENTER; SAVETMPS;
		1282	PUSHMARK (SP);
888	EXTEND (SP, len + 1);	1283	EXTEND (SP, len + 1);
889	// we re-bless the reference to get overload and other niceties right	1284	// we re-bless the reference to get overload and other niceties right
890	PUSHs (*av_fetch (av, 0, 1));	1285	PUSHs (*av_fetch (av, 0, 1));
891	PUSHs (sv_cbor);	1286	PUSHs (sv_cbor);
892		1287
…		…
900	SPAGAIN;	1295	SPAGAIN;
901		1296
902	if (SvTRUE (ERRSV))	1297	if (SvTRUE (ERRSV))
903	{	1298	{
904	FREETMPS; LEAVE;	1299	FREETMPS; LEAVE;
905	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1300	ERR_ERRSV;
906	}	1301	}
907		1302
908	SvREFCNT_dec (sv);	1303	SvREFCNT_dec_NN (sv);
909	sv = SvREFCNT_inc (POPs);	1304	sv = SvREFCNT_inc (POPs);
910		1305
911	PUTBACK;	1306	PUTBACK;
912		1307
913	FREETMPS; LEAVE;	1308	FREETMPS; LEAVE;
914	}	1309	}
915	break;	1310	break;
916		1311
917	default:	1312	default:
		1313	filter:
918	{	1314	{
		1315	SV *tag_sv = newSVuv (tag);
		1316
919	sv = decode_sv (dec);	1317	sv = decode_sv (dec);
920		1318
921	dSP;	1319	dSP;
922	ENTER; SAVETMPS; PUSHMARK (SP);	1320	ENTER; SAVETMPS;
		1321	PUSHMARK (SP);
923	EXTEND (SP, 2);	1322	EXTEND (SP, 2);
924	PUSHs (newSVuv (tag));	1323	PUSHs (tag_sv);
925	PUSHs (sv);	1324	PUSHs (sv);
926		1325
927	PUTBACK;	1326	PUTBACK;
928	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);	1327	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
929	SPAGAIN;	1328	SPAGAIN;
930		1329
931	if (SvTRUE (ERRSV))	1330	if (SvTRUE (ERRSV))
932	{	1331	{
		1332	SvREFCNT_dec_NN (tag_sv);
933	FREETMPS; LEAVE;	1333	FREETMPS; LEAVE;
934	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1334	ERR_ERRSV;
935	}	1335	}
936		1336
937	if (count)	1337	if (count)
938	{	1338	{
		1339	SvREFCNT_dec_NN (tag_sv);
939	SvREFCNT_dec (sv);	1340	SvREFCNT_dec_NN (sv);
940	sv = SvREFCNT_inc (POPs);	1341	sv = SvREFCNT_inc_NN (TOPs);
		1342	SP -= count;
941	}	1343	}
942	else	1344	else
943	{	1345	{
944	AV *av = newAV ();	1346	AV *av = newAV ();
945	av_push (av, newSVuv (tag));	1347	av_push (av, tag_sv);
946	av_push (av, sv);	1348	av_push (av, sv);
947		1349
948	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1350	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
949	? cbor_tagged_stash	1351	? cbor_tagged_stash
950	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1352	: gv_stashpv ("CBOR::XS::Tagged" , 1);
…		…
968	static SV *	1370	static SV *
969	decode_sv (dec_t *dec)	1371	decode_sv (dec_t *dec)
970	{	1372	{
971	WANT (1);	1373	WANT (1);
972		1374
973	switch (*dec->cur >> 5)	1375	switch (*dec->cur >> MAJOR_SHIFT)
974	{	1376	{
975	case 0: // unsigned int	1377	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
976	return newSVuv (decode_uint (dec));	1378	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
977	case 1: // negative int	1379	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
978	return newSViv (-1 - (IV)decode_uint (dec));	1380	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
979	case 2: // octet string	1381	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
980	return decode_str (dec, 0);	1382	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
981	case 3: // utf-8 string	1383	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
982	return decode_str (dec, 1);	1384
983	case 4: // array	1385	case MAJOR_MISC >> MAJOR_SHIFT:
984	return decode_av (dec);
985	case 5: // map
986	return decode_hv (dec);
987	case 6: // tag
988	return decode_tagged (dec);
989	case 7: // misc
990	switch (*dec->cur++ & 31)	1386	switch (*dec->cur++ & MINOR_MASK)
991	{	1387	{
992	case 20:	1388	case SIMPLE_FALSE:
993	#if CBOR_SLOW	1389	#if CBOR_SLOW
994	types_false = get_bool ("Types::Serialiser::false");	1390	types_false = get_bool ("Types::Serialiser::false");
995	#endif	1391	#endif
996	return newSVsv (types_false);	1392	return newSVsv (types_false);
997	case 21:	1393	case SIMPLE_TRUE:
998	#if CBOR_SLOW	1394	#if CBOR_SLOW
999	types_true = get_bool ("Types::Serialiser::true");	1395	types_true = get_bool ("Types::Serialiser::true");
1000	#endif	1396	#endif
1001	return newSVsv (types_true);	1397	return newSVsv (types_true);
1002	case 22:	1398	case SIMPLE_NULL:
1003	return newSVsv (&PL_sv_undef);	1399	return newSVsv (&PL_sv_undef);
1004	case 23:	1400	case SIMPLE_UNDEF:
1005	#if CBOR_SLOW	1401	#if CBOR_SLOW
1006	types_error = get_bool ("Types::Serialiser::error");	1402	types_error = get_bool ("Types::Serialiser::error");
1007	#endif	1403	#endif
1008	return newSVsv (types_error);	1404	return newSVsv (types_error);
1009		1405
1010	case 25:	1406	case MISC_FLOAT16:
1011	{	1407	{
1012	WANT (2);	1408	WANT (2);
1013		1409
1014	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1410	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1015	dec->cur += 2;	1411	dec->cur += 2;
1016		1412
1017	return newSVnv (ecb_binary16_to_float (fp));	1413	return newSVnv (ecb_binary16_to_float (fp));
1018	}	1414	}
1019		1415
1020	case 26:	1416	case MISC_FLOAT32:
1021	{	1417	{
1022	uint32_t fp;	1418	uint32_t fp;
1023	WANT (4);	1419	WANT (4);
1024	memcpy (&fp, dec->cur, 4);	1420	memcpy (&fp, dec->cur, 4);
1025	dec->cur += 4;	1421	dec->cur += 4;
…		…
1028	fp = ecb_bswap32 (fp);	1424	fp = ecb_bswap32 (fp);
1029		1425
1030	return newSVnv (ecb_binary32_to_float (fp));	1426	return newSVnv (ecb_binary32_to_float (fp));
1031	}	1427	}
1032		1428
1033	case 27:	1429	case MISC_FLOAT64:
1034	{	1430	{
1035	uint64_t fp;	1431	uint64_t fp;
1036	WANT (8);	1432	WANT (8);
1037	memcpy (&fp, dec->cur, 8);	1433	memcpy (&fp, dec->cur, 8);
1038	dec->cur += 8;	1434	dec->cur += 8;
…		…
1041	fp = ecb_bswap64 (fp);	1437	fp = ecb_bswap64 (fp);
1042		1438
1043	return newSVnv (ecb_binary64_to_double (fp));	1439	return newSVnv (ecb_binary64_to_double (fp));
1044	}	1440	}
1045		1441
1046	// 0..19 unassigned	1442	// 0..19 unassigned simple
1047	// 24 reserved + unassigned (reserved values are not encodable)	1443	// 24 reserved + unassigned simple (reserved values are not encodable)
		1444	// 28-30 unassigned misc
		1445	// 31 break code
1048	default:	1446	default:
1049	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1447	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1050	}	1448	}
1051		1449
1052	break;	1450	break;
1053	}	1451	}
1054		1452
…		…
1057	}	1455	}
1058		1456
1059	static SV *	1457	static SV *
1060	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1458	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1061	{	1459	{
1062	dec_t dec = { };	1460	dec_t dec = { 0 };
1063	SV *sv;	1461	SV *sv;
1064	STRLEN len;	1462	STRLEN len;
1065	char *data = SvPVbyte (string, len);	1463	char *data = SvPVbyte (string, len);
1066		1464
1067	if (len > cbor->max_size && cbor->max_size)	1465	if (len > cbor->max_size && cbor->max_size)
…		…
1081	if (dec.cur != dec.end && !dec.err)	1479	if (dec.cur != dec.end && !dec.err)
1082	dec.err = "garbage after CBOR object";	1480	dec.err = "garbage after CBOR object";
1083		1481
1084	if (dec.err)	1482	if (dec.err)
1085	{	1483	{
		1484	if (dec.shareable)
		1485	{
		1486	// need to break cyclic links, which would all be in shareable
		1487	int i;
		1488	SV **svp;
		1489
		1490	for (i = av_len (dec.shareable) + 1; i--; )
		1491	if ((svp = av_fetch (dec.shareable, i, 0)))
		1492	sv_setsv (*svp, &PL_sv_undef);
		1493	}
		1494
1086	SvREFCNT_dec (sv);	1495	SvREFCNT_dec_NN (sv);
		1496
		1497	if (dec.err_sv)
		1498	sv_2mortal (dec.err_sv);
		1499
1087	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);	1500	croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur);
1088	}	1501	}
1089		1502
1090	sv = sv_2mortal (sv);	1503	sv = sv_2mortal (sv);
1091		1504
1092	return sv;	1505	return sv;
1093	}	1506	}
1094		1507
		1508	/////////////////////////////////////////////////////////////////////////////
		1509	// incremental parser
		1510
		1511	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1512
		1513	// returns 0 for notyet, 1 for success or error
		1514	static int
		1515	incr_parse (CBOR *self, SV *cborstr)
		1516	{
		1517	STRLEN cur;
		1518	SvPV (cborstr, cur);
		1519
		1520	while (ecb_expect_true (self->incr_need <= cur))
		1521	{
		1522	// table of integer count bytes
		1523	static I8 incr_len[MINOR_MASK + 1] = {
		1524	0, 0, 0, 0, 0, 0, 0, 0,
		1525	0, 0, 0, 0, 0, 0, 0, 0,
		1526	0, 0, 0, 0, 0, 0, 0, 0,
		1527	1, 2, 4, 8,-1,-1,-1,-2
		1528	};
		1529
		1530	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1531	U8 m = *p & MINOR_MASK;
		1532	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1533	I8 ilen = incr_len[m];
		1534
		1535	self->incr_need = self->incr_pos + 1;
		1536
		1537	if (ecb_expect_false (ilen < 0))
		1538	{
		1539	if (m != MINOR_INDEF)
		1540	return 1; // error
		1541
		1542	if (*p == (MAJOR_MISC | MINOR_INDEF))
		1543	{
		1544	if (count >= 0)
		1545	return 1; // error
		1546
		1547	count = 1;
		1548	}
		1549	else
		1550	{
		1551	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1552	count = -1;
		1553	}
		1554	}
		1555	else
		1556	{
		1557	self->incr_need += ilen;
		1558	if (ecb_expect_false (self->incr_need > cur))
		1559	return 0;
		1560
		1561	int major = *p >> MAJOR_SHIFT;
		1562
		1563	switch (major)
		1564	{
		1565	case MAJOR_TAG >> MAJOR_SHIFT:
		1566	++count; // tags merely prefix another value
		1567	break;
		1568
		1569	case MAJOR_BYTES >> MAJOR_SHIFT:
		1570	case MAJOR_TEXT >> MAJOR_SHIFT:
		1571	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1572	case MAJOR_MAP >> MAJOR_SHIFT:
		1573	{
		1574	UV len;
		1575
		1576	if (ecb_expect_false (ilen))
		1577	{
		1578	len = 0;
		1579
		1580	do {
		1581	len = (len << 8) | *++p;
		1582	} while (--ilen);
		1583	}
		1584	else
		1585	len = m;
		1586
		1587	switch (major)
		1588	{
		1589	case MAJOR_BYTES >> MAJOR_SHIFT:
		1590	case MAJOR_TEXT >> MAJOR_SHIFT:
		1591	self->incr_need += len;
		1592	if (ecb_expect_false (self->incr_need > cur))
		1593	return 0;
		1594
		1595	break;
		1596
		1597	case MAJOR_MAP >> MAJOR_SHIFT:
		1598	len <<= 1;
		1599	/* FALLTHROUGH */
		1600	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1601	if (len)
		1602	{
		1603	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1604	count = len + 1;
		1605	}
		1606	break;
		1607	}
		1608	}
		1609	}
		1610	}
		1611
		1612	self->incr_pos = self->incr_need;
		1613
		1614	if (count > 0)
		1615	{
		1616	while (!--count)
		1617	{
		1618	if (!AvFILLp (self->incr_count))
		1619	return 1; // done
		1620
		1621	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1622	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1623	}
		1624
		1625	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1626	}
		1627	}
		1628
		1629	return 0;
		1630	}
		1631
		1632
1095	/////////////////////////////////////////////////////////////////////////////	1633	/////////////////////////////////////////////////////////////////////////////
1096	// XS interface functions	1634	// XS interface functions
1097		1635
1098	MODULE = CBOR::XS PACKAGE = CBOR::XS	1636	MODULE = CBOR::XS PACKAGE = CBOR::XS
1099		1637
…		…
1111		1649
1112	default_filter = newSVpv ("CBOR::XS::default_filter", 0);	1650	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1113		1651
1114	sv_cbor = newSVpv ("CBOR", 0);	1652	sv_cbor = newSVpv ("CBOR", 0);
1115	SvREADONLY_on (sv_cbor);	1653	SvREADONLY_on (sv_cbor);
		1654
		1655	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1116	}	1656	}
1117		1657
1118	PROTOTYPES: DISABLE	1658	PROTOTYPES: DISABLE
1119		1659
1120	void CLONE (...)	1660	void CLONE (...)
…		…
1139	void shrink (CBOR *self, int enable = 1)	1679	void shrink (CBOR *self, int enable = 1)
1140	ALIAS:	1680	ALIAS:
1141	shrink = F_SHRINK	1681	shrink = F_SHRINK
1142	allow_unknown = F_ALLOW_UNKNOWN	1682	allow_unknown = F_ALLOW_UNKNOWN
1143	allow_sharing = F_ALLOW_SHARING	1683	allow_sharing = F_ALLOW_SHARING
1144	allow_stringref = F_ALLOW_STRINGREF	1684	allow_cycles = F_ALLOW_CYCLES
		1685	forbid_objects = F_FORBID_OBJECTS
		1686	pack_strings = F_PACK_STRINGS
		1687	text_keys = F_TEXT_KEYS
		1688	text_strings = F_TEXT_STRINGS
		1689	validate_utf8 = F_VALIDATE_UTF8
1145	PPCODE:	1690	PPCODE:
1146	{	1691	{
1147	if (enable)	1692	if (enable)
1148	self->flags |= ix;	1693	self->flags |= ix;
1149	else	1694	else
…		…
1155	void get_shrink (CBOR *self)	1700	void get_shrink (CBOR *self)
1156	ALIAS:	1701	ALIAS:
1157	get_shrink = F_SHRINK	1702	get_shrink = F_SHRINK
1158	get_allow_unknown = F_ALLOW_UNKNOWN	1703	get_allow_unknown = F_ALLOW_UNKNOWN
1159	get_allow_sharing = F_ALLOW_SHARING	1704	get_allow_sharing = F_ALLOW_SHARING
1160	get_allow_stringref = F_ALLOW_STRINGREF	1705	get_allow_cycles = F_ALLOW_CYCLES
		1706	get_forbid_objects = F_FORBID_OBJECTS
		1707	get_pack_strings = F_PACK_STRINGS
		1708	get_text_keys = F_TEXT_KEYS
		1709	get_text_strings = F_TEXT_STRINGS
		1710	get_validate_utf8 = F_VALIDATE_UTF8
1161	PPCODE:	1711	PPCODE:
1162	XPUSHs (boolSV (self->flags & ix));	1712	XPUSHs (boolSV (self->flags & ix));
1163		1713
1164	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1714	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1165	PPCODE:	1715	PPCODE:
…		…
1214	EXTEND (SP, 2);	1764	EXTEND (SP, 2);
1215	PUSHs (sv);	1765	PUSHs (sv);
1216	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1766	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1217	}	1767	}
1218		1768
		1769	void incr_parse (CBOR *self, SV *cborstr)
		1770	ALIAS:
		1771	incr_parse_multiple = 1
		1772	PPCODE:
		1773	{
		1774	if (SvUTF8 (cborstr))
		1775	sv_utf8_downgrade (cborstr, 0);
		1776
		1777	if (!self->incr_count)
		1778	{
		1779	self->incr_count = newAV ();
		1780	self->incr_pos = 0;
		1781	self->incr_need = 1;
		1782
		1783	av_push (self->incr_count, newSViv (1));
		1784	}
		1785
		1786	do
		1787	{
		1788	if (!incr_parse (self, cborstr))
		1789	{
		1790	if (self->incr_need > self->max_size && self->max_size)
		1791	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1792	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1793
		1794	break;
		1795	}
		1796
		1797	SV *sv;
		1798	char *offset;
		1799
		1800	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1801	XPUSHs (sv);
		1802
		1803	sv_chop (cborstr, offset);
		1804
		1805	av_clear (self->incr_count);
		1806	av_push (self->incr_count, newSViv (1));
		1807
		1808	self->incr_pos = 0;
		1809	self->incr_need = self->incr_pos + 1;
		1810	}
		1811	while (ix);
		1812	}
		1813
		1814	void incr_reset (CBOR *self)
		1815	CODE:
		1816	{
		1817	SvREFCNT_dec (self->incr_count);
		1818	self->incr_count = 0;
		1819	}
		1820
1219	void DESTROY (CBOR *self)	1821	void DESTROY (CBOR *self)
1220	PPCODE:	1822	PPCODE:
1221	cbor_free (self);	1823	cbor_free (self);
1222		1824
1223	PROTOTYPES: ENABLE	1825	PROTOTYPES: ENABLE
1224		1826
1225	void encode_cbor (SV *scalar)	1827	void encode_cbor (SV *scalar)
		1828	ALIAS:
		1829	encode_cbor = 0
		1830	encode_cbor_sharing = F_ALLOW_SHARING
1226	PPCODE:	1831	PPCODE:
1227	{	1832	{
1228	CBOR cbor;	1833	CBOR cbor;
1229	cbor_init (&cbor);	1834	cbor_init (&cbor);
		1835	cbor.flags |= ix;
1230	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1836	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1231	XPUSHs (scalar);	1837	XPUSHs (scalar);
1232	}	1838	}
1233		1839
1234	void decode_cbor (SV *cborstr)	1840	void decode_cbor (SV *cborstr)
…		…
1238	cbor_init (&cbor);	1844	cbor_init (&cbor);
1239	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1845	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1240	XPUSHs (cborstr);	1846	XPUSHs (cborstr);
1241	}	1847	}
1242		1848
		1849	#ifdef __AFL_COMPILER
		1850
		1851	void
		1852	afl_init ()
		1853	CODE:
		1854	__AFL_INIT ();
		1855
		1856	int
		1857	afl_loop (unsigned int count = 10000)
		1858	CODE:
		1859	RETVAL = __AFL_LOOP (count);
		1860	OUTPUT:
		1861	RETVAL
		1862
		1863	#endif
		1864

Diff Legend

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